repo_name
stringclasses 6
values | pr_number
int64 99
20.3k
| pr_title
stringlengths 8
158
| pr_description
stringlengths 0
6.54k
| author
stringlengths 4
18
| date_created
unknown | date_merged
unknown | previous_commit
stringlengths 40
40
| pr_commit
stringlengths 40
40
| query
stringlengths 37
6.57k
| filepath
stringlengths 8
153
| before_content
stringlengths 0
876M
| after_content
stringlengths 0
876M
| label
int64 -1
1
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/mobilebert/test_tokenization_mobilebert.py | # coding=utf-8
# Copyright 2022 Leon Derczynski. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the MobileBERT tokenizer. """
import os
import unittest
from transformers import MobileBertTokenizer, MobileBertTokenizerFast
from transformers.models.bert.tokenization_bert import (
VOCAB_FILES_NAMES,
BasicTokenizer,
WordpieceTokenizer,
_is_control,
_is_punctuation,
_is_whitespace,
)
from transformers.testing_utils import require_tokenizers, slow
from ...test_tokenization_common import TokenizerTesterMixin, filter_non_english
# Copied from transformers.tests.models.bert.test_modeling_bert.py with Bert->MobileBert and pathfix
@require_tokenizers
class MobileBERTTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
tokenizer_class = MobileBertTokenizer
rust_tokenizer_class = MobileBertTokenizerFast
test_rust_tokenizer = True
space_between_special_tokens = True
from_pretrained_filter = filter_non_english
pre_trained_model_path = "google/mobilebert-uncased"
def setUp(self):
super().setUp()
vocab_tokens = [
"[UNK]",
"[CLS]",
"[SEP]",
"[PAD]",
"[MASK]",
"want",
"##want",
"##ed",
"wa",
"un",
"runn",
"##ing",
",",
"low",
"lowest",
]
self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
self.tokenizers_list = [
(tokenizer_def[0], self.pre_trained_model_path, tokenizer_def[2]) # else the 'google/' prefix is stripped
for tokenizer_def in self.tokenizers_list
]
def get_input_output_texts(self, tokenizer):
input_text = "UNwant\u00E9d,running"
output_text = "unwanted, running"
return input_text, output_text
def test_full_tokenizer(self):
tokenizer = self.tokenizer_class(self.vocab_file)
tokens = tokenizer.tokenize("UNwant\u00E9d,running")
self.assertListEqual(tokens, ["un", "##want", "##ed", ",", "runn", "##ing"])
self.assertListEqual(tokenizer.convert_tokens_to_ids(tokens), [9, 6, 7, 12, 10, 11])
def test_rust_and_python_full_tokenizers(self):
if not self.test_rust_tokenizer:
return
tokenizer = self.get_tokenizer()
rust_tokenizer = self.get_rust_tokenizer()
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
# With lower casing
tokenizer = self.get_tokenizer(do_lower_case=True)
rust_tokenizer = self.get_rust_tokenizer(do_lower_case=True)
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
def test_chinese(self):
tokenizer = BasicTokenizer()
self.assertListEqual(tokenizer.tokenize("ah\u535A\u63A8zz"), ["ah", "\u535A", "\u63A8", "zz"])
def test_basic_tokenizer_lower(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["hello", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hällo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["h\u00E9llo"])
def test_basic_tokenizer_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_default(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_no_lower(self):
tokenizer = BasicTokenizer(do_lower_case=False)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["HeLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HäLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HaLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_respects_never_split_tokens(self):
tokenizer = BasicTokenizer(do_lower_case=False, never_split=["[UNK]"])
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? [UNK]"), ["HeLLo", "!", "how", "Are", "yoU", "?", "[UNK]"]
)
def test_wordpiece_tokenizer(self):
vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing"]
vocab = {}
for i, token in enumerate(vocab_tokens):
vocab[token] = i
tokenizer = WordpieceTokenizer(vocab=vocab, unk_token="[UNK]")
self.assertListEqual(tokenizer.tokenize(""), [])
self.assertListEqual(tokenizer.tokenize("unwanted running"), ["un", "##want", "##ed", "runn", "##ing"])
self.assertListEqual(tokenizer.tokenize("unwantedX running"), ["[UNK]", "runn", "##ing"])
def test_is_whitespace(self):
self.assertTrue(_is_whitespace(" "))
self.assertTrue(_is_whitespace("\t"))
self.assertTrue(_is_whitespace("\r"))
self.assertTrue(_is_whitespace("\n"))
self.assertTrue(_is_whitespace("\u00A0"))
self.assertFalse(_is_whitespace("A"))
self.assertFalse(_is_whitespace("-"))
def test_is_control(self):
self.assertTrue(_is_control("\u0005"))
self.assertFalse(_is_control("A"))
self.assertFalse(_is_control(" "))
self.assertFalse(_is_control("\t"))
self.assertFalse(_is_control("\r"))
def test_is_punctuation(self):
self.assertTrue(_is_punctuation("-"))
self.assertTrue(_is_punctuation("$"))
self.assertTrue(_is_punctuation("`"))
self.assertTrue(_is_punctuation("."))
self.assertFalse(_is_punctuation("A"))
self.assertFalse(_is_punctuation(" "))
def test_clean_text(self):
tokenizer = self.get_tokenizer()
rust_tokenizer = self.get_rust_tokenizer()
# Example taken from the issue https://github.com/huggingface/tokenizers/issues/340
self.assertListEqual([tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], [], ["[UNK]"]])
self.assertListEqual(
[rust_tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], [], ["[UNK]"]]
)
@slow
def test_sequence_builders(self):
tokenizer = self.tokenizer_class.from_pretrained("google/mobilebert-uncased")
text = tokenizer.encode("sequence builders", add_special_tokens=False)
text_2 = tokenizer.encode("multi-sequence build", add_special_tokens=False)
encoded_sentence = tokenizer.build_inputs_with_special_tokens(text)
encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)
assert encoded_sentence == [101] + text + [102]
assert encoded_pair == [101] + text + [102] + text_2 + [102]
def test_offsets_with_special_characters(self):
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
sentence = f"A, naïve {tokenizer_r.mask_token} AllenNLP sentence."
tokens = tokenizer_r.encode_plus(
sentence,
return_attention_mask=False,
return_token_type_ids=False,
return_offsets_mapping=True,
add_special_tokens=True,
)
do_lower_case = tokenizer_r.do_lower_case if hasattr(tokenizer_r, "do_lower_case") else False
expected_results = (
[
((0, 0), tokenizer_r.cls_token),
((0, 1), "A"),
((1, 2), ","),
((3, 5), "na"),
((5, 6), "##ï"),
((6, 8), "##ve"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "Allen"),
((21, 23), "##NL"),
((23, 24), "##P"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
if not do_lower_case
else [
((0, 0), tokenizer_r.cls_token),
((0, 1), "a"),
((1, 2), ","),
((3, 8), "naive"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "allen"),
((21, 23), "##nl"),
((23, 24), "##p"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
)
self.assertEqual(
[e[1] for e in expected_results], tokenizer_r.convert_ids_to_tokens(tokens["input_ids"])
)
self.assertEqual([e[0] for e in expected_results], tokens["offset_mapping"])
def test_change_tokenize_chinese_chars(self):
list_of_commun_chinese_char = ["的", "人", "有"]
text_with_chinese_char = "".join(list_of_commun_chinese_char)
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
kwargs["tokenize_chinese_chars"] = True
tokenizer_p = self.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
ids_without_spe_char_p = tokenizer_p.encode(text_with_chinese_char, add_special_tokens=False)
ids_without_spe_char_r = tokenizer_r.encode(text_with_chinese_char, add_special_tokens=False)
tokens_without_spe_char_r = tokenizer_r.convert_ids_to_tokens(ids_without_spe_char_r)
tokens_without_spe_char_p = tokenizer_p.convert_ids_to_tokens(ids_without_spe_char_p)
# it is expected that each Chinese character is not preceded by "##"
self.assertListEqual(tokens_without_spe_char_p, list_of_commun_chinese_char)
self.assertListEqual(tokens_without_spe_char_r, list_of_commun_chinese_char)
kwargs["tokenize_chinese_chars"] = False
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
tokenizer_p = self.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
ids_without_spe_char_r = tokenizer_r.encode(text_with_chinese_char, add_special_tokens=False)
ids_without_spe_char_p = tokenizer_p.encode(text_with_chinese_char, add_special_tokens=False)
tokens_without_spe_char_r = tokenizer_r.convert_ids_to_tokens(ids_without_spe_char_r)
tokens_without_spe_char_p = tokenizer_p.convert_ids_to_tokens(ids_without_spe_char_p)
# it is expected that only the first Chinese character is not preceded by "##".
expected_tokens = [
f"##{token}" if idx != 0 else token for idx, token in enumerate(list_of_commun_chinese_char)
]
self.assertListEqual(tokens_without_spe_char_p, expected_tokens)
self.assertListEqual(tokens_without_spe_char_r, expected_tokens)
| # coding=utf-8
# Copyright 2022 Leon Derczynski. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the MobileBERT tokenizer. """
import os
import unittest
from transformers import MobileBertTokenizer, MobileBertTokenizerFast
from transformers.models.bert.tokenization_bert import (
VOCAB_FILES_NAMES,
BasicTokenizer,
WordpieceTokenizer,
_is_control,
_is_punctuation,
_is_whitespace,
)
from transformers.testing_utils import require_tokenizers, slow
from ...test_tokenization_common import TokenizerTesterMixin, filter_non_english
# Copied from transformers.tests.models.bert.test_modeling_bert.py with Bert->MobileBert and pathfix
@require_tokenizers
class MobileBERTTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
tokenizer_class = MobileBertTokenizer
rust_tokenizer_class = MobileBertTokenizerFast
test_rust_tokenizer = True
space_between_special_tokens = True
from_pretrained_filter = filter_non_english
pre_trained_model_path = "google/mobilebert-uncased"
def setUp(self):
super().setUp()
vocab_tokens = [
"[UNK]",
"[CLS]",
"[SEP]",
"[PAD]",
"[MASK]",
"want",
"##want",
"##ed",
"wa",
"un",
"runn",
"##ing",
",",
"low",
"lowest",
]
self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
self.tokenizers_list = [
(tokenizer_def[0], self.pre_trained_model_path, tokenizer_def[2]) # else the 'google/' prefix is stripped
for tokenizer_def in self.tokenizers_list
]
def get_input_output_texts(self, tokenizer):
input_text = "UNwant\u00E9d,running"
output_text = "unwanted, running"
return input_text, output_text
def test_full_tokenizer(self):
tokenizer = self.tokenizer_class(self.vocab_file)
tokens = tokenizer.tokenize("UNwant\u00E9d,running")
self.assertListEqual(tokens, ["un", "##want", "##ed", ",", "runn", "##ing"])
self.assertListEqual(tokenizer.convert_tokens_to_ids(tokens), [9, 6, 7, 12, 10, 11])
def test_rust_and_python_full_tokenizers(self):
if not self.test_rust_tokenizer:
return
tokenizer = self.get_tokenizer()
rust_tokenizer = self.get_rust_tokenizer()
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
# With lower casing
tokenizer = self.get_tokenizer(do_lower_case=True)
rust_tokenizer = self.get_rust_tokenizer(do_lower_case=True)
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
def test_chinese(self):
tokenizer = BasicTokenizer()
self.assertListEqual(tokenizer.tokenize("ah\u535A\u63A8zz"), ["ah", "\u535A", "\u63A8", "zz"])
def test_basic_tokenizer_lower(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["hello", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hällo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["h\u00E9llo"])
def test_basic_tokenizer_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_default(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_no_lower(self):
tokenizer = BasicTokenizer(do_lower_case=False)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["HeLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HäLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HaLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_respects_never_split_tokens(self):
tokenizer = BasicTokenizer(do_lower_case=False, never_split=["[UNK]"])
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? [UNK]"), ["HeLLo", "!", "how", "Are", "yoU", "?", "[UNK]"]
)
def test_wordpiece_tokenizer(self):
vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing"]
vocab = {}
for i, token in enumerate(vocab_tokens):
vocab[token] = i
tokenizer = WordpieceTokenizer(vocab=vocab, unk_token="[UNK]")
self.assertListEqual(tokenizer.tokenize(""), [])
self.assertListEqual(tokenizer.tokenize("unwanted running"), ["un", "##want", "##ed", "runn", "##ing"])
self.assertListEqual(tokenizer.tokenize("unwantedX running"), ["[UNK]", "runn", "##ing"])
def test_is_whitespace(self):
self.assertTrue(_is_whitespace(" "))
self.assertTrue(_is_whitespace("\t"))
self.assertTrue(_is_whitespace("\r"))
self.assertTrue(_is_whitespace("\n"))
self.assertTrue(_is_whitespace("\u00A0"))
self.assertFalse(_is_whitespace("A"))
self.assertFalse(_is_whitespace("-"))
def test_is_control(self):
self.assertTrue(_is_control("\u0005"))
self.assertFalse(_is_control("A"))
self.assertFalse(_is_control(" "))
self.assertFalse(_is_control("\t"))
self.assertFalse(_is_control("\r"))
def test_is_punctuation(self):
self.assertTrue(_is_punctuation("-"))
self.assertTrue(_is_punctuation("$"))
self.assertTrue(_is_punctuation("`"))
self.assertTrue(_is_punctuation("."))
self.assertFalse(_is_punctuation("A"))
self.assertFalse(_is_punctuation(" "))
def test_clean_text(self):
tokenizer = self.get_tokenizer()
rust_tokenizer = self.get_rust_tokenizer()
# Example taken from the issue https://github.com/huggingface/tokenizers/issues/340
self.assertListEqual([tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], [], ["[UNK]"]])
self.assertListEqual(
[rust_tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], [], ["[UNK]"]]
)
@slow
def test_sequence_builders(self):
tokenizer = self.tokenizer_class.from_pretrained("google/mobilebert-uncased")
text = tokenizer.encode("sequence builders", add_special_tokens=False)
text_2 = tokenizer.encode("multi-sequence build", add_special_tokens=False)
encoded_sentence = tokenizer.build_inputs_with_special_tokens(text)
encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)
assert encoded_sentence == [101] + text + [102]
assert encoded_pair == [101] + text + [102] + text_2 + [102]
def test_offsets_with_special_characters(self):
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
sentence = f"A, naïve {tokenizer_r.mask_token} AllenNLP sentence."
tokens = tokenizer_r.encode_plus(
sentence,
return_attention_mask=False,
return_token_type_ids=False,
return_offsets_mapping=True,
add_special_tokens=True,
)
do_lower_case = tokenizer_r.do_lower_case if hasattr(tokenizer_r, "do_lower_case") else False
expected_results = (
[
((0, 0), tokenizer_r.cls_token),
((0, 1), "A"),
((1, 2), ","),
((3, 5), "na"),
((5, 6), "##ï"),
((6, 8), "##ve"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "Allen"),
((21, 23), "##NL"),
((23, 24), "##P"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
if not do_lower_case
else [
((0, 0), tokenizer_r.cls_token),
((0, 1), "a"),
((1, 2), ","),
((3, 8), "naive"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "allen"),
((21, 23), "##nl"),
((23, 24), "##p"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
)
self.assertEqual(
[e[1] for e in expected_results], tokenizer_r.convert_ids_to_tokens(tokens["input_ids"])
)
self.assertEqual([e[0] for e in expected_results], tokens["offset_mapping"])
def test_change_tokenize_chinese_chars(self):
list_of_commun_chinese_char = ["的", "人", "有"]
text_with_chinese_char = "".join(list_of_commun_chinese_char)
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
kwargs["tokenize_chinese_chars"] = True
tokenizer_p = self.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
ids_without_spe_char_p = tokenizer_p.encode(text_with_chinese_char, add_special_tokens=False)
ids_without_spe_char_r = tokenizer_r.encode(text_with_chinese_char, add_special_tokens=False)
tokens_without_spe_char_r = tokenizer_r.convert_ids_to_tokens(ids_without_spe_char_r)
tokens_without_spe_char_p = tokenizer_p.convert_ids_to_tokens(ids_without_spe_char_p)
# it is expected that each Chinese character is not preceded by "##"
self.assertListEqual(tokens_without_spe_char_p, list_of_commun_chinese_char)
self.assertListEqual(tokens_without_spe_char_r, list_of_commun_chinese_char)
kwargs["tokenize_chinese_chars"] = False
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
tokenizer_p = self.tokenizer_class.from_pretrained(pretrained_name, **kwargs)
ids_without_spe_char_r = tokenizer_r.encode(text_with_chinese_char, add_special_tokens=False)
ids_without_spe_char_p = tokenizer_p.encode(text_with_chinese_char, add_special_tokens=False)
tokens_without_spe_char_r = tokenizer_r.convert_ids_to_tokens(ids_without_spe_char_r)
tokens_without_spe_char_p = tokenizer_p.convert_ids_to_tokens(ids_without_spe_char_p)
# it is expected that only the first Chinese character is not preceded by "##".
expected_tokens = [
f"##{token}" if idx != 0 else token for idx, token in enumerate(list_of_commun_chinese_char)
]
self.assertListEqual(tokens_without_spe_char_p, expected_tokens)
self.assertListEqual(tokens_without_spe_char_r, expected_tokens)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/trajectory_transformer/convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch.py | # coding=utf-8
# Copyright 2022 The Trajectory Transformers paper authors and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" TrajectoryTransformer pytorch checkpoint conversion"""
import torch
import trajectory.utils as utils
from transformers import TrajectoryTransformerModel
class Parser(utils.Parser):
dataset: str = "halfcheetah-medium-expert-v2"
config: str = "config.offline"
def convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch(logbase, dataset, loadpath, epoch, device):
"""Converting Sequential blocks to ModuleList"""
gpt, gpt_epoch = utils.load_model(logbase, dataset, loadpath, epoch=epoch, device=device)
trajectory_transformer = TrajectoryTransformerModel(gpt.config)
trajectory_transformer.tok_emb.load_state_dict(gpt.tok_emb.state_dict())
trajectory_transformer.pos_emb = gpt.pos_emb
trajectory_transformer.drop.load_state_dict(gpt.drop.state_dict())
trajectory_transformer.ln_f.load_state_dict(gpt.ln_f.state_dict())
trajectory_transformer.head.load_state_dict(gpt.head.state_dict())
for i, block in enumerate(gpt.blocks):
trajectory_transformer.blocks[i].ln1.load_state_dict(gpt.blocks[i].ln1.state_dict())
trajectory_transformer.blocks[i].ln2.load_state_dict(gpt.blocks[i].ln2.state_dict())
trajectory_transformer.blocks[i].attn.load_state_dict(gpt.blocks[i].attn.state_dict())
trajectory_transformer.blocks[i].l1.load_state_dict(gpt.blocks[i].mlp[0].state_dict())
trajectory_transformer.blocks[i].act.load_state_dict(gpt.blocks[i].mlp[1].state_dict())
trajectory_transformer.blocks[i].l2.load_state_dict(gpt.blocks[i].mlp[2].state_dict())
trajectory_transformer.blocks[i].drop.load_state_dict(gpt.blocks[i].mlp[3].state_dict())
torch.save(trajectory_transformer.state_dict(), "pytorch_model.bin")
if __name__ == "__main__":
"""
To run this script you will need to install the original repository to run the original model. You can find it
here: https://github.com/jannerm/trajectory-transformer From this repository code you can also download the
original pytorch checkpoints.
Run with the command:
```sh
>>> python convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch.py --dataset <dataset_name>
... --gpt_loadpath <path_to_original_pytorch_checkpoint>
```
"""
args = Parser().parse_args("plan")
convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch(
args.logbase, args.dataset, args.gpt_loadpath, args.gpt_epoch, args.device
)
| # coding=utf-8
# Copyright 2022 The Trajectory Transformers paper authors and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" TrajectoryTransformer pytorch checkpoint conversion"""
import torch
import trajectory.utils as utils
from transformers import TrajectoryTransformerModel
class Parser(utils.Parser):
dataset: str = "halfcheetah-medium-expert-v2"
config: str = "config.offline"
def convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch(logbase, dataset, loadpath, epoch, device):
"""Converting Sequential blocks to ModuleList"""
gpt, gpt_epoch = utils.load_model(logbase, dataset, loadpath, epoch=epoch, device=device)
trajectory_transformer = TrajectoryTransformerModel(gpt.config)
trajectory_transformer.tok_emb.load_state_dict(gpt.tok_emb.state_dict())
trajectory_transformer.pos_emb = gpt.pos_emb
trajectory_transformer.drop.load_state_dict(gpt.drop.state_dict())
trajectory_transformer.ln_f.load_state_dict(gpt.ln_f.state_dict())
trajectory_transformer.head.load_state_dict(gpt.head.state_dict())
for i, block in enumerate(gpt.blocks):
trajectory_transformer.blocks[i].ln1.load_state_dict(gpt.blocks[i].ln1.state_dict())
trajectory_transformer.blocks[i].ln2.load_state_dict(gpt.blocks[i].ln2.state_dict())
trajectory_transformer.blocks[i].attn.load_state_dict(gpt.blocks[i].attn.state_dict())
trajectory_transformer.blocks[i].l1.load_state_dict(gpt.blocks[i].mlp[0].state_dict())
trajectory_transformer.blocks[i].act.load_state_dict(gpt.blocks[i].mlp[1].state_dict())
trajectory_transformer.blocks[i].l2.load_state_dict(gpt.blocks[i].mlp[2].state_dict())
trajectory_transformer.blocks[i].drop.load_state_dict(gpt.blocks[i].mlp[3].state_dict())
torch.save(trajectory_transformer.state_dict(), "pytorch_model.bin")
if __name__ == "__main__":
"""
To run this script you will need to install the original repository to run the original model. You can find it
here: https://github.com/jannerm/trajectory-transformer From this repository code you can also download the
original pytorch checkpoints.
Run with the command:
```sh
>>> python convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch.py --dataset <dataset_name>
... --gpt_loadpath <path_to_original_pytorch_checkpoint>
```
"""
args = Parser().parse_args("plan")
convert_trajectory_transformer_original_pytorch_checkpoint_to_pytorch(
args.logbase, args.dataset, args.gpt_loadpath, args.gpt_epoch, args.device
)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/bert/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/jax-projects/hybrid_clip/modeling_hybrid_clip.py | # coding=utf-8
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Optional, Tuple
import flax.linen as nn
import jax
import jax.numpy as jnp
from configuration_hybrid_clip import HybridCLIPConfig
from flax.core.frozen_dict import FrozenDict
from transformers import FLAX_MODEL_MAPPING, FlaxCLIPVisionModel
from transformers.modeling_flax_utils import FlaxPreTrainedModel
from transformers.models.clip.modeling_flax_clip import FlaxCLIPOutput
from transformers.utils import logging
logger = logging.get_logger(__name__)
class FlaxHybridCLIPModule(nn.Module):
config: HybridCLIPConfig
dtype: jnp.dtype = jnp.float32
def setup(self):
text_config = self.config.text_config
vision_config = self.config.vision_config
self.projection_dim = self.config.projection_dim
self.text_embed_dim = text_config.hidden_size
self.vision_embed_dim = vision_config.hidden_size
text_module = FLAX_MODEL_MAPPING[self.config.text_config.__class__].module_class
vision_module = FLAX_MODEL_MAPPING.get(self.config.vision_config.__class__, FlaxCLIPVisionModel).module_class
self.text_model = text_module(text_config, dtype=self.dtype)
self.vision_model = vision_module(vision_config, dtype=self.dtype)
self.visual_projection = nn.Dense(
self.projection_dim,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(0.02),
use_bias=False,
)
self.text_projection = nn.Dense(
self.projection_dim,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(0.02),
use_bias=False,
)
self.logit_scale = self.param("logit_scale", jax.nn.initializers.ones, [])
def __call__(
self,
input_ids=None,
pixel_values=None,
attention_mask=None,
position_ids=None,
token_type_ids=None,
deterministic: bool = True,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
return_dict = return_dict if return_dict is not None else self.config.return_dict
vision_outputs = self.vision_model(
pixel_values=pixel_values,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
text_outputs = self.text_model(
input_ids=input_ids,
attention_mask=attention_mask,
token_type_ids=token_type_ids,
position_ids=position_ids,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
image_embeds = vision_outputs[1]
image_embeds = self.visual_projection(image_embeds)
text_embeds = text_outputs[1]
text_embeds = self.text_projection(text_embeds)
# normalized features
image_embeds = image_embeds / jnp.linalg.norm(image_embeds, axis=-1, keepdims=True)
text_embeds = text_embeds / jnp.linalg.norm(text_embeds, axis=-1, keepdims=True)
# cosine similarity as logits
logit_scale = jnp.exp(self.logit_scale)
logits_per_text = jnp.matmul(text_embeds, image_embeds.T) * logit_scale
logits_per_image = logits_per_text.T
if not return_dict:
return (logits_per_image, logits_per_text, text_embeds, image_embeds, text_outputs, vision_outputs)
return FlaxCLIPOutput(
logits_per_image=logits_per_image,
logits_per_text=logits_per_text,
text_embeds=text_embeds,
image_embeds=image_embeds,
text_model_output=text_outputs,
vision_model_output=vision_outputs,
)
class FlaxHybridCLIP(FlaxPreTrainedModel):
config_class = HybridCLIPConfig
module_class = FlaxHybridCLIPModule
def __init__(
self,
config: HybridCLIPConfig,
input_shape: Optional[Tuple] = None,
seed: int = 0,
dtype: jnp.dtype = jnp.float32,
**kwargs,
):
if input_shape is None:
input_shape = ((1, 1), (1, config.vision_config.image_size, config.vision_config.image_size, 3))
module = self.module_class(config=config, dtype=dtype, **kwargs)
super().__init__(config, module, input_shape=input_shape, seed=seed, dtype=dtype)
def init_weights(self, rng: jax.random.PRNGKey, input_shape: Tuple, params: FrozenDict = None) -> FrozenDict:
# init input tensor
input_ids = jnp.zeros(input_shape[0], dtype="i4")
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_shape[0])
token_type_ids = jnp.ones_like(input_ids)
attention_mask = jnp.ones_like(input_ids)
pixel_values = jax.random.normal(rng, input_shape[1])
params_rng, dropout_rng = jax.random.split(rng)
rngs = {"params": params_rng, "dropout": dropout_rng}
return self.module.init(rngs, input_ids, pixel_values, attention_mask, position_ids, token_type_ids)["params"]
def __call__(
self,
input_ids,
pixel_values,
attention_mask=None,
position_ids=None,
token_type_ids=None,
params: dict = None,
dropout_rng: jax.random.PRNGKey = None,
train: bool = False,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
):
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.return_dict
if position_ids is None:
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
if attention_mask is None:
attention_mask = jnp.ones_like(input_ids)
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
return self.module.apply(
{"params": params or self.params},
jnp.array(input_ids, dtype="i4"),
jnp.array(pixel_values, dtype=jnp.float32),
jnp.array(attention_mask, dtype="i4"),
jnp.array(position_ids, dtype="i4"),
jnp.array(token_type_ids, dtype="i4"),
not train,
output_attentions,
output_hidden_states,
return_dict,
rngs=rngs,
)
def get_text_features(
self,
input_ids,
attention_mask=None,
position_ids=None,
token_type_ids=None,
params: dict = None,
dropout_rng: jax.random.PRNGKey = None,
train=False,
):
r"""
Args:
input_ids (:obj:`numpy.ndarray` of shape :obj:`(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.PreTrainedTokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
for details.
`What are input IDs? <../glossary.html#input-ids>`__
Returns:
text_features (:obj:`jnp.ndarray` of shape :obj:`(batch_size, output_dim`): The text embeddings
obtained by applying the projection layer to the pooled output of text model.
"""
if position_ids is None:
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
if attention_mask is None:
attention_mask = jnp.ones_like(input_ids)
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
def _get_features(module, input_ids, attention_mask, position_ids, token_type_ids, deterministic):
text_outputs = module.text_model(
input_ids=input_ids,
attention_mask=attention_mask,
position_ids=position_ids,
token_type_ids=token_type_ids,
deterministic=deterministic,
)
pooled_output = text_outputs[1]
text_features = module.text_projection(pooled_output)
return text_features
return self.module.apply(
{"params": params or self.params},
jnp.array(input_ids, dtype="i4"),
jnp.array(attention_mask, dtype="i4"),
jnp.array(position_ids, dtype="i4"),
jnp.array(token_type_ids, dtype="i4"),
not train,
method=_get_features,
rngs=rngs,
)
def get_image_features(
self, pixel_values, params: dict = None, dropout_rng: jax.random.PRNGKey = None, train=False
):
r"""
Args:
pixel_values (:obj:`numpy.ndarray` of shape :obj:`(batch_size, num_channels, height, width)`):
Pixel values. Padding will be ignored by default should you provide it. Pixel values can be obtained
using :class:`~transformers.ImageFeatureExtractionMixin`. See
:meth:`transformers.ImageFeatureExtractionMixin.__call__` for details.
Returns:
image_features (:obj:`jnp.ndarray` of shape :obj:`(batch_size, output_dim`): The image embeddings
obtained by applying the projection layer to the pooled output of vision model.
"""
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
def _get_features(module, pixel_values, deterministic):
vision_outputs = module.vision_model(pixel_values=pixel_values, deterministic=deterministic)
pooled_output = vision_outputs[1] # pooled_output
image_features = module.visual_projection(pooled_output)
return image_features
return self.module.apply(
{"params": params or self.params},
jnp.array(pixel_values, dtype=jnp.float32),
not train,
method=_get_features,
rngs=rngs,
)
@classmethod
def from_text_vision_pretrained(
cls,
text_model_name_or_path: str = None,
vision_model_name_or_path: str = None,
*model_args,
**kwargs,
) -> FlaxPreTrainedModel:
"""
Params:
text_model_name_or_path (:obj: `str`, `optional`):
Information necessary to initiate the text model. Can be either:
- A string, the `model id` of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like ``bert-base-uncased``, or namespaced under
a user or organization name, like ``dbmdz/bert-base-german-cased``.
- A path to a `directory` containing model weights saved using
:func:`~transformers.FlaxPreTrainedModel.save_pretrained`, e.g., ``./my_model_directory/``.
- A path or url to a `PyTorch checkpoint folder` (e.g, ``./pt_model``). In
this case, ``from_pt`` should be set to :obj:`True` and a configuration object should be provided
as ``config`` argument. This loading path is slower than converting the PyTorch checkpoint in
a Flax model using the provided conversion scripts and loading the Flax model afterwards.
vision_model_name_or_path (:obj: `str`, `optional`, defaults to `None`):
Information necessary to initiate the vision model. Can be either:
- A string, the `model id` of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like ``bert-base-uncased``, or namespaced under
a user or organization name, like ``dbmdz/bert-base-german-cased``.
- A path to a `directory` containing model weights saved using
:func:`~transformers.FlaxPreTrainedModel.save_pretrained`, e.g., ``./my_model_directory/``.
- A path or url to a `PyTorch checkpoint folder` (e.g, ``./pt_model``). In
this case, ``from_pt`` should be set to :obj:`True` and a configuration object should be provided
as ``config`` argument. This loading path is slower than converting the PyTorch checkpoint in
a Flax model using the provided conversion scripts and loading the Flax model afterwards.
model_args (remaining positional arguments, `optional`):
All remaning positional arguments will be passed to the underlying model's ``__init__`` method.
kwargs (remaining dictionary of keyword arguments, `optional`):
Can be used to update the configuration object (after it being loaded) and initiate the model (e.g.,
:obj:`output_attentions=True`).
- To update the text configuration, use the prefix `text_` for each configuration parameter.
- To update the vision configuration, use the prefix `vision_` for each configuration parameter.
- To update the parent model configuration, do not use a prefix for each configuration parameter.
Behaves differently depending on whether a :obj:`config` is provided or automatically loaded.
Example::
>>> from transformers import FlaxHybridCLIP
>>> # initialize a model from pretrained BERT and CLIP models. Note that the projection layers will be randomly initialized.
>>> # If using CLIP's vision model the vision projection layer will be initialized using pre-trained weights
>>> model = FlaxHybridCLIP.from_text_vision_pretrained('bert-base-uncased', 'openai/clip-vit-base-patch32')
>>> # saving model after fine-tuning
>>> model.save_pretrained("./bert-clip")
>>> # load fine-tuned model
>>> model = FlaxHybridCLIP.from_pretrained("./bert-clip")
"""
kwargs_text = {
argument[len("text_") :]: value for argument, value in kwargs.items() if argument.startswith("text_")
}
kwargs_vision = {
argument[len("vision_") :]: value for argument, value in kwargs.items() if argument.startswith("vision_")
}
# remove text, vision kwargs from kwargs
for key in kwargs_text.keys():
del kwargs["text_" + key]
for key in kwargs_vision.keys():
del kwargs["vision_" + key]
# Load and initialize the text and vision model
text_model = kwargs_text.pop("model", None)
if text_model is None:
assert (
text_model_name_or_path is not None
), "If `model` is not defined as an argument, a `text_model_name_or_path` has to be defined"
from transformers import FlaxAutoModel
if "config" not in kwargs_text:
from transformers import AutoConfig
text_config = AutoConfig.from_pretrained(text_model_name_or_path)
kwargs_text["config"] = text_config
text_model = FlaxAutoModel.from_pretrained(text_model_name_or_path, *model_args, **kwargs_text)
vision_model = kwargs_vision.pop("model", None)
if vision_model is None:
assert (
vision_model_name_or_path is not None
), "If `model` is not defined as an argument, a `vision_model_name_or_path` has to be defined"
from transformers import FlaxAutoModel
if "config" not in kwargs_vision:
from transformers import AutoConfig
vision_config = AutoConfig.from_pretrained(vision_model_name_or_path)
kwargs_vision["config"] = vision_config
vision_model = FlaxAutoModel.from_pretrained(vision_model_name_or_path, *model_args, **kwargs_vision)
# instantiate config with corresponding kwargs
dtype = kwargs.pop("dtype", jnp.float32)
config = HybridCLIPConfig.from_text_vision_configs(text_model.config, vision_model.config, **kwargs)
# init model
model = cls(config, *model_args, dtype=dtype, **kwargs)
if vision_config.model_type == "clip":
model.params["vision_model"]["vision_model"] = vision_model.params["vision_model"]
model.params["visual_projection"]["kernel"] = vision_model.params["visual_projection"]["kernel"]
else:
model.params["vision_model"] = vision_model.params
model.params["text_model"] = text_model.params
return model
| # coding=utf-8
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Optional, Tuple
import flax.linen as nn
import jax
import jax.numpy as jnp
from configuration_hybrid_clip import HybridCLIPConfig
from flax.core.frozen_dict import FrozenDict
from transformers import FLAX_MODEL_MAPPING, FlaxCLIPVisionModel
from transformers.modeling_flax_utils import FlaxPreTrainedModel
from transformers.models.clip.modeling_flax_clip import FlaxCLIPOutput
from transformers.utils import logging
logger = logging.get_logger(__name__)
class FlaxHybridCLIPModule(nn.Module):
config: HybridCLIPConfig
dtype: jnp.dtype = jnp.float32
def setup(self):
text_config = self.config.text_config
vision_config = self.config.vision_config
self.projection_dim = self.config.projection_dim
self.text_embed_dim = text_config.hidden_size
self.vision_embed_dim = vision_config.hidden_size
text_module = FLAX_MODEL_MAPPING[self.config.text_config.__class__].module_class
vision_module = FLAX_MODEL_MAPPING.get(self.config.vision_config.__class__, FlaxCLIPVisionModel).module_class
self.text_model = text_module(text_config, dtype=self.dtype)
self.vision_model = vision_module(vision_config, dtype=self.dtype)
self.visual_projection = nn.Dense(
self.projection_dim,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(0.02),
use_bias=False,
)
self.text_projection = nn.Dense(
self.projection_dim,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(0.02),
use_bias=False,
)
self.logit_scale = self.param("logit_scale", jax.nn.initializers.ones, [])
def __call__(
self,
input_ids=None,
pixel_values=None,
attention_mask=None,
position_ids=None,
token_type_ids=None,
deterministic: bool = True,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
return_dict = return_dict if return_dict is not None else self.config.return_dict
vision_outputs = self.vision_model(
pixel_values=pixel_values,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
text_outputs = self.text_model(
input_ids=input_ids,
attention_mask=attention_mask,
token_type_ids=token_type_ids,
position_ids=position_ids,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
image_embeds = vision_outputs[1]
image_embeds = self.visual_projection(image_embeds)
text_embeds = text_outputs[1]
text_embeds = self.text_projection(text_embeds)
# normalized features
image_embeds = image_embeds / jnp.linalg.norm(image_embeds, axis=-1, keepdims=True)
text_embeds = text_embeds / jnp.linalg.norm(text_embeds, axis=-1, keepdims=True)
# cosine similarity as logits
logit_scale = jnp.exp(self.logit_scale)
logits_per_text = jnp.matmul(text_embeds, image_embeds.T) * logit_scale
logits_per_image = logits_per_text.T
if not return_dict:
return (logits_per_image, logits_per_text, text_embeds, image_embeds, text_outputs, vision_outputs)
return FlaxCLIPOutput(
logits_per_image=logits_per_image,
logits_per_text=logits_per_text,
text_embeds=text_embeds,
image_embeds=image_embeds,
text_model_output=text_outputs,
vision_model_output=vision_outputs,
)
class FlaxHybridCLIP(FlaxPreTrainedModel):
config_class = HybridCLIPConfig
module_class = FlaxHybridCLIPModule
def __init__(
self,
config: HybridCLIPConfig,
input_shape: Optional[Tuple] = None,
seed: int = 0,
dtype: jnp.dtype = jnp.float32,
**kwargs,
):
if input_shape is None:
input_shape = ((1, 1), (1, config.vision_config.image_size, config.vision_config.image_size, 3))
module = self.module_class(config=config, dtype=dtype, **kwargs)
super().__init__(config, module, input_shape=input_shape, seed=seed, dtype=dtype)
def init_weights(self, rng: jax.random.PRNGKey, input_shape: Tuple, params: FrozenDict = None) -> FrozenDict:
# init input tensor
input_ids = jnp.zeros(input_shape[0], dtype="i4")
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_shape[0])
token_type_ids = jnp.ones_like(input_ids)
attention_mask = jnp.ones_like(input_ids)
pixel_values = jax.random.normal(rng, input_shape[1])
params_rng, dropout_rng = jax.random.split(rng)
rngs = {"params": params_rng, "dropout": dropout_rng}
return self.module.init(rngs, input_ids, pixel_values, attention_mask, position_ids, token_type_ids)["params"]
def __call__(
self,
input_ids,
pixel_values,
attention_mask=None,
position_ids=None,
token_type_ids=None,
params: dict = None,
dropout_rng: jax.random.PRNGKey = None,
train: bool = False,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
):
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.return_dict
if position_ids is None:
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
if attention_mask is None:
attention_mask = jnp.ones_like(input_ids)
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
return self.module.apply(
{"params": params or self.params},
jnp.array(input_ids, dtype="i4"),
jnp.array(pixel_values, dtype=jnp.float32),
jnp.array(attention_mask, dtype="i4"),
jnp.array(position_ids, dtype="i4"),
jnp.array(token_type_ids, dtype="i4"),
not train,
output_attentions,
output_hidden_states,
return_dict,
rngs=rngs,
)
def get_text_features(
self,
input_ids,
attention_mask=None,
position_ids=None,
token_type_ids=None,
params: dict = None,
dropout_rng: jax.random.PRNGKey = None,
train=False,
):
r"""
Args:
input_ids (:obj:`numpy.ndarray` of shape :obj:`(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using :class:`~transformers.PreTrainedTokenizer`. See
:meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__`
for details.
`What are input IDs? <../glossary.html#input-ids>`__
Returns:
text_features (:obj:`jnp.ndarray` of shape :obj:`(batch_size, output_dim`): The text embeddings
obtained by applying the projection layer to the pooled output of text model.
"""
if position_ids is None:
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
if attention_mask is None:
attention_mask = jnp.ones_like(input_ids)
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
def _get_features(module, input_ids, attention_mask, position_ids, token_type_ids, deterministic):
text_outputs = module.text_model(
input_ids=input_ids,
attention_mask=attention_mask,
position_ids=position_ids,
token_type_ids=token_type_ids,
deterministic=deterministic,
)
pooled_output = text_outputs[1]
text_features = module.text_projection(pooled_output)
return text_features
return self.module.apply(
{"params": params or self.params},
jnp.array(input_ids, dtype="i4"),
jnp.array(attention_mask, dtype="i4"),
jnp.array(position_ids, dtype="i4"),
jnp.array(token_type_ids, dtype="i4"),
not train,
method=_get_features,
rngs=rngs,
)
def get_image_features(
self, pixel_values, params: dict = None, dropout_rng: jax.random.PRNGKey = None, train=False
):
r"""
Args:
pixel_values (:obj:`numpy.ndarray` of shape :obj:`(batch_size, num_channels, height, width)`):
Pixel values. Padding will be ignored by default should you provide it. Pixel values can be obtained
using :class:`~transformers.ImageFeatureExtractionMixin`. See
:meth:`transformers.ImageFeatureExtractionMixin.__call__` for details.
Returns:
image_features (:obj:`jnp.ndarray` of shape :obj:`(batch_size, output_dim`): The image embeddings
obtained by applying the projection layer to the pooled output of vision model.
"""
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
def _get_features(module, pixel_values, deterministic):
vision_outputs = module.vision_model(pixel_values=pixel_values, deterministic=deterministic)
pooled_output = vision_outputs[1] # pooled_output
image_features = module.visual_projection(pooled_output)
return image_features
return self.module.apply(
{"params": params or self.params},
jnp.array(pixel_values, dtype=jnp.float32),
not train,
method=_get_features,
rngs=rngs,
)
@classmethod
def from_text_vision_pretrained(
cls,
text_model_name_or_path: str = None,
vision_model_name_or_path: str = None,
*model_args,
**kwargs,
) -> FlaxPreTrainedModel:
"""
Params:
text_model_name_or_path (:obj: `str`, `optional`):
Information necessary to initiate the text model. Can be either:
- A string, the `model id` of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like ``bert-base-uncased``, or namespaced under
a user or organization name, like ``dbmdz/bert-base-german-cased``.
- A path to a `directory` containing model weights saved using
:func:`~transformers.FlaxPreTrainedModel.save_pretrained`, e.g., ``./my_model_directory/``.
- A path or url to a `PyTorch checkpoint folder` (e.g, ``./pt_model``). In
this case, ``from_pt`` should be set to :obj:`True` and a configuration object should be provided
as ``config`` argument. This loading path is slower than converting the PyTorch checkpoint in
a Flax model using the provided conversion scripts and loading the Flax model afterwards.
vision_model_name_or_path (:obj: `str`, `optional`, defaults to `None`):
Information necessary to initiate the vision model. Can be either:
- A string, the `model id` of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like ``bert-base-uncased``, or namespaced under
a user or organization name, like ``dbmdz/bert-base-german-cased``.
- A path to a `directory` containing model weights saved using
:func:`~transformers.FlaxPreTrainedModel.save_pretrained`, e.g., ``./my_model_directory/``.
- A path or url to a `PyTorch checkpoint folder` (e.g, ``./pt_model``). In
this case, ``from_pt`` should be set to :obj:`True` and a configuration object should be provided
as ``config`` argument. This loading path is slower than converting the PyTorch checkpoint in
a Flax model using the provided conversion scripts and loading the Flax model afterwards.
model_args (remaining positional arguments, `optional`):
All remaning positional arguments will be passed to the underlying model's ``__init__`` method.
kwargs (remaining dictionary of keyword arguments, `optional`):
Can be used to update the configuration object (after it being loaded) and initiate the model (e.g.,
:obj:`output_attentions=True`).
- To update the text configuration, use the prefix `text_` for each configuration parameter.
- To update the vision configuration, use the prefix `vision_` for each configuration parameter.
- To update the parent model configuration, do not use a prefix for each configuration parameter.
Behaves differently depending on whether a :obj:`config` is provided or automatically loaded.
Example::
>>> from transformers import FlaxHybridCLIP
>>> # initialize a model from pretrained BERT and CLIP models. Note that the projection layers will be randomly initialized.
>>> # If using CLIP's vision model the vision projection layer will be initialized using pre-trained weights
>>> model = FlaxHybridCLIP.from_text_vision_pretrained('bert-base-uncased', 'openai/clip-vit-base-patch32')
>>> # saving model after fine-tuning
>>> model.save_pretrained("./bert-clip")
>>> # load fine-tuned model
>>> model = FlaxHybridCLIP.from_pretrained("./bert-clip")
"""
kwargs_text = {
argument[len("text_") :]: value for argument, value in kwargs.items() if argument.startswith("text_")
}
kwargs_vision = {
argument[len("vision_") :]: value for argument, value in kwargs.items() if argument.startswith("vision_")
}
# remove text, vision kwargs from kwargs
for key in kwargs_text.keys():
del kwargs["text_" + key]
for key in kwargs_vision.keys():
del kwargs["vision_" + key]
# Load and initialize the text and vision model
text_model = kwargs_text.pop("model", None)
if text_model is None:
assert (
text_model_name_or_path is not None
), "If `model` is not defined as an argument, a `text_model_name_or_path` has to be defined"
from transformers import FlaxAutoModel
if "config" not in kwargs_text:
from transformers import AutoConfig
text_config = AutoConfig.from_pretrained(text_model_name_or_path)
kwargs_text["config"] = text_config
text_model = FlaxAutoModel.from_pretrained(text_model_name_or_path, *model_args, **kwargs_text)
vision_model = kwargs_vision.pop("model", None)
if vision_model is None:
assert (
vision_model_name_or_path is not None
), "If `model` is not defined as an argument, a `vision_model_name_or_path` has to be defined"
from transformers import FlaxAutoModel
if "config" not in kwargs_vision:
from transformers import AutoConfig
vision_config = AutoConfig.from_pretrained(vision_model_name_or_path)
kwargs_vision["config"] = vision_config
vision_model = FlaxAutoModel.from_pretrained(vision_model_name_or_path, *model_args, **kwargs_vision)
# instantiate config with corresponding kwargs
dtype = kwargs.pop("dtype", jnp.float32)
config = HybridCLIPConfig.from_text_vision_configs(text_model.config, vision_model.config, **kwargs)
# init model
model = cls(config, *model_args, dtype=dtype, **kwargs)
if vision_config.model_type == "clip":
model.params["vision_model"]["vision_model"] = vision_model.params["vision_model"]
model.params["visual_projection"]["kernel"] = vision_model.params["visual_projection"]["kernel"]
else:
model.params["vision_model"] = vision_model.params
model.params["text_model"] = text_model.params
return model
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/ibert/configuration_ibert.py | # coding=utf-8
# Copyright 2021 The I-BERT Authors (Sehoon Kim, Amir Gholami, Zhewei Yao,
# Michael Mahoney, Kurt Keutzer - UC Berkeley) and The HuggingFace Inc. team.
# Copyright (c) 20121, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" I-BERT configuration"""
from collections import OrderedDict
from typing import Mapping
from transformers.onnx import OnnxConfig
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"kssteven/ibert-roberta-base": "https://huggingface.co/kssteven/ibert-roberta-base/resolve/main/config.json",
"kssteven/ibert-roberta-large": "https://huggingface.co/kssteven/ibert-roberta-large/resolve/main/config.json",
"kssteven/ibert-roberta-large-mnli": (
"https://huggingface.co/kssteven/ibert-roberta-large-mnli/resolve/main/config.json"
),
}
class IBertConfig(PretrainedConfig):
"""
This is the configuration class to store the configuration of a [`IBertModel`]. It is used to instantiate a I-BERT
model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
defaults will yield a similar configuration to that of the IBERT
[kssteven/ibert-roberta-base](https://huggingface.co/kssteven/ibert-roberta-base) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 30522):
Vocabulary size of the I-BERT model. Defines the number of different tokens that can be represented by the
`inputs_ids` passed when calling [`IBertModel`]
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
intermediate_size (`int`, *optional*, defaults to 3072):
Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
hidden_act (`str` or `Callable`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
max_position_embeddings (`int`, *optional*, defaults to 512):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
type_vocab_size (`int`, *optional*, defaults to 2):
The vocabulary size of the `token_type_ids` passed when calling [`IBertModel`]
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
position_embedding_type (`str`, *optional*, defaults to `"absolute"`):
Type of position embedding. Choose one of `"absolute"`, `"relative_key"`, `"relative_key_query"`. For
positional embeddings use `"absolute"`. For more information on `"relative_key"`, please refer to
[Self-Attention with Relative Position Representations (Shaw et al.)](https://arxiv.org/abs/1803.02155).
For more information on `"relative_key_query"`, please refer to *Method 4* in [Improve Transformer Models
with Better Relative Position Embeddings (Huang et al.)](https://arxiv.org/abs/2009.13658).
quant_mode (`bool`, *optional*, defaults to `False`):
Whether to quantize the model or not.
force_dequant (`str`, *optional*, defaults to `"none"`):
Force dequantize specific nonlinear layer. Dequatized layers are then executed with full precision.
`"none"`, `"gelu"`, `"softmax"`, `"layernorm"` and `"nonlinear"` are supported. As deafult, it is set as
`"none"`, which does not dequantize any layers. Please specify `"gelu"`, `"softmax"`, or `"layernorm"` to
dequantize GELU, Softmax, or LayerNorm, respectively. `"nonlinear"` will dequantize all nonlinear layers,
i.e., GELU, Softmax, and LayerNorm.
"""
model_type = "ibert"
def __init__(
self,
vocab_size=30522,
hidden_size=768,
num_hidden_layers=12,
num_attention_heads=12,
intermediate_size=3072,
hidden_act="gelu",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=512,
type_vocab_size=2,
initializer_range=0.02,
layer_norm_eps=1e-12,
pad_token_id=1,
bos_token_id=0,
eos_token_id=2,
position_embedding_type="absolute",
quant_mode=False,
force_dequant="none",
**kwargs,
):
super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.hidden_act = hidden_act
self.intermediate_size = intermediate_size
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.type_vocab_size = type_vocab_size
self.initializer_range = initializer_range
self.layer_norm_eps = layer_norm_eps
self.position_embedding_type = position_embedding_type
self.quant_mode = quant_mode
self.force_dequant = force_dequant
class IBertOnnxConfig(OnnxConfig):
@property
def inputs(self) -> Mapping[str, Mapping[int, str]]:
if self.task == "multiple-choice":
dynamic_axis = {0: "batch", 1: "choice", 2: "sequence"}
else:
dynamic_axis = {0: "batch", 1: "sequence"}
return OrderedDict(
[
("input_ids", dynamic_axis),
("attention_mask", dynamic_axis),
]
)
| # coding=utf-8
# Copyright 2021 The I-BERT Authors (Sehoon Kim, Amir Gholami, Zhewei Yao,
# Michael Mahoney, Kurt Keutzer - UC Berkeley) and The HuggingFace Inc. team.
# Copyright (c) 20121, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" I-BERT configuration"""
from collections import OrderedDict
from typing import Mapping
from transformers.onnx import OnnxConfig
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"kssteven/ibert-roberta-base": "https://huggingface.co/kssteven/ibert-roberta-base/resolve/main/config.json",
"kssteven/ibert-roberta-large": "https://huggingface.co/kssteven/ibert-roberta-large/resolve/main/config.json",
"kssteven/ibert-roberta-large-mnli": (
"https://huggingface.co/kssteven/ibert-roberta-large-mnli/resolve/main/config.json"
),
}
class IBertConfig(PretrainedConfig):
"""
This is the configuration class to store the configuration of a [`IBertModel`]. It is used to instantiate a I-BERT
model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
defaults will yield a similar configuration to that of the IBERT
[kssteven/ibert-roberta-base](https://huggingface.co/kssteven/ibert-roberta-base) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 30522):
Vocabulary size of the I-BERT model. Defines the number of different tokens that can be represented by the
`inputs_ids` passed when calling [`IBertModel`]
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
intermediate_size (`int`, *optional*, defaults to 3072):
Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
hidden_act (`str` or `Callable`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
max_position_embeddings (`int`, *optional*, defaults to 512):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
type_vocab_size (`int`, *optional*, defaults to 2):
The vocabulary size of the `token_type_ids` passed when calling [`IBertModel`]
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
position_embedding_type (`str`, *optional*, defaults to `"absolute"`):
Type of position embedding. Choose one of `"absolute"`, `"relative_key"`, `"relative_key_query"`. For
positional embeddings use `"absolute"`. For more information on `"relative_key"`, please refer to
[Self-Attention with Relative Position Representations (Shaw et al.)](https://arxiv.org/abs/1803.02155).
For more information on `"relative_key_query"`, please refer to *Method 4* in [Improve Transformer Models
with Better Relative Position Embeddings (Huang et al.)](https://arxiv.org/abs/2009.13658).
quant_mode (`bool`, *optional*, defaults to `False`):
Whether to quantize the model or not.
force_dequant (`str`, *optional*, defaults to `"none"`):
Force dequantize specific nonlinear layer. Dequatized layers are then executed with full precision.
`"none"`, `"gelu"`, `"softmax"`, `"layernorm"` and `"nonlinear"` are supported. As deafult, it is set as
`"none"`, which does not dequantize any layers. Please specify `"gelu"`, `"softmax"`, or `"layernorm"` to
dequantize GELU, Softmax, or LayerNorm, respectively. `"nonlinear"` will dequantize all nonlinear layers,
i.e., GELU, Softmax, and LayerNorm.
"""
model_type = "ibert"
def __init__(
self,
vocab_size=30522,
hidden_size=768,
num_hidden_layers=12,
num_attention_heads=12,
intermediate_size=3072,
hidden_act="gelu",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=512,
type_vocab_size=2,
initializer_range=0.02,
layer_norm_eps=1e-12,
pad_token_id=1,
bos_token_id=0,
eos_token_id=2,
position_embedding_type="absolute",
quant_mode=False,
force_dequant="none",
**kwargs,
):
super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.hidden_act = hidden_act
self.intermediate_size = intermediate_size
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.type_vocab_size = type_vocab_size
self.initializer_range = initializer_range
self.layer_norm_eps = layer_norm_eps
self.position_embedding_type = position_embedding_type
self.quant_mode = quant_mode
self.force_dequant = force_dequant
class IBertOnnxConfig(OnnxConfig):
@property
def inputs(self) -> Mapping[str, Mapping[int, str]]:
if self.task == "multiple-choice":
dynamic_axis = {0: "batch", 1: "choice", 2: "sequence"}
else:
dynamic_axis = {0: "batch", 1: "sequence"}
return OrderedDict(
[
("input_ids", dynamic_axis),
("attention_mask", dynamic_axis),
]
)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/pegasus_x/configuration_pegasus_x.py | # coding=utf-8
# Copyright 2022, Google and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PEGASUS-X model configuration"""
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"google/pegasus-x-base": "https://huggingface.co/google/pegasus-x-base/resolve/main/config.json",
"google/pegasus-x-large": "https://huggingface.co/google/pegasus-x-large/resolve/main/config.json",
# See all PEGASUS-X models at https://huggingface.co/models?filter=pegasus-x
}
class PegasusXConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`PegasusXModel`]. It is used to instantiate a
PEGASUS-X model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the PEGASUS-X
[google/pegasus-x-large](https://huggingface.co/google/pegasus-x-large) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 96103):
Vocabulary size of the PEGASUS-X model. Defines the number of different tokens that can be represented by
the `inputs_ids` passed when calling [`PegasusXModel`].
d_model (`int`, *optional*, defaults to 1024):
Dimension of the layers and the pooler layer.
encoder_layers (`int`, *optional*, defaults to 16):
Number of encoder layers.
decoder_layers (`int`, *optional*, defaults to 16):
Number of decoder layers.
encoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
decoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer decoder.
decoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimension of the "intermediate" (often named feed-forward) layer in decoder.
encoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimension of the "intermediate" (often named feed-forward) layer in decoder.
activation_function (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for the attention probabilities.
activation_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for activations inside the fully connected layer.
max_position_embeddings (`int`, *optional*, defaults to 16384):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
init_std (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
encoder_layerdrop: (`float`, *optional*, defaults to 0.0):
The LayerDrop probability for the encoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
for more details.
decoder_layerdrop: (`float`, *optional*, defaults to 0.0):
The LayerDrop probability for the decoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
for more details.
use_cache (`bool`, *optional*, defaults to `True`):
Whether or not the model should return the last key/values attentions (not used by all models)
forced_eos_token_id (`int`, *optional*, defaults to 1):
The id of the token to force as the last generated token when `max_length` is reached. Usually set to
`eos_token_id`.
num_global_tokens (`int`, *optional*, defaults to 128):
Number of global tokens to use for the encoder
block_size (`int`, *optional*, defaults to 512):
Block size for encoder local attention. Sequence length should be an exact multiple of block size.
block_size must be a multiple of 2 if stagger_local_block is True
stagger_local_block (`bool`, *optional*, defaults to `True`):
Whether to stagger every other local attention by half a block
Example:
```python
>>> from transformers import PegasusXConfig, PegasusXModel
>>> # Initializing a PEGASUS google/pegasus-x-large style configuration
>>> configuration = PegasusXConfig()
>>> # Initializing a model (with random weights) from the google/pegasus-x-large style configuration
>>> model = PegasusXModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "pegasus_x"
keys_to_ignore_at_inference = ["past_key_values"]
attribute_map = {"num_attention_heads": "encoder_attention_heads", "hidden_size": "d_model"}
def __init__(
self,
vocab_size=96103,
max_position_embeddings=16384,
encoder_layers=16,
encoder_ffn_dim=4096,
encoder_attention_heads=16,
decoder_layers=16,
decoder_ffn_dim=4096,
decoder_attention_heads=16,
encoder_layerdrop=0.0,
decoder_layerdrop=0.0,
use_cache=True,
is_encoder_decoder=True,
activation_function="gelu",
d_model=1024,
dropout=0.1,
attention_dropout=0.0,
activation_dropout=0.0,
init_std=0.02,
decoder_start_token_id=0,
scale_embedding=True,
pad_token_id=0,
eos_token_id=1,
forced_eos_token_id=1,
num_global_tokens=32,
block_size=512,
stagger_local_blocks=True,
**kwargs,
):
self.vocab_size = vocab_size
self.max_position_embeddings = max_position_embeddings
self.d_model = d_model
self.encoder_ffn_dim = encoder_ffn_dim
self.encoder_layers = encoder_layers
self.encoder_attention_heads = encoder_attention_heads
self.decoder_ffn_dim = decoder_ffn_dim
self.decoder_layers = decoder_layers
self.decoder_attention_heads = decoder_attention_heads
self.dropout = dropout
self.attention_dropout = attention_dropout
self.activation_dropout = activation_dropout
self.activation_function = activation_function
self.init_std = init_std
self.encoder_layerdrop = encoder_layerdrop
self.decoder_layerdrop = decoder_layerdrop
self.use_cache = use_cache
self.num_hidden_layers = encoder_layers
self.scale_embedding = scale_embedding # scale factor will be sqrt(d_model) if True
self.num_global_tokens = num_global_tokens
self.block_size = block_size
self.stagger_local_blocks = stagger_local_blocks
super().__init__(
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
is_encoder_decoder=is_encoder_decoder,
decoder_start_token_id=decoder_start_token_id,
forced_eos_token_id=forced_eos_token_id,
**kwargs,
)
@property
def num_attention_heads(self) -> int:
return self.encoder_attention_heads
@property
def hidden_size(self) -> int:
return self.d_model
| # coding=utf-8
# Copyright 2022, Google and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PEGASUS-X model configuration"""
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"google/pegasus-x-base": "https://huggingface.co/google/pegasus-x-base/resolve/main/config.json",
"google/pegasus-x-large": "https://huggingface.co/google/pegasus-x-large/resolve/main/config.json",
# See all PEGASUS-X models at https://huggingface.co/models?filter=pegasus-x
}
class PegasusXConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`PegasusXModel`]. It is used to instantiate a
PEGASUS-X model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the PEGASUS-X
[google/pegasus-x-large](https://huggingface.co/google/pegasus-x-large) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 96103):
Vocabulary size of the PEGASUS-X model. Defines the number of different tokens that can be represented by
the `inputs_ids` passed when calling [`PegasusXModel`].
d_model (`int`, *optional*, defaults to 1024):
Dimension of the layers and the pooler layer.
encoder_layers (`int`, *optional*, defaults to 16):
Number of encoder layers.
decoder_layers (`int`, *optional*, defaults to 16):
Number of decoder layers.
encoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
decoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer decoder.
decoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimension of the "intermediate" (often named feed-forward) layer in decoder.
encoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimension of the "intermediate" (often named feed-forward) layer in decoder.
activation_function (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for the attention probabilities.
activation_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for activations inside the fully connected layer.
max_position_embeddings (`int`, *optional*, defaults to 16384):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
init_std (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
encoder_layerdrop: (`float`, *optional*, defaults to 0.0):
The LayerDrop probability for the encoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
for more details.
decoder_layerdrop: (`float`, *optional*, defaults to 0.0):
The LayerDrop probability for the decoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
for more details.
use_cache (`bool`, *optional*, defaults to `True`):
Whether or not the model should return the last key/values attentions (not used by all models)
forced_eos_token_id (`int`, *optional*, defaults to 1):
The id of the token to force as the last generated token when `max_length` is reached. Usually set to
`eos_token_id`.
num_global_tokens (`int`, *optional*, defaults to 128):
Number of global tokens to use for the encoder
block_size (`int`, *optional*, defaults to 512):
Block size for encoder local attention. Sequence length should be an exact multiple of block size.
block_size must be a multiple of 2 if stagger_local_block is True
stagger_local_block (`bool`, *optional*, defaults to `True`):
Whether to stagger every other local attention by half a block
Example:
```python
>>> from transformers import PegasusXConfig, PegasusXModel
>>> # Initializing a PEGASUS google/pegasus-x-large style configuration
>>> configuration = PegasusXConfig()
>>> # Initializing a model (with random weights) from the google/pegasus-x-large style configuration
>>> model = PegasusXModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "pegasus_x"
keys_to_ignore_at_inference = ["past_key_values"]
attribute_map = {"num_attention_heads": "encoder_attention_heads", "hidden_size": "d_model"}
def __init__(
self,
vocab_size=96103,
max_position_embeddings=16384,
encoder_layers=16,
encoder_ffn_dim=4096,
encoder_attention_heads=16,
decoder_layers=16,
decoder_ffn_dim=4096,
decoder_attention_heads=16,
encoder_layerdrop=0.0,
decoder_layerdrop=0.0,
use_cache=True,
is_encoder_decoder=True,
activation_function="gelu",
d_model=1024,
dropout=0.1,
attention_dropout=0.0,
activation_dropout=0.0,
init_std=0.02,
decoder_start_token_id=0,
scale_embedding=True,
pad_token_id=0,
eos_token_id=1,
forced_eos_token_id=1,
num_global_tokens=32,
block_size=512,
stagger_local_blocks=True,
**kwargs,
):
self.vocab_size = vocab_size
self.max_position_embeddings = max_position_embeddings
self.d_model = d_model
self.encoder_ffn_dim = encoder_ffn_dim
self.encoder_layers = encoder_layers
self.encoder_attention_heads = encoder_attention_heads
self.decoder_ffn_dim = decoder_ffn_dim
self.decoder_layers = decoder_layers
self.decoder_attention_heads = decoder_attention_heads
self.dropout = dropout
self.attention_dropout = attention_dropout
self.activation_dropout = activation_dropout
self.activation_function = activation_function
self.init_std = init_std
self.encoder_layerdrop = encoder_layerdrop
self.decoder_layerdrop = decoder_layerdrop
self.use_cache = use_cache
self.num_hidden_layers = encoder_layers
self.scale_embedding = scale_embedding # scale factor will be sqrt(d_model) if True
self.num_global_tokens = num_global_tokens
self.block_size = block_size
self.stagger_local_blocks = stagger_local_blocks
super().__init__(
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
is_encoder_decoder=is_encoder_decoder,
decoder_start_token_id=decoder_start_token_id,
forced_eos_token_id=forced_eos_token_id,
**kwargs,
)
@property
def num_attention_heads(self) -> int:
return self.encoder_attention_heads
@property
def hidden_size(self) -> int:
return self.d_model
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/swin2sr/test_modeling_swin2sr.py | # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch Swin2SR model. """
import inspect
import unittest
from transformers import Swin2SRConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import Swin2SRForImageSuperResolution, Swin2SRModel
from transformers.models.swin2sr.modeling_swin2sr import SWIN2SR_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import Swin2SRImageProcessor
class Swin2SRModelTester:
def __init__(
self,
parent,
batch_size=13,
image_size=32,
patch_size=1,
num_channels=3,
embed_dim=16,
depths=[1, 2, 1],
num_heads=[2, 2, 4],
window_size=2,
mlp_ratio=2.0,
qkv_bias=True,
hidden_dropout_prob=0.0,
attention_probs_dropout_prob=0.0,
drop_path_rate=0.1,
hidden_act="gelu",
use_absolute_embeddings=False,
patch_norm=True,
initializer_range=0.02,
layer_norm_eps=1e-5,
is_training=True,
scope=None,
use_labels=False,
upscale=2,
):
self.parent = parent
self.batch_size = batch_size
self.image_size = image_size
self.patch_size = patch_size
self.num_channels = num_channels
self.embed_dim = embed_dim
self.depths = depths
self.num_heads = num_heads
self.window_size = window_size
self.mlp_ratio = mlp_ratio
self.qkv_bias = qkv_bias
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.drop_path_rate = drop_path_rate
self.hidden_act = hidden_act
self.use_absolute_embeddings = use_absolute_embeddings
self.patch_norm = patch_norm
self.layer_norm_eps = layer_norm_eps
self.initializer_range = initializer_range
self.is_training = is_training
self.scope = scope
self.use_labels = use_labels
self.upscale = upscale
# here we set some attributes to make tests pass
self.num_hidden_layers = len(depths)
self.hidden_size = embed_dim
self.seq_length = (image_size // patch_size) ** 2
def prepare_config_and_inputs(self):
pixel_values = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size])
labels = None
if self.use_labels:
labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
config = self.get_config()
return config, pixel_values, labels
def get_config(self):
return Swin2SRConfig(
image_size=self.image_size,
patch_size=self.patch_size,
num_channels=self.num_channels,
embed_dim=self.embed_dim,
depths=self.depths,
num_heads=self.num_heads,
window_size=self.window_size,
mlp_ratio=self.mlp_ratio,
qkv_bias=self.qkv_bias,
hidden_dropout_prob=self.hidden_dropout_prob,
attention_probs_dropout_prob=self.attention_probs_dropout_prob,
drop_path_rate=self.drop_path_rate,
hidden_act=self.hidden_act,
use_absolute_embeddings=self.use_absolute_embeddings,
path_norm=self.patch_norm,
layer_norm_eps=self.layer_norm_eps,
initializer_range=self.initializer_range,
upscale=self.upscale,
)
def create_and_check_model(self, config, pixel_values, labels):
model = Swin2SRModel(config=config)
model.to(torch_device)
model.eval()
result = model(pixel_values)
self.parent.assertEqual(
result.last_hidden_state.shape, (self.batch_size, self.embed_dim, self.image_size, self.image_size)
)
def create_and_check_for_image_super_resolution(self, config, pixel_values, labels):
model = Swin2SRForImageSuperResolution(config)
model.to(torch_device)
model.eval()
result = model(pixel_values)
expected_image_size = self.image_size * self.upscale
self.parent.assertEqual(
result.reconstruction.shape, (self.batch_size, self.num_channels, expected_image_size, expected_image_size)
)
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
config, pixel_values, labels = config_and_inputs
inputs_dict = {"pixel_values": pixel_values}
return config, inputs_dict
@require_torch
class Swin2SRModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (Swin2SRModel, Swin2SRForImageSuperResolution) if is_torch_available() else ()
pipeline_model_mapping = {"feature-extraction": Swin2SRModel} if is_torch_available() else {}
fx_compatible = False
test_pruning = False
test_resize_embeddings = False
test_head_masking = False
test_torchscript = False
def setUp(self):
self.model_tester = Swin2SRModelTester(self)
self.config_tester = ConfigTester(self, config_class=Swin2SRConfig, embed_dim=37)
def test_config(self):
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def test_model(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*config_and_inputs)
def test_model_for_image_super_resolution(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_super_resolution(*config_and_inputs)
@unittest.skip(reason="Swin2SR does not use inputs_embeds")
def test_inputs_embeds(self):
pass
@unittest.skip(reason="Swin2SR does not support training yet")
def test_training(self):
pass
@unittest.skip(reason="Swin2SR does not support training yet")
def test_training_gradient_checkpointing(self):
pass
def test_model_common_attributes(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
self.assertIsInstance(model.get_input_embeddings(), (nn.Module))
x = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(x, nn.Linear))
def test_forward_signature(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
signature = inspect.signature(model.forward)
# signature.parameters is an OrderedDict => so arg_names order is deterministic
arg_names = [*signature.parameters.keys()]
expected_arg_names = ["pixel_values"]
self.assertListEqual(arg_names[:1], expected_arg_names)
@slow
def test_model_from_pretrained(self):
for model_name in SWIN2SR_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
model = Swin2SRModel.from_pretrained(model_name)
self.assertIsNotNone(model)
# overwriting because of `logit_scale` parameter
def test_initialization(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
configs_no_init = _config_zero_init(config)
for model_class in self.all_model_classes:
model = model_class(config=configs_no_init)
for name, param in model.named_parameters():
if "logit_scale" in name:
continue
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item(),
[0.0, 1.0],
msg=f"Parameter {name} of model {model_class} seems not properly initialized",
)
def test_attention_outputs(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
config.return_dict = True
for model_class in self.all_model_classes:
inputs_dict["output_attentions"] = True
inputs_dict["output_hidden_states"] = False
config.return_dict = True
model = model_class(config)
model.to(torch_device)
model.eval()
with torch.no_grad():
outputs = model(**self._prepare_for_class(inputs_dict, model_class))
attentions = outputs.attentions
expected_num_attentions = len(self.model_tester.depths)
self.assertEqual(len(attentions), expected_num_attentions)
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
config.output_attentions = True
window_size_squared = config.window_size**2
model = model_class(config)
model.to(torch_device)
model.eval()
with torch.no_grad():
outputs = model(**self._prepare_for_class(inputs_dict, model_class))
attentions = outputs.attentions
self.assertEqual(len(attentions), expected_num_attentions)
self.assertListEqual(
list(attentions[0].shape[-3:]),
[self.model_tester.num_heads[0], window_size_squared, window_size_squared],
)
out_len = len(outputs)
# Check attention is always last and order is fine
inputs_dict["output_attentions"] = True
inputs_dict["output_hidden_states"] = True
model = model_class(config)
model.to(torch_device)
model.eval()
with torch.no_grad():
outputs = model(**self._prepare_for_class(inputs_dict, model_class))
self.assertEqual(out_len + 1, len(outputs))
self_attentions = outputs.attentions
self.assertEqual(len(self_attentions), expected_num_attentions)
self.assertListEqual(
list(self_attentions[0].shape[-3:]),
[self.model_tester.num_heads[0], window_size_squared, window_size_squared],
)
@require_vision
@require_torch
@slow
class Swin2SRModelIntegrationTest(unittest.TestCase):
def test_inference_image_super_resolution_head(self):
processor = Swin2SRImageProcessor()
model = Swin2SRForImageSuperResolution.from_pretrained("caidas/swin2SR-classical-sr-x2-64").to(torch_device)
image = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png")
inputs = processor(images=image, return_tensors="pt").to(torch_device)
# forward pass
with torch.no_grad():
outputs = model(**inputs)
# verify the logits
expected_shape = torch.Size([1, 3, 976, 1296])
self.assertEqual(outputs.reconstruction.shape, expected_shape)
expected_slice = torch.tensor(
[[0.5458, 0.5546, 0.5638], [0.5526, 0.5565, 0.5651], [0.5396, 0.5426, 0.5621]]
).to(torch_device)
self.assertTrue(torch.allclose(outputs.reconstruction[0, 0, :3, :3], expected_slice, atol=1e-4))
| # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch Swin2SR model. """
import inspect
import unittest
from transformers import Swin2SRConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import Swin2SRForImageSuperResolution, Swin2SRModel
from transformers.models.swin2sr.modeling_swin2sr import SWIN2SR_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import Swin2SRImageProcessor
class Swin2SRModelTester:
def __init__(
self,
parent,
batch_size=13,
image_size=32,
patch_size=1,
num_channels=3,
embed_dim=16,
depths=[1, 2, 1],
num_heads=[2, 2, 4],
window_size=2,
mlp_ratio=2.0,
qkv_bias=True,
hidden_dropout_prob=0.0,
attention_probs_dropout_prob=0.0,
drop_path_rate=0.1,
hidden_act="gelu",
use_absolute_embeddings=False,
patch_norm=True,
initializer_range=0.02,
layer_norm_eps=1e-5,
is_training=True,
scope=None,
use_labels=False,
upscale=2,
):
self.parent = parent
self.batch_size = batch_size
self.image_size = image_size
self.patch_size = patch_size
self.num_channels = num_channels
self.embed_dim = embed_dim
self.depths = depths
self.num_heads = num_heads
self.window_size = window_size
self.mlp_ratio = mlp_ratio
self.qkv_bias = qkv_bias
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.drop_path_rate = drop_path_rate
self.hidden_act = hidden_act
self.use_absolute_embeddings = use_absolute_embeddings
self.patch_norm = patch_norm
self.layer_norm_eps = layer_norm_eps
self.initializer_range = initializer_range
self.is_training = is_training
self.scope = scope
self.use_labels = use_labels
self.upscale = upscale
# here we set some attributes to make tests pass
self.num_hidden_layers = len(depths)
self.hidden_size = embed_dim
self.seq_length = (image_size // patch_size) ** 2
def prepare_config_and_inputs(self):
pixel_values = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size])
labels = None
if self.use_labels:
labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
config = self.get_config()
return config, pixel_values, labels
def get_config(self):
return Swin2SRConfig(
image_size=self.image_size,
patch_size=self.patch_size,
num_channels=self.num_channels,
embed_dim=self.embed_dim,
depths=self.depths,
num_heads=self.num_heads,
window_size=self.window_size,
mlp_ratio=self.mlp_ratio,
qkv_bias=self.qkv_bias,
hidden_dropout_prob=self.hidden_dropout_prob,
attention_probs_dropout_prob=self.attention_probs_dropout_prob,
drop_path_rate=self.drop_path_rate,
hidden_act=self.hidden_act,
use_absolute_embeddings=self.use_absolute_embeddings,
path_norm=self.patch_norm,
layer_norm_eps=self.layer_norm_eps,
initializer_range=self.initializer_range,
upscale=self.upscale,
)
def create_and_check_model(self, config, pixel_values, labels):
model = Swin2SRModel(config=config)
model.to(torch_device)
model.eval()
result = model(pixel_values)
self.parent.assertEqual(
result.last_hidden_state.shape, (self.batch_size, self.embed_dim, self.image_size, self.image_size)
)
def create_and_check_for_image_super_resolution(self, config, pixel_values, labels):
model = Swin2SRForImageSuperResolution(config)
model.to(torch_device)
model.eval()
result = model(pixel_values)
expected_image_size = self.image_size * self.upscale
self.parent.assertEqual(
result.reconstruction.shape, (self.batch_size, self.num_channels, expected_image_size, expected_image_size)
)
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
config, pixel_values, labels = config_and_inputs
inputs_dict = {"pixel_values": pixel_values}
return config, inputs_dict
@require_torch
class Swin2SRModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (Swin2SRModel, Swin2SRForImageSuperResolution) if is_torch_available() else ()
pipeline_model_mapping = {"feature-extraction": Swin2SRModel} if is_torch_available() else {}
fx_compatible = False
test_pruning = False
test_resize_embeddings = False
test_head_masking = False
test_torchscript = False
def setUp(self):
self.model_tester = Swin2SRModelTester(self)
self.config_tester = ConfigTester(self, config_class=Swin2SRConfig, embed_dim=37)
def test_config(self):
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def test_model(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*config_and_inputs)
def test_model_for_image_super_resolution(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_super_resolution(*config_and_inputs)
@unittest.skip(reason="Swin2SR does not use inputs_embeds")
def test_inputs_embeds(self):
pass
@unittest.skip(reason="Swin2SR does not support training yet")
def test_training(self):
pass
@unittest.skip(reason="Swin2SR does not support training yet")
def test_training_gradient_checkpointing(self):
pass
def test_model_common_attributes(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
self.assertIsInstance(model.get_input_embeddings(), (nn.Module))
x = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(x, nn.Linear))
def test_forward_signature(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
signature = inspect.signature(model.forward)
# signature.parameters is an OrderedDict => so arg_names order is deterministic
arg_names = [*signature.parameters.keys()]
expected_arg_names = ["pixel_values"]
self.assertListEqual(arg_names[:1], expected_arg_names)
@slow
def test_model_from_pretrained(self):
for model_name in SWIN2SR_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
model = Swin2SRModel.from_pretrained(model_name)
self.assertIsNotNone(model)
# overwriting because of `logit_scale` parameter
def test_initialization(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
configs_no_init = _config_zero_init(config)
for model_class in self.all_model_classes:
model = model_class(config=configs_no_init)
for name, param in model.named_parameters():
if "logit_scale" in name:
continue
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item(),
[0.0, 1.0],
msg=f"Parameter {name} of model {model_class} seems not properly initialized",
)
def test_attention_outputs(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
config.return_dict = True
for model_class in self.all_model_classes:
inputs_dict["output_attentions"] = True
inputs_dict["output_hidden_states"] = False
config.return_dict = True
model = model_class(config)
model.to(torch_device)
model.eval()
with torch.no_grad():
outputs = model(**self._prepare_for_class(inputs_dict, model_class))
attentions = outputs.attentions
expected_num_attentions = len(self.model_tester.depths)
self.assertEqual(len(attentions), expected_num_attentions)
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
config.output_attentions = True
window_size_squared = config.window_size**2
model = model_class(config)
model.to(torch_device)
model.eval()
with torch.no_grad():
outputs = model(**self._prepare_for_class(inputs_dict, model_class))
attentions = outputs.attentions
self.assertEqual(len(attentions), expected_num_attentions)
self.assertListEqual(
list(attentions[0].shape[-3:]),
[self.model_tester.num_heads[0], window_size_squared, window_size_squared],
)
out_len = len(outputs)
# Check attention is always last and order is fine
inputs_dict["output_attentions"] = True
inputs_dict["output_hidden_states"] = True
model = model_class(config)
model.to(torch_device)
model.eval()
with torch.no_grad():
outputs = model(**self._prepare_for_class(inputs_dict, model_class))
self.assertEqual(out_len + 1, len(outputs))
self_attentions = outputs.attentions
self.assertEqual(len(self_attentions), expected_num_attentions)
self.assertListEqual(
list(self_attentions[0].shape[-3:]),
[self.model_tester.num_heads[0], window_size_squared, window_size_squared],
)
@require_vision
@require_torch
@slow
class Swin2SRModelIntegrationTest(unittest.TestCase):
def test_inference_image_super_resolution_head(self):
processor = Swin2SRImageProcessor()
model = Swin2SRForImageSuperResolution.from_pretrained("caidas/swin2SR-classical-sr-x2-64").to(torch_device)
image = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png")
inputs = processor(images=image, return_tensors="pt").to(torch_device)
# forward pass
with torch.no_grad():
outputs = model(**inputs)
# verify the logits
expected_shape = torch.Size([1, 3, 976, 1296])
self.assertEqual(outputs.reconstruction.shape, expected_shape)
expected_slice = torch.tensor(
[[0.5458, 0.5546, 0.5638], [0.5526, 0.5565, 0.5651], [0.5396, 0.5426, 0.5621]]
).to(torch_device)
self.assertTrue(torch.allclose(outputs.reconstruction[0, 0, :3, :3], expected_slice, atol=1e-4))
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/en/attention.mdx | <!--Copyright 2023 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Attention mechanisms
Most transformer models use full attention in the sense that the attention matrix is square. It can be a big
computational bottleneck when you have long texts. Longformer and reformer are models that try to be more efficient and
use a sparse version of the attention matrix to speed up training.
## LSH attention
[Reformer](#reformer) uses LSH attention. In the softmax(QK^t), only the biggest elements (in the softmax
dimension) of the matrix QK^t are going to give useful contributions. So for each query q in Q, we can consider only
the keys k in K that are close to q. A hash function is used to determine if q and k are close. The attention mask is
modified to mask the current token (except at the first position), because it will give a query and a key equal (so
very similar to each other). Since the hash can be a bit random, several hash functions are used in practice
(determined by a n_rounds parameter) and then are averaged together.
## Local attention
[Longformer](#longformer) uses local attention: often, the local context (e.g., what are the two tokens to the
left and right?) is enough to take action for a given token. Also, by stacking attention layers that have a small
window, the last layer will have a receptive field of more than just the tokens in the window, allowing them to build a
representation of the whole sentence.
Some preselected input tokens are also given global attention: for those few tokens, the attention matrix can access
all tokens and this process is symmetric: all other tokens have access to those specific tokens (on top of the ones in
their local window). This is shown in Figure 2d of the paper, see below for a sample attention mask:
<div class="flex justify-center">
<img scale="50 %" align="center" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/local_attention_mask.png"/>
</div>
Using those attention matrices with less parameters then allows the model to have inputs having a bigger sequence
length.
## Other tricks
### Axial positional encodings
[Reformer](#reformer) uses axial positional encodings: in traditional transformer models, the positional encoding
E is a matrix of size \\(l\\) by \\(d\\), \\(l\\) being the sequence length and \\(d\\) the dimension of the
hidden state. If you have very long texts, this matrix can be huge and take way too much space on the GPU. To alleviate
that, axial positional encodings consist of factorizing that big matrix E in two smaller matrices E1 and E2, with
dimensions \\(l_{1} \times d_{1}\\) and \\(l_{2} \times d_{2}\\), such that \\(l_{1} \times l_{2} = l\\) and
\\(d_{1} + d_{2} = d\\) (with the product for the lengths, this ends up being way smaller). The embedding for time
step \\(j\\) in E is obtained by concatenating the embeddings for timestep \\(j \% l1\\) in E1 and \\(j // l1\\)
in E2.
| <!--Copyright 2023 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Attention mechanisms
Most transformer models use full attention in the sense that the attention matrix is square. It can be a big
computational bottleneck when you have long texts. Longformer and reformer are models that try to be more efficient and
use a sparse version of the attention matrix to speed up training.
## LSH attention
[Reformer](#reformer) uses LSH attention. In the softmax(QK^t), only the biggest elements (in the softmax
dimension) of the matrix QK^t are going to give useful contributions. So for each query q in Q, we can consider only
the keys k in K that are close to q. A hash function is used to determine if q and k are close. The attention mask is
modified to mask the current token (except at the first position), because it will give a query and a key equal (so
very similar to each other). Since the hash can be a bit random, several hash functions are used in practice
(determined by a n_rounds parameter) and then are averaged together.
## Local attention
[Longformer](#longformer) uses local attention: often, the local context (e.g., what are the two tokens to the
left and right?) is enough to take action for a given token. Also, by stacking attention layers that have a small
window, the last layer will have a receptive field of more than just the tokens in the window, allowing them to build a
representation of the whole sentence.
Some preselected input tokens are also given global attention: for those few tokens, the attention matrix can access
all tokens and this process is symmetric: all other tokens have access to those specific tokens (on top of the ones in
their local window). This is shown in Figure 2d of the paper, see below for a sample attention mask:
<div class="flex justify-center">
<img scale="50 %" align="center" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/local_attention_mask.png"/>
</div>
Using those attention matrices with less parameters then allows the model to have inputs having a bigger sequence
length.
## Other tricks
### Axial positional encodings
[Reformer](#reformer) uses axial positional encodings: in traditional transformer models, the positional encoding
E is a matrix of size \\(l\\) by \\(d\\), \\(l\\) being the sequence length and \\(d\\) the dimension of the
hidden state. If you have very long texts, this matrix can be huge and take way too much space on the GPU. To alleviate
that, axial positional encodings consist of factorizing that big matrix E in two smaller matrices E1 and E2, with
dimensions \\(l_{1} \times d_{1}\\) and \\(l_{2} \times d_{2}\\), such that \\(l_{1} \times l_{2} = l\\) and
\\(d_{1} + d_{2} = d\\) (with the product for the lengths, this ends up being way smaller). The embedding for time
step \\(j\\) in E is obtained by concatenating the embeddings for timestep \\(j \% l1\\) in E1 and \\(j // l1\\)
in E2.
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/upernet/configuration_upernet.py | # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" UperNet model configuration"""
import copy
from ...configuration_utils import PretrainedConfig
from ...utils import logging
from ..auto.configuration_auto import CONFIG_MAPPING
logger = logging.get_logger(__name__)
class UperNetConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of an [`UperNetForSemanticSegmentation`]. It is used to
instantiate an UperNet model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the UperNet
[openmmlab/upernet-convnext-tiny](https://huggingface.co/openmmlab/upernet-convnext-tiny) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
backbone_config (`PretrainedConfig` or `dict`, *optional*, defaults to `ResNetConfig()`):
The configuration of the backbone model.
hidden_size (`int`, *optional*, defaults to 512):
The number of hidden units in the convolutional layers.
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
pool_scales (`Tuple[int]`, *optional*, defaults to `[1, 2, 3, 6]`):
Pooling scales used in Pooling Pyramid Module applied on the last feature map.
use_auxiliary_head (`bool`, *optional*, defaults to `True`):
Whether to use an auxiliary head during training.
auxiliary_loss_weight (`float`, *optional*, defaults to 0.4):
Weight of the cross-entropy loss of the auxiliary head.
auxiliary_channels (`int`, *optional*, defaults to 256):
Number of channels to use in the auxiliary head.
auxiliary_num_convs (`int`, *optional*, defaults to 1):
Number of convolutional layers to use in the auxiliary head.
auxiliary_concat_input (`bool`, *optional*, defaults to `False`):
Whether to concatenate the output of the auxiliary head with the input before the classification layer.
loss_ignore_index (`int`, *optional*, defaults to 255):
The index that is ignored by the loss function.
Examples:
```python
>>> from transformers import UperNetConfig, UperNetForSemanticSegmentation
>>> # Initializing a configuration
>>> configuration = UperNetConfig()
>>> # Initializing a model (with random weights) from the configuration
>>> model = UperNetForSemanticSegmentation(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "upernet"
def __init__(
self,
backbone_config=None,
hidden_size=512,
initializer_range=0.02,
pool_scales=[1, 2, 3, 6],
use_auxiliary_head=True,
auxiliary_loss_weight=0.4,
auxiliary_in_channels=384,
auxiliary_channels=256,
auxiliary_num_convs=1,
auxiliary_concat_input=False,
loss_ignore_index=255,
**kwargs,
):
super().__init__(**kwargs)
if backbone_config is None:
logger.info("`backbone_config` is `None`. Initializing the config with the default `ResNet` backbone.")
backbone_config = CONFIG_MAPPING["resnet"](out_features=["stage1", "stage2", "stage3", "stage4"])
elif isinstance(backbone_config, dict):
backbone_model_type = backbone_config.get("model_type")
config_class = CONFIG_MAPPING[backbone_model_type]
backbone_config = config_class.from_dict(backbone_config)
self.backbone_config = backbone_config
self.hidden_size = hidden_size
self.initializer_range = initializer_range
self.pool_scales = pool_scales
self.use_auxiliary_head = use_auxiliary_head
self.auxiliary_loss_weight = auxiliary_loss_weight
self.auxiliary_in_channels = auxiliary_in_channels
self.auxiliary_channels = auxiliary_channels
self.auxiliary_num_convs = auxiliary_num_convs
self.auxiliary_concat_input = auxiliary_concat_input
self.loss_ignore_index = loss_ignore_index
def to_dict(self):
"""
Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`]. Returns:
`Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
"""
output = copy.deepcopy(self.__dict__)
output["backbone_config"] = self.backbone_config.to_dict()
output["model_type"] = self.__class__.model_type
return output
| # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" UperNet model configuration"""
import copy
from ...configuration_utils import PretrainedConfig
from ...utils import logging
from ..auto.configuration_auto import CONFIG_MAPPING
logger = logging.get_logger(__name__)
class UperNetConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of an [`UperNetForSemanticSegmentation`]. It is used to
instantiate an UperNet model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the UperNet
[openmmlab/upernet-convnext-tiny](https://huggingface.co/openmmlab/upernet-convnext-tiny) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
backbone_config (`PretrainedConfig` or `dict`, *optional*, defaults to `ResNetConfig()`):
The configuration of the backbone model.
hidden_size (`int`, *optional*, defaults to 512):
The number of hidden units in the convolutional layers.
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
pool_scales (`Tuple[int]`, *optional*, defaults to `[1, 2, 3, 6]`):
Pooling scales used in Pooling Pyramid Module applied on the last feature map.
use_auxiliary_head (`bool`, *optional*, defaults to `True`):
Whether to use an auxiliary head during training.
auxiliary_loss_weight (`float`, *optional*, defaults to 0.4):
Weight of the cross-entropy loss of the auxiliary head.
auxiliary_channels (`int`, *optional*, defaults to 256):
Number of channels to use in the auxiliary head.
auxiliary_num_convs (`int`, *optional*, defaults to 1):
Number of convolutional layers to use in the auxiliary head.
auxiliary_concat_input (`bool`, *optional*, defaults to `False`):
Whether to concatenate the output of the auxiliary head with the input before the classification layer.
loss_ignore_index (`int`, *optional*, defaults to 255):
The index that is ignored by the loss function.
Examples:
```python
>>> from transformers import UperNetConfig, UperNetForSemanticSegmentation
>>> # Initializing a configuration
>>> configuration = UperNetConfig()
>>> # Initializing a model (with random weights) from the configuration
>>> model = UperNetForSemanticSegmentation(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "upernet"
def __init__(
self,
backbone_config=None,
hidden_size=512,
initializer_range=0.02,
pool_scales=[1, 2, 3, 6],
use_auxiliary_head=True,
auxiliary_loss_weight=0.4,
auxiliary_in_channels=384,
auxiliary_channels=256,
auxiliary_num_convs=1,
auxiliary_concat_input=False,
loss_ignore_index=255,
**kwargs,
):
super().__init__(**kwargs)
if backbone_config is None:
logger.info("`backbone_config` is `None`. Initializing the config with the default `ResNet` backbone.")
backbone_config = CONFIG_MAPPING["resnet"](out_features=["stage1", "stage2", "stage3", "stage4"])
elif isinstance(backbone_config, dict):
backbone_model_type = backbone_config.get("model_type")
config_class = CONFIG_MAPPING[backbone_model_type]
backbone_config = config_class.from_dict(backbone_config)
self.backbone_config = backbone_config
self.hidden_size = hidden_size
self.initializer_range = initializer_range
self.pool_scales = pool_scales
self.use_auxiliary_head = use_auxiliary_head
self.auxiliary_loss_weight = auxiliary_loss_weight
self.auxiliary_in_channels = auxiliary_in_channels
self.auxiliary_channels = auxiliary_channels
self.auxiliary_num_convs = auxiliary_num_convs
self.auxiliary_concat_input = auxiliary_concat_input
self.loss_ignore_index = loss_ignore_index
def to_dict(self):
"""
Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`]. Returns:
`Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
"""
output = copy.deepcopy(self.__dict__)
output["backbone_config"] = self.backbone_config.to_dict()
output["model_type"] = self.__class__.model_type
return output
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/distillation/run_squad_w_distillation.py | # coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" This is the exact same script as `examples/question-answering/run_squad.py` (as of 2020, January 8th) with an additional and optional step of distillation."""
import argparse
import glob
import logging
import os
import random
import timeit
import numpy as np
import torch
from torch import nn
from torch.utils.data import DataLoader, RandomSampler, SequentialSampler
from torch.utils.data.distributed import DistributedSampler
from tqdm import tqdm, trange
import transformers
from transformers import (
WEIGHTS_NAME,
AdamW,
BertConfig,
BertForQuestionAnswering,
BertTokenizer,
DistilBertConfig,
DistilBertForQuestionAnswering,
DistilBertTokenizer,
RobertaConfig,
RobertaForQuestionAnswering,
RobertaTokenizer,
XLMConfig,
XLMForQuestionAnswering,
XLMTokenizer,
XLNetConfig,
XLNetForQuestionAnswering,
XLNetTokenizer,
get_linear_schedule_with_warmup,
squad_convert_examples_to_features,
)
from transformers.data.metrics.squad_metrics import (
compute_predictions_log_probs,
compute_predictions_logits,
squad_evaluate,
)
from transformers.data.processors.squad import SquadResult, SquadV1Processor, SquadV2Processor
from transformers.trainer_utils import is_main_process
try:
from torch.utils.tensorboard import SummaryWriter
except ImportError:
from tensorboardX import SummaryWriter
logger = logging.getLogger(__name__)
MODEL_CLASSES = {
"bert": (BertConfig, BertForQuestionAnswering, BertTokenizer),
"xlnet": (XLNetConfig, XLNetForQuestionAnswering, XLNetTokenizer),
"xlm": (XLMConfig, XLMForQuestionAnswering, XLMTokenizer),
"distilbert": (DistilBertConfig, DistilBertForQuestionAnswering, DistilBertTokenizer),
"roberta": (RobertaConfig, RobertaForQuestionAnswering, RobertaTokenizer),
}
def set_seed(args):
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.n_gpu > 0:
torch.cuda.manual_seed_all(args.seed)
def to_list(tensor):
return tensor.detach().cpu().tolist()
def train(args, train_dataset, model, tokenizer, teacher=None):
"""Train the model"""
if args.local_rank in [-1, 0]:
tb_writer = SummaryWriter()
args.train_batch_size = args.per_gpu_train_batch_size * max(1, args.n_gpu)
train_sampler = RandomSampler(train_dataset) if args.local_rank == -1 else DistributedSampler(train_dataset)
train_dataloader = DataLoader(train_dataset, sampler=train_sampler, batch_size=args.train_batch_size)
if args.max_steps > 0:
t_total = args.max_steps
args.num_train_epochs = args.max_steps // (len(train_dataloader) // args.gradient_accumulation_steps) + 1
else:
t_total = len(train_dataloader) // args.gradient_accumulation_steps * args.num_train_epochs
# Prepare optimizer and schedule (linear warmup and decay)
no_decay = ["bias", "LayerNorm.weight"]
optimizer_grouped_parameters = [
{
"params": [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)],
"weight_decay": args.weight_decay,
},
{"params": [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)], "weight_decay": 0.0},
]
optimizer = AdamW(optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon)
scheduler = get_linear_schedule_with_warmup(
optimizer, num_warmup_steps=args.warmup_steps, num_training_steps=t_total
)
# Check if saved optimizer or scheduler states exist
if os.path.isfile(os.path.join(args.model_name_or_path, "optimizer.pt")) and os.path.isfile(
os.path.join(args.model_name_or_path, "scheduler.pt")
):
# Load in optimizer and scheduler states
optimizer.load_state_dict(torch.load(os.path.join(args.model_name_or_path, "optimizer.pt")))
scheduler.load_state_dict(torch.load(os.path.join(args.model_name_or_path, "scheduler.pt")))
if args.fp16:
try:
from apex import amp
except ImportError:
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use fp16 training.")
model, optimizer = amp.initialize(model, optimizer, opt_level=args.fp16_opt_level)
# multi-gpu training (should be after apex fp16 initialization)
if args.n_gpu > 1:
model = nn.DataParallel(model)
# Distributed training (should be after apex fp16 initialization)
if args.local_rank != -1:
model = nn.parallel.DistributedDataParallel(
model, device_ids=[args.local_rank], output_device=args.local_rank, find_unused_parameters=True
)
# Train!
logger.info("***** Running training *****")
logger.info(" Num examples = %d", len(train_dataset))
logger.info(" Num Epochs = %d", args.num_train_epochs)
logger.info(" Instantaneous batch size per GPU = %d", args.per_gpu_train_batch_size)
logger.info(
" Total train batch size (w. parallel, distributed & accumulation) = %d",
args.train_batch_size
* args.gradient_accumulation_steps
* (torch.distributed.get_world_size() if args.local_rank != -1 else 1),
)
logger.info(" Gradient Accumulation steps = %d", args.gradient_accumulation_steps)
logger.info(" Total optimization steps = %d", t_total)
global_step = 1
epochs_trained = 0
steps_trained_in_current_epoch = 0
# Check if continuing training from a checkpoint
if os.path.exists(args.model_name_or_path):
try:
# set global_step to gobal_step of last saved checkpoint from model path
checkpoint_suffix = args.model_name_or_path.split("-")[-1].split("/")[0]
global_step = int(checkpoint_suffix)
epochs_trained = global_step // (len(train_dataloader) // args.gradient_accumulation_steps)
steps_trained_in_current_epoch = global_step % (len(train_dataloader) // args.gradient_accumulation_steps)
logger.info(" Continuing training from checkpoint, will skip to saved global_step")
logger.info(" Continuing training from epoch %d", epochs_trained)
logger.info(" Continuing training from global step %d", global_step)
logger.info(" Will skip the first %d steps in the first epoch", steps_trained_in_current_epoch)
except ValueError:
logger.info(" Starting fine-tuning.")
tr_loss, logging_loss = 0.0, 0.0
model.zero_grad()
train_iterator = trange(
epochs_trained, int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0]
)
# Added here for reproductibility
set_seed(args)
for _ in train_iterator:
epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
for step, batch in enumerate(epoch_iterator):
# Skip past any already trained steps if resuming training
if steps_trained_in_current_epoch > 0:
steps_trained_in_current_epoch -= 1
continue
model.train()
if teacher is not None:
teacher.eval()
batch = tuple(t.to(args.device) for t in batch)
inputs = {
"input_ids": batch[0],
"attention_mask": batch[1],
"start_positions": batch[3],
"end_positions": batch[4],
}
if args.model_type != "distilbert":
inputs["token_type_ids"] = None if args.model_type == "xlm" else batch[2]
if args.model_type in ["xlnet", "xlm"]:
inputs.update({"cls_index": batch[5], "p_mask": batch[6]})
if args.version_2_with_negative:
inputs.update({"is_impossible": batch[7]})
outputs = model(**inputs)
loss, start_logits_stu, end_logits_stu = outputs
# Distillation loss
if teacher is not None:
if "token_type_ids" not in inputs:
inputs["token_type_ids"] = None if args.teacher_type == "xlm" else batch[2]
with torch.no_grad():
start_logits_tea, end_logits_tea = teacher(
input_ids=inputs["input_ids"],
token_type_ids=inputs["token_type_ids"],
attention_mask=inputs["attention_mask"],
)
assert start_logits_tea.size() == start_logits_stu.size()
assert end_logits_tea.size() == end_logits_stu.size()
loss_fct = nn.KLDivLoss(reduction="batchmean")
loss_start = loss_fct(
nn.functional.log_softmax(start_logits_stu / args.temperature, dim=-1),
nn.functional.softmax(start_logits_tea / args.temperature, dim=-1),
) * (args.temperature**2)
loss_end = loss_fct(
nn.functional.log_softmax(end_logits_stu / args.temperature, dim=-1),
nn.functional.softmax(end_logits_tea / args.temperature, dim=-1),
) * (args.temperature**2)
loss_ce = (loss_start + loss_end) / 2.0
loss = args.alpha_ce * loss_ce + args.alpha_squad * loss
if args.n_gpu > 1:
loss = loss.mean() # mean() to average on multi-gpu parallel (not distributed) training
if args.gradient_accumulation_steps > 1:
loss = loss / args.gradient_accumulation_steps
if args.fp16:
with amp.scale_loss(loss, optimizer) as scaled_loss:
scaled_loss.backward()
else:
loss.backward()
tr_loss += loss.item()
if (step + 1) % args.gradient_accumulation_steps == 0:
if args.fp16:
nn.utils.clip_grad_norm_(amp.master_params(optimizer), args.max_grad_norm)
else:
nn.utils.clip_grad_norm_(model.parameters(), args.max_grad_norm)
optimizer.step()
scheduler.step() # Update learning rate schedule
model.zero_grad()
global_step += 1
# Log metrics
if args.local_rank in [-1, 0] and args.logging_steps > 0 and global_step % args.logging_steps == 0:
# Only evaluate when single GPU otherwise metrics may not average well
if args.local_rank == -1 and args.evaluate_during_training:
results = evaluate(args, model, tokenizer)
for key, value in results.items():
tb_writer.add_scalar("eval_{}".format(key), value, global_step)
tb_writer.add_scalar("lr", scheduler.get_lr()[0], global_step)
tb_writer.add_scalar("loss", (tr_loss - logging_loss) / args.logging_steps, global_step)
logging_loss = tr_loss
if args.local_rank in [-1, 0] and args.save_steps > 0 and global_step % args.save_steps == 0:
# Save model checkpoint
output_dir = os.path.join(args.output_dir, "checkpoint-{}".format(global_step))
if not os.path.exists(output_dir):
os.makedirs(output_dir)
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(output_dir)
tokenizer.save_pretrained(output_dir)
torch.save(args, os.path.join(output_dir, "training_args.bin"))
logger.info("Saving model checkpoint to %s", output_dir)
torch.save(optimizer.state_dict(), os.path.join(output_dir, "optimizer.pt"))
torch.save(scheduler.state_dict(), os.path.join(output_dir, "scheduler.pt"))
logger.info("Saving optimizer and scheduler states to %s", output_dir)
if args.max_steps > 0 and global_step > args.max_steps:
epoch_iterator.close()
break
if args.max_steps > 0 and global_step > args.max_steps:
train_iterator.close()
break
if args.local_rank in [-1, 0]:
tb_writer.close()
return global_step, tr_loss / global_step
def evaluate(args, model, tokenizer, prefix=""):
dataset, examples, features = load_and_cache_examples(args, tokenizer, evaluate=True, output_examples=True)
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(dataset)
eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu evaluate
if args.n_gpu > 1 and not isinstance(model, nn.DataParallel):
model = nn.DataParallel(model)
# Eval!
logger.info("***** Running evaluation {} *****".format(prefix))
logger.info(" Num examples = %d", len(dataset))
logger.info(" Batch size = %d", args.eval_batch_size)
all_results = []
start_time = timeit.default_timer()
for batch in tqdm(eval_dataloader, desc="Evaluating"):
model.eval()
batch = tuple(t.to(args.device) for t in batch)
with torch.no_grad():
inputs = {"input_ids": batch[0], "attention_mask": batch[1]}
if args.model_type != "distilbert":
inputs["token_type_ids"] = None if args.model_type == "xlm" else batch[2] # XLM don't use segment_ids
example_indices = batch[3]
if args.model_type in ["xlnet", "xlm"]:
inputs.update({"cls_index": batch[4], "p_mask": batch[5]})
outputs = model(**inputs)
for i, example_index in enumerate(example_indices):
eval_feature = features[example_index.item()]
unique_id = int(eval_feature.unique_id)
output = [to_list(output[i]) for output in outputs]
# Some models (XLNet, XLM) use 5 arguments for their predictions, while the other "simpler"
# models only use two.
if len(output) >= 5:
start_logits = output[0]
start_top_index = output[1]
end_logits = output[2]
end_top_index = output[3]
cls_logits = output[4]
result = SquadResult(
unique_id,
start_logits,
end_logits,
start_top_index=start_top_index,
end_top_index=end_top_index,
cls_logits=cls_logits,
)
else:
start_logits, end_logits = output
result = SquadResult(unique_id, start_logits, end_logits)
all_results.append(result)
evalTime = timeit.default_timer() - start_time
logger.info(" Evaluation done in total %f secs (%f sec per example)", evalTime, evalTime / len(dataset))
# Compute predictions
output_prediction_file = os.path.join(args.output_dir, "predictions_{}.json".format(prefix))
output_nbest_file = os.path.join(args.output_dir, "nbest_predictions_{}.json".format(prefix))
if args.version_2_with_negative:
output_null_log_odds_file = os.path.join(args.output_dir, "null_odds_{}.json".format(prefix))
else:
output_null_log_odds_file = None
if args.model_type in ["xlnet", "xlm"]:
# XLNet uses a more complex post-processing procedure
predictions = compute_predictions_log_probs(
examples,
features,
all_results,
args.n_best_size,
args.max_answer_length,
output_prediction_file,
output_nbest_file,
output_null_log_odds_file,
model.config.start_n_top,
model.config.end_n_top,
args.version_2_with_negative,
tokenizer,
args.verbose_logging,
)
else:
predictions = compute_predictions_logits(
examples,
features,
all_results,
args.n_best_size,
args.max_answer_length,
args.do_lower_case,
output_prediction_file,
output_nbest_file,
output_null_log_odds_file,
args.verbose_logging,
args.version_2_with_negative,
args.null_score_diff_threshold,
tokenizer,
)
# Compute the F1 and exact scores.
results = squad_evaluate(examples, predictions)
return results
def load_and_cache_examples(args, tokenizer, evaluate=False, output_examples=False):
if args.local_rank not in [-1, 0] and not evaluate:
# Make sure only the first process in distributed training process the dataset, and the others will use the cache
torch.distributed.barrier()
# Load data features from cache or dataset file
input_file = args.predict_file if evaluate else args.train_file
cached_features_file = os.path.join(
os.path.dirname(input_file),
"cached_distillation_{}_{}_{}".format(
"dev" if evaluate else "train",
list(filter(None, args.model_name_or_path.split("/"))).pop(),
str(args.max_seq_length),
),
)
if os.path.exists(cached_features_file) and not args.overwrite_cache:
logger.info("Loading features from cached file %s", cached_features_file)
features_and_dataset = torch.load(cached_features_file)
try:
features, dataset, examples = (
features_and_dataset["features"],
features_and_dataset["dataset"],
features_and_dataset["examples"],
)
except KeyError:
raise DeprecationWarning(
"You seem to be loading features from an older version of this script please delete the "
"file %s in order for it to be created again" % cached_features_file
)
else:
logger.info("Creating features from dataset file at %s", input_file)
processor = SquadV2Processor() if args.version_2_with_negative else SquadV1Processor()
if evaluate:
examples = processor.get_dev_examples(args.data_dir, filename=args.predict_file)
else:
examples = processor.get_train_examples(args.data_dir, filename=args.train_file)
features, dataset = squad_convert_examples_to_features(
examples=examples,
tokenizer=tokenizer,
max_seq_length=args.max_seq_length,
doc_stride=args.doc_stride,
max_query_length=args.max_query_length,
is_training=not evaluate,
return_dataset="pt",
threads=args.threads,
)
if args.local_rank in [-1, 0]:
logger.info("Saving features into cached file %s", cached_features_file)
torch.save({"features": features, "dataset": dataset, "examples": examples}, cached_features_file)
if args.local_rank == 0 and not evaluate:
# Make sure only the first process in distributed training process the dataset, and the others will use the cache
torch.distributed.barrier()
if output_examples:
return dataset, examples, features
return dataset
def main():
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--model_type",
default=None,
type=str,
required=True,
help="Model type selected in the list: " + ", ".join(MODEL_CLASSES.keys()),
)
parser.add_argument(
"--model_name_or_path",
default=None,
type=str,
required=True,
help="Path to pretrained model or model identifier from huggingface.co/models",
)
parser.add_argument(
"--output_dir",
default=None,
type=str,
required=True,
help="The output directory where the model checkpoints and predictions will be written.",
)
# Distillation parameters (optional)
parser.add_argument(
"--teacher_type",
default=None,
type=str,
help=(
"Teacher type. Teacher tokenizer and student (model) tokenizer must output the same tokenization. Only for"
" distillation."
),
)
parser.add_argument(
"--teacher_name_or_path",
default=None,
type=str,
help="Path to the already SQuAD fine-tuned teacher model. Only for distillation.",
)
parser.add_argument(
"--alpha_ce", default=0.5, type=float, help="Distillation loss linear weight. Only for distillation."
)
parser.add_argument(
"--alpha_squad", default=0.5, type=float, help="True SQuAD loss linear weight. Only for distillation."
)
parser.add_argument(
"--temperature", default=2.0, type=float, help="Distillation temperature. Only for distillation."
)
# Other parameters
parser.add_argument(
"--data_dir",
default=None,
type=str,
help="The input data dir. Should contain the .json files for the task."
+ "If no data dir or train/predict files are specified, will run with tensorflow_datasets.",
)
parser.add_argument(
"--train_file",
default=None,
type=str,
help="The input training file. If a data dir is specified, will look for the file there"
+ "If no data dir or train/predict files are specified, will run with tensorflow_datasets.",
)
parser.add_argument(
"--predict_file",
default=None,
type=str,
help="The input evaluation file. If a data dir is specified, will look for the file there"
+ "If no data dir or train/predict files are specified, will run with tensorflow_datasets.",
)
parser.add_argument(
"--config_name", default="", type=str, help="Pretrained config name or path if not the same as model_name"
)
parser.add_argument(
"--tokenizer_name",
default="",
type=str,
help="Pretrained tokenizer name or path if not the same as model_name",
)
parser.add_argument(
"--cache_dir",
default="",
type=str,
help="Where do you want to store the pre-trained models downloaded from huggingface.co",
)
parser.add_argument(
"--version_2_with_negative",
action="store_true",
help="If true, the SQuAD examples contain some that do not have an answer.",
)
parser.add_argument(
"--null_score_diff_threshold",
type=float,
default=0.0,
help="If null_score - best_non_null is greater than the threshold predict null.",
)
parser.add_argument(
"--max_seq_length",
default=384,
type=int,
help=(
"The maximum total input sequence length after WordPiece tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded."
),
)
parser.add_argument(
"--doc_stride",
default=128,
type=int,
help="When splitting up a long document into chunks, how much stride to take between chunks.",
)
parser.add_argument(
"--max_query_length",
default=64,
type=int,
help=(
"The maximum number of tokens for the question. Questions longer than this will "
"be truncated to this length."
),
)
parser.add_argument("--do_train", action="store_true", help="Whether to run training.")
parser.add_argument("--do_eval", action="store_true", help="Whether to run eval on the dev set.")
parser.add_argument(
"--evaluate_during_training", action="store_true", help="Rul evaluation during training at each logging step."
)
parser.add_argument(
"--do_lower_case", action="store_true", help="Set this flag if you are using an uncased model."
)
parser.add_argument("--per_gpu_train_batch_size", default=8, type=int, help="Batch size per GPU/CPU for training.")
parser.add_argument(
"--per_gpu_eval_batch_size", default=8, type=int, help="Batch size per GPU/CPU for evaluation."
)
parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.")
parser.add_argument(
"--gradient_accumulation_steps",
type=int,
default=1,
help="Number of updates steps to accumulate before performing a backward/update pass.",
)
parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight decay if we apply some.")
parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.")
parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
parser.add_argument(
"--num_train_epochs", default=3.0, type=float, help="Total number of training epochs to perform."
)
parser.add_argument(
"--max_steps",
default=-1,
type=int,
help="If > 0: set total number of training steps to perform. Override num_train_epochs.",
)
parser.add_argument("--warmup_steps", default=0, type=int, help="Linear warmup over warmup_steps.")
parser.add_argument(
"--n_best_size",
default=20,
type=int,
help="The total number of n-best predictions to generate in the nbest_predictions.json output file.",
)
parser.add_argument(
"--max_answer_length",
default=30,
type=int,
help=(
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
),
)
parser.add_argument(
"--verbose_logging",
action="store_true",
help=(
"If true, all of the warnings related to data processing will be printed. "
"A number of warnings are expected for a normal SQuAD evaluation."
),
)
parser.add_argument("--logging_steps", type=int, default=50, help="Log every X updates steps.")
parser.add_argument("--save_steps", type=int, default=50, help="Save checkpoint every X updates steps.")
parser.add_argument(
"--eval_all_checkpoints",
action="store_true",
help="Evaluate all checkpoints starting with the same prefix as model_name ending and ending with step number",
)
parser.add_argument("--no_cuda", action="store_true", help="Whether not to use CUDA when available")
parser.add_argument(
"--overwrite_output_dir", action="store_true", help="Overwrite the content of the output directory"
)
parser.add_argument(
"--overwrite_cache", action="store_true", help="Overwrite the cached training and evaluation sets"
)
parser.add_argument("--seed", type=int, default=42, help="random seed for initialization")
parser.add_argument("--local_rank", type=int, default=-1, help="local_rank for distributed training on gpus")
parser.add_argument(
"--fp16",
action="store_true",
help="Whether to use 16-bit (mixed) precision (through NVIDIA apex) instead of 32-bit",
)
parser.add_argument(
"--fp16_opt_level",
type=str,
default="O1",
help=(
"For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html"
),
)
parser.add_argument("--server_ip", type=str, default="", help="Can be used for distant debugging.")
parser.add_argument("--server_port", type=str, default="", help="Can be used for distant debugging.")
parser.add_argument("--threads", type=int, default=1, help="multiple threads for converting example to features")
args = parser.parse_args()
if (
os.path.exists(args.output_dir)
and os.listdir(args.output_dir)
and args.do_train
and not args.overwrite_output_dir
):
raise ValueError(
"Output directory ({}) already exists and is not empty. Use --overwrite_output_dir to overcome.".format(
args.output_dir
)
)
# Setup distant debugging if needed
if args.server_ip and args.server_port:
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
import ptvsd
print("Waiting for debugger attach")
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
ptvsd.wait_for_attach()
# Setup CUDA, GPU & distributed training
if args.local_rank == -1 or args.no_cuda:
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
else: # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.cuda.set_device(args.local_rank)
device = torch.device("cuda", args.local_rank)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
level=logging.INFO if args.local_rank in [-1, 0] else logging.WARN,
)
logger.warning(
"Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
args.local_rank,
device,
args.n_gpu,
bool(args.local_rank != -1),
args.fp16,
)
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(args.local_rank):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Set seed
set_seed(args)
# Load pretrained model and tokenizer
if args.local_rank not in [-1, 0]:
# Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier()
args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(
args.config_name if args.config_name else args.model_name_or_path,
cache_dir=args.cache_dir if args.cache_dir else None,
)
tokenizer = tokenizer_class.from_pretrained(
args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None,
)
model = model_class.from_pretrained(
args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None,
)
if args.teacher_type is not None:
assert args.teacher_name_or_path is not None
assert args.alpha_ce > 0.0
assert args.alpha_ce + args.alpha_squad > 0.0
assert args.teacher_type != "distilbert", "We constraint teachers not to be of type DistilBERT."
teacher_config_class, teacher_model_class, _ = MODEL_CLASSES[args.teacher_type]
teacher_config = teacher_config_class.from_pretrained(
args.teacher_name_or_path, cache_dir=args.cache_dir if args.cache_dir else None
)
teacher = teacher_model_class.from_pretrained(
args.teacher_name_or_path, config=teacher_config, cache_dir=args.cache_dir if args.cache_dir else None
)
teacher.to(args.device)
else:
teacher = None
if args.local_rank == 0:
# Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier()
model.to(args.device)
logger.info("Training/evaluation parameters %s", args)
# Before we do anything with models, we want to ensure that we get fp16 execution of torch.einsum if args.fp16 is set.
# Otherwise it'll default to "promote" mode, and we'll get fp32 operations. Note that running `--fp16_opt_level="O2"` will
# remove the need for this code, but it is still valid.
if args.fp16:
try:
import apex
apex.amp.register_half_function(torch, "einsum")
except ImportError:
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use fp16 training.")
# Training
if args.do_train:
train_dataset = load_and_cache_examples(args, tokenizer, evaluate=False, output_examples=False)
global_step, tr_loss = train(args, train_dataset, model, tokenizer, teacher=teacher)
logger.info(" global_step = %s, average loss = %s", global_step, tr_loss)
# Save the trained model and the tokenizer
if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
logger.info("Saving model checkpoint to %s", args.output_dir)
# Save a trained model, configuration and tokenizer using `save_pretrained()`.
# They can then be reloaded using `from_pretrained()`
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(args.output_dir)
tokenizer.save_pretrained(args.output_dir)
# Good practice: save your training arguments together with the trained model
torch.save(args, os.path.join(args.output_dir, "training_args.bin"))
# Load a trained model and vocabulary that you have fine-tuned
model = model_class.from_pretrained(args.output_dir)
tokenizer = tokenizer_class.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
model.to(args.device)
# Evaluation - we can ask to evaluate all the checkpoints (sub-directories) in a directory
results = {}
if args.do_eval and args.local_rank in [-1, 0]:
if args.do_train:
logger.info("Loading checkpoints saved during training for evaluation")
checkpoints = [args.output_dir]
if args.eval_all_checkpoints:
checkpoints = [
os.path.dirname(c) for c in sorted(glob.glob(args.output_dir + "/**/" + WEIGHTS_NAME, recursive=True))
]
logger.info("Evaluate the following checkpoints: %s", checkpoints)
for checkpoint in checkpoints:
# Reload the model
global_step = checkpoint.split("-")[-1] if len(checkpoints) > 1 else ""
model = model_class.from_pretrained(checkpoint)
model.to(args.device)
# Evaluate
result = evaluate(args, model, tokenizer, prefix=global_step)
result = {k + ("_{}".format(global_step) if global_step else ""): v for k, v in result.items()}
results.update(result)
logger.info("Results: {}".format(results))
return results
if __name__ == "__main__":
main()
| # coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" This is the exact same script as `examples/question-answering/run_squad.py` (as of 2020, January 8th) with an additional and optional step of distillation."""
import argparse
import glob
import logging
import os
import random
import timeit
import numpy as np
import torch
from torch import nn
from torch.utils.data import DataLoader, RandomSampler, SequentialSampler
from torch.utils.data.distributed import DistributedSampler
from tqdm import tqdm, trange
import transformers
from transformers import (
WEIGHTS_NAME,
AdamW,
BertConfig,
BertForQuestionAnswering,
BertTokenizer,
DistilBertConfig,
DistilBertForQuestionAnswering,
DistilBertTokenizer,
RobertaConfig,
RobertaForQuestionAnswering,
RobertaTokenizer,
XLMConfig,
XLMForQuestionAnswering,
XLMTokenizer,
XLNetConfig,
XLNetForQuestionAnswering,
XLNetTokenizer,
get_linear_schedule_with_warmup,
squad_convert_examples_to_features,
)
from transformers.data.metrics.squad_metrics import (
compute_predictions_log_probs,
compute_predictions_logits,
squad_evaluate,
)
from transformers.data.processors.squad import SquadResult, SquadV1Processor, SquadV2Processor
from transformers.trainer_utils import is_main_process
try:
from torch.utils.tensorboard import SummaryWriter
except ImportError:
from tensorboardX import SummaryWriter
logger = logging.getLogger(__name__)
MODEL_CLASSES = {
"bert": (BertConfig, BertForQuestionAnswering, BertTokenizer),
"xlnet": (XLNetConfig, XLNetForQuestionAnswering, XLNetTokenizer),
"xlm": (XLMConfig, XLMForQuestionAnswering, XLMTokenizer),
"distilbert": (DistilBertConfig, DistilBertForQuestionAnswering, DistilBertTokenizer),
"roberta": (RobertaConfig, RobertaForQuestionAnswering, RobertaTokenizer),
}
def set_seed(args):
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.n_gpu > 0:
torch.cuda.manual_seed_all(args.seed)
def to_list(tensor):
return tensor.detach().cpu().tolist()
def train(args, train_dataset, model, tokenizer, teacher=None):
"""Train the model"""
if args.local_rank in [-1, 0]:
tb_writer = SummaryWriter()
args.train_batch_size = args.per_gpu_train_batch_size * max(1, args.n_gpu)
train_sampler = RandomSampler(train_dataset) if args.local_rank == -1 else DistributedSampler(train_dataset)
train_dataloader = DataLoader(train_dataset, sampler=train_sampler, batch_size=args.train_batch_size)
if args.max_steps > 0:
t_total = args.max_steps
args.num_train_epochs = args.max_steps // (len(train_dataloader) // args.gradient_accumulation_steps) + 1
else:
t_total = len(train_dataloader) // args.gradient_accumulation_steps * args.num_train_epochs
# Prepare optimizer and schedule (linear warmup and decay)
no_decay = ["bias", "LayerNorm.weight"]
optimizer_grouped_parameters = [
{
"params": [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)],
"weight_decay": args.weight_decay,
},
{"params": [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)], "weight_decay": 0.0},
]
optimizer = AdamW(optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon)
scheduler = get_linear_schedule_with_warmup(
optimizer, num_warmup_steps=args.warmup_steps, num_training_steps=t_total
)
# Check if saved optimizer or scheduler states exist
if os.path.isfile(os.path.join(args.model_name_or_path, "optimizer.pt")) and os.path.isfile(
os.path.join(args.model_name_or_path, "scheduler.pt")
):
# Load in optimizer and scheduler states
optimizer.load_state_dict(torch.load(os.path.join(args.model_name_or_path, "optimizer.pt")))
scheduler.load_state_dict(torch.load(os.path.join(args.model_name_or_path, "scheduler.pt")))
if args.fp16:
try:
from apex import amp
except ImportError:
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use fp16 training.")
model, optimizer = amp.initialize(model, optimizer, opt_level=args.fp16_opt_level)
# multi-gpu training (should be after apex fp16 initialization)
if args.n_gpu > 1:
model = nn.DataParallel(model)
# Distributed training (should be after apex fp16 initialization)
if args.local_rank != -1:
model = nn.parallel.DistributedDataParallel(
model, device_ids=[args.local_rank], output_device=args.local_rank, find_unused_parameters=True
)
# Train!
logger.info("***** Running training *****")
logger.info(" Num examples = %d", len(train_dataset))
logger.info(" Num Epochs = %d", args.num_train_epochs)
logger.info(" Instantaneous batch size per GPU = %d", args.per_gpu_train_batch_size)
logger.info(
" Total train batch size (w. parallel, distributed & accumulation) = %d",
args.train_batch_size
* args.gradient_accumulation_steps
* (torch.distributed.get_world_size() if args.local_rank != -1 else 1),
)
logger.info(" Gradient Accumulation steps = %d", args.gradient_accumulation_steps)
logger.info(" Total optimization steps = %d", t_total)
global_step = 1
epochs_trained = 0
steps_trained_in_current_epoch = 0
# Check if continuing training from a checkpoint
if os.path.exists(args.model_name_or_path):
try:
# set global_step to gobal_step of last saved checkpoint from model path
checkpoint_suffix = args.model_name_or_path.split("-")[-1].split("/")[0]
global_step = int(checkpoint_suffix)
epochs_trained = global_step // (len(train_dataloader) // args.gradient_accumulation_steps)
steps_trained_in_current_epoch = global_step % (len(train_dataloader) // args.gradient_accumulation_steps)
logger.info(" Continuing training from checkpoint, will skip to saved global_step")
logger.info(" Continuing training from epoch %d", epochs_trained)
logger.info(" Continuing training from global step %d", global_step)
logger.info(" Will skip the first %d steps in the first epoch", steps_trained_in_current_epoch)
except ValueError:
logger.info(" Starting fine-tuning.")
tr_loss, logging_loss = 0.0, 0.0
model.zero_grad()
train_iterator = trange(
epochs_trained, int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0]
)
# Added here for reproductibility
set_seed(args)
for _ in train_iterator:
epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
for step, batch in enumerate(epoch_iterator):
# Skip past any already trained steps if resuming training
if steps_trained_in_current_epoch > 0:
steps_trained_in_current_epoch -= 1
continue
model.train()
if teacher is not None:
teacher.eval()
batch = tuple(t.to(args.device) for t in batch)
inputs = {
"input_ids": batch[0],
"attention_mask": batch[1],
"start_positions": batch[3],
"end_positions": batch[4],
}
if args.model_type != "distilbert":
inputs["token_type_ids"] = None if args.model_type == "xlm" else batch[2]
if args.model_type in ["xlnet", "xlm"]:
inputs.update({"cls_index": batch[5], "p_mask": batch[6]})
if args.version_2_with_negative:
inputs.update({"is_impossible": batch[7]})
outputs = model(**inputs)
loss, start_logits_stu, end_logits_stu = outputs
# Distillation loss
if teacher is not None:
if "token_type_ids" not in inputs:
inputs["token_type_ids"] = None if args.teacher_type == "xlm" else batch[2]
with torch.no_grad():
start_logits_tea, end_logits_tea = teacher(
input_ids=inputs["input_ids"],
token_type_ids=inputs["token_type_ids"],
attention_mask=inputs["attention_mask"],
)
assert start_logits_tea.size() == start_logits_stu.size()
assert end_logits_tea.size() == end_logits_stu.size()
loss_fct = nn.KLDivLoss(reduction="batchmean")
loss_start = loss_fct(
nn.functional.log_softmax(start_logits_stu / args.temperature, dim=-1),
nn.functional.softmax(start_logits_tea / args.temperature, dim=-1),
) * (args.temperature**2)
loss_end = loss_fct(
nn.functional.log_softmax(end_logits_stu / args.temperature, dim=-1),
nn.functional.softmax(end_logits_tea / args.temperature, dim=-1),
) * (args.temperature**2)
loss_ce = (loss_start + loss_end) / 2.0
loss = args.alpha_ce * loss_ce + args.alpha_squad * loss
if args.n_gpu > 1:
loss = loss.mean() # mean() to average on multi-gpu parallel (not distributed) training
if args.gradient_accumulation_steps > 1:
loss = loss / args.gradient_accumulation_steps
if args.fp16:
with amp.scale_loss(loss, optimizer) as scaled_loss:
scaled_loss.backward()
else:
loss.backward()
tr_loss += loss.item()
if (step + 1) % args.gradient_accumulation_steps == 0:
if args.fp16:
nn.utils.clip_grad_norm_(amp.master_params(optimizer), args.max_grad_norm)
else:
nn.utils.clip_grad_norm_(model.parameters(), args.max_grad_norm)
optimizer.step()
scheduler.step() # Update learning rate schedule
model.zero_grad()
global_step += 1
# Log metrics
if args.local_rank in [-1, 0] and args.logging_steps > 0 and global_step % args.logging_steps == 0:
# Only evaluate when single GPU otherwise metrics may not average well
if args.local_rank == -1 and args.evaluate_during_training:
results = evaluate(args, model, tokenizer)
for key, value in results.items():
tb_writer.add_scalar("eval_{}".format(key), value, global_step)
tb_writer.add_scalar("lr", scheduler.get_lr()[0], global_step)
tb_writer.add_scalar("loss", (tr_loss - logging_loss) / args.logging_steps, global_step)
logging_loss = tr_loss
if args.local_rank in [-1, 0] and args.save_steps > 0 and global_step % args.save_steps == 0:
# Save model checkpoint
output_dir = os.path.join(args.output_dir, "checkpoint-{}".format(global_step))
if not os.path.exists(output_dir):
os.makedirs(output_dir)
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(output_dir)
tokenizer.save_pretrained(output_dir)
torch.save(args, os.path.join(output_dir, "training_args.bin"))
logger.info("Saving model checkpoint to %s", output_dir)
torch.save(optimizer.state_dict(), os.path.join(output_dir, "optimizer.pt"))
torch.save(scheduler.state_dict(), os.path.join(output_dir, "scheduler.pt"))
logger.info("Saving optimizer and scheduler states to %s", output_dir)
if args.max_steps > 0 and global_step > args.max_steps:
epoch_iterator.close()
break
if args.max_steps > 0 and global_step > args.max_steps:
train_iterator.close()
break
if args.local_rank in [-1, 0]:
tb_writer.close()
return global_step, tr_loss / global_step
def evaluate(args, model, tokenizer, prefix=""):
dataset, examples, features = load_and_cache_examples(args, tokenizer, evaluate=True, output_examples=True)
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(dataset)
eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu evaluate
if args.n_gpu > 1 and not isinstance(model, nn.DataParallel):
model = nn.DataParallel(model)
# Eval!
logger.info("***** Running evaluation {} *****".format(prefix))
logger.info(" Num examples = %d", len(dataset))
logger.info(" Batch size = %d", args.eval_batch_size)
all_results = []
start_time = timeit.default_timer()
for batch in tqdm(eval_dataloader, desc="Evaluating"):
model.eval()
batch = tuple(t.to(args.device) for t in batch)
with torch.no_grad():
inputs = {"input_ids": batch[0], "attention_mask": batch[1]}
if args.model_type != "distilbert":
inputs["token_type_ids"] = None if args.model_type == "xlm" else batch[2] # XLM don't use segment_ids
example_indices = batch[3]
if args.model_type in ["xlnet", "xlm"]:
inputs.update({"cls_index": batch[4], "p_mask": batch[5]})
outputs = model(**inputs)
for i, example_index in enumerate(example_indices):
eval_feature = features[example_index.item()]
unique_id = int(eval_feature.unique_id)
output = [to_list(output[i]) for output in outputs]
# Some models (XLNet, XLM) use 5 arguments for their predictions, while the other "simpler"
# models only use two.
if len(output) >= 5:
start_logits = output[0]
start_top_index = output[1]
end_logits = output[2]
end_top_index = output[3]
cls_logits = output[4]
result = SquadResult(
unique_id,
start_logits,
end_logits,
start_top_index=start_top_index,
end_top_index=end_top_index,
cls_logits=cls_logits,
)
else:
start_logits, end_logits = output
result = SquadResult(unique_id, start_logits, end_logits)
all_results.append(result)
evalTime = timeit.default_timer() - start_time
logger.info(" Evaluation done in total %f secs (%f sec per example)", evalTime, evalTime / len(dataset))
# Compute predictions
output_prediction_file = os.path.join(args.output_dir, "predictions_{}.json".format(prefix))
output_nbest_file = os.path.join(args.output_dir, "nbest_predictions_{}.json".format(prefix))
if args.version_2_with_negative:
output_null_log_odds_file = os.path.join(args.output_dir, "null_odds_{}.json".format(prefix))
else:
output_null_log_odds_file = None
if args.model_type in ["xlnet", "xlm"]:
# XLNet uses a more complex post-processing procedure
predictions = compute_predictions_log_probs(
examples,
features,
all_results,
args.n_best_size,
args.max_answer_length,
output_prediction_file,
output_nbest_file,
output_null_log_odds_file,
model.config.start_n_top,
model.config.end_n_top,
args.version_2_with_negative,
tokenizer,
args.verbose_logging,
)
else:
predictions = compute_predictions_logits(
examples,
features,
all_results,
args.n_best_size,
args.max_answer_length,
args.do_lower_case,
output_prediction_file,
output_nbest_file,
output_null_log_odds_file,
args.verbose_logging,
args.version_2_with_negative,
args.null_score_diff_threshold,
tokenizer,
)
# Compute the F1 and exact scores.
results = squad_evaluate(examples, predictions)
return results
def load_and_cache_examples(args, tokenizer, evaluate=False, output_examples=False):
if args.local_rank not in [-1, 0] and not evaluate:
# Make sure only the first process in distributed training process the dataset, and the others will use the cache
torch.distributed.barrier()
# Load data features from cache or dataset file
input_file = args.predict_file if evaluate else args.train_file
cached_features_file = os.path.join(
os.path.dirname(input_file),
"cached_distillation_{}_{}_{}".format(
"dev" if evaluate else "train",
list(filter(None, args.model_name_or_path.split("/"))).pop(),
str(args.max_seq_length),
),
)
if os.path.exists(cached_features_file) and not args.overwrite_cache:
logger.info("Loading features from cached file %s", cached_features_file)
features_and_dataset = torch.load(cached_features_file)
try:
features, dataset, examples = (
features_and_dataset["features"],
features_and_dataset["dataset"],
features_and_dataset["examples"],
)
except KeyError:
raise DeprecationWarning(
"You seem to be loading features from an older version of this script please delete the "
"file %s in order for it to be created again" % cached_features_file
)
else:
logger.info("Creating features from dataset file at %s", input_file)
processor = SquadV2Processor() if args.version_2_with_negative else SquadV1Processor()
if evaluate:
examples = processor.get_dev_examples(args.data_dir, filename=args.predict_file)
else:
examples = processor.get_train_examples(args.data_dir, filename=args.train_file)
features, dataset = squad_convert_examples_to_features(
examples=examples,
tokenizer=tokenizer,
max_seq_length=args.max_seq_length,
doc_stride=args.doc_stride,
max_query_length=args.max_query_length,
is_training=not evaluate,
return_dataset="pt",
threads=args.threads,
)
if args.local_rank in [-1, 0]:
logger.info("Saving features into cached file %s", cached_features_file)
torch.save({"features": features, "dataset": dataset, "examples": examples}, cached_features_file)
if args.local_rank == 0 and not evaluate:
# Make sure only the first process in distributed training process the dataset, and the others will use the cache
torch.distributed.barrier()
if output_examples:
return dataset, examples, features
return dataset
def main():
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--model_type",
default=None,
type=str,
required=True,
help="Model type selected in the list: " + ", ".join(MODEL_CLASSES.keys()),
)
parser.add_argument(
"--model_name_or_path",
default=None,
type=str,
required=True,
help="Path to pretrained model or model identifier from huggingface.co/models",
)
parser.add_argument(
"--output_dir",
default=None,
type=str,
required=True,
help="The output directory where the model checkpoints and predictions will be written.",
)
# Distillation parameters (optional)
parser.add_argument(
"--teacher_type",
default=None,
type=str,
help=(
"Teacher type. Teacher tokenizer and student (model) tokenizer must output the same tokenization. Only for"
" distillation."
),
)
parser.add_argument(
"--teacher_name_or_path",
default=None,
type=str,
help="Path to the already SQuAD fine-tuned teacher model. Only for distillation.",
)
parser.add_argument(
"--alpha_ce", default=0.5, type=float, help="Distillation loss linear weight. Only for distillation."
)
parser.add_argument(
"--alpha_squad", default=0.5, type=float, help="True SQuAD loss linear weight. Only for distillation."
)
parser.add_argument(
"--temperature", default=2.0, type=float, help="Distillation temperature. Only for distillation."
)
# Other parameters
parser.add_argument(
"--data_dir",
default=None,
type=str,
help="The input data dir. Should contain the .json files for the task."
+ "If no data dir or train/predict files are specified, will run with tensorflow_datasets.",
)
parser.add_argument(
"--train_file",
default=None,
type=str,
help="The input training file. If a data dir is specified, will look for the file there"
+ "If no data dir or train/predict files are specified, will run with tensorflow_datasets.",
)
parser.add_argument(
"--predict_file",
default=None,
type=str,
help="The input evaluation file. If a data dir is specified, will look for the file there"
+ "If no data dir or train/predict files are specified, will run with tensorflow_datasets.",
)
parser.add_argument(
"--config_name", default="", type=str, help="Pretrained config name or path if not the same as model_name"
)
parser.add_argument(
"--tokenizer_name",
default="",
type=str,
help="Pretrained tokenizer name or path if not the same as model_name",
)
parser.add_argument(
"--cache_dir",
default="",
type=str,
help="Where do you want to store the pre-trained models downloaded from huggingface.co",
)
parser.add_argument(
"--version_2_with_negative",
action="store_true",
help="If true, the SQuAD examples contain some that do not have an answer.",
)
parser.add_argument(
"--null_score_diff_threshold",
type=float,
default=0.0,
help="If null_score - best_non_null is greater than the threshold predict null.",
)
parser.add_argument(
"--max_seq_length",
default=384,
type=int,
help=(
"The maximum total input sequence length after WordPiece tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded."
),
)
parser.add_argument(
"--doc_stride",
default=128,
type=int,
help="When splitting up a long document into chunks, how much stride to take between chunks.",
)
parser.add_argument(
"--max_query_length",
default=64,
type=int,
help=(
"The maximum number of tokens for the question. Questions longer than this will "
"be truncated to this length."
),
)
parser.add_argument("--do_train", action="store_true", help="Whether to run training.")
parser.add_argument("--do_eval", action="store_true", help="Whether to run eval on the dev set.")
parser.add_argument(
"--evaluate_during_training", action="store_true", help="Rul evaluation during training at each logging step."
)
parser.add_argument(
"--do_lower_case", action="store_true", help="Set this flag if you are using an uncased model."
)
parser.add_argument("--per_gpu_train_batch_size", default=8, type=int, help="Batch size per GPU/CPU for training.")
parser.add_argument(
"--per_gpu_eval_batch_size", default=8, type=int, help="Batch size per GPU/CPU for evaluation."
)
parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.")
parser.add_argument(
"--gradient_accumulation_steps",
type=int,
default=1,
help="Number of updates steps to accumulate before performing a backward/update pass.",
)
parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight decay if we apply some.")
parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.")
parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
parser.add_argument(
"--num_train_epochs", default=3.0, type=float, help="Total number of training epochs to perform."
)
parser.add_argument(
"--max_steps",
default=-1,
type=int,
help="If > 0: set total number of training steps to perform. Override num_train_epochs.",
)
parser.add_argument("--warmup_steps", default=0, type=int, help="Linear warmup over warmup_steps.")
parser.add_argument(
"--n_best_size",
default=20,
type=int,
help="The total number of n-best predictions to generate in the nbest_predictions.json output file.",
)
parser.add_argument(
"--max_answer_length",
default=30,
type=int,
help=(
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
),
)
parser.add_argument(
"--verbose_logging",
action="store_true",
help=(
"If true, all of the warnings related to data processing will be printed. "
"A number of warnings are expected for a normal SQuAD evaluation."
),
)
parser.add_argument("--logging_steps", type=int, default=50, help="Log every X updates steps.")
parser.add_argument("--save_steps", type=int, default=50, help="Save checkpoint every X updates steps.")
parser.add_argument(
"--eval_all_checkpoints",
action="store_true",
help="Evaluate all checkpoints starting with the same prefix as model_name ending and ending with step number",
)
parser.add_argument("--no_cuda", action="store_true", help="Whether not to use CUDA when available")
parser.add_argument(
"--overwrite_output_dir", action="store_true", help="Overwrite the content of the output directory"
)
parser.add_argument(
"--overwrite_cache", action="store_true", help="Overwrite the cached training and evaluation sets"
)
parser.add_argument("--seed", type=int, default=42, help="random seed for initialization")
parser.add_argument("--local_rank", type=int, default=-1, help="local_rank for distributed training on gpus")
parser.add_argument(
"--fp16",
action="store_true",
help="Whether to use 16-bit (mixed) precision (through NVIDIA apex) instead of 32-bit",
)
parser.add_argument(
"--fp16_opt_level",
type=str,
default="O1",
help=(
"For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html"
),
)
parser.add_argument("--server_ip", type=str, default="", help="Can be used for distant debugging.")
parser.add_argument("--server_port", type=str, default="", help="Can be used for distant debugging.")
parser.add_argument("--threads", type=int, default=1, help="multiple threads for converting example to features")
args = parser.parse_args()
if (
os.path.exists(args.output_dir)
and os.listdir(args.output_dir)
and args.do_train
and not args.overwrite_output_dir
):
raise ValueError(
"Output directory ({}) already exists and is not empty. Use --overwrite_output_dir to overcome.".format(
args.output_dir
)
)
# Setup distant debugging if needed
if args.server_ip and args.server_port:
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
import ptvsd
print("Waiting for debugger attach")
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
ptvsd.wait_for_attach()
# Setup CUDA, GPU & distributed training
if args.local_rank == -1 or args.no_cuda:
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
else: # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.cuda.set_device(args.local_rank)
device = torch.device("cuda", args.local_rank)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
level=logging.INFO if args.local_rank in [-1, 0] else logging.WARN,
)
logger.warning(
"Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
args.local_rank,
device,
args.n_gpu,
bool(args.local_rank != -1),
args.fp16,
)
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(args.local_rank):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Set seed
set_seed(args)
# Load pretrained model and tokenizer
if args.local_rank not in [-1, 0]:
# Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier()
args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(
args.config_name if args.config_name else args.model_name_or_path,
cache_dir=args.cache_dir if args.cache_dir else None,
)
tokenizer = tokenizer_class.from_pretrained(
args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None,
)
model = model_class.from_pretrained(
args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None,
)
if args.teacher_type is not None:
assert args.teacher_name_or_path is not None
assert args.alpha_ce > 0.0
assert args.alpha_ce + args.alpha_squad > 0.0
assert args.teacher_type != "distilbert", "We constraint teachers not to be of type DistilBERT."
teacher_config_class, teacher_model_class, _ = MODEL_CLASSES[args.teacher_type]
teacher_config = teacher_config_class.from_pretrained(
args.teacher_name_or_path, cache_dir=args.cache_dir if args.cache_dir else None
)
teacher = teacher_model_class.from_pretrained(
args.teacher_name_or_path, config=teacher_config, cache_dir=args.cache_dir if args.cache_dir else None
)
teacher.to(args.device)
else:
teacher = None
if args.local_rank == 0:
# Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier()
model.to(args.device)
logger.info("Training/evaluation parameters %s", args)
# Before we do anything with models, we want to ensure that we get fp16 execution of torch.einsum if args.fp16 is set.
# Otherwise it'll default to "promote" mode, and we'll get fp32 operations. Note that running `--fp16_opt_level="O2"` will
# remove the need for this code, but it is still valid.
if args.fp16:
try:
import apex
apex.amp.register_half_function(torch, "einsum")
except ImportError:
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use fp16 training.")
# Training
if args.do_train:
train_dataset = load_and_cache_examples(args, tokenizer, evaluate=False, output_examples=False)
global_step, tr_loss = train(args, train_dataset, model, tokenizer, teacher=teacher)
logger.info(" global_step = %s, average loss = %s", global_step, tr_loss)
# Save the trained model and the tokenizer
if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
logger.info("Saving model checkpoint to %s", args.output_dir)
# Save a trained model, configuration and tokenizer using `save_pretrained()`.
# They can then be reloaded using `from_pretrained()`
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(args.output_dir)
tokenizer.save_pretrained(args.output_dir)
# Good practice: save your training arguments together with the trained model
torch.save(args, os.path.join(args.output_dir, "training_args.bin"))
# Load a trained model and vocabulary that you have fine-tuned
model = model_class.from_pretrained(args.output_dir)
tokenizer = tokenizer_class.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
model.to(args.device)
# Evaluation - we can ask to evaluate all the checkpoints (sub-directories) in a directory
results = {}
if args.do_eval and args.local_rank in [-1, 0]:
if args.do_train:
logger.info("Loading checkpoints saved during training for evaluation")
checkpoints = [args.output_dir]
if args.eval_all_checkpoints:
checkpoints = [
os.path.dirname(c) for c in sorted(glob.glob(args.output_dir + "/**/" + WEIGHTS_NAME, recursive=True))
]
logger.info("Evaluate the following checkpoints: %s", checkpoints)
for checkpoint in checkpoints:
# Reload the model
global_step = checkpoint.split("-")[-1] if len(checkpoints) > 1 else ""
model = model_class.from_pretrained(checkpoint)
model.to(args.device)
# Evaluate
result = evaluate(args, model, tokenizer, prefix=global_step)
result = {k + ("_{}".format(global_step) if global_step else ""): v for k, v in result.items()}
results.update(result)
logger.info("Results: {}".format(results))
return results
if __name__ == "__main__":
main()
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./scripts/fsmt/gen-card-allenai-wmt19.py | #!/usr/bin/env python
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# Usage:
# ./gen-card-allenai-wmt19.py
import os
from pathlib import Path
def write_model_card(model_card_dir, src_lang, tgt_lang, model_name):
texts = {
"en": "Machine learning is great, isn't it?",
"ru": "Машинное обучение - это здорово, не так ли?",
"de": "Maschinelles Lernen ist großartig, nicht wahr?",
}
# BLUE scores as follows:
# "pair": [fairseq, transformers]
scores = {
"wmt19-de-en-6-6-base": [0, 38.37],
"wmt19-de-en-6-6-big": [0, 39.90],
}
pair = f"{src_lang}-{tgt_lang}"
readme = f"""
---
language:
- {src_lang}
- {tgt_lang}
thumbnail:
tags:
- translation
- wmt19
- allenai
license: apache-2.0
datasets:
- wmt19
metrics:
- bleu
---
# FSMT
## Model description
This is a ported version of fairseq-based [wmt19 transformer](https://github.com/jungokasai/deep-shallow/) for {src_lang}-{tgt_lang}.
For more details, please, see [Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation](https://arxiv.org/abs/2006.10369).
2 models are available:
* [wmt19-de-en-6-6-big](https://huggingface.co/allenai/wmt19-de-en-6-6-big)
* [wmt19-de-en-6-6-base](https://huggingface.co/allenai/wmt19-de-en-6-6-base)
## Intended uses & limitations
#### How to use
```python
from transformers import FSMTForConditionalGeneration, FSMTTokenizer
mname = "allenai/{model_name}"
tokenizer = FSMTTokenizer.from_pretrained(mname)
model = FSMTForConditionalGeneration.from_pretrained(mname)
input = "{texts[src_lang]}"
input_ids = tokenizer.encode(input, return_tensors="pt")
outputs = model.generate(input_ids)
decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(decoded) # {texts[tgt_lang]}
```
#### Limitations and bias
## Training data
Pretrained weights were left identical to the original model released by allenai. For more details, please, see the [paper](https://arxiv.org/abs/2006.10369).
## Eval results
Here are the BLEU scores:
model | transformers
-------|---------
{model_name} | {scores[model_name][1]}
The score was calculated using this code:
```bash
git clone https://github.com/huggingface/transformers
cd transformers
export PAIR={pair}
export DATA_DIR=data/$PAIR
export SAVE_DIR=data/$PAIR
export BS=8
export NUM_BEAMS=5
mkdir -p $DATA_DIR
sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source
sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target
echo $PAIR
PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py allenai/{model_name} $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS
```
## Data Sources
- [training, etc.](http://www.statmt.org/wmt19/)
- [test set](http://matrix.statmt.org/test_sets/newstest2019.tgz?1556572561)
### BibTeX entry and citation info
```
@misc{{kasai2020deep,
title={{Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation}},
author={{Jungo Kasai and Nikolaos Pappas and Hao Peng and James Cross and Noah A. Smith}},
year={{2020}},
eprint={{2006.10369}},
archivePrefix={{arXiv}},
primaryClass={{cs.CL}}
}}
```
"""
model_card_dir.mkdir(parents=True, exist_ok=True)
path = os.path.join(model_card_dir, "README.md")
print(f"Generating {path}")
with open(path, "w", encoding="utf-8") as f:
f.write(readme)
# make sure we are under the root of the project
repo_dir = Path(__file__).resolve().parent.parent.parent
model_cards_dir = repo_dir / "model_cards"
for model_name in ["wmt19-de-en-6-6-base", "wmt19-de-en-6-6-big"]:
model_card_dir = model_cards_dir / "allenai" / model_name
write_model_card(model_card_dir, src_lang="de", tgt_lang="en", model_name=model_name)
| #!/usr/bin/env python
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# Usage:
# ./gen-card-allenai-wmt19.py
import os
from pathlib import Path
def write_model_card(model_card_dir, src_lang, tgt_lang, model_name):
texts = {
"en": "Machine learning is great, isn't it?",
"ru": "Машинное обучение - это здорово, не так ли?",
"de": "Maschinelles Lernen ist großartig, nicht wahr?",
}
# BLUE scores as follows:
# "pair": [fairseq, transformers]
scores = {
"wmt19-de-en-6-6-base": [0, 38.37],
"wmt19-de-en-6-6-big": [0, 39.90],
}
pair = f"{src_lang}-{tgt_lang}"
readme = f"""
---
language:
- {src_lang}
- {tgt_lang}
thumbnail:
tags:
- translation
- wmt19
- allenai
license: apache-2.0
datasets:
- wmt19
metrics:
- bleu
---
# FSMT
## Model description
This is a ported version of fairseq-based [wmt19 transformer](https://github.com/jungokasai/deep-shallow/) for {src_lang}-{tgt_lang}.
For more details, please, see [Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation](https://arxiv.org/abs/2006.10369).
2 models are available:
* [wmt19-de-en-6-6-big](https://huggingface.co/allenai/wmt19-de-en-6-6-big)
* [wmt19-de-en-6-6-base](https://huggingface.co/allenai/wmt19-de-en-6-6-base)
## Intended uses & limitations
#### How to use
```python
from transformers import FSMTForConditionalGeneration, FSMTTokenizer
mname = "allenai/{model_name}"
tokenizer = FSMTTokenizer.from_pretrained(mname)
model = FSMTForConditionalGeneration.from_pretrained(mname)
input = "{texts[src_lang]}"
input_ids = tokenizer.encode(input, return_tensors="pt")
outputs = model.generate(input_ids)
decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(decoded) # {texts[tgt_lang]}
```
#### Limitations and bias
## Training data
Pretrained weights were left identical to the original model released by allenai. For more details, please, see the [paper](https://arxiv.org/abs/2006.10369).
## Eval results
Here are the BLEU scores:
model | transformers
-------|---------
{model_name} | {scores[model_name][1]}
The score was calculated using this code:
```bash
git clone https://github.com/huggingface/transformers
cd transformers
export PAIR={pair}
export DATA_DIR=data/$PAIR
export SAVE_DIR=data/$PAIR
export BS=8
export NUM_BEAMS=5
mkdir -p $DATA_DIR
sacrebleu -t wmt19 -l $PAIR --echo src > $DATA_DIR/val.source
sacrebleu -t wmt19 -l $PAIR --echo ref > $DATA_DIR/val.target
echo $PAIR
PYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py allenai/{model_name} $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS
```
## Data Sources
- [training, etc.](http://www.statmt.org/wmt19/)
- [test set](http://matrix.statmt.org/test_sets/newstest2019.tgz?1556572561)
### BibTeX entry and citation info
```
@misc{{kasai2020deep,
title={{Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation}},
author={{Jungo Kasai and Nikolaos Pappas and Hao Peng and James Cross and Noah A. Smith}},
year={{2020}},
eprint={{2006.10369}},
archivePrefix={{arXiv}},
primaryClass={{cs.CL}}
}}
```
"""
model_card_dir.mkdir(parents=True, exist_ok=True)
path = os.path.join(model_card_dir, "README.md")
print(f"Generating {path}")
with open(path, "w", encoding="utf-8") as f:
f.write(readme)
# make sure we are under the root of the project
repo_dir = Path(__file__).resolve().parent.parent.parent
model_cards_dir = repo_dir / "model_cards"
for model_name in ["wmt19-de-en-6-6-base", "wmt19-de-en-6-6-big"]:
model_card_dir = model_cards_dir / "allenai" / model_name
write_model_card(model_card_dir, src_lang="de", tgt_lang="en", model_name=model_name)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/yoso/fast_lsh_cumulation_cuda.cu | // File from https://github.com/mlpen/YOSO/blob/main/encoders/backbones/efficient_attentions/yoso/yoso_v1/cuda/fast_lsh_cumulation_cuda.cu
#include "fast_lsh_cumulation_cuda.h"
#include "common_cuda_device.h"
#include "common_cuda.h"
#include "common.h"
#include <stdio.h>
//////////////////////////////////////////////////////////////////////////////////////////////////
//////////////////////////////////////////////////////////////////////////////////////////////////
inline __device__ void fast_hadamard_transform(float *vector_buffer, int vector_dim, int dim_idx) {
int stride = vector_dim / 2;
while (stride > (WARP_SIZE / 2)) {
__syncthreads();
int sign = 1 - ((dim_idx / stride) % 2) * 2;
float val1 = vector_buffer[dim_idx];
float val2 = vector_buffer[dim_idx + sign * stride];
__syncthreads();
vector_buffer[dim_idx] = float(sign) * val1 + val2;
stride = stride / 2;
}
float val = vector_buffer[dim_idx];
#pragma unroll
for (stride = (WARP_SIZE / 2); stride > 0; stride = stride / 2) {
int sign = 1 - ((dim_idx / stride) % 2) * 2;
val = float(sign) * val + __shfl_xor_sync(FULL_MASK, val, stride);
}
vector_buffer[dim_idx] = val;
}
__global__ void fast_hash_ver1_cuda_kernel(
int *mask, // [batch_size, num_vector]
float *vector, // [batch_size, num_vector, vector_dim]
int *Dmat, // [batch_size, 3, num_part, vector_dim]
int *hash_code, // [batch_size, num_vector, num_hash_f]
int batch_size,
int num_vector,
int vector_dim,
int num_part,
int num_hash_f,
int hash_code_len
) {
int batch_idx = blockIdx.z;
int vector_idx = blockIdx.y;
int part_idx = blockIdx.x;
int dim_idx = threadIdx.x;
int batch_idx__vector_idx = batch_idx * num_vector + vector_idx;
if (mask[batch_idx__vector_idx] == 0) {
return;
}
extern __shared__ float buffer[];
float *vector_buffer = buffer;
vector_buffer[dim_idx] = vector[batch_idx__vector_idx * vector_dim + dim_idx];
vector_buffer[dim_idx] = vector_buffer[dim_idx] * (float)Dmat[((batch_idx * 3 + 0) * num_part + part_idx) * vector_dim + dim_idx];
fast_hadamard_transform(vector_buffer, vector_dim, dim_idx);
vector_buffer[dim_idx] = vector_buffer[dim_idx] * (float)Dmat[((batch_idx * 3 + 1) * num_part + part_idx) * vector_dim + dim_idx];
fast_hadamard_transform(vector_buffer, vector_dim, dim_idx);
vector_buffer[dim_idx] = vector_buffer[dim_idx] * (float)Dmat[((batch_idx * 3 + 2) * num_part + part_idx) * vector_dim + dim_idx];
fast_hadamard_transform(vector_buffer, vector_dim, dim_idx);
int num_hash_per_part = vector_dim / hash_code_len;
if (hash_code_len == 8 || hash_code_len == 16) {
int code = select(vector_buffer[dim_idx] > 0, 1 << (dim_idx % hash_code_len), 0);
for (int offset = 1; offset < hash_code_len; offset = offset * 2) {
code += __shfl_xor_sync(FULL_MASK, code, offset);
}
if (dim_idx % hash_code_len == 0) {
int hash_f_idx = part_idx * num_hash_per_part + dim_idx / hash_code_len;
if (hash_f_idx < num_hash_f) {
hash_code[batch_idx__vector_idx * num_hash_f + hash_f_idx] = code;
}
}
} else {
vector_buffer[dim_idx] = select(vector_buffer[dim_idx] > 0, 1 << (dim_idx % hash_code_len), 0);
__syncthreads();
if (dim_idx < num_hash_per_part) {
int code = 0;
for (int i = 0; i < hash_code_len; i++) {
code += vector_buffer[dim_idx * hash_code_len + i];
}
int hash_f_idx = part_idx * num_hash_per_part + dim_idx;
if (hash_f_idx < num_hash_f) {
hash_code[batch_idx__vector_idx * num_hash_f + hash_f_idx] = code;
}
}
}
}
__global__ void lsh_cumulation_ver1_step1_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
float *value, // [batch_size, num_key, value_dim]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key,
int value_dim,
int offset_warp
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
} else {
float warp_value = value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
}
__global__ void lsh_cumulation_ver1_step2_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_hash_code, // [batch_size, num_query, num_hash_f]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_query,
int value_dim,
int offset_warp
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int query_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = 0;
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
}
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] = warp_value / float(num_hash_f);
} else {
float warp_value = 0;
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] = warp_value / float(num_hash_f);
}
}
__global__ void lsh_weighted_cumulation_ver1_step1_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key,
int value_dim,
int weight_dim,
int offset_warp,
int weight_idx
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = key_weight[batch_idx__key_idx * weight_dim + weight_idx] * value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
} else {
float warp_value = key_weight[batch_idx__key_idx * weight_dim + weight_idx] * value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
}
__global__ void lsh_weighted_cumulation_ver1_step2_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_hash_code, // [batch_size, num_query, num_hash_f]
float *query_weight, // [batch_size, num_query, weight_dim]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_query,
int value_dim,
int weight_dim,
int offset_warp,
int weight_idx
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int query_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = 0;
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
}
float warp_weight = query_weight[batch_idx__query_idx * weight_dim + weight_idx];
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] += warp_weight * warp_value / float(num_hash_f);
} else {
float warp_value = 0;
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
float warp_weight = query_weight[batch_idx__query_idx * weight_dim + weight_idx];
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] += warp_weight * warp_value / float(num_hash_f);
}
}
__global__ void count_sort_step1_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key
) {
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int hash_f_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
int hash_code = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_idx];
atomicAdd(&count_sort_table[(batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + hash_code], 1);
}
__global__ void count_sort_step2_cuda_kernel(
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int batch_size,
int num_hash_f,
int hashtable_capacity
) {
int batch_idx = blockIdx.y;
int hash_f_idx = blockIdx.x;
int num_threads = blockDim.x;
int thread_id = threadIdx.x;
int batch_idx__hash_f_idx = batch_idx * num_hash_f + hash_f_idx;
extern __shared__ float buffer[];
int *table_buffer = (int*)buffer;
if (thread_id == 0) {
table_buffer[0] = 0;
}
copy_data<int>(&count_sort_table[batch_idx__hash_f_idx * hashtable_capacity], &table_buffer[1], hashtable_capacity - 1, num_threads, thread_id);
for (int table_idx_start = 0; table_idx_start < hashtable_capacity; table_idx_start = table_idx_start + num_threads) {
int thread_value = table_buffer[table_idx_start + thread_id];
int next_thread_value = 0;
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
next_thread_value = __shfl_up_sync(FULL_MASK, thread_value, offset);
if (thread_id % WARP_SIZE >= offset) {
thread_value = thread_value + next_thread_value;
}
}
table_buffer[table_idx_start + thread_id] = thread_value;
}
__syncthreads();
if (hashtable_capacity > WARP_SIZE) {
if (thread_id < WARP_SIZE) {
for (int table_idx_start = WARP_SIZE; table_idx_start < hashtable_capacity; table_idx_start = table_idx_start + WARP_SIZE) {
table_buffer[table_idx_start + thread_id] += table_buffer[table_idx_start - 1];
}
}
}
copy_data<int>(table_buffer, &count_sort_table[batch_idx__hash_f_idx * hashtable_capacity], hashtable_capacity, num_threads, thread_id);
}
__global__ void count_sort_step3_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int *key_sorted_idxes, // [batch_size, num_hash_f, num_key]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key
) {
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int hash_f_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
int batch_idx__hash_f_idx = batch_idx * num_hash_f + hash_f_idx;
int hash_code = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_idx];
int sort_idx = atomicAdd(&count_sort_table[batch_idx__hash_f_idx * hashtable_capacity + hash_code], 1);
key_sorted_idxes[batch_idx__hash_f_idx * num_key + sort_idx] = key_idx;
}
__global__ void extract_query_info_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_hash_code, // [batch_size, num_query, num_hash_f]
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int *query_info, // [batch_size, num_query, 2, num_hash_f]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_query
) {
int batch_idx = blockIdx.y;
int query_idx = blockIdx.x * blockDim.y + threadIdx.y;
int hash_f_idx = threadIdx.x;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
int hash_code = query_hash_code[batch_idx__query_idx * num_hash_f + hash_f_idx];
int batch_idx__hash_f_idx__hash_code = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + hash_code;
int key_offset = select(hash_code == 0, 0, count_sort_table[batch_idx__hash_f_idx__hash_code - 1]);
int key_count = count_sort_table[batch_idx__hash_f_idx__hash_code] - key_offset;
query_info[batch_idx__query_idx * 2 * num_hash_f + hash_f_idx] = key_offset;
query_info[(batch_idx__query_idx * 2 + 1) * num_hash_f + hash_f_idx] = key_count;
}
__global__ void lsh_weighted_cumulation_ver2_step2_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_info, // [batch_size, num_query, 2, num_hash_f]
int *key_sorted_idxes, // [batch_size, num_hash_f, num_key]
float *query_weight, // [batch_size, num_query, weight_dim]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int num_query,
int num_key,
int value_dim,
int weight_dim
) {
int batch_idx = blockIdx.z;
int hash_f_idx = blockIdx.y;
int query_idx = blockIdx.x;
int num_threads = blockDim.y * blockDim.x;
int thread_id = threadIdx.y * blockDim.x + threadIdx.x;
int num_warps = blockDim.y;
int warp_idx = threadIdx.y;
int warp_thread_idx = threadIdx.x;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
int key_offset = query_info[batch_idx__query_idx * 2 * num_hash_f + hash_f_idx];
int key_count = query_info[(batch_idx__query_idx * 2 + 1) * num_hash_f + hash_f_idx];
if (key_count == 0) {
return;
}
extern __shared__ float buffer[];
if (key_count == 1) {
if (warp_idx == 0) {
int key_idx = key_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_key + key_offset];
int batch_idx__key_idx = batch_idx * num_key + key_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx] * key_weight[batch_idx__key_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value[batch_idx__key_idx * value_dim + value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
} else {
float *weight_buffer = buffer;
int *key_idxes_buffer = (int*)&buffer[weight_dim];
copy_data_nonblocking<float>(&query_weight[batch_idx__query_idx * weight_dim], weight_buffer, weight_dim, num_threads, thread_id);
while (key_count > 0) {
int work_size = min(WARP_SIZE, key_count);
copy_data_nonblocking<int>(&key_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_key + key_offset], key_idxes_buffer, work_size, num_threads, thread_id);
__syncthreads();
for (int work_offset = 0; work_offset < WARP_SIZE; work_offset = work_offset + num_warps) {
int work_idx = work_offset + warp_idx;
if (work_idx < key_count) {
int key_idx = key_idxes_buffer[work_idx];
int batch_idx__key_idx = batch_idx * num_key + key_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = weight_buffer[weight_dim_idx] * key_weight[batch_idx__key_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value[batch_idx__key_idx * value_dim + value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
}
key_count = key_count - work_size;
key_offset = key_offset + work_size;
}
}
}
__global__ void lsh_weighted_cumulation_ver3_step2_cuda_kernel(
int *query_sorted_idxes, // [batch_size, num_hash_f, num_query]
int *key_mask, // [batch_size, num_key]
int *key_info, // [batch_size, num_key, 2, num_hash_f]
float *query_weight, // [batch_size, num_query, weight_dim]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int num_query,
int num_key,
int value_dim,
int weight_dim
) {
int batch_idx = blockIdx.z;
int hash_f_idx = blockIdx.y;
int key_idx = blockIdx.x;
int num_threads = blockDim.y * blockDim.x;
int thread_id = threadIdx.y * blockDim.x + threadIdx.x;
int num_warps = blockDim.y;
int warp_idx = threadIdx.y;
int warp_thread_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
int query_offset = key_info[batch_idx__key_idx * 2 * num_hash_f + hash_f_idx];
int query_count = key_info[(batch_idx__key_idx * 2 + 1) * num_hash_f + hash_f_idx];
if (query_count == 0) {
return;
}
extern __shared__ float buffer[];
if (query_count == 1) {
if (warp_idx == 0) {
int query_idx = query_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_query + query_offset];
int batch_idx__query_idx = batch_idx * num_query + query_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = key_weight[batch_idx__key_idx * weight_dim + weight_dim_idx] * query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value[batch_idx__key_idx * value_dim + value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
} else {
float *weight_buffer = buffer;
float *value_buffer = &buffer[weight_dim];
int *query_idxes_buffer = (int*)&buffer[weight_dim + value_dim];
copy_data_nonblocking<float>(&key_weight[batch_idx__key_idx * weight_dim], weight_buffer, weight_dim, num_threads, thread_id);
copy_data_nonblocking<float>(&value[batch_idx__key_idx * value_dim], value_buffer, value_dim, num_threads, thread_id);
while (query_count > 0) {
int work_size = min(WARP_SIZE, query_count);
copy_data_nonblocking<int>(&query_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_query + query_offset], query_idxes_buffer, work_size, num_threads, thread_id);
__syncthreads();
for (int work_offset = 0; work_offset < WARP_SIZE; work_offset = work_offset + num_warps) {
int work_idx = work_offset + warp_idx;
if (work_idx < query_count) {
int query_idx = query_idxes_buffer[work_idx];
int batch_idx__query_idx = batch_idx * num_query + query_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = weight_buffer[weight_dim_idx] * query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value_buffer[value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
}
query_count = query_count - work_size;
query_offset = query_offset + work_size;
}
}
}
__global__ void lsh_weighted_cumulation_ver4_step2_cuda_kernel(
int *query_sorted_idxes, // [batch_size, num_hash_f, num_query]
int *key_mask, // [batch_size, num_key]
int *key_info, // [batch_size, num_key, 2, num_hash_f]
float *query_weight, // [batch_size, num_query, weight_dim]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int num_query,
int num_key,
int value_dim,
int weight_dim
) {
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x;
int num_threads = blockDim.y * blockDim.x;
int thread_id = threadIdx.y * blockDim.x + threadIdx.x;
int num_warps = blockDim.y;
int warp_idx = threadIdx.y;
int warp_thread_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
extern __shared__ float buffer[];
float *weight_buffer = buffer;
float *value_buffer = &buffer[weight_dim];
int *key_info_buffer = (int*)&buffer[weight_dim + value_dim];
copy_data_nonblocking<float>(&key_weight[batch_idx__key_idx * weight_dim], weight_buffer, weight_dim, num_threads, thread_id);
copy_data_nonblocking<float>(&value[batch_idx__key_idx * value_dim], value_buffer, value_dim, num_threads, thread_id);
copy_data_nonblocking<int>(&key_info[batch_idx__key_idx * 2 * num_hash_f], key_info_buffer, 2 * num_hash_f, num_threads, thread_id);
int *query_offset_buffer = key_info_buffer;
int *query_count_buffer = &key_info_buffer[num_hash_f];
const int hashtable_size = 1024 + OPTIMAL_THREADS_PER_BLOCK;
__shared__ int hashtable_query[hashtable_size];
__shared__ int hashtable_count[hashtable_size];
__shared__ int inserted_query[hashtable_size];
__shared__ int query_counter[1];
int hash_f_idx_base = 0;
while (true) {
init_buffer_nonblocking<int>(EMPTY_VALUE, hashtable_query, hashtable_size, num_threads, thread_id);
init_buffer_nonblocking<int>(0, hashtable_count, hashtable_size, num_threads, thread_id);
init_buffer_nonblocking<int>(EMPTY_VALUE, inserted_query, hashtable_size, num_threads, thread_id);
init_buffer_nonblocking<int>(0, query_counter, 1, num_threads, thread_id);
__syncthreads();
while (hash_f_idx_base < num_hash_f) {
int hash_f_idx = hash_f_idx_base + warp_idx;
int batch_idx__hash_f_idx = batch_idx * num_hash_f + hash_f_idx;
int stop_flag = 0;
int query_offset = query_offset_buffer[hash_f_idx];
int query_count = query_count_buffer[hash_f_idx];
while (query_count > 0) {
int work_size = min(query_count, WARP_SIZE);
// try inserting query to set and check whether the query is new
int found_new_query = 0;
int query_idx = -1;
if (warp_thread_idx < work_size) {
query_idx = query_sorted_idxes[batch_idx__hash_f_idx * num_query + query_offset + warp_thread_idx];
int slot = set_insert<int>(hashtable_query, hashtable_size, query_idx);
if (slot >= 0) {
found_new_query = atomicAdd(&hashtable_count[slot], 1) == 0;
}
}
// compute cumulative offset
int position_offset = found_new_query;
int next_position_offset = 0;
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
next_position_offset = __shfl_up_sync(FULL_MASK, position_offset, offset);
if (thread_id % WARP_SIZE >= offset) {
position_offset = position_offset + next_position_offset;
}
}
// get the inserted query list end index
int inserted_query_base = 0;
if (thread_id % WARP_SIZE == WARP_SIZE - 1) {
inserted_query_base = atomicAdd(query_counter, position_offset);
}
inserted_query_base = __shfl_sync(FULL_MASK, inserted_query_base, WARP_SIZE - 1);
// insert new queries to list
int insert_idx = inserted_query_base + position_offset - 1;
if (found_new_query) {
inserted_query[insert_idx] = query_idx;
}
// remove inserted queries from list
query_offset_buffer[hash_f_idx] += work_size;
query_count_buffer[hash_f_idx] -= work_size;
query_offset += work_size;
query_count -= work_size;
// if list is almost full, stop inserting
if (inserted_query_base + OPTIMAL_THREADS_PER_BLOCK > hashtable_size) {
stop_flag = 1;
break;
}
}
if (stop_flag) {
break;
}
hash_f_idx_base = hash_f_idx_base + num_warps;
}
__syncthreads();
int num_distint_query = query_counter[0];
if (num_distint_query > 0) {
for (int idx_base = 0; idx_base < num_distint_query; idx_base = idx_base + num_warps) {
int idx = idx_base + warp_idx;
if (idx < num_distint_query) {
int query_idx = inserted_query[idx];
int batch_idx__query_idx = batch_idx * num_query + query_idx;
int slot = set_lookup<int>(hashtable_query, hashtable_size, query_idx);
int duplicate_count = hashtable_count[slot];
float weight = 0;
for (int weight_idx_base = 0; weight_idx_base < weight_dim; weight_idx_base = weight_idx_base + WARP_SIZE) {
int weight_dim_idx = weight_idx_base + warp_thread_idx;
float val = weight_buffer[weight_dim_idx] * query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = (float)duplicate_count * weight / float(num_hash_f);
for (int value_idx_base = 0; value_idx_base < value_dim; value_idx_base = value_idx_base + WARP_SIZE) {
int value_dim_idx = value_idx_base + warp_thread_idx;
float val = value_buffer[value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
}
} else {
// all computation is completed if num_distint_query == 0
break;
}
__syncthreads();
}
}
| // File from https://github.com/mlpen/YOSO/blob/main/encoders/backbones/efficient_attentions/yoso/yoso_v1/cuda/fast_lsh_cumulation_cuda.cu
#include "fast_lsh_cumulation_cuda.h"
#include "common_cuda_device.h"
#include "common_cuda.h"
#include "common.h"
#include <stdio.h>
//////////////////////////////////////////////////////////////////////////////////////////////////
//////////////////////////////////////////////////////////////////////////////////////////////////
inline __device__ void fast_hadamard_transform(float *vector_buffer, int vector_dim, int dim_idx) {
int stride = vector_dim / 2;
while (stride > (WARP_SIZE / 2)) {
__syncthreads();
int sign = 1 - ((dim_idx / stride) % 2) * 2;
float val1 = vector_buffer[dim_idx];
float val2 = vector_buffer[dim_idx + sign * stride];
__syncthreads();
vector_buffer[dim_idx] = float(sign) * val1 + val2;
stride = stride / 2;
}
float val = vector_buffer[dim_idx];
#pragma unroll
for (stride = (WARP_SIZE / 2); stride > 0; stride = stride / 2) {
int sign = 1 - ((dim_idx / stride) % 2) * 2;
val = float(sign) * val + __shfl_xor_sync(FULL_MASK, val, stride);
}
vector_buffer[dim_idx] = val;
}
__global__ void fast_hash_ver1_cuda_kernel(
int *mask, // [batch_size, num_vector]
float *vector, // [batch_size, num_vector, vector_dim]
int *Dmat, // [batch_size, 3, num_part, vector_dim]
int *hash_code, // [batch_size, num_vector, num_hash_f]
int batch_size,
int num_vector,
int vector_dim,
int num_part,
int num_hash_f,
int hash_code_len
) {
int batch_idx = blockIdx.z;
int vector_idx = blockIdx.y;
int part_idx = blockIdx.x;
int dim_idx = threadIdx.x;
int batch_idx__vector_idx = batch_idx * num_vector + vector_idx;
if (mask[batch_idx__vector_idx] == 0) {
return;
}
extern __shared__ float buffer[];
float *vector_buffer = buffer;
vector_buffer[dim_idx] = vector[batch_idx__vector_idx * vector_dim + dim_idx];
vector_buffer[dim_idx] = vector_buffer[dim_idx] * (float)Dmat[((batch_idx * 3 + 0) * num_part + part_idx) * vector_dim + dim_idx];
fast_hadamard_transform(vector_buffer, vector_dim, dim_idx);
vector_buffer[dim_idx] = vector_buffer[dim_idx] * (float)Dmat[((batch_idx * 3 + 1) * num_part + part_idx) * vector_dim + dim_idx];
fast_hadamard_transform(vector_buffer, vector_dim, dim_idx);
vector_buffer[dim_idx] = vector_buffer[dim_idx] * (float)Dmat[((batch_idx * 3 + 2) * num_part + part_idx) * vector_dim + dim_idx];
fast_hadamard_transform(vector_buffer, vector_dim, dim_idx);
int num_hash_per_part = vector_dim / hash_code_len;
if (hash_code_len == 8 || hash_code_len == 16) {
int code = select(vector_buffer[dim_idx] > 0, 1 << (dim_idx % hash_code_len), 0);
for (int offset = 1; offset < hash_code_len; offset = offset * 2) {
code += __shfl_xor_sync(FULL_MASK, code, offset);
}
if (dim_idx % hash_code_len == 0) {
int hash_f_idx = part_idx * num_hash_per_part + dim_idx / hash_code_len;
if (hash_f_idx < num_hash_f) {
hash_code[batch_idx__vector_idx * num_hash_f + hash_f_idx] = code;
}
}
} else {
vector_buffer[dim_idx] = select(vector_buffer[dim_idx] > 0, 1 << (dim_idx % hash_code_len), 0);
__syncthreads();
if (dim_idx < num_hash_per_part) {
int code = 0;
for (int i = 0; i < hash_code_len; i++) {
code += vector_buffer[dim_idx * hash_code_len + i];
}
int hash_f_idx = part_idx * num_hash_per_part + dim_idx;
if (hash_f_idx < num_hash_f) {
hash_code[batch_idx__vector_idx * num_hash_f + hash_f_idx] = code;
}
}
}
}
__global__ void lsh_cumulation_ver1_step1_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
float *value, // [batch_size, num_key, value_dim]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key,
int value_dim,
int offset_warp
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
} else {
float warp_value = value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
}
__global__ void lsh_cumulation_ver1_step2_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_hash_code, // [batch_size, num_query, num_hash_f]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_query,
int value_dim,
int offset_warp
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int query_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = 0;
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
}
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] = warp_value / float(num_hash_f);
} else {
float warp_value = 0;
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] = warp_value / float(num_hash_f);
}
}
__global__ void lsh_weighted_cumulation_ver1_step1_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key,
int value_dim,
int weight_dim,
int offset_warp,
int weight_idx
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = key_weight[batch_idx__key_idx * weight_dim + weight_idx] * value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
} else {
float warp_value = key_weight[batch_idx__key_idx * weight_dim + weight_idx] * value[batch_idx__key_idx * value_dim + offset_warp + warp_thread_idx];
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = key_hash_code[batch_idx__key_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
atomicAdd(&hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx], warp_value);
}
}
}
__global__ void lsh_weighted_cumulation_ver1_step2_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_hash_code, // [batch_size, num_query, num_hash_f]
float *query_weight, // [batch_size, num_query, weight_dim]
float *hashtable_value, // [batch_size, num_hash_f, hashtable_capacity, WARP_SIZE]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_query,
int value_dim,
int weight_dim,
int offset_warp,
int weight_idx
) {
int warp_thread_idx = threadIdx.x;
int batch_idx = blockIdx.y;
int query_idx = blockIdx.x * blockDim.y + threadIdx.y;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
if (num_hash_f > WARP_SIZE) {
float warp_value = 0;
for (int hash_f_start = 0; hash_f_start < num_hash_f; hash_f_start = hash_f_start + WARP_SIZE) {
int warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + hash_f_start + warp_thread_idx];
#pragma unroll
for (int hash_f_offset = 0; hash_f_offset < WARP_SIZE; hash_f_offset++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_offset);
int hashtable_idx = (batch_idx * num_hash_f + (hash_f_start + hash_f_offset)) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
}
float warp_weight = query_weight[batch_idx__query_idx * weight_dim + weight_idx];
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] += warp_weight * warp_value / float(num_hash_f);
} else {
float warp_value = 0;
int warp_hashcode = 0;
if (warp_thread_idx < num_hash_f) {
warp_hashcode = query_hash_code[batch_idx__query_idx * num_hash_f + warp_thread_idx];
}
for (int hash_f_idx = 0; hash_f_idx < num_hash_f; hash_f_idx++) {
int current_hashcode = warp_hashcode;
current_hashcode = __shfl_sync(FULL_MASK, current_hashcode, hash_f_idx);
int hashtable_idx = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + current_hashcode;
warp_value = warp_value + hashtable_value[hashtable_idx * WARP_SIZE + warp_thread_idx];
}
float warp_weight = query_weight[batch_idx__query_idx * weight_dim + weight_idx];
cumulation_value[batch_idx__query_idx * value_dim + offset_warp + warp_thread_idx] += warp_weight * warp_value / float(num_hash_f);
}
}
__global__ void count_sort_step1_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key
) {
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int hash_f_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
int hash_code = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_idx];
atomicAdd(&count_sort_table[(batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + hash_code], 1);
}
__global__ void count_sort_step2_cuda_kernel(
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int batch_size,
int num_hash_f,
int hashtable_capacity
) {
int batch_idx = blockIdx.y;
int hash_f_idx = blockIdx.x;
int num_threads = blockDim.x;
int thread_id = threadIdx.x;
int batch_idx__hash_f_idx = batch_idx * num_hash_f + hash_f_idx;
extern __shared__ float buffer[];
int *table_buffer = (int*)buffer;
if (thread_id == 0) {
table_buffer[0] = 0;
}
copy_data<int>(&count_sort_table[batch_idx__hash_f_idx * hashtable_capacity], &table_buffer[1], hashtable_capacity - 1, num_threads, thread_id);
for (int table_idx_start = 0; table_idx_start < hashtable_capacity; table_idx_start = table_idx_start + num_threads) {
int thread_value = table_buffer[table_idx_start + thread_id];
int next_thread_value = 0;
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
next_thread_value = __shfl_up_sync(FULL_MASK, thread_value, offset);
if (thread_id % WARP_SIZE >= offset) {
thread_value = thread_value + next_thread_value;
}
}
table_buffer[table_idx_start + thread_id] = thread_value;
}
__syncthreads();
if (hashtable_capacity > WARP_SIZE) {
if (thread_id < WARP_SIZE) {
for (int table_idx_start = WARP_SIZE; table_idx_start < hashtable_capacity; table_idx_start = table_idx_start + WARP_SIZE) {
table_buffer[table_idx_start + thread_id] += table_buffer[table_idx_start - 1];
}
}
}
copy_data<int>(table_buffer, &count_sort_table[batch_idx__hash_f_idx * hashtable_capacity], hashtable_capacity, num_threads, thread_id);
}
__global__ void count_sort_step3_cuda_kernel(
int *key_mask, // [batch_size, num_key]
int *key_hash_code, // [batch_size, num_key, num_hash_f]
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int *key_sorted_idxes, // [batch_size, num_hash_f, num_key]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_key
) {
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x * blockDim.y + threadIdx.y;
int hash_f_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
int batch_idx__hash_f_idx = batch_idx * num_hash_f + hash_f_idx;
int hash_code = key_hash_code[batch_idx__key_idx * num_hash_f + hash_f_idx];
int sort_idx = atomicAdd(&count_sort_table[batch_idx__hash_f_idx * hashtable_capacity + hash_code], 1);
key_sorted_idxes[batch_idx__hash_f_idx * num_key + sort_idx] = key_idx;
}
__global__ void extract_query_info_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_hash_code, // [batch_size, num_query, num_hash_f]
int *count_sort_table, // [batch_size, num_hash_f, hashtable_capacity]
int *query_info, // [batch_size, num_query, 2, num_hash_f]
int batch_size,
int num_hash_f,
int hashtable_capacity,
int num_query
) {
int batch_idx = blockIdx.y;
int query_idx = blockIdx.x * blockDim.y + threadIdx.y;
int hash_f_idx = threadIdx.x;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
int hash_code = query_hash_code[batch_idx__query_idx * num_hash_f + hash_f_idx];
int batch_idx__hash_f_idx__hash_code = (batch_idx * num_hash_f + hash_f_idx) * hashtable_capacity + hash_code;
int key_offset = select(hash_code == 0, 0, count_sort_table[batch_idx__hash_f_idx__hash_code - 1]);
int key_count = count_sort_table[batch_idx__hash_f_idx__hash_code] - key_offset;
query_info[batch_idx__query_idx * 2 * num_hash_f + hash_f_idx] = key_offset;
query_info[(batch_idx__query_idx * 2 + 1) * num_hash_f + hash_f_idx] = key_count;
}
__global__ void lsh_weighted_cumulation_ver2_step2_cuda_kernel(
int *query_mask, // [batch_size, num_query]
int *query_info, // [batch_size, num_query, 2, num_hash_f]
int *key_sorted_idxes, // [batch_size, num_hash_f, num_key]
float *query_weight, // [batch_size, num_query, weight_dim]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int num_query,
int num_key,
int value_dim,
int weight_dim
) {
int batch_idx = blockIdx.z;
int hash_f_idx = blockIdx.y;
int query_idx = blockIdx.x;
int num_threads = blockDim.y * blockDim.x;
int thread_id = threadIdx.y * blockDim.x + threadIdx.x;
int num_warps = blockDim.y;
int warp_idx = threadIdx.y;
int warp_thread_idx = threadIdx.x;
int batch_idx__query_idx = batch_idx * num_query + query_idx;
if (query_mask[batch_idx__query_idx] == 0) {
return;
}
int key_offset = query_info[batch_idx__query_idx * 2 * num_hash_f + hash_f_idx];
int key_count = query_info[(batch_idx__query_idx * 2 + 1) * num_hash_f + hash_f_idx];
if (key_count == 0) {
return;
}
extern __shared__ float buffer[];
if (key_count == 1) {
if (warp_idx == 0) {
int key_idx = key_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_key + key_offset];
int batch_idx__key_idx = batch_idx * num_key + key_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx] * key_weight[batch_idx__key_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value[batch_idx__key_idx * value_dim + value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
} else {
float *weight_buffer = buffer;
int *key_idxes_buffer = (int*)&buffer[weight_dim];
copy_data_nonblocking<float>(&query_weight[batch_idx__query_idx * weight_dim], weight_buffer, weight_dim, num_threads, thread_id);
while (key_count > 0) {
int work_size = min(WARP_SIZE, key_count);
copy_data_nonblocking<int>(&key_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_key + key_offset], key_idxes_buffer, work_size, num_threads, thread_id);
__syncthreads();
for (int work_offset = 0; work_offset < WARP_SIZE; work_offset = work_offset + num_warps) {
int work_idx = work_offset + warp_idx;
if (work_idx < key_count) {
int key_idx = key_idxes_buffer[work_idx];
int batch_idx__key_idx = batch_idx * num_key + key_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = weight_buffer[weight_dim_idx] * key_weight[batch_idx__key_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value[batch_idx__key_idx * value_dim + value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
}
key_count = key_count - work_size;
key_offset = key_offset + work_size;
}
}
}
__global__ void lsh_weighted_cumulation_ver3_step2_cuda_kernel(
int *query_sorted_idxes, // [batch_size, num_hash_f, num_query]
int *key_mask, // [batch_size, num_key]
int *key_info, // [batch_size, num_key, 2, num_hash_f]
float *query_weight, // [batch_size, num_query, weight_dim]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int num_query,
int num_key,
int value_dim,
int weight_dim
) {
int batch_idx = blockIdx.z;
int hash_f_idx = blockIdx.y;
int key_idx = blockIdx.x;
int num_threads = blockDim.y * blockDim.x;
int thread_id = threadIdx.y * blockDim.x + threadIdx.x;
int num_warps = blockDim.y;
int warp_idx = threadIdx.y;
int warp_thread_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
int query_offset = key_info[batch_idx__key_idx * 2 * num_hash_f + hash_f_idx];
int query_count = key_info[(batch_idx__key_idx * 2 + 1) * num_hash_f + hash_f_idx];
if (query_count == 0) {
return;
}
extern __shared__ float buffer[];
if (query_count == 1) {
if (warp_idx == 0) {
int query_idx = query_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_query + query_offset];
int batch_idx__query_idx = batch_idx * num_query + query_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = key_weight[batch_idx__key_idx * weight_dim + weight_dim_idx] * query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value[batch_idx__key_idx * value_dim + value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
} else {
float *weight_buffer = buffer;
float *value_buffer = &buffer[weight_dim];
int *query_idxes_buffer = (int*)&buffer[weight_dim + value_dim];
copy_data_nonblocking<float>(&key_weight[batch_idx__key_idx * weight_dim], weight_buffer, weight_dim, num_threads, thread_id);
copy_data_nonblocking<float>(&value[batch_idx__key_idx * value_dim], value_buffer, value_dim, num_threads, thread_id);
while (query_count > 0) {
int work_size = min(WARP_SIZE, query_count);
copy_data_nonblocking<int>(&query_sorted_idxes[(batch_idx * num_hash_f + hash_f_idx) * num_query + query_offset], query_idxes_buffer, work_size, num_threads, thread_id);
__syncthreads();
for (int work_offset = 0; work_offset < WARP_SIZE; work_offset = work_offset + num_warps) {
int work_idx = work_offset + warp_idx;
if (work_idx < query_count) {
int query_idx = query_idxes_buffer[work_idx];
int batch_idx__query_idx = batch_idx * num_query + query_idx;
float weight = 0;
for (int weight_offset = 0; weight_offset < weight_dim; weight_offset = weight_offset + WARP_SIZE) {
int weight_dim_idx = weight_offset + warp_thread_idx;
float val = weight_buffer[weight_dim_idx] * query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = weight / float(num_hash_f);
for (int value_offset = 0; value_offset < value_dim; value_offset = value_offset + WARP_SIZE) {
int value_dim_idx = value_offset + warp_thread_idx;
float val = value_buffer[value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
}
query_count = query_count - work_size;
query_offset = query_offset + work_size;
}
}
}
__global__ void lsh_weighted_cumulation_ver4_step2_cuda_kernel(
int *query_sorted_idxes, // [batch_size, num_hash_f, num_query]
int *key_mask, // [batch_size, num_key]
int *key_info, // [batch_size, num_key, 2, num_hash_f]
float *query_weight, // [batch_size, num_query, weight_dim]
float *key_weight, // [batch_size, num_key, weight_dim]
float *value, // [batch_size, num_key, value_dim]
float *cumulation_value, // [batch_size, num_query, value_dim]
int batch_size,
int num_hash_f,
int num_query,
int num_key,
int value_dim,
int weight_dim
) {
int batch_idx = blockIdx.y;
int key_idx = blockIdx.x;
int num_threads = blockDim.y * blockDim.x;
int thread_id = threadIdx.y * blockDim.x + threadIdx.x;
int num_warps = blockDim.y;
int warp_idx = threadIdx.y;
int warp_thread_idx = threadIdx.x;
int batch_idx__key_idx = batch_idx * num_key + key_idx;
if (key_mask[batch_idx__key_idx] == 0) {
return;
}
extern __shared__ float buffer[];
float *weight_buffer = buffer;
float *value_buffer = &buffer[weight_dim];
int *key_info_buffer = (int*)&buffer[weight_dim + value_dim];
copy_data_nonblocking<float>(&key_weight[batch_idx__key_idx * weight_dim], weight_buffer, weight_dim, num_threads, thread_id);
copy_data_nonblocking<float>(&value[batch_idx__key_idx * value_dim], value_buffer, value_dim, num_threads, thread_id);
copy_data_nonblocking<int>(&key_info[batch_idx__key_idx * 2 * num_hash_f], key_info_buffer, 2 * num_hash_f, num_threads, thread_id);
int *query_offset_buffer = key_info_buffer;
int *query_count_buffer = &key_info_buffer[num_hash_f];
const int hashtable_size = 1024 + OPTIMAL_THREADS_PER_BLOCK;
__shared__ int hashtable_query[hashtable_size];
__shared__ int hashtable_count[hashtable_size];
__shared__ int inserted_query[hashtable_size];
__shared__ int query_counter[1];
int hash_f_idx_base = 0;
while (true) {
init_buffer_nonblocking<int>(EMPTY_VALUE, hashtable_query, hashtable_size, num_threads, thread_id);
init_buffer_nonblocking<int>(0, hashtable_count, hashtable_size, num_threads, thread_id);
init_buffer_nonblocking<int>(EMPTY_VALUE, inserted_query, hashtable_size, num_threads, thread_id);
init_buffer_nonblocking<int>(0, query_counter, 1, num_threads, thread_id);
__syncthreads();
while (hash_f_idx_base < num_hash_f) {
int hash_f_idx = hash_f_idx_base + warp_idx;
int batch_idx__hash_f_idx = batch_idx * num_hash_f + hash_f_idx;
int stop_flag = 0;
int query_offset = query_offset_buffer[hash_f_idx];
int query_count = query_count_buffer[hash_f_idx];
while (query_count > 0) {
int work_size = min(query_count, WARP_SIZE);
// try inserting query to set and check whether the query is new
int found_new_query = 0;
int query_idx = -1;
if (warp_thread_idx < work_size) {
query_idx = query_sorted_idxes[batch_idx__hash_f_idx * num_query + query_offset + warp_thread_idx];
int slot = set_insert<int>(hashtable_query, hashtable_size, query_idx);
if (slot >= 0) {
found_new_query = atomicAdd(&hashtable_count[slot], 1) == 0;
}
}
// compute cumulative offset
int position_offset = found_new_query;
int next_position_offset = 0;
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
next_position_offset = __shfl_up_sync(FULL_MASK, position_offset, offset);
if (thread_id % WARP_SIZE >= offset) {
position_offset = position_offset + next_position_offset;
}
}
// get the inserted query list end index
int inserted_query_base = 0;
if (thread_id % WARP_SIZE == WARP_SIZE - 1) {
inserted_query_base = atomicAdd(query_counter, position_offset);
}
inserted_query_base = __shfl_sync(FULL_MASK, inserted_query_base, WARP_SIZE - 1);
// insert new queries to list
int insert_idx = inserted_query_base + position_offset - 1;
if (found_new_query) {
inserted_query[insert_idx] = query_idx;
}
// remove inserted queries from list
query_offset_buffer[hash_f_idx] += work_size;
query_count_buffer[hash_f_idx] -= work_size;
query_offset += work_size;
query_count -= work_size;
// if list is almost full, stop inserting
if (inserted_query_base + OPTIMAL_THREADS_PER_BLOCK > hashtable_size) {
stop_flag = 1;
break;
}
}
if (stop_flag) {
break;
}
hash_f_idx_base = hash_f_idx_base + num_warps;
}
__syncthreads();
int num_distint_query = query_counter[0];
if (num_distint_query > 0) {
for (int idx_base = 0; idx_base < num_distint_query; idx_base = idx_base + num_warps) {
int idx = idx_base + warp_idx;
if (idx < num_distint_query) {
int query_idx = inserted_query[idx];
int batch_idx__query_idx = batch_idx * num_query + query_idx;
int slot = set_lookup<int>(hashtable_query, hashtable_size, query_idx);
int duplicate_count = hashtable_count[slot];
float weight = 0;
for (int weight_idx_base = 0; weight_idx_base < weight_dim; weight_idx_base = weight_idx_base + WARP_SIZE) {
int weight_dim_idx = weight_idx_base + warp_thread_idx;
float val = weight_buffer[weight_dim_idx] * query_weight[batch_idx__query_idx * weight_dim + weight_dim_idx];
#pragma unroll
for (int offset = 1; offset < WARP_SIZE; offset = offset << 1) {
val += __shfl_xor_sync(FULL_MASK, val, offset);
}
weight = weight + val;
}
weight = (float)duplicate_count * weight / float(num_hash_f);
for (int value_idx_base = 0; value_idx_base < value_dim; value_idx_base = value_idx_base + WARP_SIZE) {
int value_dim_idx = value_idx_base + warp_thread_idx;
float val = value_buffer[value_dim_idx];
atomicAdd(&cumulation_value[batch_idx__query_idx * value_dim + value_dim_idx], weight * val);
}
}
}
} else {
// all computation is completed if num_distint_query == 0
break;
}
__syncthreads();
}
}
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/de/pipeline_tutorial.mdx | <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Pipelines für Inferenzen
Die [`pipeline`] macht es einfach, jedes beliebige Modell aus dem [Hub](https://huggingface.co/models) für die Inferenz auf jede Sprache, Computer Vision, Sprache und multimodale Aufgaben zu verwenden. Selbst wenn Sie keine Erfahrung mit einer bestimmten Modalität haben oder nicht mit dem zugrundeliegenden Code hinter den Modellen vertraut sind, können Sie sie mit der [`pipeline`] für Inferenzen verwenden! In diesem Beispiel lernen Sie, wie:
* Eine [`pipeline`] für Inferenz zu verwenden.
* Einen bestimmten Tokenizer oder ein bestimmtes Modell zu verwenden.
* Eine [`pipeline`] für Audio-, Vision- und multimodale Aufgaben zu verwenden.
<Tip>
Eine vollständige Liste der unterstützten Aufgaben und verfügbaren Parameter finden Sie in der [`pipeline`]-Dokumentation.
</Tip>
## Verwendung von Pipelines
Obwohl jede Aufgabe eine zugehörige [`pipeline`] hat, ist es einfacher, die allgemeine [`pipeline`]-Abstraktion zu verwenden, die alle aufgabenspezifischen Pipelines enthält. Die [`pipeline`] lädt automatisch ein Standardmodell und eine Vorverarbeitungsklasse, die für Ihre Aufgabe inferenzfähig ist.
1. Beginnen Sie mit der Erstellung einer [`pipeline`] und geben Sie eine Inferenzaufgabe an:
```py
>>> from transformers import pipeline
>>> generator = pipeline(task="text-generation")
```
2. Übergeben Sie Ihren Eingabetext an die [`pipeline`]:
```py
>>> generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone"
... ) # doctest: +SKIP
[{'generated_text': 'Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone, Seven for the Iron-priests at the door to the east, and thirteen for the Lord Kings at the end of the mountain'}]
```
Wenn Sie mehr als eine Eingabe haben, übergeben Sie die Eingabe als Liste:
```py
>>> generator(
... [
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone",
... "Nine for Mortal Men, doomed to die, One for the Dark Lord on his dark throne",
... ]
... ) # doctest: +SKIP
```
Alle zusätzlichen Parameter für Ihre Aufgabe können auch in die [`pipeline`] aufgenommen werden. Die Aufgabe `Text-Generierung` hat eine [`~generation.GenerationMixin.generate`]-Methode mit mehreren Parametern zur Steuerung der Ausgabe. Wenn Sie zum Beispiel mehr als eine Ausgabe erzeugen wollen, setzen Sie den Parameter `num_return_sequences`:
```py
>>> generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone",
... num_return_sequences=2,
... ) # doctest: +SKIP
```
### Wählen Sie ein Modell und einen Tokenizer
Die [`pipeline`] akzeptiert jedes Modell aus dem [Hub] (https://huggingface.co/models). Auf dem Hub gibt es Tags, mit denen Sie nach einem Modell filtern können, das Sie für Ihre Aufgabe verwenden möchten. Sobald Sie ein passendes Modell ausgewählt haben, laden Sie es mit der entsprechenden `AutoModelFor` und [`AutoTokenizer`] Klasse. Laden Sie zum Beispiel die Klasse [`AutoModelForCausalLM`] für eine kausale Sprachmodellierungsaufgabe:
```py
>>> from transformers import AutoTokenizer, AutoModelForCausalLM
>>> tokenizer = AutoTokenizer.from_pretrained("distilgpt2")
>>> model = AutoModelForCausalLM.from_pretrained("distilgpt2")
```
Erstellen Sie eine [`pipeline`] für Ihre Aufgabe, und geben Sie das Modell und den Tokenizer an, die Sie geladen haben:
```py
>>> from transformers import pipeline
>>> generator = pipeline(task="text-generation", model=model, tokenizer=tokenizer)
```
Übergeben Sie Ihren Eingabetext an die [`pipeline`] , um einen Text zu erzeugen:
```py
>>> generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone"
... ) # doctest: +SKIP
[{'generated_text': 'Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone, Seven for the Dragon-lords (for them to rule in a world ruled by their rulers, and all who live within the realm'}]
```
## Audio-Pipeline
Die [`pipeline`] unterstützt auch Audioaufgaben wie Audioklassifizierung und automatische Spracherkennung.
Lassen Sie uns zum Beispiel die Emotion in diesem Audioclip klassifizieren:
```py
>>> from datasets import load_dataset
>>> import torch
>>> torch.manual_seed(42) # doctest: +IGNORE_RESULT
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> audio_file = ds[0]["audio"]["path"]
```
Finden Sie ein [Audioklassifikation](https://huggingface.co/models?pipeline_tag=audio-classification) Modell auf dem Model Hub für Emotionserkennung und laden Sie es in die [`pipeline`]:
```py
>>> from transformers import pipeline
>>> audio_classifier = pipeline(
... task="audio-classification", model="ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition"
... )
```
Übergeben Sie die Audiodatei an die [`pipeline`]:
```py
>>> preds = audio_classifier(audio_file)
>>> preds = [{"score": round(pred["score"], 4), "label": pred["label"]} for pred in preds]
>>> preds
[{'score': 0.1315, 'label': 'calm'}, {'score': 0.1307, 'label': 'neutral'}, {'score': 0.1274, 'label': 'sad'}, {'score': 0.1261, 'label': 'fearful'}, {'score': 0.1242, 'label': 'happy'}]
```
## Bildverarbeitungs-Pipeline
Die Verwendung einer [`pipeline`] für Bildverarbeitungsaufgaben ist praktisch identisch.
Geben Sie Ihre Aufgabe an und übergeben Sie Ihr Bild an den Klassifikator. Das Bild kann ein Link oder ein lokaler Pfad zu dem Bild sein. Zum Beispiel: Welche Katzenart ist unten abgebildet?
```py
>>> from transformers import pipeline
>>> vision_classifier = pipeline(task="image-classification")
>>> preds = vision_classifier(
... images="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
... )
>>> preds = [{"score": round(pred["score"], 4), "label": pred["label"]} for pred in preds]
>>> preds
[{'score': 0.4335, 'label': 'lynx, catamount'}, {'score': 0.0348, 'label': 'cougar, puma, catamount, mountain lion, painter, panther, Felis concolor'}, {'score': 0.0324, 'label': 'snow leopard, ounce, Panthera uncia'}, {'score': 0.0239, 'label': 'Egyptian cat'}, {'score': 0.0229, 'label': 'tiger cat'}]
```
## Multimodale Pipeline
Die [`pipeline`] unterstützt mehr als eine Modalität. Eine Aufgabe zur Beantwortung visueller Fragen (VQA) kombiniert zum Beispiel Text und Bild. Verwenden Sie einen beliebigen Bildlink und eine Frage, die Sie zu dem Bild stellen möchten. Das Bild kann eine URL oder ein lokaler Pfad zu dem Bild sein.
Wenn Sie zum Beispiel das gleiche Bild wie in der obigen Vision-Pipeline verwenden:
```py
>>> image = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
>>> question = "Where is the cat?"
```
Erstellen Sie eine Pipeline für "vqa" und übergeben Sie ihr das Bild und die Frage:
```py
>>> from transformers import pipeline
>>> vqa = pipeline(task="vqa")
>>> preds = vqa(image=image, question=question)
>>> preds = [{"score": round(pred["score"], 4), "answer": pred["answer"]} for pred in preds]
>>> preds
[{'score': 0.9112, 'answer': 'snow'}, {'score': 0.8796, 'answer': 'in snow'}, {'score': 0.6717, 'answer': 'outside'}, {'score': 0.0291, 'answer': 'on ground'}, {'score': 0.027, 'answer': 'ground'}]
```
| <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Pipelines für Inferenzen
Die [`pipeline`] macht es einfach, jedes beliebige Modell aus dem [Hub](https://huggingface.co/models) für die Inferenz auf jede Sprache, Computer Vision, Sprache und multimodale Aufgaben zu verwenden. Selbst wenn Sie keine Erfahrung mit einer bestimmten Modalität haben oder nicht mit dem zugrundeliegenden Code hinter den Modellen vertraut sind, können Sie sie mit der [`pipeline`] für Inferenzen verwenden! In diesem Beispiel lernen Sie, wie:
* Eine [`pipeline`] für Inferenz zu verwenden.
* Einen bestimmten Tokenizer oder ein bestimmtes Modell zu verwenden.
* Eine [`pipeline`] für Audio-, Vision- und multimodale Aufgaben zu verwenden.
<Tip>
Eine vollständige Liste der unterstützten Aufgaben und verfügbaren Parameter finden Sie in der [`pipeline`]-Dokumentation.
</Tip>
## Verwendung von Pipelines
Obwohl jede Aufgabe eine zugehörige [`pipeline`] hat, ist es einfacher, die allgemeine [`pipeline`]-Abstraktion zu verwenden, die alle aufgabenspezifischen Pipelines enthält. Die [`pipeline`] lädt automatisch ein Standardmodell und eine Vorverarbeitungsklasse, die für Ihre Aufgabe inferenzfähig ist.
1. Beginnen Sie mit der Erstellung einer [`pipeline`] und geben Sie eine Inferenzaufgabe an:
```py
>>> from transformers import pipeline
>>> generator = pipeline(task="text-generation")
```
2. Übergeben Sie Ihren Eingabetext an die [`pipeline`]:
```py
>>> generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone"
... ) # doctest: +SKIP
[{'generated_text': 'Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone, Seven for the Iron-priests at the door to the east, and thirteen for the Lord Kings at the end of the mountain'}]
```
Wenn Sie mehr als eine Eingabe haben, übergeben Sie die Eingabe als Liste:
```py
>>> generator(
... [
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone",
... "Nine for Mortal Men, doomed to die, One for the Dark Lord on his dark throne",
... ]
... ) # doctest: +SKIP
```
Alle zusätzlichen Parameter für Ihre Aufgabe können auch in die [`pipeline`] aufgenommen werden. Die Aufgabe `Text-Generierung` hat eine [`~generation.GenerationMixin.generate`]-Methode mit mehreren Parametern zur Steuerung der Ausgabe. Wenn Sie zum Beispiel mehr als eine Ausgabe erzeugen wollen, setzen Sie den Parameter `num_return_sequences`:
```py
>>> generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone",
... num_return_sequences=2,
... ) # doctest: +SKIP
```
### Wählen Sie ein Modell und einen Tokenizer
Die [`pipeline`] akzeptiert jedes Modell aus dem [Hub] (https://huggingface.co/models). Auf dem Hub gibt es Tags, mit denen Sie nach einem Modell filtern können, das Sie für Ihre Aufgabe verwenden möchten. Sobald Sie ein passendes Modell ausgewählt haben, laden Sie es mit der entsprechenden `AutoModelFor` und [`AutoTokenizer`] Klasse. Laden Sie zum Beispiel die Klasse [`AutoModelForCausalLM`] für eine kausale Sprachmodellierungsaufgabe:
```py
>>> from transformers import AutoTokenizer, AutoModelForCausalLM
>>> tokenizer = AutoTokenizer.from_pretrained("distilgpt2")
>>> model = AutoModelForCausalLM.from_pretrained("distilgpt2")
```
Erstellen Sie eine [`pipeline`] für Ihre Aufgabe, und geben Sie das Modell und den Tokenizer an, die Sie geladen haben:
```py
>>> from transformers import pipeline
>>> generator = pipeline(task="text-generation", model=model, tokenizer=tokenizer)
```
Übergeben Sie Ihren Eingabetext an die [`pipeline`] , um einen Text zu erzeugen:
```py
>>> generator(
... "Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone"
... ) # doctest: +SKIP
[{'generated_text': 'Three Rings for the Elven-kings under the sky, Seven for the Dwarf-lords in their halls of stone, Seven for the Dragon-lords (for them to rule in a world ruled by their rulers, and all who live within the realm'}]
```
## Audio-Pipeline
Die [`pipeline`] unterstützt auch Audioaufgaben wie Audioklassifizierung und automatische Spracherkennung.
Lassen Sie uns zum Beispiel die Emotion in diesem Audioclip klassifizieren:
```py
>>> from datasets import load_dataset
>>> import torch
>>> torch.manual_seed(42) # doctest: +IGNORE_RESULT
>>> ds = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation")
>>> audio_file = ds[0]["audio"]["path"]
```
Finden Sie ein [Audioklassifikation](https://huggingface.co/models?pipeline_tag=audio-classification) Modell auf dem Model Hub für Emotionserkennung und laden Sie es in die [`pipeline`]:
```py
>>> from transformers import pipeline
>>> audio_classifier = pipeline(
... task="audio-classification", model="ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition"
... )
```
Übergeben Sie die Audiodatei an die [`pipeline`]:
```py
>>> preds = audio_classifier(audio_file)
>>> preds = [{"score": round(pred["score"], 4), "label": pred["label"]} for pred in preds]
>>> preds
[{'score': 0.1315, 'label': 'calm'}, {'score': 0.1307, 'label': 'neutral'}, {'score': 0.1274, 'label': 'sad'}, {'score': 0.1261, 'label': 'fearful'}, {'score': 0.1242, 'label': 'happy'}]
```
## Bildverarbeitungs-Pipeline
Die Verwendung einer [`pipeline`] für Bildverarbeitungsaufgaben ist praktisch identisch.
Geben Sie Ihre Aufgabe an und übergeben Sie Ihr Bild an den Klassifikator. Das Bild kann ein Link oder ein lokaler Pfad zu dem Bild sein. Zum Beispiel: Welche Katzenart ist unten abgebildet?
```py
>>> from transformers import pipeline
>>> vision_classifier = pipeline(task="image-classification")
>>> preds = vision_classifier(
... images="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
... )
>>> preds = [{"score": round(pred["score"], 4), "label": pred["label"]} for pred in preds]
>>> preds
[{'score': 0.4335, 'label': 'lynx, catamount'}, {'score': 0.0348, 'label': 'cougar, puma, catamount, mountain lion, painter, panther, Felis concolor'}, {'score': 0.0324, 'label': 'snow leopard, ounce, Panthera uncia'}, {'score': 0.0239, 'label': 'Egyptian cat'}, {'score': 0.0229, 'label': 'tiger cat'}]
```
## Multimodale Pipeline
Die [`pipeline`] unterstützt mehr als eine Modalität. Eine Aufgabe zur Beantwortung visueller Fragen (VQA) kombiniert zum Beispiel Text und Bild. Verwenden Sie einen beliebigen Bildlink und eine Frage, die Sie zu dem Bild stellen möchten. Das Bild kann eine URL oder ein lokaler Pfad zu dem Bild sein.
Wenn Sie zum Beispiel das gleiche Bild wie in der obigen Vision-Pipeline verwenden:
```py
>>> image = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
>>> question = "Where is the cat?"
```
Erstellen Sie eine Pipeline für "vqa" und übergeben Sie ihr das Bild und die Frage:
```py
>>> from transformers import pipeline
>>> vqa = pipeline(task="vqa")
>>> preds = vqa(image=image, question=question)
>>> preds = [{"score": round(pred["score"], 4), "answer": pred["answer"]} for pred in preds]
>>> preds
[{'score': 0.9112, 'answer': 'snow'}, {'score': 0.8796, 'answer': 'in snow'}, {'score': 0.6717, 'answer': 'outside'}, {'score': 0.0291, 'answer': 'on ground'}, {'score': 0.027, 'answer': 'ground'}]
```
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/robust-speech-event/eval.py | #!/usr/bin/env python3
import argparse
import re
from typing import Dict
import torch
from datasets import Audio, Dataset, load_dataset, load_metric
from transformers import AutoFeatureExtractor, pipeline
def log_results(result: Dataset, args: Dict[str, str]):
"""DO NOT CHANGE. This function computes and logs the result metrics."""
log_outputs = args.log_outputs
dataset_id = "_".join(args.dataset.split("/") + [args.config, args.split])
# load metric
wer = load_metric("wer")
cer = load_metric("cer")
# compute metrics
wer_result = wer.compute(references=result["target"], predictions=result["prediction"])
cer_result = cer.compute(references=result["target"], predictions=result["prediction"])
# print & log results
result_str = f"WER: {wer_result}\nCER: {cer_result}"
print(result_str)
with open(f"{dataset_id}_eval_results.txt", "w") as f:
f.write(result_str)
# log all results in text file. Possibly interesting for analysis
if log_outputs is not None:
pred_file = f"log_{dataset_id}_predictions.txt"
target_file = f"log_{dataset_id}_targets.txt"
with open(pred_file, "w") as p, open(target_file, "w") as t:
# mapping function to write output
def write_to_file(batch, i):
p.write(f"{i}" + "\n")
p.write(batch["prediction"] + "\n")
t.write(f"{i}" + "\n")
t.write(batch["target"] + "\n")
result.map(write_to_file, with_indices=True)
def normalize_text(text: str) -> str:
"""DO ADAPT FOR YOUR USE CASE. this function normalizes the target text."""
chars_to_ignore_regex = '[,?.!\-\;\:"“%‘”�—’…–]' # noqa: W605 IMPORTANT: this should correspond to the chars that were ignored during training
text = re.sub(chars_to_ignore_regex, "", text.lower())
# In addition, we can normalize the target text, e.g. removing new lines characters etc...
# note that order is important here!
token_sequences_to_ignore = ["\n\n", "\n", " ", " "]
for t in token_sequences_to_ignore:
text = " ".join(text.split(t))
return text
def main(args):
# load dataset
dataset = load_dataset(args.dataset, args.config, split=args.split, use_auth_token=True)
# for testing: only process the first two examples as a test
# dataset = dataset.select(range(10))
# load processor
feature_extractor = AutoFeatureExtractor.from_pretrained(args.model_id)
sampling_rate = feature_extractor.sampling_rate
# resample audio
dataset = dataset.cast_column("audio", Audio(sampling_rate=sampling_rate))
# load eval pipeline
if args.device is None:
args.device = 0 if torch.cuda.is_available() else -1
asr = pipeline("automatic-speech-recognition", model=args.model_id, device=args.device)
# map function to decode audio
def map_to_pred(batch):
prediction = asr(
batch["audio"]["array"], chunk_length_s=args.chunk_length_s, stride_length_s=args.stride_length_s
)
batch["prediction"] = prediction["text"]
batch["target"] = normalize_text(batch["sentence"])
return batch
# run inference on all examples
result = dataset.map(map_to_pred, remove_columns=dataset.column_names)
# compute and log_results
# do not change function below
log_results(result, args)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--model_id", type=str, required=True, help="Model identifier. Should be loadable with 🤗 Transformers"
)
parser.add_argument(
"--dataset",
type=str,
required=True,
help="Dataset name to evaluate the `model_id`. Should be loadable with 🤗 Datasets",
)
parser.add_argument(
"--config", type=str, required=True, help="Config of the dataset. *E.g.* `'en'` for Common Voice"
)
parser.add_argument("--split", type=str, required=True, help="Split of the dataset. *E.g.* `'test'`")
parser.add_argument(
"--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
)
parser.add_argument(
"--stride_length_s", type=float, default=None, help="Stride of the audio chunks. Defaults to 1 second."
)
parser.add_argument(
"--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
)
parser.add_argument(
"--device",
type=int,
default=None,
help="The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.",
)
args = parser.parse_args()
main(args)
| #!/usr/bin/env python3
import argparse
import re
from typing import Dict
import torch
from datasets import Audio, Dataset, load_dataset, load_metric
from transformers import AutoFeatureExtractor, pipeline
def log_results(result: Dataset, args: Dict[str, str]):
"""DO NOT CHANGE. This function computes and logs the result metrics."""
log_outputs = args.log_outputs
dataset_id = "_".join(args.dataset.split("/") + [args.config, args.split])
# load metric
wer = load_metric("wer")
cer = load_metric("cer")
# compute metrics
wer_result = wer.compute(references=result["target"], predictions=result["prediction"])
cer_result = cer.compute(references=result["target"], predictions=result["prediction"])
# print & log results
result_str = f"WER: {wer_result}\nCER: {cer_result}"
print(result_str)
with open(f"{dataset_id}_eval_results.txt", "w") as f:
f.write(result_str)
# log all results in text file. Possibly interesting for analysis
if log_outputs is not None:
pred_file = f"log_{dataset_id}_predictions.txt"
target_file = f"log_{dataset_id}_targets.txt"
with open(pred_file, "w") as p, open(target_file, "w") as t:
# mapping function to write output
def write_to_file(batch, i):
p.write(f"{i}" + "\n")
p.write(batch["prediction"] + "\n")
t.write(f"{i}" + "\n")
t.write(batch["target"] + "\n")
result.map(write_to_file, with_indices=True)
def normalize_text(text: str) -> str:
"""DO ADAPT FOR YOUR USE CASE. this function normalizes the target text."""
chars_to_ignore_regex = '[,?.!\-\;\:"“%‘”�—’…–]' # noqa: W605 IMPORTANT: this should correspond to the chars that were ignored during training
text = re.sub(chars_to_ignore_regex, "", text.lower())
# In addition, we can normalize the target text, e.g. removing new lines characters etc...
# note that order is important here!
token_sequences_to_ignore = ["\n\n", "\n", " ", " "]
for t in token_sequences_to_ignore:
text = " ".join(text.split(t))
return text
def main(args):
# load dataset
dataset = load_dataset(args.dataset, args.config, split=args.split, use_auth_token=True)
# for testing: only process the first two examples as a test
# dataset = dataset.select(range(10))
# load processor
feature_extractor = AutoFeatureExtractor.from_pretrained(args.model_id)
sampling_rate = feature_extractor.sampling_rate
# resample audio
dataset = dataset.cast_column("audio", Audio(sampling_rate=sampling_rate))
# load eval pipeline
if args.device is None:
args.device = 0 if torch.cuda.is_available() else -1
asr = pipeline("automatic-speech-recognition", model=args.model_id, device=args.device)
# map function to decode audio
def map_to_pred(batch):
prediction = asr(
batch["audio"]["array"], chunk_length_s=args.chunk_length_s, stride_length_s=args.stride_length_s
)
batch["prediction"] = prediction["text"]
batch["target"] = normalize_text(batch["sentence"])
return batch
# run inference on all examples
result = dataset.map(map_to_pred, remove_columns=dataset.column_names)
# compute and log_results
# do not change function below
log_results(result, args)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--model_id", type=str, required=True, help="Model identifier. Should be loadable with 🤗 Transformers"
)
parser.add_argument(
"--dataset",
type=str,
required=True,
help="Dataset name to evaluate the `model_id`. Should be loadable with 🤗 Datasets",
)
parser.add_argument(
"--config", type=str, required=True, help="Config of the dataset. *E.g.* `'en'` for Common Voice"
)
parser.add_argument("--split", type=str, required=True, help="Split of the dataset. *E.g.* `'test'`")
parser.add_argument(
"--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
)
parser.add_argument(
"--stride_length_s", type=float, default=None, help="Stride of the audio chunks. Defaults to 1 second."
)
parser.add_argument(
"--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
)
parser.add_argument(
"--device",
type=int,
default=None,
help="The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.",
)
args = parser.parse_args()
main(args)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/cvt/modeling_tf_cvt.py | # coding=utf-8
# Copyright 2022 Microsoft Research and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" TF 2.0 Cvt model."""
import collections.abc
from dataclasses import dataclass
from typing import Dict, Optional, Tuple, Union
import tensorflow as tf
from ...modeling_tf_outputs import TFImageClassifierOutputWithNoAttention
from ...modeling_tf_utils import (
TFModelInputType,
TFPreTrainedModel,
TFSequenceClassificationLoss,
get_initializer,
keras_serializable,
unpack_inputs,
)
from ...tf_utils import shape_list, stable_softmax
from ...utils import (
ModelOutput,
add_start_docstrings,
add_start_docstrings_to_model_forward,
logging,
replace_return_docstrings,
)
from .configuration_cvt import CvtConfig
logger = logging.get_logger(__name__)
# General docstring
_CONFIG_FOR_DOC = "CvtConfig"
TF_CVT_PRETRAINED_MODEL_ARCHIVE_LIST = [
"microsoft/cvt-13",
"microsoft/cvt-13-384",
"microsoft/cvt-13-384-22k",
"microsoft/cvt-21",
"microsoft/cvt-21-384",
"microsoft/cvt-21-384-22k",
# See all Cvt models at https://huggingface.co/models?filter=cvt
]
@dataclass
class TFBaseModelOutputWithCLSToken(ModelOutput):
"""
Base class for model's outputs.
Args:
last_hidden_state (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
cls_token_value (`tf.Tensor` of shape `(batch_size, 1, hidden_size)`):
Classification token at the output of the last layer of the model.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of shape
`(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus
the initial embedding outputs.
"""
last_hidden_state: tf.Tensor = None
cls_token_value: tf.Tensor = None
hidden_states: Optional[Tuple[tf.Tensor]] = None
class TFCvtDropPath(tf.keras.layers.Layer):
"""Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
References:
(1) github.com:rwightman/pytorch-image-models
"""
def __init__(self, drop_prob: float, **kwargs):
super().__init__(**kwargs)
self.drop_prob = drop_prob
def call(self, x: tf.Tensor, training=None):
if self.drop_prob == 0.0 or not training:
return x
keep_prob = 1 - self.drop_prob
shape = (tf.shape(x)[0],) + (1,) * (len(tf.shape(x)) - 1)
random_tensor = keep_prob + tf.random.uniform(shape, 0, 1)
random_tensor = tf.floor(random_tensor)
return (x / keep_prob) * random_tensor
class TFCvtEmbeddings(tf.keras.layers.Layer):
"""Construct the Convolutional Token Embeddings."""
def __init__(
self,
config: CvtConfig,
patch_size: int,
embed_dim: int,
stride: int,
padding: int,
dropout_rate: float,
**kwargs,
):
super().__init__(**kwargs)
self.convolution_embeddings = TFCvtConvEmbeddings(
config,
patch_size=patch_size,
embed_dim=embed_dim,
stride=stride,
padding=padding,
name="convolution_embeddings",
)
self.dropout = tf.keras.layers.Dropout(dropout_rate)
def call(self, pixel_values: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.convolution_embeddings(pixel_values)
hidden_state = self.dropout(hidden_state, training=training)
return hidden_state
class TFCvtConvEmbeddings(tf.keras.layers.Layer):
"""Image to Convolution Embeddings. This convolutional operation aims to model local spatial contexts."""
def __init__(self, config: CvtConfig, patch_size: int, embed_dim: int, stride: int, padding: int, **kwargs):
super().__init__(**kwargs)
self.padding = tf.keras.layers.ZeroPadding2D(padding=padding)
self.patch_size = patch_size if isinstance(patch_size, collections.abc.Iterable) else (patch_size, patch_size)
self.projection = tf.keras.layers.Conv2D(
filters=embed_dim,
kernel_size=patch_size,
strides=stride,
padding="valid",
data_format="channels_last",
kernel_initializer=get_initializer(config.initializer_range),
name="projection",
)
# Using the same default epsilon as PyTorch
self.normalization = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="normalization")
def call(self, pixel_values: tf.Tensor) -> tf.Tensor:
if isinstance(pixel_values, dict):
pixel_values = pixel_values["pixel_values"]
pixel_values = self.projection(self.padding(pixel_values))
# "batch_size, height, width, num_channels -> batch_size, (height*width), num_channels"
batch_size, height, width, num_channels = shape_list(pixel_values)
hidden_size = height * width
pixel_values = tf.reshape(pixel_values, shape=(batch_size, hidden_size, num_channels))
pixel_values = self.normalization(pixel_values)
# "batch_size, (height*width), num_channels -> batch_size, height, width, num_channels"
pixel_values = tf.reshape(pixel_values, shape=(batch_size, height, width, num_channels))
return pixel_values
class TFCvtSelfAttentionConvProjection(tf.keras.layers.Layer):
"""Convolutional projection layer."""
def __init__(self, config: CvtConfig, embed_dim: int, kernel_size: int, stride: int, padding: int, **kwargs):
super().__init__(**kwargs)
self.padding = tf.keras.layers.ZeroPadding2D(padding=padding)
self.convolution = tf.keras.layers.Conv2D(
filters=embed_dim,
kernel_size=kernel_size,
kernel_initializer=get_initializer(config.initializer_range),
padding="valid",
strides=stride,
use_bias=False,
name="convolution",
groups=embed_dim,
)
# Using the same default epsilon as PyTorch, TF uses (1 - pytorch momentum)
self.normalization = tf.keras.layers.BatchNormalization(epsilon=1e-5, momentum=0.9, name="normalization")
def call(self, hidden_state: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.convolution(self.padding(hidden_state))
hidden_state = self.normalization(hidden_state, training=training)
return hidden_state
class TFCvtSelfAttentionLinearProjection(tf.keras.layers.Layer):
"""Linear projection layer used to flatten tokens into 1D."""
def call(self, hidden_state: tf.Tensor) -> tf.Tensor:
# "batch_size, height, width, num_channels -> batch_size, (height*width), num_channels"
batch_size, height, width, num_channels = shape_list(hidden_state)
hidden_size = height * width
hidden_state = tf.reshape(hidden_state, shape=(batch_size, hidden_size, num_channels))
return hidden_state
class TFCvtSelfAttentionProjection(tf.keras.layers.Layer):
"""Convolutional Projection for Attention."""
def __init__(
self,
config: CvtConfig,
embed_dim: int,
kernel_size: int,
stride: int,
padding: int,
projection_method: str = "dw_bn",
**kwargs,
):
super().__init__(**kwargs)
if projection_method == "dw_bn":
self.convolution_projection = TFCvtSelfAttentionConvProjection(
config, embed_dim, kernel_size, stride, padding, name="convolution_projection"
)
self.linear_projection = TFCvtSelfAttentionLinearProjection()
def call(self, hidden_state: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.convolution_projection(hidden_state, training=training)
hidden_state = self.linear_projection(hidden_state)
return hidden_state
class TFCvtSelfAttention(tf.keras.layers.Layer):
"""
Self-attention layer. A depth-wise separable convolution operation (Convolutional Projection), is applied for
query, key, and value embeddings.
"""
def __init__(
self,
config: CvtConfig,
num_heads: int,
embed_dim: int,
kernel_size: int,
stride_q: int,
stride_kv: int,
padding_q: int,
padding_kv: int,
qkv_projection_method: str,
qkv_bias: bool,
attention_drop_rate: float,
with_cls_token: bool = True,
**kwargs,
):
super().__init__(**kwargs)
self.scale = embed_dim**-0.5
self.with_cls_token = with_cls_token
self.embed_dim = embed_dim
self.num_heads = num_heads
self.convolution_projection_query = TFCvtSelfAttentionProjection(
config,
embed_dim,
kernel_size,
stride_q,
padding_q,
projection_method="linear" if qkv_projection_method == "avg" else qkv_projection_method,
name="convolution_projection_query",
)
self.convolution_projection_key = TFCvtSelfAttentionProjection(
config,
embed_dim,
kernel_size,
stride_kv,
padding_kv,
projection_method=qkv_projection_method,
name="convolution_projection_key",
)
self.convolution_projection_value = TFCvtSelfAttentionProjection(
config,
embed_dim,
kernel_size,
stride_kv,
padding_kv,
projection_method=qkv_projection_method,
name="convolution_projection_value",
)
self.projection_query = tf.keras.layers.Dense(
units=embed_dim,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=qkv_bias,
bias_initializer="zeros",
name="projection_query",
)
self.projection_key = tf.keras.layers.Dense(
units=embed_dim,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=qkv_bias,
bias_initializer="zeros",
name="projection_key",
)
self.projection_value = tf.keras.layers.Dense(
units=embed_dim,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=qkv_bias,
bias_initializer="zeros",
name="projection_value",
)
self.dropout = tf.keras.layers.Dropout(attention_drop_rate)
def rearrange_for_multi_head_attention(self, hidden_state: tf.Tensor) -> tf.Tensor:
batch_size, hidden_size, _ = shape_list(hidden_state)
head_dim = self.embed_dim // self.num_heads
hidden_state = tf.reshape(hidden_state, shape=(batch_size, hidden_size, self.num_heads, head_dim))
hidden_state = tf.transpose(hidden_state, perm=(0, 2, 1, 3))
return hidden_state
def call(self, hidden_state: tf.Tensor, height: int, width: int, training: bool = False) -> tf.Tensor:
if self.with_cls_token:
cls_token, hidden_state = tf.split(hidden_state, [1, height * width], 1)
# "batch_size, (height*width), num_channels -> batch_size, height, width, num_channels"
batch_size, hidden_size, num_channels = shape_list(hidden_state)
hidden_state = tf.reshape(hidden_state, shape=(batch_size, height, width, num_channels))
key = self.convolution_projection_key(hidden_state, training=training)
query = self.convolution_projection_query(hidden_state, training=training)
value = self.convolution_projection_value(hidden_state, training=training)
if self.with_cls_token:
query = tf.concat((cls_token, query), axis=1)
key = tf.concat((cls_token, key), axis=1)
value = tf.concat((cls_token, value), axis=1)
head_dim = self.embed_dim // self.num_heads
query = self.rearrange_for_multi_head_attention(self.projection_query(query))
key = self.rearrange_for_multi_head_attention(self.projection_key(key))
value = self.rearrange_for_multi_head_attention(self.projection_value(value))
attention_score = tf.matmul(query, key, transpose_b=True) * self.scale
attention_probs = stable_softmax(logits=attention_score, axis=-1)
attention_probs = self.dropout(attention_probs, training=training)
context = tf.matmul(attention_probs, value)
# "batch_size, num_heads, hidden_size, head_dim -> batch_size, hidden_size, (num_heads*head_dim)"
_, _, hidden_size, _ = shape_list(context)
context = tf.transpose(context, perm=(0, 2, 1, 3))
context = tf.reshape(context, (batch_size, hidden_size, self.num_heads * head_dim))
return context
class TFCvtSelfOutput(tf.keras.layers.Layer):
"""Output of the Attention layer ."""
def __init__(self, config: CvtConfig, embed_dim: int, drop_rate: float, **kwargs):
super().__init__(**kwargs)
self.dense = tf.keras.layers.Dense(
units=embed_dim, kernel_initializer=get_initializer(config.initializer_range), name="dense"
)
self.dropout = tf.keras.layers.Dropout(drop_rate)
def call(self, hidden_state: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.dense(inputs=hidden_state)
hidden_state = self.dropout(inputs=hidden_state, training=training)
return hidden_state
class TFCvtAttention(tf.keras.layers.Layer):
"""Attention layer. First chunk of the convolutional transformer block."""
def __init__(
self,
config: CvtConfig,
num_heads: int,
embed_dim: int,
kernel_size: int,
stride_q: int,
stride_kv: int,
padding_q: int,
padding_kv: int,
qkv_projection_method: str,
qkv_bias: bool,
attention_drop_rate: float,
drop_rate: float,
with_cls_token: bool = True,
**kwargs,
):
super().__init__(**kwargs)
self.attention = TFCvtSelfAttention(
config,
num_heads,
embed_dim,
kernel_size,
stride_q,
stride_kv,
padding_q,
padding_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
with_cls_token,
name="attention",
)
self.dense_output = TFCvtSelfOutput(config, embed_dim, drop_rate, name="output")
def prune_heads(self, heads):
raise NotImplementedError
def call(self, hidden_state: tf.Tensor, height: int, width: int, training: bool = False):
self_output = self.attention(hidden_state, height, width, training=training)
attention_output = self.dense_output(self_output, training=training)
return attention_output
class TFCvtIntermediate(tf.keras.layers.Layer):
"""Intermediate dense layer. Second chunk of the convolutional transformer block."""
def __init__(self, config: CvtConfig, embed_dim: int, mlp_ratio: int, **kwargs):
super().__init__(**kwargs)
self.dense = tf.keras.layers.Dense(
units=int(embed_dim * mlp_ratio),
kernel_initializer=get_initializer(config.initializer_range),
activation="gelu",
name="dense",
)
def call(self, hidden_state: tf.Tensor) -> tf.Tensor:
hidden_state = self.dense(hidden_state)
return hidden_state
class TFCvtOutput(tf.keras.layers.Layer):
"""
Output of the Convolutional Transformer Block (last chunk). It consists of a MLP and a residual connection.
"""
def __init__(self, config: CvtConfig, embed_dim: int, drop_rate: int, **kwargs):
super().__init__(**kwargs)
self.dense = tf.keras.layers.Dense(
units=embed_dim, kernel_initializer=get_initializer(config.initializer_range), name="dense"
)
self.dropout = tf.keras.layers.Dropout(drop_rate)
def call(self, hidden_state: tf.Tensor, input_tensor: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.dense(inputs=hidden_state)
hidden_state = self.dropout(inputs=hidden_state, training=training)
hidden_state = hidden_state + input_tensor
return hidden_state
class TFCvtLayer(tf.keras.layers.Layer):
"""
Convolutional Transformer Block composed by attention layers, normalization and multi-layer perceptrons (mlps). It
consists of 3 chunks : an attention layer, an intermediate dense layer and an output layer. This corresponds to the
`Block` class in the original implementation.
"""
def __init__(
self,
config: CvtConfig,
num_heads: int,
embed_dim: int,
kernel_size: int,
stride_q: int,
stride_kv: int,
padding_q: int,
padding_kv: int,
qkv_projection_method: str,
qkv_bias: bool,
attention_drop_rate: float,
drop_rate: float,
mlp_ratio: float,
drop_path_rate: float,
with_cls_token: bool = True,
**kwargs,
):
super().__init__(**kwargs)
self.attention = TFCvtAttention(
config,
num_heads,
embed_dim,
kernel_size,
stride_q,
stride_kv,
padding_q,
padding_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
with_cls_token,
name="attention",
)
self.intermediate = TFCvtIntermediate(config, embed_dim, mlp_ratio, name="intermediate")
self.dense_output = TFCvtOutput(config, embed_dim, drop_rate, name="output")
# Using `layers.Activation` instead of `tf.identity` to better control `training` behaviour.
self.drop_path = (
TFCvtDropPath(drop_path_rate, name="drop_path")
if drop_path_rate > 0.0
else tf.keras.layers.Activation("linear", name="drop_path")
)
# Using the same default epsilon as PyTorch
self.layernorm_before = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="layernorm_before")
self.layernorm_after = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="layernorm_after")
def call(self, hidden_state: tf.Tensor, height: int, width: int, training: bool = False) -> tf.Tensor:
# in Cvt, layernorm is applied before self-attention
attention_output = self.attention(self.layernorm_before(hidden_state), height, width, training=training)
attention_output = self.drop_path(attention_output, training=training)
# first residual connection
hidden_state = attention_output + hidden_state
# in Cvt, layernorm is also applied after self-attention
layer_output = self.layernorm_after(hidden_state)
layer_output = self.intermediate(layer_output)
# second residual connection is done here
layer_output = self.dense_output(layer_output, hidden_state)
layer_output = self.drop_path(layer_output, training=training)
return layer_output
class TFCvtStage(tf.keras.layers.Layer):
"""
Cvt stage (encoder block). Each stage has 2 parts :
- (1) A Convolutional Token Embedding layer
- (2) A Convolutional Transformer Block (layer).
The classification token is added only in the last stage.
Args:
config ([`CvtConfig`]): Model configuration class.
stage (`int`): Stage number.
"""
def __init__(self, config: CvtConfig, stage: int, **kwargs):
super().__init__(**kwargs)
self.config = config
self.stage = stage
if self.config.cls_token[self.stage]:
self.cls_token = self.add_weight(
shape=(1, 1, self.config.embed_dim[-1]),
initializer=get_initializer(self.config.initializer_range),
trainable=True,
name="cvt.encoder.stages.2.cls_token",
)
self.embedding = TFCvtEmbeddings(
self.config,
patch_size=config.patch_sizes[self.stage],
stride=config.patch_stride[self.stage],
embed_dim=config.embed_dim[self.stage],
padding=config.patch_padding[self.stage],
dropout_rate=config.drop_rate[self.stage],
name="embedding",
)
drop_path_rates = tf.linspace(0.0, config.drop_path_rate[self.stage], config.depth[stage])
drop_path_rates = [x.numpy().item() for x in drop_path_rates]
self.layers = [
TFCvtLayer(
config,
num_heads=config.num_heads[self.stage],
embed_dim=config.embed_dim[self.stage],
kernel_size=config.kernel_qkv[self.stage],
stride_q=config.stride_q[self.stage],
stride_kv=config.stride_kv[self.stage],
padding_q=config.padding_q[self.stage],
padding_kv=config.padding_kv[self.stage],
qkv_projection_method=config.qkv_projection_method[self.stage],
qkv_bias=config.qkv_bias[self.stage],
attention_drop_rate=config.attention_drop_rate[self.stage],
drop_rate=config.drop_rate[self.stage],
mlp_ratio=config.mlp_ratio[self.stage],
drop_path_rate=drop_path_rates[self.stage],
with_cls_token=config.cls_token[self.stage],
name=f"layers.{j}",
)
for j in range(config.depth[self.stage])
]
def call(self, hidden_state: tf.Tensor, training: bool = False):
cls_token = None
hidden_state = self.embedding(hidden_state, training)
# "batch_size, height, width, num_channels -> batch_size, (height*width), num_channels"
batch_size, height, width, num_channels = shape_list(hidden_state)
hidden_size = height * width
hidden_state = tf.reshape(hidden_state, shape=(batch_size, hidden_size, num_channels))
if self.config.cls_token[self.stage]:
cls_token = tf.repeat(self.cls_token, repeats=batch_size, axis=0)
hidden_state = tf.concat((cls_token, hidden_state), axis=1)
for layer in self.layers:
layer_outputs = layer(hidden_state, height, width, training=training)
hidden_state = layer_outputs
if self.config.cls_token[self.stage]:
cls_token, hidden_state = tf.split(hidden_state, [1, height * width], 1)
# "batch_size, (height*width), num_channels -> batch_size, height, width, num_channels"
hidden_state = tf.reshape(hidden_state, shape=(batch_size, height, width, num_channels))
return hidden_state, cls_token
class TFCvtEncoder(tf.keras.layers.Layer):
"""
Convolutional Vision Transformer encoder. CVT has 3 stages of encoder blocks with their respective number of layers
(depth) being 1, 2 and 10.
Args:
config ([`CvtConfig`]): Model configuration class.
"""
config_class = CvtConfig
def __init__(self, config: CvtConfig, **kwargs):
super().__init__(**kwargs)
self.config = config
self.stages = [
TFCvtStage(config, stage_idx, name=f"stages.{stage_idx}") for stage_idx in range(len(config.depth))
]
def call(
self,
pixel_values: TFModelInputType,
output_hidden_states: Optional[bool] = False,
return_dict: Optional[bool] = True,
training: Optional[bool] = False,
) -> Union[TFBaseModelOutputWithCLSToken, Tuple[tf.Tensor]]:
all_hidden_states = () if output_hidden_states else None
hidden_state = pixel_values
# When running on CPU, `tf.keras.layers.Conv2D` doesn't support (batch_size, num_channels, height, width)
# as input format. So change the input format to (batch_size, height, width, num_channels).
hidden_state = tf.transpose(hidden_state, perm=(0, 2, 3, 1))
cls_token = None
for _, (stage_module) in enumerate(self.stages):
hidden_state, cls_token = stage_module(hidden_state, training=training)
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_state,)
# Change back to (batch_size, num_channels, height, width) format to have uniformity in the modules
hidden_state = tf.transpose(hidden_state, perm=(0, 3, 1, 2))
if output_hidden_states:
all_hidden_states = tuple([tf.transpose(hs, perm=(0, 3, 1, 2)) for hs in all_hidden_states])
if not return_dict:
return tuple(v for v in [hidden_state, cls_token, all_hidden_states] if v is not None)
return TFBaseModelOutputWithCLSToken(
last_hidden_state=hidden_state,
cls_token_value=cls_token,
hidden_states=all_hidden_states,
)
@keras_serializable
class TFCvtMainLayer(tf.keras.layers.Layer):
"""Construct the Cvt model."""
config_class = CvtConfig
def __init__(self, config: CvtConfig, **kwargs):
super().__init__(**kwargs)
self.config = config
self.encoder = TFCvtEncoder(config, name="encoder")
@unpack_inputs
def call(
self,
pixel_values: Optional[TFModelInputType] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
training: Optional[bool] = False,
) -> Union[TFBaseModelOutputWithCLSToken, Tuple[tf.Tensor]]:
if pixel_values is None:
raise ValueError("You have to specify pixel_values")
encoder_outputs = self.encoder(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
training=training,
)
sequence_output = encoder_outputs[0]
if not return_dict:
return (sequence_output,) + encoder_outputs[1:]
return TFBaseModelOutputWithCLSToken(
last_hidden_state=sequence_output,
cls_token_value=encoder_outputs.cls_token_value,
hidden_states=encoder_outputs.hidden_states,
)
class TFCvtPreTrainedModel(TFPreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = CvtConfig
base_model_prefix = "cvt"
main_input_name = "pixel_values"
@property
def dummy_inputs(self) -> Dict[str, tf.Tensor]:
"""
Dummy inputs to build the network.
Returns:
`Dict[str, tf.Tensor]`: The dummy inputs.
"""
VISION_DUMMY_INPUTS = tf.random.uniform(shape=(3, self.config.num_channels, 224, 224), dtype=tf.float32)
return {"pixel_values": tf.constant(VISION_DUMMY_INPUTS)}
@tf.function(
input_signature=[
{
"pixel_values": tf.TensorSpec((None, None, None, None), tf.float32, name="pixel_values"),
}
]
)
def serving(self, inputs):
"""
Method used for serving the model.
Args:
inputs (`Dict[str, tf.Tensor]`):
The input of the saved model as a dictionary of tensors.
"""
output = self.call(inputs)
return self.serving_output(output)
TFCVT_START_DOCSTRING = r"""
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use it
as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and
behavior.
<Tip>
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all the
tensors in the first argument of the model call function: `model(inputs)`.
</Tip>
Args:
config ([`CvtConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~TFPreTrainedModel.from_pretrained`] method to load the model weights.
"""
TFCVT_INPUTS_DOCSTRING = r"""
Args:
pixel_values (`np.ndarray`, `tf.Tensor`, `List[tf.Tensor]` ``Dict[str, tf.Tensor]` or `Dict[str, np.ndarray]` and each example must have the shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See [`CvtImageProcessor.__call__`]
for details.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. This argument can be used in
eager mode, in graph mode the value will always be set to True.
training (`bool`, *optional*, defaults to `False``):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
@add_start_docstrings(
"The bare Cvt Model transformer outputting raw hidden-states without any specific head on top.",
TFCVT_START_DOCSTRING,
)
class TFCvtModel(TFCvtPreTrainedModel):
def __init__(self, config: CvtConfig, *inputs, **kwargs):
super().__init__(config, *inputs, **kwargs)
self.cvt = TFCvtMainLayer(config, name="cvt")
@unpack_inputs
@add_start_docstrings_to_model_forward(TFCVT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=TFBaseModelOutputWithCLSToken, config_class=_CONFIG_FOR_DOC)
def call(
self,
pixel_values: Optional[tf.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
training: Optional[bool] = False,
) -> Union[TFBaseModelOutputWithCLSToken, Tuple[tf.Tensor]]:
r"""
Returns:
Examples:
```python
>>> from transformers import AutoImageProcessor, TFCvtModel
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> image_processor = AutoImageProcessor.from_pretrained("microsoft/cvt-13")
>>> model = TFCvtModel.from_pretrained("microsoft/cvt-13")
>>> inputs = image_processor(images=image, return_tensors="tf")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
```"""
if pixel_values is None:
raise ValueError("You have to specify pixel_values")
outputs = self.cvt(
pixel_values=pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
training=training,
)
if not return_dict:
return (outputs[0],) + outputs[1:]
return TFBaseModelOutputWithCLSToken(
last_hidden_state=outputs.last_hidden_state,
cls_token_value=outputs.cls_token_value,
hidden_states=outputs.hidden_states,
)
def serving_output(self, output: TFBaseModelOutputWithCLSToken) -> TFBaseModelOutputWithCLSToken:
return TFBaseModelOutputWithCLSToken(
last_hidden_state=output.last_hidden_state,
cls_token_value=output.cls_token_value,
hidden_states=output.hidden_states,
)
@add_start_docstrings(
"""
Cvt Model transformer with an image classification head on top (a linear layer on top of the final hidden state of
the [CLS] token) e.g. for ImageNet.
""",
TFCVT_START_DOCSTRING,
)
class TFCvtForImageClassification(TFCvtPreTrainedModel, TFSequenceClassificationLoss):
def __init__(self, config: CvtConfig, *inputs, **kwargs):
super().__init__(config, *inputs, **kwargs)
self.num_labels = config.num_labels
self.cvt = TFCvtMainLayer(config, name="cvt")
# Using same default epsilon as in the original implementation.
self.layernorm = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="layernorm")
# Classifier head
self.classifier = tf.keras.layers.Dense(
units=config.num_labels,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=True,
bias_initializer="zeros",
name="classifier",
)
@unpack_inputs
@add_start_docstrings_to_model_forward(TFCVT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=TFImageClassifierOutputWithNoAttention, config_class=_CONFIG_FOR_DOC)
def call(
self,
pixel_values: Optional[tf.Tensor] = None,
labels: Optional[tf.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
training: Optional[bool] = False,
) -> Union[TFImageClassifierOutputWithNoAttention, Tuple[tf.Tensor]]:
r"""
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size,)`, *optional*):
Labels for computing the image classification/regression loss. Indices should be in `[0, ...,
config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
Returns:
Examples:
```python
>>> from transformers import AutoImageProcessor, TFCvtForImageClassification
>>> import tensorflow as tf
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> image_processor = AutoImageProcessor.from_pretrained("microsoft/cvt-13")
>>> model = TFCvtForImageClassification.from_pretrained("microsoft/cvt-13")
>>> inputs = image_processor(images=image, return_tensors="tf")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_class_idx = tf.math.argmax(logits, axis=-1)[0]
>>> print("Predicted class:", model.config.id2label[int(predicted_class_idx)])
```"""
outputs = self.cvt(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
training=training,
)
sequence_output = outputs[0]
cls_token = outputs[1]
if self.config.cls_token[-1]:
sequence_output = self.layernorm(cls_token)
else:
# rearrange "batch_size, num_channels, height, width -> batch_size, (height*width), num_channels"
batch_size, num_channels, height, width = shape_list(sequence_output)
sequence_output = tf.reshape(sequence_output, shape=(batch_size, num_channels, height * width))
sequence_output = tf.transpose(sequence_output, perm=(0, 2, 1))
sequence_output = self.layernorm(sequence_output)
sequence_output_mean = tf.reduce_mean(sequence_output, axis=1)
logits = self.classifier(sequence_output_mean)
loss = None if labels is None else self.hf_compute_loss(labels=labels, logits=logits)
if not return_dict:
output = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return TFImageClassifierOutputWithNoAttention(loss=loss, logits=logits, hidden_states=outputs.hidden_states)
def serving_output(self, output: TFImageClassifierOutputWithNoAttention) -> TFImageClassifierOutputWithNoAttention:
return TFImageClassifierOutputWithNoAttention(logits=output.logits, hidden_states=output.hidden_states)
| # coding=utf-8
# Copyright 2022 Microsoft Research and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" TF 2.0 Cvt model."""
import collections.abc
from dataclasses import dataclass
from typing import Dict, Optional, Tuple, Union
import tensorflow as tf
from ...modeling_tf_outputs import TFImageClassifierOutputWithNoAttention
from ...modeling_tf_utils import (
TFModelInputType,
TFPreTrainedModel,
TFSequenceClassificationLoss,
get_initializer,
keras_serializable,
unpack_inputs,
)
from ...tf_utils import shape_list, stable_softmax
from ...utils import (
ModelOutput,
add_start_docstrings,
add_start_docstrings_to_model_forward,
logging,
replace_return_docstrings,
)
from .configuration_cvt import CvtConfig
logger = logging.get_logger(__name__)
# General docstring
_CONFIG_FOR_DOC = "CvtConfig"
TF_CVT_PRETRAINED_MODEL_ARCHIVE_LIST = [
"microsoft/cvt-13",
"microsoft/cvt-13-384",
"microsoft/cvt-13-384-22k",
"microsoft/cvt-21",
"microsoft/cvt-21-384",
"microsoft/cvt-21-384-22k",
# See all Cvt models at https://huggingface.co/models?filter=cvt
]
@dataclass
class TFBaseModelOutputWithCLSToken(ModelOutput):
"""
Base class for model's outputs.
Args:
last_hidden_state (`tf.Tensor` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
cls_token_value (`tf.Tensor` of shape `(batch_size, 1, hidden_size)`):
Classification token at the output of the last layer of the model.
hidden_states (`tuple(tf.Tensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `tf.Tensor` (one for the output of the embeddings + one for the output of each layer) of shape
`(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus
the initial embedding outputs.
"""
last_hidden_state: tf.Tensor = None
cls_token_value: tf.Tensor = None
hidden_states: Optional[Tuple[tf.Tensor]] = None
class TFCvtDropPath(tf.keras.layers.Layer):
"""Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
References:
(1) github.com:rwightman/pytorch-image-models
"""
def __init__(self, drop_prob: float, **kwargs):
super().__init__(**kwargs)
self.drop_prob = drop_prob
def call(self, x: tf.Tensor, training=None):
if self.drop_prob == 0.0 or not training:
return x
keep_prob = 1 - self.drop_prob
shape = (tf.shape(x)[0],) + (1,) * (len(tf.shape(x)) - 1)
random_tensor = keep_prob + tf.random.uniform(shape, 0, 1)
random_tensor = tf.floor(random_tensor)
return (x / keep_prob) * random_tensor
class TFCvtEmbeddings(tf.keras.layers.Layer):
"""Construct the Convolutional Token Embeddings."""
def __init__(
self,
config: CvtConfig,
patch_size: int,
embed_dim: int,
stride: int,
padding: int,
dropout_rate: float,
**kwargs,
):
super().__init__(**kwargs)
self.convolution_embeddings = TFCvtConvEmbeddings(
config,
patch_size=patch_size,
embed_dim=embed_dim,
stride=stride,
padding=padding,
name="convolution_embeddings",
)
self.dropout = tf.keras.layers.Dropout(dropout_rate)
def call(self, pixel_values: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.convolution_embeddings(pixel_values)
hidden_state = self.dropout(hidden_state, training=training)
return hidden_state
class TFCvtConvEmbeddings(tf.keras.layers.Layer):
"""Image to Convolution Embeddings. This convolutional operation aims to model local spatial contexts."""
def __init__(self, config: CvtConfig, patch_size: int, embed_dim: int, stride: int, padding: int, **kwargs):
super().__init__(**kwargs)
self.padding = tf.keras.layers.ZeroPadding2D(padding=padding)
self.patch_size = patch_size if isinstance(patch_size, collections.abc.Iterable) else (patch_size, patch_size)
self.projection = tf.keras.layers.Conv2D(
filters=embed_dim,
kernel_size=patch_size,
strides=stride,
padding="valid",
data_format="channels_last",
kernel_initializer=get_initializer(config.initializer_range),
name="projection",
)
# Using the same default epsilon as PyTorch
self.normalization = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="normalization")
def call(self, pixel_values: tf.Tensor) -> tf.Tensor:
if isinstance(pixel_values, dict):
pixel_values = pixel_values["pixel_values"]
pixel_values = self.projection(self.padding(pixel_values))
# "batch_size, height, width, num_channels -> batch_size, (height*width), num_channels"
batch_size, height, width, num_channels = shape_list(pixel_values)
hidden_size = height * width
pixel_values = tf.reshape(pixel_values, shape=(batch_size, hidden_size, num_channels))
pixel_values = self.normalization(pixel_values)
# "batch_size, (height*width), num_channels -> batch_size, height, width, num_channels"
pixel_values = tf.reshape(pixel_values, shape=(batch_size, height, width, num_channels))
return pixel_values
class TFCvtSelfAttentionConvProjection(tf.keras.layers.Layer):
"""Convolutional projection layer."""
def __init__(self, config: CvtConfig, embed_dim: int, kernel_size: int, stride: int, padding: int, **kwargs):
super().__init__(**kwargs)
self.padding = tf.keras.layers.ZeroPadding2D(padding=padding)
self.convolution = tf.keras.layers.Conv2D(
filters=embed_dim,
kernel_size=kernel_size,
kernel_initializer=get_initializer(config.initializer_range),
padding="valid",
strides=stride,
use_bias=False,
name="convolution",
groups=embed_dim,
)
# Using the same default epsilon as PyTorch, TF uses (1 - pytorch momentum)
self.normalization = tf.keras.layers.BatchNormalization(epsilon=1e-5, momentum=0.9, name="normalization")
def call(self, hidden_state: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.convolution(self.padding(hidden_state))
hidden_state = self.normalization(hidden_state, training=training)
return hidden_state
class TFCvtSelfAttentionLinearProjection(tf.keras.layers.Layer):
"""Linear projection layer used to flatten tokens into 1D."""
def call(self, hidden_state: tf.Tensor) -> tf.Tensor:
# "batch_size, height, width, num_channels -> batch_size, (height*width), num_channels"
batch_size, height, width, num_channels = shape_list(hidden_state)
hidden_size = height * width
hidden_state = tf.reshape(hidden_state, shape=(batch_size, hidden_size, num_channels))
return hidden_state
class TFCvtSelfAttentionProjection(tf.keras.layers.Layer):
"""Convolutional Projection for Attention."""
def __init__(
self,
config: CvtConfig,
embed_dim: int,
kernel_size: int,
stride: int,
padding: int,
projection_method: str = "dw_bn",
**kwargs,
):
super().__init__(**kwargs)
if projection_method == "dw_bn":
self.convolution_projection = TFCvtSelfAttentionConvProjection(
config, embed_dim, kernel_size, stride, padding, name="convolution_projection"
)
self.linear_projection = TFCvtSelfAttentionLinearProjection()
def call(self, hidden_state: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.convolution_projection(hidden_state, training=training)
hidden_state = self.linear_projection(hidden_state)
return hidden_state
class TFCvtSelfAttention(tf.keras.layers.Layer):
"""
Self-attention layer. A depth-wise separable convolution operation (Convolutional Projection), is applied for
query, key, and value embeddings.
"""
def __init__(
self,
config: CvtConfig,
num_heads: int,
embed_dim: int,
kernel_size: int,
stride_q: int,
stride_kv: int,
padding_q: int,
padding_kv: int,
qkv_projection_method: str,
qkv_bias: bool,
attention_drop_rate: float,
with_cls_token: bool = True,
**kwargs,
):
super().__init__(**kwargs)
self.scale = embed_dim**-0.5
self.with_cls_token = with_cls_token
self.embed_dim = embed_dim
self.num_heads = num_heads
self.convolution_projection_query = TFCvtSelfAttentionProjection(
config,
embed_dim,
kernel_size,
stride_q,
padding_q,
projection_method="linear" if qkv_projection_method == "avg" else qkv_projection_method,
name="convolution_projection_query",
)
self.convolution_projection_key = TFCvtSelfAttentionProjection(
config,
embed_dim,
kernel_size,
stride_kv,
padding_kv,
projection_method=qkv_projection_method,
name="convolution_projection_key",
)
self.convolution_projection_value = TFCvtSelfAttentionProjection(
config,
embed_dim,
kernel_size,
stride_kv,
padding_kv,
projection_method=qkv_projection_method,
name="convolution_projection_value",
)
self.projection_query = tf.keras.layers.Dense(
units=embed_dim,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=qkv_bias,
bias_initializer="zeros",
name="projection_query",
)
self.projection_key = tf.keras.layers.Dense(
units=embed_dim,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=qkv_bias,
bias_initializer="zeros",
name="projection_key",
)
self.projection_value = tf.keras.layers.Dense(
units=embed_dim,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=qkv_bias,
bias_initializer="zeros",
name="projection_value",
)
self.dropout = tf.keras.layers.Dropout(attention_drop_rate)
def rearrange_for_multi_head_attention(self, hidden_state: tf.Tensor) -> tf.Tensor:
batch_size, hidden_size, _ = shape_list(hidden_state)
head_dim = self.embed_dim // self.num_heads
hidden_state = tf.reshape(hidden_state, shape=(batch_size, hidden_size, self.num_heads, head_dim))
hidden_state = tf.transpose(hidden_state, perm=(0, 2, 1, 3))
return hidden_state
def call(self, hidden_state: tf.Tensor, height: int, width: int, training: bool = False) -> tf.Tensor:
if self.with_cls_token:
cls_token, hidden_state = tf.split(hidden_state, [1, height * width], 1)
# "batch_size, (height*width), num_channels -> batch_size, height, width, num_channels"
batch_size, hidden_size, num_channels = shape_list(hidden_state)
hidden_state = tf.reshape(hidden_state, shape=(batch_size, height, width, num_channels))
key = self.convolution_projection_key(hidden_state, training=training)
query = self.convolution_projection_query(hidden_state, training=training)
value = self.convolution_projection_value(hidden_state, training=training)
if self.with_cls_token:
query = tf.concat((cls_token, query), axis=1)
key = tf.concat((cls_token, key), axis=1)
value = tf.concat((cls_token, value), axis=1)
head_dim = self.embed_dim // self.num_heads
query = self.rearrange_for_multi_head_attention(self.projection_query(query))
key = self.rearrange_for_multi_head_attention(self.projection_key(key))
value = self.rearrange_for_multi_head_attention(self.projection_value(value))
attention_score = tf.matmul(query, key, transpose_b=True) * self.scale
attention_probs = stable_softmax(logits=attention_score, axis=-1)
attention_probs = self.dropout(attention_probs, training=training)
context = tf.matmul(attention_probs, value)
# "batch_size, num_heads, hidden_size, head_dim -> batch_size, hidden_size, (num_heads*head_dim)"
_, _, hidden_size, _ = shape_list(context)
context = tf.transpose(context, perm=(0, 2, 1, 3))
context = tf.reshape(context, (batch_size, hidden_size, self.num_heads * head_dim))
return context
class TFCvtSelfOutput(tf.keras.layers.Layer):
"""Output of the Attention layer ."""
def __init__(self, config: CvtConfig, embed_dim: int, drop_rate: float, **kwargs):
super().__init__(**kwargs)
self.dense = tf.keras.layers.Dense(
units=embed_dim, kernel_initializer=get_initializer(config.initializer_range), name="dense"
)
self.dropout = tf.keras.layers.Dropout(drop_rate)
def call(self, hidden_state: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.dense(inputs=hidden_state)
hidden_state = self.dropout(inputs=hidden_state, training=training)
return hidden_state
class TFCvtAttention(tf.keras.layers.Layer):
"""Attention layer. First chunk of the convolutional transformer block."""
def __init__(
self,
config: CvtConfig,
num_heads: int,
embed_dim: int,
kernel_size: int,
stride_q: int,
stride_kv: int,
padding_q: int,
padding_kv: int,
qkv_projection_method: str,
qkv_bias: bool,
attention_drop_rate: float,
drop_rate: float,
with_cls_token: bool = True,
**kwargs,
):
super().__init__(**kwargs)
self.attention = TFCvtSelfAttention(
config,
num_heads,
embed_dim,
kernel_size,
stride_q,
stride_kv,
padding_q,
padding_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
with_cls_token,
name="attention",
)
self.dense_output = TFCvtSelfOutput(config, embed_dim, drop_rate, name="output")
def prune_heads(self, heads):
raise NotImplementedError
def call(self, hidden_state: tf.Tensor, height: int, width: int, training: bool = False):
self_output = self.attention(hidden_state, height, width, training=training)
attention_output = self.dense_output(self_output, training=training)
return attention_output
class TFCvtIntermediate(tf.keras.layers.Layer):
"""Intermediate dense layer. Second chunk of the convolutional transformer block."""
def __init__(self, config: CvtConfig, embed_dim: int, mlp_ratio: int, **kwargs):
super().__init__(**kwargs)
self.dense = tf.keras.layers.Dense(
units=int(embed_dim * mlp_ratio),
kernel_initializer=get_initializer(config.initializer_range),
activation="gelu",
name="dense",
)
def call(self, hidden_state: tf.Tensor) -> tf.Tensor:
hidden_state = self.dense(hidden_state)
return hidden_state
class TFCvtOutput(tf.keras.layers.Layer):
"""
Output of the Convolutional Transformer Block (last chunk). It consists of a MLP and a residual connection.
"""
def __init__(self, config: CvtConfig, embed_dim: int, drop_rate: int, **kwargs):
super().__init__(**kwargs)
self.dense = tf.keras.layers.Dense(
units=embed_dim, kernel_initializer=get_initializer(config.initializer_range), name="dense"
)
self.dropout = tf.keras.layers.Dropout(drop_rate)
def call(self, hidden_state: tf.Tensor, input_tensor: tf.Tensor, training: bool = False) -> tf.Tensor:
hidden_state = self.dense(inputs=hidden_state)
hidden_state = self.dropout(inputs=hidden_state, training=training)
hidden_state = hidden_state + input_tensor
return hidden_state
class TFCvtLayer(tf.keras.layers.Layer):
"""
Convolutional Transformer Block composed by attention layers, normalization and multi-layer perceptrons (mlps). It
consists of 3 chunks : an attention layer, an intermediate dense layer and an output layer. This corresponds to the
`Block` class in the original implementation.
"""
def __init__(
self,
config: CvtConfig,
num_heads: int,
embed_dim: int,
kernel_size: int,
stride_q: int,
stride_kv: int,
padding_q: int,
padding_kv: int,
qkv_projection_method: str,
qkv_bias: bool,
attention_drop_rate: float,
drop_rate: float,
mlp_ratio: float,
drop_path_rate: float,
with_cls_token: bool = True,
**kwargs,
):
super().__init__(**kwargs)
self.attention = TFCvtAttention(
config,
num_heads,
embed_dim,
kernel_size,
stride_q,
stride_kv,
padding_q,
padding_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
with_cls_token,
name="attention",
)
self.intermediate = TFCvtIntermediate(config, embed_dim, mlp_ratio, name="intermediate")
self.dense_output = TFCvtOutput(config, embed_dim, drop_rate, name="output")
# Using `layers.Activation` instead of `tf.identity` to better control `training` behaviour.
self.drop_path = (
TFCvtDropPath(drop_path_rate, name="drop_path")
if drop_path_rate > 0.0
else tf.keras.layers.Activation("linear", name="drop_path")
)
# Using the same default epsilon as PyTorch
self.layernorm_before = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="layernorm_before")
self.layernorm_after = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="layernorm_after")
def call(self, hidden_state: tf.Tensor, height: int, width: int, training: bool = False) -> tf.Tensor:
# in Cvt, layernorm is applied before self-attention
attention_output = self.attention(self.layernorm_before(hidden_state), height, width, training=training)
attention_output = self.drop_path(attention_output, training=training)
# first residual connection
hidden_state = attention_output + hidden_state
# in Cvt, layernorm is also applied after self-attention
layer_output = self.layernorm_after(hidden_state)
layer_output = self.intermediate(layer_output)
# second residual connection is done here
layer_output = self.dense_output(layer_output, hidden_state)
layer_output = self.drop_path(layer_output, training=training)
return layer_output
class TFCvtStage(tf.keras.layers.Layer):
"""
Cvt stage (encoder block). Each stage has 2 parts :
- (1) A Convolutional Token Embedding layer
- (2) A Convolutional Transformer Block (layer).
The classification token is added only in the last stage.
Args:
config ([`CvtConfig`]): Model configuration class.
stage (`int`): Stage number.
"""
def __init__(self, config: CvtConfig, stage: int, **kwargs):
super().__init__(**kwargs)
self.config = config
self.stage = stage
if self.config.cls_token[self.stage]:
self.cls_token = self.add_weight(
shape=(1, 1, self.config.embed_dim[-1]),
initializer=get_initializer(self.config.initializer_range),
trainable=True,
name="cvt.encoder.stages.2.cls_token",
)
self.embedding = TFCvtEmbeddings(
self.config,
patch_size=config.patch_sizes[self.stage],
stride=config.patch_stride[self.stage],
embed_dim=config.embed_dim[self.stage],
padding=config.patch_padding[self.stage],
dropout_rate=config.drop_rate[self.stage],
name="embedding",
)
drop_path_rates = tf.linspace(0.0, config.drop_path_rate[self.stage], config.depth[stage])
drop_path_rates = [x.numpy().item() for x in drop_path_rates]
self.layers = [
TFCvtLayer(
config,
num_heads=config.num_heads[self.stage],
embed_dim=config.embed_dim[self.stage],
kernel_size=config.kernel_qkv[self.stage],
stride_q=config.stride_q[self.stage],
stride_kv=config.stride_kv[self.stage],
padding_q=config.padding_q[self.stage],
padding_kv=config.padding_kv[self.stage],
qkv_projection_method=config.qkv_projection_method[self.stage],
qkv_bias=config.qkv_bias[self.stage],
attention_drop_rate=config.attention_drop_rate[self.stage],
drop_rate=config.drop_rate[self.stage],
mlp_ratio=config.mlp_ratio[self.stage],
drop_path_rate=drop_path_rates[self.stage],
with_cls_token=config.cls_token[self.stage],
name=f"layers.{j}",
)
for j in range(config.depth[self.stage])
]
def call(self, hidden_state: tf.Tensor, training: bool = False):
cls_token = None
hidden_state = self.embedding(hidden_state, training)
# "batch_size, height, width, num_channels -> batch_size, (height*width), num_channels"
batch_size, height, width, num_channels = shape_list(hidden_state)
hidden_size = height * width
hidden_state = tf.reshape(hidden_state, shape=(batch_size, hidden_size, num_channels))
if self.config.cls_token[self.stage]:
cls_token = tf.repeat(self.cls_token, repeats=batch_size, axis=0)
hidden_state = tf.concat((cls_token, hidden_state), axis=1)
for layer in self.layers:
layer_outputs = layer(hidden_state, height, width, training=training)
hidden_state = layer_outputs
if self.config.cls_token[self.stage]:
cls_token, hidden_state = tf.split(hidden_state, [1, height * width], 1)
# "batch_size, (height*width), num_channels -> batch_size, height, width, num_channels"
hidden_state = tf.reshape(hidden_state, shape=(batch_size, height, width, num_channels))
return hidden_state, cls_token
class TFCvtEncoder(tf.keras.layers.Layer):
"""
Convolutional Vision Transformer encoder. CVT has 3 stages of encoder blocks with their respective number of layers
(depth) being 1, 2 and 10.
Args:
config ([`CvtConfig`]): Model configuration class.
"""
config_class = CvtConfig
def __init__(self, config: CvtConfig, **kwargs):
super().__init__(**kwargs)
self.config = config
self.stages = [
TFCvtStage(config, stage_idx, name=f"stages.{stage_idx}") for stage_idx in range(len(config.depth))
]
def call(
self,
pixel_values: TFModelInputType,
output_hidden_states: Optional[bool] = False,
return_dict: Optional[bool] = True,
training: Optional[bool] = False,
) -> Union[TFBaseModelOutputWithCLSToken, Tuple[tf.Tensor]]:
all_hidden_states = () if output_hidden_states else None
hidden_state = pixel_values
# When running on CPU, `tf.keras.layers.Conv2D` doesn't support (batch_size, num_channels, height, width)
# as input format. So change the input format to (batch_size, height, width, num_channels).
hidden_state = tf.transpose(hidden_state, perm=(0, 2, 3, 1))
cls_token = None
for _, (stage_module) in enumerate(self.stages):
hidden_state, cls_token = stage_module(hidden_state, training=training)
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_state,)
# Change back to (batch_size, num_channels, height, width) format to have uniformity in the modules
hidden_state = tf.transpose(hidden_state, perm=(0, 3, 1, 2))
if output_hidden_states:
all_hidden_states = tuple([tf.transpose(hs, perm=(0, 3, 1, 2)) for hs in all_hidden_states])
if not return_dict:
return tuple(v for v in [hidden_state, cls_token, all_hidden_states] if v is not None)
return TFBaseModelOutputWithCLSToken(
last_hidden_state=hidden_state,
cls_token_value=cls_token,
hidden_states=all_hidden_states,
)
@keras_serializable
class TFCvtMainLayer(tf.keras.layers.Layer):
"""Construct the Cvt model."""
config_class = CvtConfig
def __init__(self, config: CvtConfig, **kwargs):
super().__init__(**kwargs)
self.config = config
self.encoder = TFCvtEncoder(config, name="encoder")
@unpack_inputs
def call(
self,
pixel_values: Optional[TFModelInputType] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
training: Optional[bool] = False,
) -> Union[TFBaseModelOutputWithCLSToken, Tuple[tf.Tensor]]:
if pixel_values is None:
raise ValueError("You have to specify pixel_values")
encoder_outputs = self.encoder(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
training=training,
)
sequence_output = encoder_outputs[0]
if not return_dict:
return (sequence_output,) + encoder_outputs[1:]
return TFBaseModelOutputWithCLSToken(
last_hidden_state=sequence_output,
cls_token_value=encoder_outputs.cls_token_value,
hidden_states=encoder_outputs.hidden_states,
)
class TFCvtPreTrainedModel(TFPreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = CvtConfig
base_model_prefix = "cvt"
main_input_name = "pixel_values"
@property
def dummy_inputs(self) -> Dict[str, tf.Tensor]:
"""
Dummy inputs to build the network.
Returns:
`Dict[str, tf.Tensor]`: The dummy inputs.
"""
VISION_DUMMY_INPUTS = tf.random.uniform(shape=(3, self.config.num_channels, 224, 224), dtype=tf.float32)
return {"pixel_values": tf.constant(VISION_DUMMY_INPUTS)}
@tf.function(
input_signature=[
{
"pixel_values": tf.TensorSpec((None, None, None, None), tf.float32, name="pixel_values"),
}
]
)
def serving(self, inputs):
"""
Method used for serving the model.
Args:
inputs (`Dict[str, tf.Tensor]`):
The input of the saved model as a dictionary of tensors.
"""
output = self.call(inputs)
return self.serving_output(output)
TFCVT_START_DOCSTRING = r"""
This model inherits from [`TFPreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a [tf.keras.Model](https://www.tensorflow.org/api_docs/python/tf/keras/Model) subclass. Use it
as a regular TF 2.0 Keras Model and refer to the TF 2.0 documentation for all matter related to general usage and
behavior.
<Tip>
TF 2.0 models accepts two formats as inputs:
- having all inputs as keyword arguments (like PyTorch models), or
- having all inputs as a list, tuple or dict in the first positional arguments.
This second option is useful when using [`tf.keras.Model.fit`] method which currently requires having all the
tensors in the first argument of the model call function: `model(inputs)`.
</Tip>
Args:
config ([`CvtConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~TFPreTrainedModel.from_pretrained`] method to load the model weights.
"""
TFCVT_INPUTS_DOCSTRING = r"""
Args:
pixel_values (`np.ndarray`, `tf.Tensor`, `List[tf.Tensor]` ``Dict[str, tf.Tensor]` or `Dict[str, np.ndarray]` and each example must have the shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See [`CvtImageProcessor.__call__`]
for details.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail. This argument can be used only in eager mode, in graph mode the value in the config will be
used instead.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. This argument can be used in
eager mode, in graph mode the value will always be set to True.
training (`bool`, *optional*, defaults to `False``):
Whether or not to use the model in training mode (some modules like dropout modules have different
behaviors between training and evaluation).
"""
@add_start_docstrings(
"The bare Cvt Model transformer outputting raw hidden-states without any specific head on top.",
TFCVT_START_DOCSTRING,
)
class TFCvtModel(TFCvtPreTrainedModel):
def __init__(self, config: CvtConfig, *inputs, **kwargs):
super().__init__(config, *inputs, **kwargs)
self.cvt = TFCvtMainLayer(config, name="cvt")
@unpack_inputs
@add_start_docstrings_to_model_forward(TFCVT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=TFBaseModelOutputWithCLSToken, config_class=_CONFIG_FOR_DOC)
def call(
self,
pixel_values: Optional[tf.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
training: Optional[bool] = False,
) -> Union[TFBaseModelOutputWithCLSToken, Tuple[tf.Tensor]]:
r"""
Returns:
Examples:
```python
>>> from transformers import AutoImageProcessor, TFCvtModel
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> image_processor = AutoImageProcessor.from_pretrained("microsoft/cvt-13")
>>> model = TFCvtModel.from_pretrained("microsoft/cvt-13")
>>> inputs = image_processor(images=image, return_tensors="tf")
>>> outputs = model(**inputs)
>>> last_hidden_states = outputs.last_hidden_state
```"""
if pixel_values is None:
raise ValueError("You have to specify pixel_values")
outputs = self.cvt(
pixel_values=pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
training=training,
)
if not return_dict:
return (outputs[0],) + outputs[1:]
return TFBaseModelOutputWithCLSToken(
last_hidden_state=outputs.last_hidden_state,
cls_token_value=outputs.cls_token_value,
hidden_states=outputs.hidden_states,
)
def serving_output(self, output: TFBaseModelOutputWithCLSToken) -> TFBaseModelOutputWithCLSToken:
return TFBaseModelOutputWithCLSToken(
last_hidden_state=output.last_hidden_state,
cls_token_value=output.cls_token_value,
hidden_states=output.hidden_states,
)
@add_start_docstrings(
"""
Cvt Model transformer with an image classification head on top (a linear layer on top of the final hidden state of
the [CLS] token) e.g. for ImageNet.
""",
TFCVT_START_DOCSTRING,
)
class TFCvtForImageClassification(TFCvtPreTrainedModel, TFSequenceClassificationLoss):
def __init__(self, config: CvtConfig, *inputs, **kwargs):
super().__init__(config, *inputs, **kwargs)
self.num_labels = config.num_labels
self.cvt = TFCvtMainLayer(config, name="cvt")
# Using same default epsilon as in the original implementation.
self.layernorm = tf.keras.layers.LayerNormalization(epsilon=1e-5, name="layernorm")
# Classifier head
self.classifier = tf.keras.layers.Dense(
units=config.num_labels,
kernel_initializer=get_initializer(config.initializer_range),
use_bias=True,
bias_initializer="zeros",
name="classifier",
)
@unpack_inputs
@add_start_docstrings_to_model_forward(TFCVT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=TFImageClassifierOutputWithNoAttention, config_class=_CONFIG_FOR_DOC)
def call(
self,
pixel_values: Optional[tf.Tensor] = None,
labels: Optional[tf.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
training: Optional[bool] = False,
) -> Union[TFImageClassifierOutputWithNoAttention, Tuple[tf.Tensor]]:
r"""
labels (`tf.Tensor` or `np.ndarray` of shape `(batch_size,)`, *optional*):
Labels for computing the image classification/regression loss. Indices should be in `[0, ...,
config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
Returns:
Examples:
```python
>>> from transformers import AutoImageProcessor, TFCvtForImageClassification
>>> import tensorflow as tf
>>> from PIL import Image
>>> import requests
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = Image.open(requests.get(url, stream=True).raw)
>>> image_processor = AutoImageProcessor.from_pretrained("microsoft/cvt-13")
>>> model = TFCvtForImageClassification.from_pretrained("microsoft/cvt-13")
>>> inputs = image_processor(images=image, return_tensors="tf")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> # model predicts one of the 1000 ImageNet classes
>>> predicted_class_idx = tf.math.argmax(logits, axis=-1)[0]
>>> print("Predicted class:", model.config.id2label[int(predicted_class_idx)])
```"""
outputs = self.cvt(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
training=training,
)
sequence_output = outputs[0]
cls_token = outputs[1]
if self.config.cls_token[-1]:
sequence_output = self.layernorm(cls_token)
else:
# rearrange "batch_size, num_channels, height, width -> batch_size, (height*width), num_channels"
batch_size, num_channels, height, width = shape_list(sequence_output)
sequence_output = tf.reshape(sequence_output, shape=(batch_size, num_channels, height * width))
sequence_output = tf.transpose(sequence_output, perm=(0, 2, 1))
sequence_output = self.layernorm(sequence_output)
sequence_output_mean = tf.reduce_mean(sequence_output, axis=1)
logits = self.classifier(sequence_output_mean)
loss = None if labels is None else self.hf_compute_loss(labels=labels, logits=logits)
if not return_dict:
output = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return TFImageClassifierOutputWithNoAttention(loss=loss, logits=logits, hidden_states=outputs.hidden_states)
def serving_output(self, output: TFImageClassifierOutputWithNoAttention) -> TFImageClassifierOutputWithNoAttention:
return TFImageClassifierOutputWithNoAttention(logits=output.logits, hidden_states=output.hidden_states)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./.git/objects/a2/cc185401e7764367ae52d1654491cf863ae69e | x+)JMU00e040031Q,+dx6M9{wk+qIODĒĒ">K q {YCIQbf^*XiQc52Mw
X''$%&gk4:tki.-[pz 6=)%EI%) mpɸIޱtiD9i+N-4b\p9nNn76-% qW03s&lKK2sA|LJNNvɢv
=
OA | x+)JMU00e040031Q,+dx6M9{wk+qIODĒĒ">K q {YCIQbf^*XiQc52Mw
X''$%&gk4:tki.-[pz 6=)%EI%) mpɸIޱtiD9i+N-4b\p9nNn76-% qW03s&lKK2sA|LJNNvɢv
=
OA | -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/tensorflow/question-answering/run_qa.py | #!/usr/bin/env python
# coding=utf-8
# Copyright 2020 The HuggingFace Team All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Fine-tuning the library models for question answering.
"""
# You can also adapt this script on your own question answering task. Pointers for this are left as comments.
import json
import logging
import os
import sys
from dataclasses import dataclass, field
from pathlib import Path
from typing import Optional
import evaluate
import tensorflow as tf
from datasets import load_dataset
from utils_qa import postprocess_qa_predictions
import transformers
from transformers import (
AutoConfig,
AutoTokenizer,
EvalPrediction,
HfArgumentParser,
PreTrainedTokenizerFast,
PushToHubCallback,
TFAutoModelForQuestionAnswering,
TFTrainingArguments,
create_optimizer,
set_seed,
)
from transformers.utils import CONFIG_NAME, TF2_WEIGHTS_NAME, check_min_version, send_example_telemetry
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.27.0.dev0")
logger = logging.getLogger(__name__)
# region Arguments
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
"""
model_name_or_path: str = field(
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
tokenizer_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None,
metadata={"help": "Path to directory to store the pretrained models downloaded from huggingface.co"},
)
model_revision: str = field(
default="main",
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
)
use_auth_token: bool = field(
default=False,
metadata={
"help": (
"Will use the token generated when running `huggingface-cli login` (necessary to use this script "
"with private models)."
)
},
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
"""
dataset_name: Optional[str] = field(
default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
)
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
validation_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."},
)
test_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input test data file to evaluate the perplexity on (a text file)."},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
preprocessing_num_workers: Optional[int] = field(
default=None,
metadata={"help": "The number of processes to use for the preprocessing."},
)
max_seq_length: int = field(
default=384,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": (
"Whether to pad all samples to `max_seq_length`. If False, will pad the samples dynamically when"
" batching to the maximum length in the batch (which can be faster on GPU but will be slower on TPU)."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
version_2_with_negative: bool = field(
default=False, metadata={"help": "If true, some of the examples do not have an answer."}
)
null_score_diff_threshold: float = field(
default=0.0,
metadata={
"help": (
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
)
},
)
doc_stride: int = field(
default=128,
metadata={"help": "When splitting up a long document into chunks, how much stride to take between chunks."},
)
n_best_size: int = field(
default=20,
metadata={"help": "The total number of n-best predictions to generate when looking for an answer."},
)
max_answer_length: int = field(
default=30,
metadata={
"help": (
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
)
},
)
def __post_init__(self):
if (
self.dataset_name is None
and self.train_file is None
and self.validation_file is None
and self.test_file is None
):
raise ValueError("Need either a dataset name or a training/validation file/test_file.")
else:
if self.train_file is not None:
extension = self.train_file.split(".")[-1]
assert extension in ["csv", "json"], "`train_file` should be a csv or a json file."
if self.validation_file is not None:
extension = self.validation_file.split(".")[-1]
assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
if self.test_file is not None:
extension = self.test_file.split(".")[-1]
assert extension in ["csv", "json"], "`test_file` should be a csv or a json file."
# endregion
# region Helper classes
class SavePretrainedCallback(tf.keras.callbacks.Callback):
# Hugging Face models have a save_pretrained() method that saves both the weights and the necessary
# metadata to allow them to be loaded as a pretrained model in future. This is a simple Keras callback
# that saves the model with this method after each epoch.
def __init__(self, output_dir, **kwargs):
super().__init__()
self.output_dir = output_dir
def on_epoch_end(self, epoch, logs=None):
self.model.save_pretrained(self.output_dir)
# endregion
def main():
# region Argument parsing
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TFTrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
# information sent is the one passed as arguments along with your Python/PyTorch versions.
send_example_telemetry("run_qa", model_args, data_args, framework="tensorflow")
output_dir = Path(training_args.output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
# endregion
# region Checkpoints
checkpoint = None
if len(os.listdir(training_args.output_dir)) > 0 and not training_args.overwrite_output_dir:
if (output_dir / CONFIG_NAME).is_file() and (output_dir / TF2_WEIGHTS_NAME).is_file():
checkpoint = output_dir
logger.info(
f"Checkpoint detected, resuming training from checkpoint in {training_args.output_dir}. To avoid this"
" behavior, change the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
)
else:
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty. "
"Use --overwrite_output_dir to continue regardless."
)
# endregion
# region Logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
handlers=[logging.StreamHandler(sys.stdout)],
)
logger.setLevel(logging.INFO if training_args.should_log else logging.WARN)
# Set the verbosity to info of the Transformers logger (on main process only):
if training_args.should_log:
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info(f"Training/evaluation parameters {training_args}")
# endregion
# Set seed before initializing model.
set_seed(training_args.seed)
# region Load Data
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
datasets = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
cache_dir=model_args.cache_dir,
use_auth_token=True if model_args.use_auth_token else None,
)
else:
data_files = {}
if data_args.train_file is not None:
data_files["train"] = data_args.train_file
extension = data_args.train_file.split(".")[-1]
if data_args.validation_file is not None:
data_files["validation"] = data_args.validation_file
extension = data_args.validation_file.split(".")[-1]
if data_args.test_file is not None:
data_files["test"] = data_args.test_file
extension = data_args.test_file.split(".")[-1]
datasets = load_dataset(
extension,
data_files=data_files,
field="data",
cache_dir=model_args.cache_dir,
use_auth_token=True if model_args.use_auth_token else None,
)
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# endregion
# region Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
config = AutoConfig.from_pretrained(
model_args.config_name if model_args.config_name else model_args.model_name_or_path,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
tokenizer = AutoTokenizer.from_pretrained(
model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path,
cache_dir=model_args.cache_dir,
use_fast=True,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
# endregion
# region Tokenizer check: this script requires a fast tokenizer.
if not isinstance(tokenizer, PreTrainedTokenizerFast):
raise ValueError(
"This example script only works for models that have a fast tokenizer. Checkout the big table of models at"
" https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet"
" this requirement"
)
# endregion
# region Preprocessing the datasets
# Preprocessing is slightly different for training and evaluation.
if training_args.do_train:
column_names = datasets["train"].column_names
elif training_args.do_eval:
column_names = datasets["validation"].column_names
else:
column_names = datasets["test"].column_names
question_column_name = "question" if "question" in column_names else column_names[0]
context_column_name = "context" if "context" in column_names else column_names[1]
answer_column_name = "answers" if "answers" in column_names else column_names[2]
# Padding side determines if we do (question|context) or (context|question).
pad_on_right = tokenizer.padding_side == "right"
if data_args.max_seq_length > tokenizer.model_max_length:
logger.warning(
f"The max_seq_length passed ({data_args.max_seq_length}) is larger than the maximum length for the"
f"model ({tokenizer.model_max_length}). Using max_seq_length={tokenizer.model_max_length}."
)
max_seq_length = min(data_args.max_seq_length, tokenizer.model_max_length)
if data_args.pad_to_max_length or isinstance(training_args.strategy, tf.distribute.TPUStrategy):
logger.info("Padding all batches to max length because argument was set or we're on TPU.")
padding = "max_length"
else:
padding = False
# Training preprocessing
def prepare_train_features(examples):
# Some of the questions have lots of whitespace on the left, which is not useful and will make the
# truncation of the context fail (the tokenized question will take a lots of space). So we remove that
# left whitespace
examples[question_column_name] = [q.lstrip() for q in examples[question_column_name]]
# Tokenize our examples with truncation and maybe padding, but keep the overflows using a stride. This results
# in one example possible giving several features when a context is long, each of those features having a
# context that overlaps a bit the context of the previous feature.
tokenized_examples = tokenizer(
examples[question_column_name if pad_on_right else context_column_name],
examples[context_column_name if pad_on_right else question_column_name],
truncation="only_second" if pad_on_right else "only_first",
max_length=max_seq_length,
stride=data_args.doc_stride,
return_overflowing_tokens=True,
return_offsets_mapping=True,
padding=padding,
)
# Since one example might give us several features if it has a long context, we need a map from a feature to
# its corresponding example. This key gives us just that.
sample_mapping = tokenized_examples.pop("overflow_to_sample_mapping")
# The offset mappings will give us a map from token to character position in the original context. This will
# help us compute the start_positions and end_positions.
offset_mapping = tokenized_examples.pop("offset_mapping")
# Let's label those examples!
tokenized_examples["start_positions"] = []
tokenized_examples["end_positions"] = []
for i, offsets in enumerate(offset_mapping):
# We will label impossible answers with the index of the CLS token.
input_ids = tokenized_examples["input_ids"][i]
cls_index = input_ids.index(tokenizer.cls_token_id)
# Grab the sequence corresponding to that example (to know what is the context and what is the question).
sequence_ids = tokenized_examples.sequence_ids(i)
# One example can give several spans, this is the index of the example containing this span of text.
sample_index = sample_mapping[i]
answers = examples[answer_column_name][sample_index]
# If no answers are given, set the cls_index as answer.
if len(answers["answer_start"]) == 0:
tokenized_examples["start_positions"].append(cls_index)
tokenized_examples["end_positions"].append(cls_index)
else:
# Start/end character index of the answer in the text.
start_char = answers["answer_start"][0]
end_char = start_char + len(answers["text"][0])
# Start token index of the current span in the text.
token_start_index = 0
while sequence_ids[token_start_index] != (1 if pad_on_right else 0):
token_start_index += 1
# End token index of the current span in the text.
token_end_index = len(input_ids) - 1
while sequence_ids[token_end_index] != (1 if pad_on_right else 0):
token_end_index -= 1
# Detect if the answer is out of the span (in which case this feature is labeled with the CLS index).
if not (offsets[token_start_index][0] <= start_char and offsets[token_end_index][1] >= end_char):
tokenized_examples["start_positions"].append(cls_index)
tokenized_examples["end_positions"].append(cls_index)
else:
# Otherwise move the token_start_index and token_end_index to the two ends of the answer.
# Note: we could go after the last offset if the answer is the last word (edge case).
while token_start_index < len(offsets) and offsets[token_start_index][0] <= start_char:
token_start_index += 1
tokenized_examples["start_positions"].append(token_start_index - 1)
while offsets[token_end_index][1] >= end_char:
token_end_index -= 1
tokenized_examples["end_positions"].append(token_end_index + 1)
return tokenized_examples
processed_datasets = {}
if training_args.do_train:
if "train" not in datasets:
raise ValueError("--do_train requires a train dataset")
train_dataset = datasets["train"]
if data_args.max_train_samples is not None:
# We will select sample from whole data if agument is specified
max_train_samples = min(len(train_dataset), data_args.max_train_samples)
train_dataset = train_dataset.select(range(max_train_samples))
# Create train feature from dataset
train_dataset = train_dataset.map(
prepare_train_features,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
)
if data_args.max_train_samples is not None:
# Number of samples might increase during Feature Creation, We select only specified max samples
max_train_samples = min(len(train_dataset), data_args.max_train_samples)
train_dataset = train_dataset.select(range(max_train_samples))
processed_datasets["train"] = train_dataset
# Validation preprocessing
def prepare_validation_features(examples):
# Some of the questions have lots of whitespace on the left, which is not useful and will make the
# truncation of the context fail (the tokenized question will take a lots of space). So we remove that
# left whitespace
examples[question_column_name] = [q.lstrip() for q in examples[question_column_name]]
# Tokenize our examples with truncation and maybe padding, but keep the overflows using a stride. This results
# in one example possible giving several features when a context is long, each of those features having a
# context that overlaps a bit the context of the previous feature.
tokenized_examples = tokenizer(
examples[question_column_name if pad_on_right else context_column_name],
examples[context_column_name if pad_on_right else question_column_name],
truncation="only_second" if pad_on_right else "only_first",
max_length=max_seq_length,
stride=data_args.doc_stride,
return_overflowing_tokens=True,
return_offsets_mapping=True,
padding=padding,
)
# Since one example might give us several features if it has a long context, we need a map from a feature to
# its corresponding example. This key gives us just that.
sample_mapping = tokenized_examples.pop("overflow_to_sample_mapping")
# For evaluation, we will need to convert our predictions to substrings of the context, so we keep the
# corresponding example_id and we will store the offset mappings.
tokenized_examples["example_id"] = []
for i in range(len(tokenized_examples["input_ids"])):
# Grab the sequence corresponding to that example (to know what is the context and what is the question).
sequence_ids = tokenized_examples.sequence_ids(i)
context_index = 1 if pad_on_right else 0
# One example can give several spans, this is the index of the example containing this span of text.
sample_index = sample_mapping[i]
tokenized_examples["example_id"].append(examples["id"][sample_index])
# Set to None the offset_mapping that are not part of the context so it's easy to determine if a token
# position is part of the context or not.
tokenized_examples["offset_mapping"][i] = [
(o if sequence_ids[k] == context_index else None)
for k, o in enumerate(tokenized_examples["offset_mapping"][i])
]
return tokenized_examples
if training_args.do_eval:
if "validation" not in datasets:
raise ValueError("--do_eval requires a validation dataset")
eval_examples = datasets["validation"]
if data_args.max_eval_samples is not None:
# We will select sample from whole data
max_eval_samples = min(len(eval_examples), data_args.max_eval_samples)
eval_examples = eval_examples.select(range(max_eval_samples))
# Validation Feature Creation
eval_dataset = eval_examples.map(
prepare_validation_features,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
)
if data_args.max_eval_samples is not None:
# During Feature creation dataset samples might increase, we will select required samples again
max_eval_samples = min(len(eval_dataset), data_args.max_eval_samples)
eval_dataset = eval_dataset.select(range(max_eval_samples))
processed_datasets["validation"] = eval_dataset
if training_args.do_predict:
if "test" not in datasets:
raise ValueError("--do_predict requires a test dataset")
predict_examples = datasets["test"]
if data_args.max_predict_samples is not None:
# We will select sample from whole data
predict_examples = predict_examples.select(range(data_args.max_predict_samples))
# Predict Feature Creation
predict_dataset = predict_examples.map(
prepare_validation_features,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
)
if data_args.max_predict_samples is not None:
# During Feature creation dataset samples might increase, we will select required samples again
max_predict_samples = min(len(predict_dataset), data_args.max_predict_samples)
predict_dataset = predict_dataset.select(range(max_predict_samples))
processed_datasets["test"] = predict_dataset
# endregion
# region Metrics and Post-processing:
def post_processing_function(examples, features, predictions, stage="eval"):
# Post-processing: we match the start logits and end logits to answers in the original context.
predictions = postprocess_qa_predictions(
examples=examples,
features=features,
predictions=predictions,
version_2_with_negative=data_args.version_2_with_negative,
n_best_size=data_args.n_best_size,
max_answer_length=data_args.max_answer_length,
null_score_diff_threshold=data_args.null_score_diff_threshold,
output_dir=training_args.output_dir,
prefix=stage,
)
# Format the result to the format the metric expects.
if data_args.version_2_with_negative:
formatted_predictions = [
{"id": k, "prediction_text": v, "no_answer_probability": 0.0} for k, v in predictions.items()
]
else:
formatted_predictions = [{"id": k, "prediction_text": v} for k, v in predictions.items()]
references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
return EvalPrediction(predictions=formatted_predictions, label_ids=references)
metric = evaluate.load("squad_v2" if data_args.version_2_with_negative else "squad")
def compute_metrics(p: EvalPrediction):
return metric.compute(predictions=p.predictions, references=p.label_ids)
# endregion
with training_args.strategy.scope():
dataset_options = tf.data.Options()
dataset_options.experimental_distribute.auto_shard_policy = tf.data.experimental.AutoShardPolicy.OFF
num_replicas = training_args.strategy.num_replicas_in_sync
# region Load model and prepare datasets
if checkpoint is None:
model_path = model_args.model_name_or_path
else:
model_path = checkpoint
model = TFAutoModelForQuestionAnswering.from_pretrained(
model_path,
config=config,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
if training_args.do_train:
training_dataset = model.prepare_tf_dataset(
processed_datasets["train"],
shuffle=True,
batch_size=training_args.per_device_train_batch_size * num_replicas,
tokenizer=tokenizer,
)
training_dataset = training_dataset.with_options(dataset_options)
num_train_steps = len(training_dataset) * training_args.num_train_epochs
if training_args.warmup_steps > 0:
num_warmup_steps = training_args.warmup_steps
elif training_args.warmup_ratio > 0:
num_warmup_steps = int(num_train_steps * training_args.warmup_ratio)
else:
num_warmup_steps = 0
optimizer, schedule = create_optimizer(
init_lr=training_args.learning_rate,
num_train_steps=len(training_dataset) * training_args.num_train_epochs,
num_warmup_steps=num_warmup_steps,
adam_beta1=training_args.adam_beta1,
adam_beta2=training_args.adam_beta2,
adam_epsilon=training_args.adam_epsilon,
weight_decay_rate=training_args.weight_decay,
adam_global_clipnorm=training_args.max_grad_norm,
)
# no user-specified loss = will use the model internal loss
model.compile(optimizer=optimizer, jit_compile=training_args.xla, metrics=["accuracy"])
else:
model.compile(optimizer=None, jit_compile=training_args.xla, metrics=["accuracy"])
training_dataset = None
if training_args.do_eval:
eval_dataset = model.prepare_tf_dataset(
processed_datasets["validation"],
shuffle=False,
batch_size=training_args.per_device_train_batch_size * num_replicas,
tokenizer=tokenizer,
)
eval_dataset = eval_dataset.with_options(dataset_options)
else:
eval_dataset = None
if training_args.do_predict:
predict_dataset = model.prepare_tf_dataset(
processed_datasets["test"],
shuffle=False,
batch_size=training_args.per_device_eval_batch_size * num_replicas,
tokenizer=tokenizer,
)
predict_dataset = predict_dataset.with_options(dataset_options)
else:
predict_dataset = None
# endregion
# region Preparing push_to_hub and model card
push_to_hub_model_id = training_args.push_to_hub_model_id
model_name = model_args.model_name_or_path.split("/")[-1]
if not push_to_hub_model_id:
if data_args.dataset_name is not None:
push_to_hub_model_id = f"{model_name}-finetuned-{data_args.dataset_name}"
else:
push_to_hub_model_id = f"{model_name}-finetuned-question-answering"
model_card_kwargs = {"finetuned_from": model_args.model_name_or_path, "tasks": "question-answering"}
if data_args.dataset_name is not None:
model_card_kwargs["dataset_tags"] = data_args.dataset_name
if data_args.dataset_config_name is not None:
model_card_kwargs["dataset_args"] = data_args.dataset_config_name
model_card_kwargs["dataset"] = f"{data_args.dataset_name} {data_args.dataset_config_name}"
else:
model_card_kwargs["dataset"] = data_args.dataset_name
if training_args.push_to_hub:
callbacks = [
PushToHubCallback(
output_dir=training_args.output_dir,
model_id=push_to_hub_model_id,
organization=training_args.push_to_hub_organization,
token=training_args.push_to_hub_token,
tokenizer=tokenizer,
**model_card_kwargs,
)
]
else:
callbacks = []
# endregion
# region Training and Evaluation
if training_args.do_train:
# Note that the validation and test datasets have been processed in a different way to the
# training datasets in this example, and so they don't have the same label structure.
# As such, we don't pass them directly to Keras, but instead get model predictions to evaluate
# after training.
model.fit(training_dataset, epochs=int(training_args.num_train_epochs), callbacks=callbacks)
if training_args.do_eval:
logger.info("*** Evaluation ***")
# In this example, we compute advanced metrics at the end of training, but
# if you'd like to compute metrics every epoch that are too complex to be written as
# standard Keras metrics, you can use our KerasMetricCallback. See
# https://huggingface.co/docs/transformers/main/en/main_classes/keras_callbacks
eval_predictions = model.predict(eval_dataset)
if isinstance(eval_predictions.start_logits, tf.RaggedTensor):
# If predictions are RaggedTensor, we densify them. Since they are logits, padding with 0 is a bad idea!
# The reason is that a logit of 0 can often end up as quite a high probability value, sometimes even
# the highest probability in a sample. Instead, we use a large negative value, which ensures that the
# padding positions are correctly masked.
eval_start_logits = eval_predictions.start_logits.to_tensor(default_value=-1000).numpy()
eval_end_logits = eval_predictions.end_logits.to_tensor(default_value=-1000).numpy()
else:
eval_start_logits = eval_predictions.start_logits
eval_end_logits = eval_predictions.end_logits
post_processed_eval = post_processing_function(
datasets["validation"],
processed_datasets["validation"],
(eval_start_logits, eval_end_logits),
)
metrics = compute_metrics(post_processed_eval)
logging.info("Evaluation metrics:")
for metric, value in metrics.items():
logging.info(f"{metric}: {value:.3f}")
if training_args.output_dir is not None:
output_eval_file = os.path.join(training_args.output_dir, "all_results.json")
with open(output_eval_file, "w") as writer:
writer.write(json.dumps(metrics))
# endregion
# region Prediction
if training_args.do_predict:
logger.info("*** Predict ***")
test_predictions = model.predict(predict_dataset)
if isinstance(test_predictions.start_logits, tf.RaggedTensor):
# If predictions are RaggedTensor, we densify them. Since they are logits, padding with 0 is a bad idea!
# The reason is that a logit of 0 can often end up as quite a high probability value, sometimes even
# the highest probability in a sample. Instead, we use a large negative value, which ensures that the
# padding positions are correctly masked.
test_start_logits = test_predictions.start_logits.to_tensor(default_value=-1000).numpy()
test_end_logits = test_predictions.end_logits.to_tensor(default_value=-1000).numpy()
else:
test_start_logits = test_predictions.start_logits
test_end_logits = test_predictions.end_logits
post_processed_test = post_processing_function(
datasets["test"],
processed_datasets["test"],
(test_start_logits, test_end_logits),
)
metrics = compute_metrics(post_processed_test)
logging.info("Test metrics:")
for metric, value in metrics.items():
logging.info(f"{metric}: {value:.3f}")
# endregion
if training_args.output_dir is not None and not training_args.push_to_hub:
# If we're not pushing to hub, at least save a local copy when we're done
model.save_pretrained(training_args.output_dir)
if __name__ == "__main__":
main()
| #!/usr/bin/env python
# coding=utf-8
# Copyright 2020 The HuggingFace Team All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Fine-tuning the library models for question answering.
"""
# You can also adapt this script on your own question answering task. Pointers for this are left as comments.
import json
import logging
import os
import sys
from dataclasses import dataclass, field
from pathlib import Path
from typing import Optional
import evaluate
import tensorflow as tf
from datasets import load_dataset
from utils_qa import postprocess_qa_predictions
import transformers
from transformers import (
AutoConfig,
AutoTokenizer,
EvalPrediction,
HfArgumentParser,
PreTrainedTokenizerFast,
PushToHubCallback,
TFAutoModelForQuestionAnswering,
TFTrainingArguments,
create_optimizer,
set_seed,
)
from transformers.utils import CONFIG_NAME, TF2_WEIGHTS_NAME, check_min_version, send_example_telemetry
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.27.0.dev0")
logger = logging.getLogger(__name__)
# region Arguments
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
"""
model_name_or_path: str = field(
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
tokenizer_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None,
metadata={"help": "Path to directory to store the pretrained models downloaded from huggingface.co"},
)
model_revision: str = field(
default="main",
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
)
use_auth_token: bool = field(
default=False,
metadata={
"help": (
"Will use the token generated when running `huggingface-cli login` (necessary to use this script "
"with private models)."
)
},
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
"""
dataset_name: Optional[str] = field(
default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
)
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
validation_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."},
)
test_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input test data file to evaluate the perplexity on (a text file)."},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
preprocessing_num_workers: Optional[int] = field(
default=None,
metadata={"help": "The number of processes to use for the preprocessing."},
)
max_seq_length: int = field(
default=384,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": (
"Whether to pad all samples to `max_seq_length`. If False, will pad the samples dynamically when"
" batching to the maximum length in the batch (which can be faster on GPU but will be slower on TPU)."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
version_2_with_negative: bool = field(
default=False, metadata={"help": "If true, some of the examples do not have an answer."}
)
null_score_diff_threshold: float = field(
default=0.0,
metadata={
"help": (
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
)
},
)
doc_stride: int = field(
default=128,
metadata={"help": "When splitting up a long document into chunks, how much stride to take between chunks."},
)
n_best_size: int = field(
default=20,
metadata={"help": "The total number of n-best predictions to generate when looking for an answer."},
)
max_answer_length: int = field(
default=30,
metadata={
"help": (
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
)
},
)
def __post_init__(self):
if (
self.dataset_name is None
and self.train_file is None
and self.validation_file is None
and self.test_file is None
):
raise ValueError("Need either a dataset name or a training/validation file/test_file.")
else:
if self.train_file is not None:
extension = self.train_file.split(".")[-1]
assert extension in ["csv", "json"], "`train_file` should be a csv or a json file."
if self.validation_file is not None:
extension = self.validation_file.split(".")[-1]
assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
if self.test_file is not None:
extension = self.test_file.split(".")[-1]
assert extension in ["csv", "json"], "`test_file` should be a csv or a json file."
# endregion
# region Helper classes
class SavePretrainedCallback(tf.keras.callbacks.Callback):
# Hugging Face models have a save_pretrained() method that saves both the weights and the necessary
# metadata to allow them to be loaded as a pretrained model in future. This is a simple Keras callback
# that saves the model with this method after each epoch.
def __init__(self, output_dir, **kwargs):
super().__init__()
self.output_dir = output_dir
def on_epoch_end(self, epoch, logs=None):
self.model.save_pretrained(self.output_dir)
# endregion
def main():
# region Argument parsing
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TFTrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
# information sent is the one passed as arguments along with your Python/PyTorch versions.
send_example_telemetry("run_qa", model_args, data_args, framework="tensorflow")
output_dir = Path(training_args.output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
# endregion
# region Checkpoints
checkpoint = None
if len(os.listdir(training_args.output_dir)) > 0 and not training_args.overwrite_output_dir:
if (output_dir / CONFIG_NAME).is_file() and (output_dir / TF2_WEIGHTS_NAME).is_file():
checkpoint = output_dir
logger.info(
f"Checkpoint detected, resuming training from checkpoint in {training_args.output_dir}. To avoid this"
" behavior, change the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
)
else:
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty. "
"Use --overwrite_output_dir to continue regardless."
)
# endregion
# region Logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
handlers=[logging.StreamHandler(sys.stdout)],
)
logger.setLevel(logging.INFO if training_args.should_log else logging.WARN)
# Set the verbosity to info of the Transformers logger (on main process only):
if training_args.should_log:
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info(f"Training/evaluation parameters {training_args}")
# endregion
# Set seed before initializing model.
set_seed(training_args.seed)
# region Load Data
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
datasets = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
cache_dir=model_args.cache_dir,
use_auth_token=True if model_args.use_auth_token else None,
)
else:
data_files = {}
if data_args.train_file is not None:
data_files["train"] = data_args.train_file
extension = data_args.train_file.split(".")[-1]
if data_args.validation_file is not None:
data_files["validation"] = data_args.validation_file
extension = data_args.validation_file.split(".")[-1]
if data_args.test_file is not None:
data_files["test"] = data_args.test_file
extension = data_args.test_file.split(".")[-1]
datasets = load_dataset(
extension,
data_files=data_files,
field="data",
cache_dir=model_args.cache_dir,
use_auth_token=True if model_args.use_auth_token else None,
)
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# endregion
# region Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
config = AutoConfig.from_pretrained(
model_args.config_name if model_args.config_name else model_args.model_name_or_path,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
tokenizer = AutoTokenizer.from_pretrained(
model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path,
cache_dir=model_args.cache_dir,
use_fast=True,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
# endregion
# region Tokenizer check: this script requires a fast tokenizer.
if not isinstance(tokenizer, PreTrainedTokenizerFast):
raise ValueError(
"This example script only works for models that have a fast tokenizer. Checkout the big table of models at"
" https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet"
" this requirement"
)
# endregion
# region Preprocessing the datasets
# Preprocessing is slightly different for training and evaluation.
if training_args.do_train:
column_names = datasets["train"].column_names
elif training_args.do_eval:
column_names = datasets["validation"].column_names
else:
column_names = datasets["test"].column_names
question_column_name = "question" if "question" in column_names else column_names[0]
context_column_name = "context" if "context" in column_names else column_names[1]
answer_column_name = "answers" if "answers" in column_names else column_names[2]
# Padding side determines if we do (question|context) or (context|question).
pad_on_right = tokenizer.padding_side == "right"
if data_args.max_seq_length > tokenizer.model_max_length:
logger.warning(
f"The max_seq_length passed ({data_args.max_seq_length}) is larger than the maximum length for the"
f"model ({tokenizer.model_max_length}). Using max_seq_length={tokenizer.model_max_length}."
)
max_seq_length = min(data_args.max_seq_length, tokenizer.model_max_length)
if data_args.pad_to_max_length or isinstance(training_args.strategy, tf.distribute.TPUStrategy):
logger.info("Padding all batches to max length because argument was set or we're on TPU.")
padding = "max_length"
else:
padding = False
# Training preprocessing
def prepare_train_features(examples):
# Some of the questions have lots of whitespace on the left, which is not useful and will make the
# truncation of the context fail (the tokenized question will take a lots of space). So we remove that
# left whitespace
examples[question_column_name] = [q.lstrip() for q in examples[question_column_name]]
# Tokenize our examples with truncation and maybe padding, but keep the overflows using a stride. This results
# in one example possible giving several features when a context is long, each of those features having a
# context that overlaps a bit the context of the previous feature.
tokenized_examples = tokenizer(
examples[question_column_name if pad_on_right else context_column_name],
examples[context_column_name if pad_on_right else question_column_name],
truncation="only_second" if pad_on_right else "only_first",
max_length=max_seq_length,
stride=data_args.doc_stride,
return_overflowing_tokens=True,
return_offsets_mapping=True,
padding=padding,
)
# Since one example might give us several features if it has a long context, we need a map from a feature to
# its corresponding example. This key gives us just that.
sample_mapping = tokenized_examples.pop("overflow_to_sample_mapping")
# The offset mappings will give us a map from token to character position in the original context. This will
# help us compute the start_positions and end_positions.
offset_mapping = tokenized_examples.pop("offset_mapping")
# Let's label those examples!
tokenized_examples["start_positions"] = []
tokenized_examples["end_positions"] = []
for i, offsets in enumerate(offset_mapping):
# We will label impossible answers with the index of the CLS token.
input_ids = tokenized_examples["input_ids"][i]
cls_index = input_ids.index(tokenizer.cls_token_id)
# Grab the sequence corresponding to that example (to know what is the context and what is the question).
sequence_ids = tokenized_examples.sequence_ids(i)
# One example can give several spans, this is the index of the example containing this span of text.
sample_index = sample_mapping[i]
answers = examples[answer_column_name][sample_index]
# If no answers are given, set the cls_index as answer.
if len(answers["answer_start"]) == 0:
tokenized_examples["start_positions"].append(cls_index)
tokenized_examples["end_positions"].append(cls_index)
else:
# Start/end character index of the answer in the text.
start_char = answers["answer_start"][0]
end_char = start_char + len(answers["text"][0])
# Start token index of the current span in the text.
token_start_index = 0
while sequence_ids[token_start_index] != (1 if pad_on_right else 0):
token_start_index += 1
# End token index of the current span in the text.
token_end_index = len(input_ids) - 1
while sequence_ids[token_end_index] != (1 if pad_on_right else 0):
token_end_index -= 1
# Detect if the answer is out of the span (in which case this feature is labeled with the CLS index).
if not (offsets[token_start_index][0] <= start_char and offsets[token_end_index][1] >= end_char):
tokenized_examples["start_positions"].append(cls_index)
tokenized_examples["end_positions"].append(cls_index)
else:
# Otherwise move the token_start_index and token_end_index to the two ends of the answer.
# Note: we could go after the last offset if the answer is the last word (edge case).
while token_start_index < len(offsets) and offsets[token_start_index][0] <= start_char:
token_start_index += 1
tokenized_examples["start_positions"].append(token_start_index - 1)
while offsets[token_end_index][1] >= end_char:
token_end_index -= 1
tokenized_examples["end_positions"].append(token_end_index + 1)
return tokenized_examples
processed_datasets = {}
if training_args.do_train:
if "train" not in datasets:
raise ValueError("--do_train requires a train dataset")
train_dataset = datasets["train"]
if data_args.max_train_samples is not None:
# We will select sample from whole data if agument is specified
max_train_samples = min(len(train_dataset), data_args.max_train_samples)
train_dataset = train_dataset.select(range(max_train_samples))
# Create train feature from dataset
train_dataset = train_dataset.map(
prepare_train_features,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
)
if data_args.max_train_samples is not None:
# Number of samples might increase during Feature Creation, We select only specified max samples
max_train_samples = min(len(train_dataset), data_args.max_train_samples)
train_dataset = train_dataset.select(range(max_train_samples))
processed_datasets["train"] = train_dataset
# Validation preprocessing
def prepare_validation_features(examples):
# Some of the questions have lots of whitespace on the left, which is not useful and will make the
# truncation of the context fail (the tokenized question will take a lots of space). So we remove that
# left whitespace
examples[question_column_name] = [q.lstrip() for q in examples[question_column_name]]
# Tokenize our examples with truncation and maybe padding, but keep the overflows using a stride. This results
# in one example possible giving several features when a context is long, each of those features having a
# context that overlaps a bit the context of the previous feature.
tokenized_examples = tokenizer(
examples[question_column_name if pad_on_right else context_column_name],
examples[context_column_name if pad_on_right else question_column_name],
truncation="only_second" if pad_on_right else "only_first",
max_length=max_seq_length,
stride=data_args.doc_stride,
return_overflowing_tokens=True,
return_offsets_mapping=True,
padding=padding,
)
# Since one example might give us several features if it has a long context, we need a map from a feature to
# its corresponding example. This key gives us just that.
sample_mapping = tokenized_examples.pop("overflow_to_sample_mapping")
# For evaluation, we will need to convert our predictions to substrings of the context, so we keep the
# corresponding example_id and we will store the offset mappings.
tokenized_examples["example_id"] = []
for i in range(len(tokenized_examples["input_ids"])):
# Grab the sequence corresponding to that example (to know what is the context and what is the question).
sequence_ids = tokenized_examples.sequence_ids(i)
context_index = 1 if pad_on_right else 0
# One example can give several spans, this is the index of the example containing this span of text.
sample_index = sample_mapping[i]
tokenized_examples["example_id"].append(examples["id"][sample_index])
# Set to None the offset_mapping that are not part of the context so it's easy to determine if a token
# position is part of the context or not.
tokenized_examples["offset_mapping"][i] = [
(o if sequence_ids[k] == context_index else None)
for k, o in enumerate(tokenized_examples["offset_mapping"][i])
]
return tokenized_examples
if training_args.do_eval:
if "validation" not in datasets:
raise ValueError("--do_eval requires a validation dataset")
eval_examples = datasets["validation"]
if data_args.max_eval_samples is not None:
# We will select sample from whole data
max_eval_samples = min(len(eval_examples), data_args.max_eval_samples)
eval_examples = eval_examples.select(range(max_eval_samples))
# Validation Feature Creation
eval_dataset = eval_examples.map(
prepare_validation_features,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
)
if data_args.max_eval_samples is not None:
# During Feature creation dataset samples might increase, we will select required samples again
max_eval_samples = min(len(eval_dataset), data_args.max_eval_samples)
eval_dataset = eval_dataset.select(range(max_eval_samples))
processed_datasets["validation"] = eval_dataset
if training_args.do_predict:
if "test" not in datasets:
raise ValueError("--do_predict requires a test dataset")
predict_examples = datasets["test"]
if data_args.max_predict_samples is not None:
# We will select sample from whole data
predict_examples = predict_examples.select(range(data_args.max_predict_samples))
# Predict Feature Creation
predict_dataset = predict_examples.map(
prepare_validation_features,
batched=True,
num_proc=data_args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not data_args.overwrite_cache,
)
if data_args.max_predict_samples is not None:
# During Feature creation dataset samples might increase, we will select required samples again
max_predict_samples = min(len(predict_dataset), data_args.max_predict_samples)
predict_dataset = predict_dataset.select(range(max_predict_samples))
processed_datasets["test"] = predict_dataset
# endregion
# region Metrics and Post-processing:
def post_processing_function(examples, features, predictions, stage="eval"):
# Post-processing: we match the start logits and end logits to answers in the original context.
predictions = postprocess_qa_predictions(
examples=examples,
features=features,
predictions=predictions,
version_2_with_negative=data_args.version_2_with_negative,
n_best_size=data_args.n_best_size,
max_answer_length=data_args.max_answer_length,
null_score_diff_threshold=data_args.null_score_diff_threshold,
output_dir=training_args.output_dir,
prefix=stage,
)
# Format the result to the format the metric expects.
if data_args.version_2_with_negative:
formatted_predictions = [
{"id": k, "prediction_text": v, "no_answer_probability": 0.0} for k, v in predictions.items()
]
else:
formatted_predictions = [{"id": k, "prediction_text": v} for k, v in predictions.items()]
references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
return EvalPrediction(predictions=formatted_predictions, label_ids=references)
metric = evaluate.load("squad_v2" if data_args.version_2_with_negative else "squad")
def compute_metrics(p: EvalPrediction):
return metric.compute(predictions=p.predictions, references=p.label_ids)
# endregion
with training_args.strategy.scope():
dataset_options = tf.data.Options()
dataset_options.experimental_distribute.auto_shard_policy = tf.data.experimental.AutoShardPolicy.OFF
num_replicas = training_args.strategy.num_replicas_in_sync
# region Load model and prepare datasets
if checkpoint is None:
model_path = model_args.model_name_or_path
else:
model_path = checkpoint
model = TFAutoModelForQuestionAnswering.from_pretrained(
model_path,
config=config,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
if training_args.do_train:
training_dataset = model.prepare_tf_dataset(
processed_datasets["train"],
shuffle=True,
batch_size=training_args.per_device_train_batch_size * num_replicas,
tokenizer=tokenizer,
)
training_dataset = training_dataset.with_options(dataset_options)
num_train_steps = len(training_dataset) * training_args.num_train_epochs
if training_args.warmup_steps > 0:
num_warmup_steps = training_args.warmup_steps
elif training_args.warmup_ratio > 0:
num_warmup_steps = int(num_train_steps * training_args.warmup_ratio)
else:
num_warmup_steps = 0
optimizer, schedule = create_optimizer(
init_lr=training_args.learning_rate,
num_train_steps=len(training_dataset) * training_args.num_train_epochs,
num_warmup_steps=num_warmup_steps,
adam_beta1=training_args.adam_beta1,
adam_beta2=training_args.adam_beta2,
adam_epsilon=training_args.adam_epsilon,
weight_decay_rate=training_args.weight_decay,
adam_global_clipnorm=training_args.max_grad_norm,
)
# no user-specified loss = will use the model internal loss
model.compile(optimizer=optimizer, jit_compile=training_args.xla, metrics=["accuracy"])
else:
model.compile(optimizer=None, jit_compile=training_args.xla, metrics=["accuracy"])
training_dataset = None
if training_args.do_eval:
eval_dataset = model.prepare_tf_dataset(
processed_datasets["validation"],
shuffle=False,
batch_size=training_args.per_device_train_batch_size * num_replicas,
tokenizer=tokenizer,
)
eval_dataset = eval_dataset.with_options(dataset_options)
else:
eval_dataset = None
if training_args.do_predict:
predict_dataset = model.prepare_tf_dataset(
processed_datasets["test"],
shuffle=False,
batch_size=training_args.per_device_eval_batch_size * num_replicas,
tokenizer=tokenizer,
)
predict_dataset = predict_dataset.with_options(dataset_options)
else:
predict_dataset = None
# endregion
# region Preparing push_to_hub and model card
push_to_hub_model_id = training_args.push_to_hub_model_id
model_name = model_args.model_name_or_path.split("/")[-1]
if not push_to_hub_model_id:
if data_args.dataset_name is not None:
push_to_hub_model_id = f"{model_name}-finetuned-{data_args.dataset_name}"
else:
push_to_hub_model_id = f"{model_name}-finetuned-question-answering"
model_card_kwargs = {"finetuned_from": model_args.model_name_or_path, "tasks": "question-answering"}
if data_args.dataset_name is not None:
model_card_kwargs["dataset_tags"] = data_args.dataset_name
if data_args.dataset_config_name is not None:
model_card_kwargs["dataset_args"] = data_args.dataset_config_name
model_card_kwargs["dataset"] = f"{data_args.dataset_name} {data_args.dataset_config_name}"
else:
model_card_kwargs["dataset"] = data_args.dataset_name
if training_args.push_to_hub:
callbacks = [
PushToHubCallback(
output_dir=training_args.output_dir,
model_id=push_to_hub_model_id,
organization=training_args.push_to_hub_organization,
token=training_args.push_to_hub_token,
tokenizer=tokenizer,
**model_card_kwargs,
)
]
else:
callbacks = []
# endregion
# region Training and Evaluation
if training_args.do_train:
# Note that the validation and test datasets have been processed in a different way to the
# training datasets in this example, and so they don't have the same label structure.
# As such, we don't pass them directly to Keras, but instead get model predictions to evaluate
# after training.
model.fit(training_dataset, epochs=int(training_args.num_train_epochs), callbacks=callbacks)
if training_args.do_eval:
logger.info("*** Evaluation ***")
# In this example, we compute advanced metrics at the end of training, but
# if you'd like to compute metrics every epoch that are too complex to be written as
# standard Keras metrics, you can use our KerasMetricCallback. See
# https://huggingface.co/docs/transformers/main/en/main_classes/keras_callbacks
eval_predictions = model.predict(eval_dataset)
if isinstance(eval_predictions.start_logits, tf.RaggedTensor):
# If predictions are RaggedTensor, we densify them. Since they are logits, padding with 0 is a bad idea!
# The reason is that a logit of 0 can often end up as quite a high probability value, sometimes even
# the highest probability in a sample. Instead, we use a large negative value, which ensures that the
# padding positions are correctly masked.
eval_start_logits = eval_predictions.start_logits.to_tensor(default_value=-1000).numpy()
eval_end_logits = eval_predictions.end_logits.to_tensor(default_value=-1000).numpy()
else:
eval_start_logits = eval_predictions.start_logits
eval_end_logits = eval_predictions.end_logits
post_processed_eval = post_processing_function(
datasets["validation"],
processed_datasets["validation"],
(eval_start_logits, eval_end_logits),
)
metrics = compute_metrics(post_processed_eval)
logging.info("Evaluation metrics:")
for metric, value in metrics.items():
logging.info(f"{metric}: {value:.3f}")
if training_args.output_dir is not None:
output_eval_file = os.path.join(training_args.output_dir, "all_results.json")
with open(output_eval_file, "w") as writer:
writer.write(json.dumps(metrics))
# endregion
# region Prediction
if training_args.do_predict:
logger.info("*** Predict ***")
test_predictions = model.predict(predict_dataset)
if isinstance(test_predictions.start_logits, tf.RaggedTensor):
# If predictions are RaggedTensor, we densify them. Since they are logits, padding with 0 is a bad idea!
# The reason is that a logit of 0 can often end up as quite a high probability value, sometimes even
# the highest probability in a sample. Instead, we use a large negative value, which ensures that the
# padding positions are correctly masked.
test_start_logits = test_predictions.start_logits.to_tensor(default_value=-1000).numpy()
test_end_logits = test_predictions.end_logits.to_tensor(default_value=-1000).numpy()
else:
test_start_logits = test_predictions.start_logits
test_end_logits = test_predictions.end_logits
post_processed_test = post_processing_function(
datasets["test"],
processed_datasets["test"],
(test_start_logits, test_end_logits),
)
metrics = compute_metrics(post_processed_test)
logging.info("Test metrics:")
for metric, value in metrics.items():
logging.info(f"{metric}: {value:.3f}")
# endregion
if training_args.output_dir is not None and not training_args.push_to_hub:
# If we're not pushing to hub, at least save a local copy when we're done
model.save_pretrained(training_args.output_dir)
if __name__ == "__main__":
main()
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/commands/run.py | # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from ..pipelines import Pipeline, PipelineDataFormat, get_supported_tasks, pipeline
from ..utils import logging
from . import BaseTransformersCLICommand
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
def try_infer_format_from_ext(path: str):
if not path:
return "pipe"
for ext in PipelineDataFormat.SUPPORTED_FORMATS:
if path.endswith(ext):
return ext
raise Exception(
f"Unable to determine file format from file extension {path}. "
f"Please provide the format through --format {PipelineDataFormat.SUPPORTED_FORMATS}"
)
def run_command_factory(args):
nlp = pipeline(
task=args.task,
model=args.model if args.model else None,
config=args.config,
tokenizer=args.tokenizer,
device=args.device,
)
format = try_infer_format_from_ext(args.input) if args.format == "infer" else args.format
reader = PipelineDataFormat.from_str(
format=format,
output_path=args.output,
input_path=args.input,
column=args.column if args.column else nlp.default_input_names,
overwrite=args.overwrite,
)
return RunCommand(nlp, reader)
class RunCommand(BaseTransformersCLICommand):
def __init__(self, nlp: Pipeline, reader: PipelineDataFormat):
self._nlp = nlp
self._reader = reader
@staticmethod
def register_subcommand(parser: ArgumentParser):
run_parser = parser.add_parser("run", help="Run a pipeline through the CLI")
run_parser.add_argument("--task", choices=get_supported_tasks(), help="Task to run")
run_parser.add_argument("--input", type=str, help="Path to the file to use for inference")
run_parser.add_argument("--output", type=str, help="Path to the file that will be used post to write results.")
run_parser.add_argument("--model", type=str, help="Name or path to the model to instantiate.")
run_parser.add_argument("--config", type=str, help="Name or path to the model's config to instantiate.")
run_parser.add_argument(
"--tokenizer", type=str, help="Name of the tokenizer to use. (default: same as the model name)"
)
run_parser.add_argument(
"--column",
type=str,
help="Name of the column to use as input. (For multi columns input as QA use column1,columns2)",
)
run_parser.add_argument(
"--format",
type=str,
default="infer",
choices=PipelineDataFormat.SUPPORTED_FORMATS,
help="Input format to read from",
)
run_parser.add_argument(
"--device",
type=int,
default=-1,
help="Indicate the device to run onto, -1 indicates CPU, >= 0 indicates GPU (default: -1)",
)
run_parser.add_argument("--overwrite", action="store_true", help="Allow overwriting the output file.")
run_parser.set_defaults(func=run_command_factory)
def run(self):
nlp, outputs = self._nlp, []
for entry in self._reader:
output = nlp(**entry) if self._reader.is_multi_columns else nlp(entry)
if isinstance(output, dict):
outputs.append(output)
else:
outputs += output
# Saving data
if self._nlp.binary_output:
binary_path = self._reader.save_binary(outputs)
logger.warning(f"Current pipeline requires output to be in binary format, saving at {binary_path}")
else:
self._reader.save(outputs)
| # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from ..pipelines import Pipeline, PipelineDataFormat, get_supported_tasks, pipeline
from ..utils import logging
from . import BaseTransformersCLICommand
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
def try_infer_format_from_ext(path: str):
if not path:
return "pipe"
for ext in PipelineDataFormat.SUPPORTED_FORMATS:
if path.endswith(ext):
return ext
raise Exception(
f"Unable to determine file format from file extension {path}. "
f"Please provide the format through --format {PipelineDataFormat.SUPPORTED_FORMATS}"
)
def run_command_factory(args):
nlp = pipeline(
task=args.task,
model=args.model if args.model else None,
config=args.config,
tokenizer=args.tokenizer,
device=args.device,
)
format = try_infer_format_from_ext(args.input) if args.format == "infer" else args.format
reader = PipelineDataFormat.from_str(
format=format,
output_path=args.output,
input_path=args.input,
column=args.column if args.column else nlp.default_input_names,
overwrite=args.overwrite,
)
return RunCommand(nlp, reader)
class RunCommand(BaseTransformersCLICommand):
def __init__(self, nlp: Pipeline, reader: PipelineDataFormat):
self._nlp = nlp
self._reader = reader
@staticmethod
def register_subcommand(parser: ArgumentParser):
run_parser = parser.add_parser("run", help="Run a pipeline through the CLI")
run_parser.add_argument("--task", choices=get_supported_tasks(), help="Task to run")
run_parser.add_argument("--input", type=str, help="Path to the file to use for inference")
run_parser.add_argument("--output", type=str, help="Path to the file that will be used post to write results.")
run_parser.add_argument("--model", type=str, help="Name or path to the model to instantiate.")
run_parser.add_argument("--config", type=str, help="Name or path to the model's config to instantiate.")
run_parser.add_argument(
"--tokenizer", type=str, help="Name of the tokenizer to use. (default: same as the model name)"
)
run_parser.add_argument(
"--column",
type=str,
help="Name of the column to use as input. (For multi columns input as QA use column1,columns2)",
)
run_parser.add_argument(
"--format",
type=str,
default="infer",
choices=PipelineDataFormat.SUPPORTED_FORMATS,
help="Input format to read from",
)
run_parser.add_argument(
"--device",
type=int,
default=-1,
help="Indicate the device to run onto, -1 indicates CPU, >= 0 indicates GPU (default: -1)",
)
run_parser.add_argument("--overwrite", action="store_true", help="Allow overwriting the output file.")
run_parser.set_defaults(func=run_command_factory)
def run(self):
nlp, outputs = self._nlp, []
for entry in self._reader:
output = nlp(**entry) if self._reader.is_multi_columns else nlp(entry)
if isinstance(output, dict):
outputs.append(output)
else:
outputs += output
# Saving data
if self._nlp.binary_output:
binary_path = self._reader.save_binary(outputs)
logger.warning(f"Current pipeline requires output to be in binary format, saving at {binary_path}")
else:
self._reader.save(outputs)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/pytorch/speech-recognition/requirements.txt | datasets >= 1.18.0
torch >= 1.5
torchaudio
librosa
jiwer
evaluate
| datasets >= 1.18.0
torch >= 1.5
torchaudio
librosa
jiwer
evaluate
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./README_zh-hans.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Chinese translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Chinese characters. E.g., 共 100 多种语言; 使用 transformers 库。
- Use square quotes, e.g.,「引用」
Dictionary
Hugging Face: 抱抱脸
token: 词符（并用括号标注原英文）
tokenize: 词符化（并用括号标注原英文）
tokenizer: 词符化器（并用括号标注原英文）
transformer: transformer（不翻译）
pipeline: 流水线
API: API (不翻译）
inference: 推理
Trainer: 训练器。当作为类名出现时不翻译。
pretrained/pretrain: 预训练
finetune: 微调
community: 社区
example: 当特指仓库中 example 目录时翻译为「用例」
Python data structures (e.g., list, set, dict): 翻译为列表，集合，词典，并用括号标注原英文
NLP/Natural Language Processing: 以 NLP 出现时不翻译，以 Natural Language Processing 出现时翻译为自然语言处理
checkpoint: 检查点
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<b>简体中文</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>为 Jax、PyTorch 和 TensorFlow 打造的先进的自然语言处理</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 提供了数以千计的预训练模型，支持 100 多种语言的文本分类、信息抽取、问答、摘要、翻译、文本生成。它的宗旨是让最先进的 NLP 技术人人易用。
🤗 Transformers 提供了便于快速下载和使用的API，让你可以把预训练模型用在给定文本、在你的数据集上微调然后通过 [model hub](https://huggingface.co/models) 与社区共享。同时，每个定义的 Python 模块均完全独立，方便修改和快速研究实验。
🤗 Transformers 支持三个最热门的深度学习库： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) 以及 [TensorFlow](https://www.tensorflow.org/) — 并与之无缝整合。你可以直接使用一个框架训练你的模型然后用另一个加载和推理。
## 在线演示
你可以直接在模型页面上测试大多数 [model hub](https://huggingface.co/models) 上的模型。 我们也提供了 [私有模型托管、模型版本管理以及推理API](https://huggingface.co/pricing)。
这里是一些例子：
- [用 BERT 做掩码填词](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [用 Electra 做命名实体识别](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [用 GPT-2 做文本生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [用 RoBERTa 做自然语言推理](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [用 BART 做文本摘要](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [用 DistilBERT 做问答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [用 T5 做翻译](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，由抱抱脸团队打造，是一个文本生成的官方 demo。
## 如果你在寻找由抱抱脸团队提供的定制化支持服务
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 快速上手
我们为快速使用模型提供了 `pipeline` （流水线）API。流水线聚合了预训练模型和对应的文本预处理。下面是一个快速使用流水线去判断正负面情绪的例子：
```python
>>> from transformers import pipeline
# 使用情绪分析流水线
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
第二行代码下载并缓存了流水线使用的预训练模型，而第三行代码则在给定的文本上进行了评估。这里的答案“正面” (positive) 具有 99 的置信度。
许多的 NLP 任务都有开箱即用的预训练流水线。比如说，我们可以轻松的从给定文本中抽取问题答案：
``` python
>>> from transformers import pipeline
# 使用问答流水线
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
除了给出答案，预训练模型还给出了对应的置信度分数、答案在词符化 (tokenized) 后的文本中开始和结束的位置。你可以从[这个教程](https://huggingface.co/docs/transformers/task_summary)了解更多流水线API支持的任务。
要在你的任务上下载和使用任意预训练模型也很简单，只需三行代码。这里是 PyTorch 版的示例：
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
这里是等效的 TensorFlow 代码：
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
词符化器 (tokenizer) 为所有的预训练模型提供了预处理，并可以直接对单个字符串进行调用（比如上面的例子）或对列表 (list) 调用。它会输出一个你可以在下游代码里使用或直接通过 `**` 解包表达式传给模型的词典 (dict)。
模型本身是一个常规的 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) 或 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)（取决于你的后端），可以常规方式使用。 [这个教程](https://huggingface.co/transformers/training.html)解释了如何将这样的模型整合到经典的 PyTorch 或 TensorFlow 训练循环中，或是如何使用我们的 `Trainer` 训练器）API 来在一个新的数据集上快速微调。
## 为什么要用 transformers？
1. 便于使用的先进模型：
- NLU 和 NLG 上表现优越
- 对教学和实践友好且低门槛
- 高级抽象，只需了解三个类
- 对所有模型统一的API
1. 更低计算开销，更少的碳排放：
- 研究人员可以分享已训练的模型而非每次从头开始训练
- 工程师可以减少计算用时和生产环境开销
- 数十种模型架构、两千多个预训练模型、100多种语言支持
1. 对于模型生命周期的每一个部分都面面俱到：
- 训练先进的模型，只需 3 行代码
- 模型在不同深度学习框架间任意转移，随你心意
- 为训练、评估和生产选择最适合的框架，衔接无缝
1. 为你的需求轻松定制专属模型和用例：
- 我们为每种模型架构提供了多个用例来复现原论文结果
- 模型内部结构保持透明一致
- 模型文件可单独使用，方便魔改和快速实验
## 什么情况下我不该用 transformers？
- 本库并不是模块化的神经网络工具箱。模型文件中的代码特意呈若璞玉，未经额外抽象封装，以便研究人员快速迭代魔改而不致溺于抽象和文件跳转之中。
- `Trainer` API 并非兼容任何模型，只为本库之模型优化。若是在寻找适用于通用机器学习的训练循环实现，请另觅他库。
- 尽管我们已尽力而为，[examples 目录](https://github.com/huggingface/transformers/tree/main/examples)中的脚本也仅为用例而已。对于你的特定问题，它们并不一定开箱即用，可能需要改几行代码以适之。
## 安装
### 使用 pip
这个仓库已在 Python 3.6+、Flax 0.3.2+、PyTorch 1.3.1+ 和 TensorFlow 2.3+ 下经过测试。
你可以在[虚拟环境](https://docs.python.org/3/library/venv.html)中安装 🤗 Transformers。如果你还不熟悉 Python 的虚拟环境，请阅此[用户说明](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
首先，用你打算使用的版本的 Python 创建一个虚拟环境并激活。
然后，你需要安装 Flax、PyTorch 或 TensorFlow 其中之一。关于在你使用的平台上安装这些框架，请参阅 [TensorFlow 安装页](https://www.tensorflow.org/install/), [PyTorch 安装页](https://pytorch.org/get-started/locally/#start-locally) 或 [Flax 安装页](https://github.com/google/flax#quick-install)。
当这些后端之一安装成功后， 🤗 Transformers 可依此安装：
```bash
pip install transformers
```
如果你想要试试用例或者想在正式发布前使用最新的开发中代码，你得[从源代码安装](https://huggingface.co/docs/transformers/installation#installing-from-source)。
### 使用 conda
自 Transformers 4.0.0 版始，我们有了一个 conda 频道： `huggingface`。
🤗 Transformers 可以通过 conda 依此安装：
```shell script
conda install -c huggingface transformers
```
要通过 conda 安装 Flax、PyTorch 或 TensorFlow 其中之一，请参阅它们各自安装页的说明。
## 模型架构
🤗 Transformers 支持的[**所有的模型检查点**](https://huggingface.co/models)由[用户](https://huggingface.co/users)和[组织](https://huggingface.co/organizations)上传，均与 huggingface.co [model hub](https://huggingface.co) 无缝整合。
目前的检查点数量： 
🤗 Transformers 目前支持如下的架构（模型概述请阅[这里](https://huggingface.co/docs/transformers/model_summary)）：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (来自 Google Research and the Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。
1. **[AltCLIP](https://huggingface.co/docs/transformers/model_doc/altclip)** (来自 BAAI) 伴随论文 [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://arxiv.org/abs/2211.06679) 由 Chen, Zhongzhi and Liu, Guang and Zhang, Bo-Wen and Ye, Fulong and Yang, Qinghong and Wu, Ledell 发布。
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (来自 MIT) 伴随论文 [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) 由 Yuan Gong, Yu-An Chung, James Glass 发布。
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (来自 Facebook) 伴随论文 [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) 由 Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer 发布。
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (来自 École polytechnique) 伴随论文 [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) 由 Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis 发布。
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (来自 VinAI Research) 伴随论文 [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) 由 Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen 发布。
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (来自 Microsoft) 伴随论文 [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) 由 Hangbo Bao, Li Dong, Furu Wei 发布。
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (来自 Google) 伴随论文 [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) 由 Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova 发布。
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (来自 Google) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (来自 VinAI Research) 伴随论文 [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) 由 Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen 发布。
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BioGpt](https://huggingface.co/docs/transformers/model_doc/biogpt)** (来自 Microsoft Research AI4Science) 伴随论文 [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) 由 Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu 发布。
1. **[BiT](https://huggingface.co/docs/transformers/model_doc/bit)** (来自 Google AI) 伴随论文 [Big Transfer (BiT) 由 Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby 发布。
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BLIP](https://huggingface.co/docs/transformers/model_doc/blip)** (来自 Salesforce) 伴随论文 [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://arxiv.org/abs/2201.12086) 由 Junnan Li, Dongxu Li, Caiming Xiong, Steven Hoi 发布。
1. **[BLIP-2](https://huggingface.co/docs/transformers/main/model_doc/blip-2)** (来自 Salesforce) 伴随论文 [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://arxiv.org/abs/2301.12597) 由 Junnan Li, Dongxu Li, Silvio Savarese, Steven Hoi 发布。
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (来自 Alexa) 伴随论文 [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) 由 Adrian de Wynter and Daniel J. Perry 发布。
1. **[BridgeTower](https://huggingface.co/docs/transformers/main/model_doc/bridgetower)** (from Harbin Institute of Technology/Microsoft Research Asia/Intel Labs) released with the paper [BridgeTower: Building Bridges Between Encoders in Vision-Language Representation Learning](https://arxiv.org/abs/2206.08657) by Xiao Xu, Chenfei Wu, Shachar Rosenman, Vasudev Lal, Wanxiang Che, Nan Duan.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (来自 Google Research) 伴随论文 [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) 由 Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel 发布。
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (来自 Inria/Facebook/Sorbonne) 伴随论文 [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) 由 Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot 发布。
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (来自 Google Research) 伴随论文 [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) 由 Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting 发布。
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (来自 OFA-Sys) 伴随论文 [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) 由 An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou 发布。
1. **[CLAP](https://huggingface.co/docs/transformers/main/model_doc/clap)** (来自 LAION-AI) 伴随论文 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation]https://arxiv.org/abs/2211.06687) 由 Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov 发布。
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (来自 Microsoft Research Asia) 伴随论文 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 由 Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 发布。
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (来自 YituTech) 伴随论文 [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) 由 Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan 发布。
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (来自 Facebook AI) 伴随论文 [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) 由 Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie 发布。
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (来自 Tsinghua University) 伴随论文 [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) 由 Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun 发布。
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (来自 Salesforce) 伴随论文 [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) 由 Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher 发布。
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (来自 Microsoft) 伴随论文 [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) 由 Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang 发布。
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (来自 Facebook) 伴随论文 [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) 由 Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli 发布。
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (来自 Berkeley/Facebook/Google) 伴随论文 [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) 由 Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch 发布。
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (来自 SenseTime Research) 伴随论文 [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) 由 Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai 发布。
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (来自 Facebook) 伴随论文 [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) 由 Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou 发布。
1. **[DETA](https://huggingface.co/docs/transformers/main/model_doc/deta)** (来自 The University of Texas at Austin) 伴随论文 [NMS Strikes Back](https://arxiv.org/abs/2212.06137) 由 Jeffrey Ouyang-Zhang, Jang Hyun Cho, Xingyi Zhou, Philipp Krähenbühl 发布。
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (来自 Facebook) 伴随论文 [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) 由 Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko 发布。
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (来自 Microsoft Research) 伴随论文 [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) 由 Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan 发布。
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (来自 SHI Labs) 伴随论文 [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) 由 Ali Hassani and Humphrey Shi 发布。
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (来自 HuggingFace), 伴随论文 [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) 由 Victor Sanh, Lysandre Debut and Thomas Wolf 发布。 同样的方法也应用于压缩 GPT-2 到 [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa 到 [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT 到 [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) 和德语版 DistilBERT。
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (来自 Microsoft Research) 伴随论文 [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) 由 Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei 发布。
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (来自 NAVER) 伴随论文 [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) 由 Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park 发布。
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (来自 Facebook) 伴随论文 [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) 由 Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih 发布。
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (来自 Intel Labs) 伴随论文 [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) 由 René Ranftl, Alexey Bochkovskiy, Vladlen Koltun 发布。
1. **[EfficientFormer](https://huggingface.co/docs/transformers/model_doc/efficientformer)** (来自 Snap Research) 伴随论文 [EfficientFormer: Vision Transformers at MobileNetSpeed](https://arxiv.org/abs/2206.01191) 由 Yanyu Li, Geng Yuan, Yang Wen, Ju Hu, Georgios Evangelidis, Sergey Tulyakov, Yanzhi Wang, Jian Ren 发布。
1. **[EfficientNet](https://huggingface.co/docs/transformers/main/model_doc/efficientnet)** (from Google Brain) released with the paper [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://arxiv.org/abs/1905.11946) by Mingxing Tan, Quoc V. Le.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (来自 Google Research/Stanford University) 伴随论文 [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) 由 Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning 发布。
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (来自 Google Research) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (来自 Baidu) 伴随论文 [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu 发布。
1. **[ErnieM](https://huggingface.co/docs/transformers/main/model_doc/ernie_m)** (来自 Baidu) 伴随论文 [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) 由 Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang 发布。
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (来自 CNRS) 伴随论文 [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) 由 Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab 发布。
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (来自 Facebook AI) 伴随论文 [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) 由 Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela 发布。
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (来自 Google Research) 伴随论文 [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) 由 James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon 发布。
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (来自 CMU/Google Brain) 伴随论文 [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) 由 Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le 发布。
1. **[GIT](https://huggingface.co/docs/transformers/model_doc/git)** (来自 Microsoft Research) 伴随论文 [GIT: A Generative Image-to-text Transformer for Vision and Language](https://arxiv.org/abs/2205.14100) 由 Jianfeng Wang, Zhengyuan Yang, Xiaowei Hu, Linjie Li, Kevin Lin, Zhe Gan, Zicheng Liu, Ce Liu, Lijuan Wang 发布。
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (来自 KAIST) 伴随论文 [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) 由 Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim 发布。
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (来自 OpenAI) 伴随论文 [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) 由 Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever 发布。
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (来自 EleutherAI) 随仓库 [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) 发布。作者为 Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy 发布。
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (来自 ABEJA) 由 Shinya Otani, Takayoshi Makabe, Anuj Arora, Kyo Hattori。
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (来自 OpenAI) 伴随论文 [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) 由 Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever** 发布。
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (来自 EleutherAI) 伴随论文 [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) 由 Ben Wang and Aran Komatsuzaki 发布。
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GPTSAN-japanese](https://huggingface.co/docs/transformers/main/model_doc/gptsan-japanese)** released in the repository [tanreinama/GPTSAN](https://github.com/tanreinama/GPTSAN/blob/main/report/model.md) by 坂本俊之(tanreinama).
1. **[Graphormer](https://huggingface.co/docs/transformers/model_doc/graphormer)** (from Microsoft) released with the paper [Do Transformers Really Perform Bad for Graph Representation?](https://arxiv.org/abs/2106.05234) by Chengxuan Ying, Tianle Cai, Shengjie Luo, Shuxin Zheng, Guolin Ke, Di He, Yanming Shen, Tie-Yan Liu.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (来自 UCSD, NVIDIA) 伴随论文 [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) 由 Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang 发布。
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (来自 Facebook) 伴随论文 [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) 由 Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed 发布。
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (来自 Berkeley) 伴随论文 [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) 由 Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer 发布。
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (来自 OpenAI) 伴随论文 [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) 由 Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever 发布。
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) 由 Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou 发布。
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) 由 Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou 发布。
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) 由 Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei 发布。
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) 由 Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei 发布。
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (来自 Meta AI) 伴随论文 [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) 由 Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze 发布。
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (来自 South China University of Technology) 伴随论文 [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) 由 Jiapeng Wang, Lianwen Jin, Kai Ding 发布。
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (来自 Google AI) released 伴随论文 [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) 由 Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang 发布。
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (来自 Studio Ousia) 伴随论文 [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) 由 Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto 发布。
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (来自 UNC Chapel Hill) 伴随论文 [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) 由 Hao Tan and Mohit Bansal 发布。
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (来自 Facebook) 伴随论文 [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) 由 Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert 发布。
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (来自 Facebook) 伴随论文 [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) 由 Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin 发布。
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** 用 [OPUS](http://opus.nlpl.eu/) 数据训练的机器翻译模型由 Jörg Tiedemann 发布。[Marian Framework](https://marian-nmt.github.io/) 由微软翻译团队开发。
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (来自 Microsoft Research Asia) 伴随论文 [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) 由 Junlong Li, Yiheng Xu, Lei Cui, Furu Wei 发布。
1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (来自 FAIR and UIUC) 伴随论文 [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) 由 Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar 发布。
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) 由 Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer 发布。
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) 由 Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan 发布。
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (来自 Studio Ousia) 伴随论文 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 由 Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 发布。
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (来自 CMU/Google Brain) 伴随论文 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 由 Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 发布。
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (来自 Google Inc.) 伴随论文 [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) 由 Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam 发布。
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (来自 Google Inc.) 伴随论文 [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) 由 Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen 发布。
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (来自 Apple) 伴随论文 [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) 由 Sachin Mehta and Mohammad Rastegari 发布。
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (来自 Microsoft Research) 伴随论文 [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) 由 Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu 发布。
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (来自 Google AI) 伴随论文 [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) 由 Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel 发布。
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (来自 中国人民大学 AI Box) 伴随论文 [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) 由 Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen 发布。
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (来自 SHI Labs) 伴随论文 [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) 由 Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi 发布。
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (来自华为诺亚方舟实验室) 伴随论文 [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) 由 Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu 发布。
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (来自 Meta) 伴随论文 [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) 由 the NLLB team 发布。
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (来自 the University of Wisconsin - Madison) 伴随论文 [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) 由 Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh 发布。
1. **[OneFormer](https://huggingface.co/docs/transformers/model_doc/oneformer)** (来自 SHI Labs) 伴随论文 [OneFormer: One Transformer to Rule Universal Image Segmentation](https://arxiv.org/abs/2211.06220) 由 Jitesh Jain, Jiachen Li, MangTik Chiu, Ali Hassani, Nikita Orlov, Humphrey Shi 发布。
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (来自 Meta AI) 伴随论文 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 由 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 发布。
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (来自 Google AI) 伴随论文 [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) 由 Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby 发布。
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (来自 Google) 伴随论文 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 由 Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 发布。
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (来自 Google) 伴随论文 [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) 由 Jason Phang, Yao Zhao, Peter J. Liu 发布。
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (来自 Deepmind) 伴随论文 [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) 由 Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira 发布。
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (来自 VinAI Research) 伴随论文 [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) 由 Dat Quoc Nguyen and Anh Tuan Nguyen 发布。
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (来自 UCLA NLP) 伴随论文 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 由 Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 发布。
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (来自 Sea AI Labs) 伴随论文 [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) 由 Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng 发布。
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (来自 NVIDIA) 伴随论文 [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) 由 Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius 发布。
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (来自 Facebook) 伴随论文 [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) 由 Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela 发布。
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (来自 Google Research) 伴随论文 [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) 由 Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang 发布。
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (来自 Google Research) 伴随论文 [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) 由 Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya 发布。
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (来自 Google Research) 伴随论文 [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) 由 Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder 发布。
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (来自 Facebook), 伴随论文 [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) 由 Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov 发布。
1. **[RoBERTa-PreLayerNorm](https://huggingface.co/docs/transformers/model_doc/roberta-prelayernorm)** (来自 Facebook) 伴随论文 [fairseq: A Fast, Extensible Toolkit for Sequence Modeling](https://arxiv.org/abs/1904.01038) 由 Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, Michael Auli 发布。
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (来自 WeChatAI), 伴随论文 [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) 由 HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou 发布。
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (来自 ZhuiyiTechnology), 伴随论文 [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) 由 Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu 发布。
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (来自 NVIDIA) 伴随论文 [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) 由 Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo 发布。
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SpeechT5](https://huggingface.co/docs/transformers/main/model_doc/speecht5)** (来自 Microsoft Research) 伴随论文 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) 由 Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei 发布。
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (来自 Facebook), 伴随论文 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 发布。
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (来自 Facebook) 伴随论文 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 由 Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 发布。
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (来自 Tel Aviv University) 伴随论文 [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) 由 Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy 发布。
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (来自 Berkeley) 伴随论文 [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) 由 Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer 发布。
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (来自 Microsoft) 伴随论文 [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) 由 Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo 发布。
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (来自 Microsoft) 伴随论文 [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) 由 Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo 发布。
1. **[Swin2SR](https://huggingface.co/docs/transformers/model_doc/swin2sr)** (来自 University of Würzburg) 伴随论文 [Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration](https://arxiv.org/abs/2209.11345) 由 Marcos V. Conde, Ui-Jin Choi, Maxime Burchi, Radu Timofte 发布。
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (来自 Google AI) 伴随论文 [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (来自 Google AI) 伴随论文 [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (来自 Microsoft Research) 伴随论文 [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) 由 Brandon Smock, Rohith Pesala, Robin Abraham 发布。
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (来自 Google AI) 伴随论文 [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) 由 Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos 发布。
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (来自 Microsoft Research) 伴随论文 [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) 由 Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou 发布。
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (来自 Google/CMU) 伴随论文 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 由 Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 发布。
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
1. **[TVLT](https://huggingface.co/docs/transformers/main/model_doc/tvlt)** (来自 UNC Chapel Hill) 伴随论文 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 由 Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 发布。
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (来自 Microsoft Research) 伴随论文 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 由 Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 发布。
1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (来自 Peking University) 伴随论文 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) 由 Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun 发布。
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (来自 Tsinghua University and Nankai University) 伴随论文 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 由 Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 发布。
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (来自 Multimedia Computing Group, Nanjing University) 伴随论文 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 由 Zhan Tong, Yibing Song, Jue Wang, Limin Wang 发布。
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (来自 NAVER AI Lab/Kakao Enterprise/Kakao Brain) 伴随论文 [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) 由 Wonjae Kim, Bokyung Son, Ildoo Kim 发布。
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (来自 UCLA NLP) 伴随论文 [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) 由 Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang 发布。
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/model_doc/vit_hybrid)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (来自 Meta AI) 伴随论文 [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) 由 Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick 发布。
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (来自 Meta AI) 伴随论文 [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas 发布.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (来自 Facebook AI) 伴随论文 [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) 由 Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli 发布。
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (来自 Facebook AI) 伴随论文 [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino 发布。
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (来自 Facebook AI) 伴随论文 [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) 由 Qiantong Xu, Alexei Baevski, Michael Auli 发布。
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (来自 OpenAI) 伴随论文 [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) 由 Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever 发布。
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (来自 Microsoft Research) 伴随论文 [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) 由 Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling 发布。
1. **[X-MOD](https://huggingface.co/docs/transformers/main/model_doc/xmod)** (来自 Meta AI) 伴随论文 [Lifting the Curse of Multilinguality by Pre-training Modular Transformers](http://dx.doi.org/10.18653/v1/2022.naacl-main.255) 由 Jonas Pfeiffer, Naman Goyal, Xi Lin, Xian Li, James Cross, Sebastian Riedel, Mikel Artetxe 发布。
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (来自 Facebook) 伴随论文 [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) 由 Guillaume Lample and Alexis Conneau 发布。
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (来自 Facebook AI), 伴随论文 [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) 由 Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov 发布。
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (来自 Facebook AI) 伴随论文 [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) 由 Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau 发布。
1. **[XLM-V](https://huggingface.co/docs/transformers/model_doc/xlm-v)** (来自 Meta AI) 伴随论文 [XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models](https://arxiv.org/abs/2301.10472) 由 Davis Liang, Hila Gonen, Yuning Mao, Rui Hou, Naman Goyal, Marjan Ghazvininejad, Luke Zettlemoyer, Madian Khabsa 发布。
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (来自 Google/CMU) 伴随论文 [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) 由 Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le 发布。
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (来自 Facebook AI) 伴随论文 [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) 由 Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli 发布。
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (来自 Facebook AI) 伴随论文 [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) 由 Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli 发布。
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (来自 Huazhong University of Science & Technology) 伴随论文 [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) 由 Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu 发布。
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (来自 the University of Wisconsin - Madison) 伴随论文 [You Only Sample (Almost) 由 Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh 发布。
1. 想要贡献新的模型？我们这里有一份**详细指引和模板**来引导你添加新的模型。你可以在 [`templates`](./templates) 目录中找到他们。记得查看 [贡献指南](./CONTRIBUTING.md) 并在开始写 PR 前联系维护人员或开一个新的 issue 来获得反馈。
要检查某个模型是否已有 Flax、PyTorch 或 TensorFlow 的实现，或其是否在 🤗 Tokenizers 库中有对应词符化器（tokenizer），敬请参阅[此表](https://huggingface.co/docs/transformers/index#supported-frameworks)。
这些实现均已于多个数据集测试（请参看用例脚本）并应于原版实现表现相当。你可以在用例文档的[此节](https://huggingface.co/docs/transformers/examples)中了解表现的细节。
## 了解更多
| 章节 | 描述 |
|-|-|
| [文档](https://huggingface.co/transformers/) | 完整的 API 文档和教程 |
| [任务总结](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers 支持的任务 |
| [预处理教程](https://huggingface.co/docs/transformers/preprocessing) | 使用 `Tokenizer` 来为模型准备数据 |
| [训练和微调](https://huggingface.co/docs/transformers/training) | 在 PyTorch/TensorFlow 的训练循环或 `Trainer` API 中使用 🤗 Transformers 提供的模型 |
| [快速上手：微调和用例脚本](https://github.com/huggingface/transformers/tree/main/examples) | 为各种任务提供的用例脚本 |
| [模型分享和上传](https://huggingface.co/docs/transformers/model_sharing) | 和社区上传和分享你微调的模型 |
| [迁移](https://huggingface.co/docs/transformers/migration) | 从 `pytorch-transformers` 或 `pytorch-pretrained-bert` 迁移到 🤗 Transformers |
## 引用
我们已将此库的[论文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)正式发表，如果你使用了 🤗 Transformers 库，请引用:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Chinese translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Chinese characters. E.g., 共 100 多种语言; 使用 transformers 库。
- Use square quotes, e.g.,「引用」
Dictionary
Hugging Face: 抱抱脸
token: 词符（并用括号标注原英文）
tokenize: 词符化（并用括号标注原英文）
tokenizer: 词符化器（并用括号标注原英文）
transformer: transformer（不翻译）
pipeline: 流水线
API: API (不翻译）
inference: 推理
Trainer: 训练器。当作为类名出现时不翻译。
pretrained/pretrain: 预训练
finetune: 微调
community: 社区
example: 当特指仓库中 example 目录时翻译为「用例」
Python data structures (e.g., list, set, dict): 翻译为列表，集合，词典，并用括号标注原英文
NLP/Natural Language Processing: 以 NLP 出现时不翻译，以 Natural Language Processing 出现时翻译为自然语言处理
checkpoint: 检查点
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<b>简体中文</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>为 Jax、PyTorch 和 TensorFlow 打造的先进的自然语言处理</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 提供了数以千计的预训练模型，支持 100 多种语言的文本分类、信息抽取、问答、摘要、翻译、文本生成。它的宗旨是让最先进的 NLP 技术人人易用。
🤗 Transformers 提供了便于快速下载和使用的API，让你可以把预训练模型用在给定文本、在你的数据集上微调然后通过 [model hub](https://huggingface.co/models) 与社区共享。同时，每个定义的 Python 模块均完全独立，方便修改和快速研究实验。
🤗 Transformers 支持三个最热门的深度学习库： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) 以及 [TensorFlow](https://www.tensorflow.org/) — 并与之无缝整合。你可以直接使用一个框架训练你的模型然后用另一个加载和推理。
## 在线演示
你可以直接在模型页面上测试大多数 [model hub](https://huggingface.co/models) 上的模型。 我们也提供了 [私有模型托管、模型版本管理以及推理API](https://huggingface.co/pricing)。
这里是一些例子：
- [用 BERT 做掩码填词](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [用 Electra 做命名实体识别](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [用 GPT-2 做文本生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [用 RoBERTa 做自然语言推理](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [用 BART 做文本摘要](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [用 DistilBERT 做问答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [用 T5 做翻译](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，由抱抱脸团队打造，是一个文本生成的官方 demo。
## 如果你在寻找由抱抱脸团队提供的定制化支持服务
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 快速上手
我们为快速使用模型提供了 `pipeline` （流水线）API。流水线聚合了预训练模型和对应的文本预处理。下面是一个快速使用流水线去判断正负面情绪的例子：
```python
>>> from transformers import pipeline
# 使用情绪分析流水线
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
第二行代码下载并缓存了流水线使用的预训练模型，而第三行代码则在给定的文本上进行了评估。这里的答案“正面” (positive) 具有 99 的置信度。
许多的 NLP 任务都有开箱即用的预训练流水线。比如说，我们可以轻松的从给定文本中抽取问题答案：
``` python
>>> from transformers import pipeline
# 使用问答流水线
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
除了给出答案，预训练模型还给出了对应的置信度分数、答案在词符化 (tokenized) 后的文本中开始和结束的位置。你可以从[这个教程](https://huggingface.co/docs/transformers/task_summary)了解更多流水线API支持的任务。
要在你的任务上下载和使用任意预训练模型也很简单，只需三行代码。这里是 PyTorch 版的示例：
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
这里是等效的 TensorFlow 代码：
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
词符化器 (tokenizer) 为所有的预训练模型提供了预处理，并可以直接对单个字符串进行调用（比如上面的例子）或对列表 (list) 调用。它会输出一个你可以在下游代码里使用或直接通过 `**` 解包表达式传给模型的词典 (dict)。
模型本身是一个常规的 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) 或 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)（取决于你的后端），可以常规方式使用。 [这个教程](https://huggingface.co/transformers/training.html)解释了如何将这样的模型整合到经典的 PyTorch 或 TensorFlow 训练循环中，或是如何使用我们的 `Trainer` 训练器）API 来在一个新的数据集上快速微调。
## 为什么要用 transformers？
1. 便于使用的先进模型：
- NLU 和 NLG 上表现优越
- 对教学和实践友好且低门槛
- 高级抽象，只需了解三个类
- 对所有模型统一的API
1. 更低计算开销，更少的碳排放：
- 研究人员可以分享已训练的模型而非每次从头开始训练
- 工程师可以减少计算用时和生产环境开销
- 数十种模型架构、两千多个预训练模型、100多种语言支持
1. 对于模型生命周期的每一个部分都面面俱到：
- 训练先进的模型，只需 3 行代码
- 模型在不同深度学习框架间任意转移，随你心意
- 为训练、评估和生产选择最适合的框架，衔接无缝
1. 为你的需求轻松定制专属模型和用例：
- 我们为每种模型架构提供了多个用例来复现原论文结果
- 模型内部结构保持透明一致
- 模型文件可单独使用，方便魔改和快速实验
## 什么情况下我不该用 transformers？
- 本库并不是模块化的神经网络工具箱。模型文件中的代码特意呈若璞玉，未经额外抽象封装，以便研究人员快速迭代魔改而不致溺于抽象和文件跳转之中。
- `Trainer` API 并非兼容任何模型，只为本库之模型优化。若是在寻找适用于通用机器学习的训练循环实现，请另觅他库。
- 尽管我们已尽力而为，[examples 目录](https://github.com/huggingface/transformers/tree/main/examples)中的脚本也仅为用例而已。对于你的特定问题，它们并不一定开箱即用，可能需要改几行代码以适之。
## 安装
### 使用 pip
这个仓库已在 Python 3.6+、Flax 0.3.2+、PyTorch 1.3.1+ 和 TensorFlow 2.3+ 下经过测试。
你可以在[虚拟环境](https://docs.python.org/3/library/venv.html)中安装 🤗 Transformers。如果你还不熟悉 Python 的虚拟环境，请阅此[用户说明](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
首先，用你打算使用的版本的 Python 创建一个虚拟环境并激活。
然后，你需要安装 Flax、PyTorch 或 TensorFlow 其中之一。关于在你使用的平台上安装这些框架，请参阅 [TensorFlow 安装页](https://www.tensorflow.org/install/), [PyTorch 安装页](https://pytorch.org/get-started/locally/#start-locally) 或 [Flax 安装页](https://github.com/google/flax#quick-install)。
当这些后端之一安装成功后， 🤗 Transformers 可依此安装：
```bash
pip install transformers
```
如果你想要试试用例或者想在正式发布前使用最新的开发中代码，你得[从源代码安装](https://huggingface.co/docs/transformers/installation#installing-from-source)。
### 使用 conda
自 Transformers 4.0.0 版始，我们有了一个 conda 频道： `huggingface`。
🤗 Transformers 可以通过 conda 依此安装：
```shell script
conda install -c huggingface transformers
```
要通过 conda 安装 Flax、PyTorch 或 TensorFlow 其中之一，请参阅它们各自安装页的说明。
## 模型架构
🤗 Transformers 支持的[**所有的模型检查点**](https://huggingface.co/models)由[用户](https://huggingface.co/users)和[组织](https://huggingface.co/organizations)上传，均与 huggingface.co [model hub](https://huggingface.co) 无缝整合。
目前的检查点数量： 
🤗 Transformers 目前支持如下的架构（模型概述请阅[这里](https://huggingface.co/docs/transformers/model_summary)）：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (来自 Google Research and the Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。
1. **[AltCLIP](https://huggingface.co/docs/transformers/model_doc/altclip)** (来自 BAAI) 伴随论文 [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://arxiv.org/abs/2211.06679) 由 Chen, Zhongzhi and Liu, Guang and Zhang, Bo-Wen and Ye, Fulong and Yang, Qinghong and Wu, Ledell 发布。
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (来自 MIT) 伴随论文 [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) 由 Yuan Gong, Yu-An Chung, James Glass 发布。
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (来自 Facebook) 伴随论文 [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) 由 Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer 发布。
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (来自 École polytechnique) 伴随论文 [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) 由 Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis 发布。
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (来自 VinAI Research) 伴随论文 [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) 由 Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen 发布。
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (来自 Microsoft) 伴随论文 [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) 由 Hangbo Bao, Li Dong, Furu Wei 发布。
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (来自 Google) 伴随论文 [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) 由 Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova 发布。
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (来自 Google) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (来自 VinAI Research) 伴随论文 [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) 由 Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen 发布。
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BioGpt](https://huggingface.co/docs/transformers/model_doc/biogpt)** (来自 Microsoft Research AI4Science) 伴随论文 [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) 由 Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu 发布。
1. **[BiT](https://huggingface.co/docs/transformers/model_doc/bit)** (来自 Google AI) 伴随论文 [Big Transfer (BiT) 由 Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby 发布。
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BLIP](https://huggingface.co/docs/transformers/model_doc/blip)** (来自 Salesforce) 伴随论文 [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://arxiv.org/abs/2201.12086) 由 Junnan Li, Dongxu Li, Caiming Xiong, Steven Hoi 发布。
1. **[BLIP-2](https://huggingface.co/docs/transformers/main/model_doc/blip-2)** (来自 Salesforce) 伴随论文 [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://arxiv.org/abs/2301.12597) 由 Junnan Li, Dongxu Li, Silvio Savarese, Steven Hoi 发布。
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (来自 Alexa) 伴随论文 [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) 由 Adrian de Wynter and Daniel J. Perry 发布。
1. **[BridgeTower](https://huggingface.co/docs/transformers/main/model_doc/bridgetower)** (from Harbin Institute of Technology/Microsoft Research Asia/Intel Labs) released with the paper [BridgeTower: Building Bridges Between Encoders in Vision-Language Representation Learning](https://arxiv.org/abs/2206.08657) by Xiao Xu, Chenfei Wu, Shachar Rosenman, Vasudev Lal, Wanxiang Che, Nan Duan.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (来自 Google Research) 伴随论文 [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) 由 Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel 发布。
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (来自 Inria/Facebook/Sorbonne) 伴随论文 [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) 由 Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot 发布。
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (来自 Google Research) 伴随论文 [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) 由 Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting 发布。
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (来自 OFA-Sys) 伴随论文 [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) 由 An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou 发布。
1. **[CLAP](https://huggingface.co/docs/transformers/main/model_doc/clap)** (来自 LAION-AI) 伴随论文 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation]https://arxiv.org/abs/2211.06687) 由 Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov 发布。
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (来自 Microsoft Research Asia) 伴随论文 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 由 Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 发布。
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (来自 YituTech) 伴随论文 [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) 由 Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan 发布。
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (来自 Facebook AI) 伴随论文 [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) 由 Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie 发布。
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (来自 Tsinghua University) 伴随论文 [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) 由 Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun 发布。
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (来自 Salesforce) 伴随论文 [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) 由 Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher 发布。
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (来自 Microsoft) 伴随论文 [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) 由 Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang 发布。
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (来自 Facebook) 伴随论文 [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) 由 Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli 发布。
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (来自 Berkeley/Facebook/Google) 伴随论文 [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) 由 Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch 发布。
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (来自 SenseTime Research) 伴随论文 [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) 由 Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai 发布。
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (来自 Facebook) 伴随论文 [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) 由 Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou 发布。
1. **[DETA](https://huggingface.co/docs/transformers/main/model_doc/deta)** (来自 The University of Texas at Austin) 伴随论文 [NMS Strikes Back](https://arxiv.org/abs/2212.06137) 由 Jeffrey Ouyang-Zhang, Jang Hyun Cho, Xingyi Zhou, Philipp Krähenbühl 发布。
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (来自 Facebook) 伴随论文 [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) 由 Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko 发布。
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (来自 Microsoft Research) 伴随论文 [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) 由 Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan 发布。
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (来自 SHI Labs) 伴随论文 [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) 由 Ali Hassani and Humphrey Shi 发布。
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (来自 HuggingFace), 伴随论文 [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) 由 Victor Sanh, Lysandre Debut and Thomas Wolf 发布。 同样的方法也应用于压缩 GPT-2 到 [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa 到 [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT 到 [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) 和德语版 DistilBERT。
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (来自 Microsoft Research) 伴随论文 [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) 由 Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei 发布。
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (来自 NAVER) 伴随论文 [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) 由 Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park 发布。
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (来自 Facebook) 伴随论文 [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) 由 Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih 发布。
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (来自 Intel Labs) 伴随论文 [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) 由 René Ranftl, Alexey Bochkovskiy, Vladlen Koltun 发布。
1. **[EfficientFormer](https://huggingface.co/docs/transformers/model_doc/efficientformer)** (来自 Snap Research) 伴随论文 [EfficientFormer: Vision Transformers at MobileNetSpeed](https://arxiv.org/abs/2206.01191) 由 Yanyu Li, Geng Yuan, Yang Wen, Ju Hu, Georgios Evangelidis, Sergey Tulyakov, Yanzhi Wang, Jian Ren 发布。
1. **[EfficientNet](https://huggingface.co/docs/transformers/main/model_doc/efficientnet)** (from Google Brain) released with the paper [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://arxiv.org/abs/1905.11946) by Mingxing Tan, Quoc V. Le.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (来自 Google Research/Stanford University) 伴随论文 [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) 由 Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning 发布。
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (来自 Google Research) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (来自 Baidu) 伴随论文 [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu 发布。
1. **[ErnieM](https://huggingface.co/docs/transformers/main/model_doc/ernie_m)** (来自 Baidu) 伴随论文 [ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolingual Corpora](https://arxiv.org/abs/2012.15674) 由 Xuan Ouyang, Shuohuan Wang, Chao Pang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang 发布。
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (来自 CNRS) 伴随论文 [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) 由 Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab 发布。
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (来自 Facebook AI) 伴随论文 [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) 由 Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela 发布。
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (来自 Google Research) 伴随论文 [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) 由 James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon 发布。
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (来自 CMU/Google Brain) 伴随论文 [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) 由 Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le 发布。
1. **[GIT](https://huggingface.co/docs/transformers/model_doc/git)** (来自 Microsoft Research) 伴随论文 [GIT: A Generative Image-to-text Transformer for Vision and Language](https://arxiv.org/abs/2205.14100) 由 Jianfeng Wang, Zhengyuan Yang, Xiaowei Hu, Linjie Li, Kevin Lin, Zhe Gan, Zicheng Liu, Ce Liu, Lijuan Wang 发布。
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (来自 KAIST) 伴随论文 [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) 由 Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim 发布。
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (来自 OpenAI) 伴随论文 [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) 由 Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever 发布。
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (来自 EleutherAI) 随仓库 [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) 发布。作者为 Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy 发布。
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (来自 ABEJA) 由 Shinya Otani, Takayoshi Makabe, Anuj Arora, Kyo Hattori。
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (来自 OpenAI) 伴随论文 [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) 由 Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever** 发布。
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (来自 EleutherAI) 伴随论文 [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) 由 Ben Wang and Aran Komatsuzaki 发布。
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GPTSAN-japanese](https://huggingface.co/docs/transformers/main/model_doc/gptsan-japanese)** released in the repository [tanreinama/GPTSAN](https://github.com/tanreinama/GPTSAN/blob/main/report/model.md) by 坂本俊之(tanreinama).
1. **[Graphormer](https://huggingface.co/docs/transformers/model_doc/graphormer)** (from Microsoft) released with the paper [Do Transformers Really Perform Bad for Graph Representation?](https://arxiv.org/abs/2106.05234) by Chengxuan Ying, Tianle Cai, Shengjie Luo, Shuxin Zheng, Guolin Ke, Di He, Yanming Shen, Tie-Yan Liu.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (来自 UCSD, NVIDIA) 伴随论文 [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) 由 Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang 发布。
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (来自 Facebook) 伴随论文 [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) 由 Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed 发布。
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (来自 Berkeley) 伴随论文 [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) 由 Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer 发布。
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (来自 OpenAI) 伴随论文 [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) 由 Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever 发布。
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) 由 Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou 发布。
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) 由 Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou 发布。
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) 由 Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei 发布。
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) 由 Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei 发布。
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (来自 Meta AI) 伴随论文 [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) 由 Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze 发布。
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (来自 South China University of Technology) 伴随论文 [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) 由 Jiapeng Wang, Lianwen Jin, Kai Ding 发布。
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (来自 Google AI) released 伴随论文 [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) 由 Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang 发布。
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (来自 Studio Ousia) 伴随论文 [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) 由 Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto 发布。
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (来自 UNC Chapel Hill) 伴随论文 [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) 由 Hao Tan and Mohit Bansal 发布。
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (来自 Facebook) 伴随论文 [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) 由 Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert 发布。
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (来自 Facebook) 伴随论文 [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) 由 Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin 发布。
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** 用 [OPUS](http://opus.nlpl.eu/) 数据训练的机器翻译模型由 Jörg Tiedemann 发布。[Marian Framework](https://marian-nmt.github.io/) 由微软翻译团队开发。
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (来自 Microsoft Research Asia) 伴随论文 [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) 由 Junlong Li, Yiheng Xu, Lei Cui, Furu Wei 发布。
1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (来自 FAIR and UIUC) 伴随论文 [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) 由 Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar 发布。
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) 由 Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer 发布。
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) 由 Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan 发布。
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (来自 Studio Ousia) 伴随论文 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 由 Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 发布。
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (来自 CMU/Google Brain) 伴随论文 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 由 Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 发布。
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (来自 Google Inc.) 伴随论文 [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) 由 Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam 发布。
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (来自 Google Inc.) 伴随论文 [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) 由 Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen 发布。
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (来自 Apple) 伴随论文 [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) 由 Sachin Mehta and Mohammad Rastegari 发布。
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (来自 Microsoft Research) 伴随论文 [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) 由 Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu 发布。
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (来自 Google AI) 伴随论文 [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) 由 Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel 发布。
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (来自 中国人民大学 AI Box) 伴随论文 [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) 由 Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen 发布。
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (来自 SHI Labs) 伴随论文 [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) 由 Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi 发布。
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (来自华为诺亚方舟实验室) 伴随论文 [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) 由 Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu 发布。
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (来自 Meta) 伴随论文 [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) 由 the NLLB team 发布。
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (来自 the University of Wisconsin - Madison) 伴随论文 [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) 由 Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh 发布。
1. **[OneFormer](https://huggingface.co/docs/transformers/model_doc/oneformer)** (来自 SHI Labs) 伴随论文 [OneFormer: One Transformer to Rule Universal Image Segmentation](https://arxiv.org/abs/2211.06220) 由 Jitesh Jain, Jiachen Li, MangTik Chiu, Ali Hassani, Nikita Orlov, Humphrey Shi 发布。
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (来自 Meta AI) 伴随论文 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 由 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 发布。
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (来自 Google AI) 伴随论文 [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) 由 Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby 发布。
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (来自 Google) 伴随论文 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 由 Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 发布。
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (来自 Google) 伴随论文 [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) 由 Jason Phang, Yao Zhao, Peter J. Liu 发布。
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (来自 Deepmind) 伴随论文 [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) 由 Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira 发布。
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (来自 VinAI Research) 伴随论文 [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) 由 Dat Quoc Nguyen and Anh Tuan Nguyen 发布。
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (来自 UCLA NLP) 伴随论文 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 由 Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 发布。
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (来自 Sea AI Labs) 伴随论文 [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) 由 Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng 发布。
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (来自 NVIDIA) 伴随论文 [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) 由 Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius 发布。
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (来自 Facebook) 伴随论文 [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) 由 Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela 发布。
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (来自 Google Research) 伴随论文 [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) 由 Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang 发布。
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (来自 Google Research) 伴随论文 [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) 由 Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya 发布。
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (来自 Google Research) 伴随论文 [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) 由 Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder 发布。
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (来自 Facebook), 伴随论文 [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) 由 Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov 发布。
1. **[RoBERTa-PreLayerNorm](https://huggingface.co/docs/transformers/model_doc/roberta-prelayernorm)** (来自 Facebook) 伴随论文 [fairseq: A Fast, Extensible Toolkit for Sequence Modeling](https://arxiv.org/abs/1904.01038) 由 Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, Michael Auli 发布。
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (来自 WeChatAI), 伴随论文 [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) 由 HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou 发布。
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (来自 ZhuiyiTechnology), 伴随论文 [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) 由 Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu 发布。
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (来自 NVIDIA) 伴随论文 [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) 由 Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo 发布。
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SpeechT5](https://huggingface.co/docs/transformers/main/model_doc/speecht5)** (来自 Microsoft Research) 伴随论文 [SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing](https://arxiv.org/abs/2110.07205) 由 Junyi Ao, Rui Wang, Long Zhou, Chengyi Wang, Shuo Ren, Yu Wu, Shujie Liu, Tom Ko, Qing Li, Yu Zhang, Zhihua Wei, Yao Qian, Jinyu Li, Furu Wei 发布。
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (来自 Facebook), 伴随论文 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 发布。
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (来自 Facebook) 伴随论文 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 由 Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 发布。
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (来自 Tel Aviv University) 伴随论文 [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) 由 Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy 发布。
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (来自 Berkeley) 伴随论文 [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) 由 Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer 发布。
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (来自 Microsoft) 伴随论文 [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) 由 Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo 发布。
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (来自 Microsoft) 伴随论文 [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) 由 Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo 发布。
1. **[Swin2SR](https://huggingface.co/docs/transformers/model_doc/swin2sr)** (来自 University of Würzburg) 伴随论文 [Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration](https://arxiv.org/abs/2209.11345) 由 Marcos V. Conde, Ui-Jin Choi, Maxime Burchi, Radu Timofte 发布。
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (来自 Google AI) 伴随论文 [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (来自 Google AI) 伴随论文 [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (来自 Microsoft Research) 伴随论文 [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) 由 Brandon Smock, Rohith Pesala, Robin Abraham 发布。
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (来自 Google AI) 伴随论文 [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) 由 Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos 发布。
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (来自 Microsoft Research) 伴随论文 [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) 由 Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou 发布。
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (来自 Google/CMU) 伴随论文 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 由 Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 发布。
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
1. **[TVLT](https://huggingface.co/docs/transformers/main/model_doc/tvlt)** (来自 UNC Chapel Hill) 伴随论文 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 由 Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 发布。
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (来自 Microsoft Research) 伴随论文 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 由 Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 发布。
1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (来自 Peking University) 伴随论文 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) 由 Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun 发布。
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (来自 Tsinghua University and Nankai University) 伴随论文 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 由 Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 发布。
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (来自 Multimedia Computing Group, Nanjing University) 伴随论文 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 由 Zhan Tong, Yibing Song, Jue Wang, Limin Wang 发布。
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (来自 NAVER AI Lab/Kakao Enterprise/Kakao Brain) 伴随论文 [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) 由 Wonjae Kim, Bokyung Son, Ildoo Kim 发布。
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (来自 UCLA NLP) 伴随论文 [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) 由 Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang 发布。
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/model_doc/vit_hybrid)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (来自 Meta AI) 伴随论文 [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) 由 Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick 发布。
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (来自 Meta AI) 伴随论文 [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas 发布.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (来自 Facebook AI) 伴随论文 [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) 由 Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli 发布。
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (来自 Facebook AI) 伴随论文 [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino 发布。
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (来自 Facebook AI) 伴随论文 [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) 由 Qiantong Xu, Alexei Baevski, Michael Auli 发布。
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (来自 OpenAI) 伴随论文 [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) 由 Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever 发布。
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (来自 Microsoft Research) 伴随论文 [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) 由 Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling 发布。
1. **[X-MOD](https://huggingface.co/docs/transformers/main/model_doc/xmod)** (来自 Meta AI) 伴随论文 [Lifting the Curse of Multilinguality by Pre-training Modular Transformers](http://dx.doi.org/10.18653/v1/2022.naacl-main.255) 由 Jonas Pfeiffer, Naman Goyal, Xi Lin, Xian Li, James Cross, Sebastian Riedel, Mikel Artetxe 发布。
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (来自 Facebook) 伴随论文 [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) 由 Guillaume Lample and Alexis Conneau 发布。
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (来自 Facebook AI), 伴随论文 [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) 由 Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov 发布。
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (来自 Facebook AI) 伴随论文 [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) 由 Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau 发布。
1. **[XLM-V](https://huggingface.co/docs/transformers/model_doc/xlm-v)** (来自 Meta AI) 伴随论文 [XLM-V: Overcoming the Vocabulary Bottleneck in Multilingual Masked Language Models](https://arxiv.org/abs/2301.10472) 由 Davis Liang, Hila Gonen, Yuning Mao, Rui Hou, Naman Goyal, Marjan Ghazvininejad, Luke Zettlemoyer, Madian Khabsa 发布。
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (来自 Google/CMU) 伴随论文 [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) 由 Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le 发布。
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (来自 Facebook AI) 伴随论文 [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) 由 Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli 发布。
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (来自 Facebook AI) 伴随论文 [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) 由 Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli 发布。
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (来自 Huazhong University of Science & Technology) 伴随论文 [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) 由 Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu 发布。
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (来自 the University of Wisconsin - Madison) 伴随论文 [You Only Sample (Almost) 由 Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh 发布。
1. 想要贡献新的模型？我们这里有一份**详细指引和模板**来引导你添加新的模型。你可以在 [`templates`](./templates) 目录中找到他们。记得查看 [贡献指南](./CONTRIBUTING.md) 并在开始写 PR 前联系维护人员或开一个新的 issue 来获得反馈。
要检查某个模型是否已有 Flax、PyTorch 或 TensorFlow 的实现，或其是否在 🤗 Tokenizers 库中有对应词符化器（tokenizer），敬请参阅[此表](https://huggingface.co/docs/transformers/index#supported-frameworks)。
这些实现均已于多个数据集测试（请参看用例脚本）并应于原版实现表现相当。你可以在用例文档的[此节](https://huggingface.co/docs/transformers/examples)中了解表现的细节。
## 了解更多
| 章节 | 描述 |
|-|-|
| [文档](https://huggingface.co/transformers/) | 完整的 API 文档和教程 |
| [任务总结](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers 支持的任务 |
| [预处理教程](https://huggingface.co/docs/transformers/preprocessing) | 使用 `Tokenizer` 来为模型准备数据 |
| [训练和微调](https://huggingface.co/docs/transformers/training) | 在 PyTorch/TensorFlow 的训练循环或 `Trainer` API 中使用 🤗 Transformers 提供的模型 |
| [快速上手：微调和用例脚本](https://github.com/huggingface/transformers/tree/main/examples) | 为各种任务提供的用例脚本 |
| [模型分享和上传](https://huggingface.co/docs/transformers/model_sharing) | 和社区上传和分享你微调的模型 |
| [迁移](https://huggingface.co/docs/transformers/migration) | 从 `pytorch-transformers` 或 `pytorch-pretrained-bert` 迁移到 🤗 Transformers |
## 引用
我们已将此库的[论文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)正式发表，如果你使用了 🤗 Transformers 库，请引用:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/jax-projects/dataset-streaming/run_mlm_flax_stream.py | #!/usr/bin/env python
# coding=utf-8
# Copyright 2021 The HuggingFace Team All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Fine-tuning the library models for masked language modeling (BERT, ALBERT, RoBERTa...) with whole word masking on a
text file or a dataset.
Here is the full list of checkpoints on the hub that can be fine-tuned by this script:
https://huggingface.co/models?filter=fill-mask
"""
import logging
import os
import sys
import time
from collections import defaultdict
from dataclasses import dataclass, field
# You can also adapt this script on your own masked language modeling task. Pointers for this are left as comments.
from pathlib import Path
from typing import Dict, List, Optional, Tuple
import datasets
import flax
import jax
import jax.numpy as jnp
import numpy as np
import optax
from datasets import load_dataset
from flax import jax_utils, traverse_util
from flax.training import train_state
from flax.training.common_utils import get_metrics, onehot, shard
from tqdm import tqdm
from transformers import (
CONFIG_MAPPING,
FLAX_MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoTokenizer,
FlaxAutoModelForMaskedLM,
HfArgumentParser,
PreTrainedTokenizerBase,
TensorType,
TrainingArguments,
is_tensorboard_available,
set_seed,
)
if datasets.__version__ <= "1.8.0":
raise ValueError("Make sure to upgrade `datasets` to a version >= 1.9.0 to use dataset streaming")
MODEL_CONFIG_CLASSES = list(FLAX_MODEL_FOR_MASKED_LM_MAPPING.keys())
MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune, or train from scratch.
"""
model_name_or_path: Optional[str] = field(
default=None,
metadata={
"help": (
"The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."
)
},
)
model_type: Optional[str] = field(
default=None,
metadata={"help": "If training from scratch, pass a model type from the list: " + ", ".join(MODEL_TYPES)},
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
tokenizer_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None, metadata={"help": "Where do you want to store the pretrained models downloaded from s3"}
)
use_fast_tokenizer: bool = field(
default=True,
metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
)
dtype: Optional[str] = field(
default="float32",
metadata={
"help": (
"Floating-point format in which the model weights should be initialized and trained. Choose one of"
" `[float32, float16, bfloat16]`."
)
},
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
"""
dataset_name: Optional[str] = field(
default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
)
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
validation_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."},
)
train_ref_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input train ref data file for whole word masking in Chinese."},
)
validation_ref_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input validation ref data file for whole word masking in Chinese."},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
validation_split_percentage: Optional[int] = field(
default=5,
metadata={
"help": "The percentage of the train set used as validation set in case there's no validation split"
},
)
max_seq_length: Optional[int] = field(
default=None,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated. Default to the max input length of the model."
)
},
)
preprocessing_num_workers: Optional[int] = field(
default=None,
metadata={"help": "The number of processes to use for the preprocessing."},
)
mlm_probability: float = field(
default=0.15, metadata={"help": "Ratio of tokens to mask for masked language modeling loss"}
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
},
)
line_by_line: bool = field(
default=False,
metadata={"help": "Whether distinct lines of text in the dataset are to be handled as distinct sequences."},
)
text_column_name: str = field(
default="text", metadata={"help": "The name of the column to retrieve the training text."}
)
shuffle_buffer_size: int = field(
default=10000, metadata={"help": "The number of examples to pre-load for shuffling."}
)
num_train_steps: int = field(default=50000, metadata={"help": "The number of training steps."})
num_eval_samples: int = field(default=50000, metadata={"help": "The number of samples to be used for evaluation"})
def __post_init__(self):
if self.dataset_name is None and self.train_file is None and self.validation_file is None:
raise ValueError("Need either a dataset name or a training/validation file.")
else:
if self.train_file is not None:
extension = self.train_file.split(".")[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
extension = self.validation_file.split(".")[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
@flax.struct.dataclass
class FlaxDataCollatorForLanguageModeling:
"""
Data collator used for language modeling. Inputs are dynamically padded to the maximum length of a batch if they
are not all of the same length.
Args:
tokenizer (:class:`~transformers.PreTrainedTokenizer` or :class:`~transformers.PreTrainedTokenizerFast`):
The tokenizer used for encoding the data.
mlm_probability (:obj:`float`, `optional`, defaults to 0.15):
The probability with which to (randomly) mask tokens in the input.
.. note::
For best performance, this data collator should be used with a dataset having items that are dictionaries or
BatchEncoding, with the :obj:`"special_tokens_mask"` key, as returned by a
:class:`~transformers.PreTrainedTokenizer` or a :class:`~transformers.PreTrainedTokenizerFast` with the
argument :obj:`return_special_tokens_mask=True`.
"""
tokenizer: PreTrainedTokenizerBase
mlm_probability: float = 0.15
def __post_init__(self):
if self.tokenizer.mask_token is None:
raise ValueError(
"This tokenizer does not have a mask token which is necessary for masked language modeling. "
"You should pass `mlm=False` to train on causal language modeling instead."
)
def __call__(self, examples: List[Dict[str, np.ndarray]]) -> Dict[str, np.ndarray]:
# Handle dict or lists with proper padding and conversion to tensor.
batch = self.tokenizer.pad(examples, return_tensors=TensorType.NUMPY)
# If special token mask has been preprocessed, pop it from the dict.
special_tokens_mask = batch.pop("special_tokens_mask", None)
batch["input_ids"], batch["labels"] = self.mask_tokens(
batch["input_ids"], special_tokens_mask=special_tokens_mask
)
return batch
def mask_tokens(
self, inputs: np.ndarray, special_tokens_mask: Optional[np.ndarray]
) -> Tuple[jnp.ndarray, jnp.ndarray]:
"""
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original.
"""
labels = inputs.copy()
# We sample a few tokens in each sequence for MLM training (with probability `self.mlm_probability`)
probability_matrix = np.full(labels.shape, self.mlm_probability)
special_tokens_mask = special_tokens_mask.astype("bool")
probability_matrix[special_tokens_mask] = 0.0
masked_indices = np.random.binomial(1, probability_matrix).astype("bool")
labels[~masked_indices] = -100 # We only compute loss on masked tokens
# 80% of the time, we replace masked input tokens with tokenizer.mask_token ([MASK])
indices_replaced = np.random.binomial(1, np.full(labels.shape, 0.8)).astype("bool") & masked_indices
inputs[indices_replaced] = self.tokenizer.convert_tokens_to_ids(self.tokenizer.mask_token)
# 10% of the time, we replace masked input tokens with random word
indices_random = np.random.binomial(1, np.full(labels.shape, 0.5)).astype("bool")
indices_random &= masked_indices & ~indices_replaced
random_words = np.random.randint(self.tokenizer.vocab_size, size=labels.shape, dtype="i4")
inputs[indices_random] = random_words[indices_random]
# The rest of the time (10% of the time) we keep the masked input tokens unchanged
return inputs, labels
def generate_batch_splits(samples_idx: np.ndarray, batch_size: int) -> np.ndarray:
num_samples = len(samples_idx)
samples_to_remove = num_samples % batch_size
if samples_to_remove != 0:
samples_idx = samples_idx[:-samples_to_remove]
sections_split = num_samples // batch_size
batch_idx = np.split(samples_idx, sections_split)
return batch_idx
def advance_iter_and_group_samples(train_iterator, num_samples, max_seq_length):
"""
The training iterator is advanced so that after groupifying the samples,
`num_samples` of length `max_seq_length` are returned.
"""
num_total_tokens = max_seq_length * num_samples
samples = defaultdict(list)
i = 0
while i < num_total_tokens:
tokenized_samples = next(train_iterator)
i += len(tokenized_samples["input_ids"])
# concatenate tokenized samples to list (excluding "id" and "text")
samples = {
k: samples[k] + tokenized_samples[k] for k in ["input_ids", "attention_mask", "special_tokens_mask"]
}
# Concatenated tokens are split to lists of length `max_seq_length`.
# Note that remainedr of % max_seq_length are thrown away.
def group_texts(examples):
result = {
k: [t[i : i + max_seq_length] for i in range(0, num_total_tokens, max_seq_length)]
for k, t in examples.items()
}
return result
grouped_samples = group_texts(samples)
return grouped_samples
def write_train_metric(summary_writer, train_metrics, train_time, step):
summary_writer.scalar("train_time", train_time, step)
train_metrics = get_metrics(train_metrics)
for key, vals in train_metrics.items():
tag = f"train_{key}"
for i, val in enumerate(vals):
summary_writer.scalar(tag, val, step - len(vals) + i + 1)
def write_eval_metric(summary_writer, eval_metrics, step):
for metric_name, value in eval_metrics.items():
summary_writer.scalar(f"eval_{metric_name}", value, step)
if __name__ == "__main__":
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
if (
os.path.exists(training_args.output_dir)
and os.listdir(training_args.output_dir)
and training_args.do_train
and not training_args.overwrite_output_dir
):
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty."
"Use --overwrite_output_dir to overcome."
)
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
level="INFO",
datefmt="[%X]",
)
# Log on each process the small summary:
logger = logging.getLogger(__name__)
logger.warning(
f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
+ f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
)
# Set the verbosity to info of the Transformers logger (on main process only):
logger.info(f"Training/evaluation parameters {training_args}")
# Set seed before initializing model.
set_seed(training_args.seed)
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
dataset = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
cache_dir=model_args.cache_dir,
streaming=True,
split="train",
)
if model_args.config_name:
config = AutoConfig.from_pretrained(model_args.config_name, cache_dir=model_args.cache_dir)
elif model_args.model_name_or_path:
config = AutoConfig.from_pretrained(model_args.model_name_or_path, cache_dir=model_args.cache_dir)
else:
config = CONFIG_MAPPING[model_args.model_type]()
logger.warning("You are instantiating a new config instance from scratch.")
if model_args.tokenizer_name:
tokenizer = AutoTokenizer.from_pretrained(
model_args.tokenizer_name, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
)
elif model_args.model_name_or_path:
tokenizer = AutoTokenizer.from_pretrained(
model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
)
else:
raise ValueError(
"You are instantiating a new tokenizer from scratch. This is not supported by this script."
"You can do it from another script, save it, and load it from here, using --tokenizer_name."
)
# Otherwise, we tokenize every text, then concatenate them together before splitting them in smaller parts.
# We use `return_special_tokens_mask=True` because DataCollatorForLanguageModeling (see below) is more
# efficient when it receives the `special_tokens_mask`.
def tokenize_function(examples):
return tokenizer(examples[data_args.text_column_name], return_special_tokens_mask=True)
tokenized_datasets = dataset.map(tokenize_function, batched=True, remove_columns=list(dataset.features.keys()))
shuffle_seed = training_args.seed
tokenized_datasets = tokenized_datasets.shuffle(buffer_size=data_args.shuffle_buffer_size, seed=shuffle_seed)
has_tensorboard = is_tensorboard_available()
if has_tensorboard and jax.process_index() == 0:
try:
from flax.metrics.tensorboard import SummaryWriter
except ImportError as ie:
has_tensorboard = False
logger.warning(
f"Unable to display metrics through TensorBoard because some package are not installed: {ie}"
)
summary_writer = SummaryWriter(log_dir=Path(training_args.output_dir))
# Data collator
# This one will take care of randomly masking the tokens.
data_collator = FlaxDataCollatorForLanguageModeling(tokenizer=tokenizer, mlm_probability=data_args.mlm_probability)
# Initialize our training
rng = jax.random.PRNGKey(training_args.seed)
dropout_rngs = jax.random.split(rng, jax.local_device_count())
if model_args.model_name_or_path:
model = FlaxAutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, config=config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
)
else:
model = FlaxAutoModelForMaskedLM.from_config(
config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
)
# Store some constant
num_epochs = int(training_args.num_train_epochs)
train_batch_size = int(training_args.per_device_train_batch_size) * jax.device_count()
eval_batch_size = int(training_args.per_device_eval_batch_size) * jax.device_count()
# define number steps per stream epoch
num_train_steps = data_args.num_train_steps
# Create learning rate schedule
warmup_fn = optax.linear_schedule(
init_value=0.0, end_value=training_args.learning_rate, transition_steps=training_args.warmup_steps
)
decay_fn = optax.linear_schedule(
init_value=training_args.learning_rate,
end_value=0,
transition_steps=num_train_steps - training_args.warmup_steps,
)
linear_decay_lr_schedule_fn = optax.join_schedules(
schedules=[warmup_fn, decay_fn], boundaries=[training_args.warmup_steps]
)
# We use Optax's "masking" functionality to not apply weight decay
# to bias and LayerNorm scale parameters. decay_mask_fn returns a
# mask boolean with the same structure as the parameters.
# The mask is True for parameters that should be decayed.
# Note that this mask is specifically adapted for FlaxBERT-like models.
# For other models, one should correct the layer norm parameter naming
# accordingly.
def decay_mask_fn(params):
flat_params = traverse_util.flatten_dict(params)
flat_mask = {path: (path[-1] != "bias" and path[-2:] != ("LayerNorm", "scale")) for path in flat_params}
return traverse_util.unflatten_dict(flat_mask)
# create adam optimizer
adamw = optax.adamw(
learning_rate=linear_decay_lr_schedule_fn,
b1=training_args.adam_beta1,
b2=training_args.adam_beta2,
eps=training_args.adam_epsilon,
weight_decay=training_args.weight_decay,
mask=decay_mask_fn,
)
# Setup train state
state = train_state.TrainState.create(apply_fn=model.__call__, params=model.params, tx=adamw)
# Define gradient update step fn
def train_step(state, batch, dropout_rng):
dropout_rng, new_dropout_rng = jax.random.split(dropout_rng)
def loss_fn(params):
labels = batch.pop("labels")
logits = state.apply_fn(**batch, params=params, dropout_rng=dropout_rng, train=True)[0]
# compute loss, ignore padded input tokens
label_mask = jnp.where(labels > 0, 1.0, 0.0)
loss = optax.softmax_cross_entropy(logits, onehot(labels, logits.shape[-1])) * label_mask
# take average
loss = loss.sum() / label_mask.sum()
return loss
grad_fn = jax.value_and_grad(loss_fn)
loss, grad = grad_fn(state.params)
grad = jax.lax.pmean(grad, "batch")
new_state = state.apply_gradients(grads=grad)
metrics = jax.lax.pmean(
{"loss": loss, "learning_rate": linear_decay_lr_schedule_fn(state.step)}, axis_name="batch"
)
return new_state, metrics, new_dropout_rng
# Create parallel version of the train step
p_train_step = jax.pmap(train_step, "batch", donate_argnums=(0,))
# Define eval fn
def eval_step(params, batch):
labels = batch.pop("labels")
logits = model(**batch, params=params, train=False)[0]
# compute loss, ignore padded input tokens
label_mask = jnp.where(labels > 0, 1.0, 0.0)
loss = optax.softmax_cross_entropy(logits, onehot(labels, logits.shape[-1])) * label_mask
# compute accuracy
accuracy = jnp.equal(jnp.argmax(logits, axis=-1), labels) * label_mask
# summarize metrics
metrics = {"loss": loss.sum(), "accuracy": accuracy.sum(), "normalizer": label_mask.sum()}
metrics = jax.lax.psum(metrics, axis_name="batch")
return metrics
p_eval_step = jax.pmap(eval_step, "batch", donate_argnums=(0,))
# Replicate the train state on each device
state = jax_utils.replicate(state)
train_time = 0
train_start = time.time()
train_metrics = []
eval_metrics = []
training_iter = iter(tokenized_datasets)
max_seq_length = min(data_args.max_seq_length, tokenizer.model_max_length)
eval_samples = advance_iter_and_group_samples(training_iter, data_args.num_eval_samples, max_seq_length)
steps = tqdm(range(num_train_steps), desc="Training...", position=0)
for step in range(num_train_steps):
# ======================== Training ================================
try:
samples = advance_iter_and_group_samples(training_iter, train_batch_size, max_seq_length)
except StopIteration:
# Once the end of the dataset stream is reached, the training iterator
# is reinitialized and reshuffled and a new eval dataset is randomly chosen.
shuffle_seed += 1
tokenized_datasets.set_epoch(shuffle_seed)
training_iter = iter(tokenized_datasets)
eval_dataset = advance_iter_and_group_samples(training_iter, data_args.num_eval_samples, max_seq_length)
samples = advance_iter_and_group_samples(training_iter, train_batch_size, max_seq_length)
# process input samples
model_inputs = data_collator(samples)
# Model forward
model_inputs = shard(model_inputs.data)
state, train_metric, dropout_rngs = p_train_step(state, model_inputs, dropout_rngs)
train_metrics.append(train_metric)
if step % training_args.logging_steps == 0 and step > 0:
steps.write(
f"Step... ({step} | Loss: {train_metric['loss'].mean()}, Learning Rate:"
f" {train_metric['learning_rate'].mean()})"
)
train_time += time.time() - train_start
if has_tensorboard and jax.process_index() == 0:
write_train_metric(summary_writer, train_metrics, train_time, step)
train_metrics = []
# ======================== Evaluating ==============================
if step % training_args.eval_steps == 0 and step > 0:
# Avoid using jax.numpy here in case of TPU training
eval_samples_idx = np.arange(data_args.num_eval_samples)
eval_batch_idx = generate_batch_splits(eval_samples_idx, eval_batch_size)
for i, batch_idx in enumerate(tqdm(eval_batch_idx, desc="Evaluating ...", position=1)):
# process input samples
batch_eval_samples = {k: [v[idx] for idx in batch_idx] for k, v in eval_samples.items()}
model_inputs = data_collator(batch_eval_samples)
# Model forward
model_inputs = shard(model_inputs.data)
metrics = p_eval_step(state.params, model_inputs)
eval_metrics.append(metrics)
# normalize eval metrics
eval_metrics = get_metrics(eval_metrics)
eval_metrics = jax.tree_util.tree_map(jnp.sum, eval_metrics)
eval_normalizer = eval_metrics.pop("normalizer")
eval_metrics = jax.tree_util.tree_map(lambda x: x / eval_normalizer, eval_metrics)
# Update progress bar
steps.desc = (
f"Step... ({step + 1}/{num_train_steps} | Loss: {eval_metrics['loss']}, Acc:"
f" {eval_metrics['accuracy']})"
)
if has_tensorboard and jax.process_index() == 0:
write_eval_metric(summary_writer, eval_metrics, step)
eval_metrics = []
# save checkpoint after each epoch and push checkpoint to the hub
if jax.process_index() == 0:
params = jax.device_get(jax.tree_util.tree_map(lambda x: x[0], state.params))
model.save_pretrained(
training_args.output_dir,
params=params,
push_to_hub=training_args.push_to_hub,
commit_message=f"Saving weights and logs of step {step+1}",
)
# update tqdm bar
steps.update(1)
| #!/usr/bin/env python
# coding=utf-8
# Copyright 2021 The HuggingFace Team All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Fine-tuning the library models for masked language modeling (BERT, ALBERT, RoBERTa...) with whole word masking on a
text file or a dataset.
Here is the full list of checkpoints on the hub that can be fine-tuned by this script:
https://huggingface.co/models?filter=fill-mask
"""
import logging
import os
import sys
import time
from collections import defaultdict
from dataclasses import dataclass, field
# You can also adapt this script on your own masked language modeling task. Pointers for this are left as comments.
from pathlib import Path
from typing import Dict, List, Optional, Tuple
import datasets
import flax
import jax
import jax.numpy as jnp
import numpy as np
import optax
from datasets import load_dataset
from flax import jax_utils, traverse_util
from flax.training import train_state
from flax.training.common_utils import get_metrics, onehot, shard
from tqdm import tqdm
from transformers import (
CONFIG_MAPPING,
FLAX_MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoTokenizer,
FlaxAutoModelForMaskedLM,
HfArgumentParser,
PreTrainedTokenizerBase,
TensorType,
TrainingArguments,
is_tensorboard_available,
set_seed,
)
if datasets.__version__ <= "1.8.0":
raise ValueError("Make sure to upgrade `datasets` to a version >= 1.9.0 to use dataset streaming")
MODEL_CONFIG_CLASSES = list(FLAX_MODEL_FOR_MASKED_LM_MAPPING.keys())
MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune, or train from scratch.
"""
model_name_or_path: Optional[str] = field(
default=None,
metadata={
"help": (
"The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."
)
},
)
model_type: Optional[str] = field(
default=None,
metadata={"help": "If training from scratch, pass a model type from the list: " + ", ".join(MODEL_TYPES)},
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
tokenizer_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None, metadata={"help": "Where do you want to store the pretrained models downloaded from s3"}
)
use_fast_tokenizer: bool = field(
default=True,
metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
)
dtype: Optional[str] = field(
default="float32",
metadata={
"help": (
"Floating-point format in which the model weights should be initialized and trained. Choose one of"
" `[float32, float16, bfloat16]`."
)
},
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
"""
dataset_name: Optional[str] = field(
default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
)
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
validation_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."},
)
train_ref_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input train ref data file for whole word masking in Chinese."},
)
validation_ref_file: Optional[str] = field(
default=None,
metadata={"help": "An optional input validation ref data file for whole word masking in Chinese."},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
validation_split_percentage: Optional[int] = field(
default=5,
metadata={
"help": "The percentage of the train set used as validation set in case there's no validation split"
},
)
max_seq_length: Optional[int] = field(
default=None,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated. Default to the max input length of the model."
)
},
)
preprocessing_num_workers: Optional[int] = field(
default=None,
metadata={"help": "The number of processes to use for the preprocessing."},
)
mlm_probability: float = field(
default=0.15, metadata={"help": "Ratio of tokens to mask for masked language modeling loss"}
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
},
)
line_by_line: bool = field(
default=False,
metadata={"help": "Whether distinct lines of text in the dataset are to be handled as distinct sequences."},
)
text_column_name: str = field(
default="text", metadata={"help": "The name of the column to retrieve the training text."}
)
shuffle_buffer_size: int = field(
default=10000, metadata={"help": "The number of examples to pre-load for shuffling."}
)
num_train_steps: int = field(default=50000, metadata={"help": "The number of training steps."})
num_eval_samples: int = field(default=50000, metadata={"help": "The number of samples to be used for evaluation"})
def __post_init__(self):
if self.dataset_name is None and self.train_file is None and self.validation_file is None:
raise ValueError("Need either a dataset name or a training/validation file.")
else:
if self.train_file is not None:
extension = self.train_file.split(".")[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
extension = self.validation_file.split(".")[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
@flax.struct.dataclass
class FlaxDataCollatorForLanguageModeling:
"""
Data collator used for language modeling. Inputs are dynamically padded to the maximum length of a batch if they
are not all of the same length.
Args:
tokenizer (:class:`~transformers.PreTrainedTokenizer` or :class:`~transformers.PreTrainedTokenizerFast`):
The tokenizer used for encoding the data.
mlm_probability (:obj:`float`, `optional`, defaults to 0.15):
The probability with which to (randomly) mask tokens in the input.
.. note::
For best performance, this data collator should be used with a dataset having items that are dictionaries or
BatchEncoding, with the :obj:`"special_tokens_mask"` key, as returned by a
:class:`~transformers.PreTrainedTokenizer` or a :class:`~transformers.PreTrainedTokenizerFast` with the
argument :obj:`return_special_tokens_mask=True`.
"""
tokenizer: PreTrainedTokenizerBase
mlm_probability: float = 0.15
def __post_init__(self):
if self.tokenizer.mask_token is None:
raise ValueError(
"This tokenizer does not have a mask token which is necessary for masked language modeling. "
"You should pass `mlm=False` to train on causal language modeling instead."
)
def __call__(self, examples: List[Dict[str, np.ndarray]]) -> Dict[str, np.ndarray]:
# Handle dict or lists with proper padding and conversion to tensor.
batch = self.tokenizer.pad(examples, return_tensors=TensorType.NUMPY)
# If special token mask has been preprocessed, pop it from the dict.
special_tokens_mask = batch.pop("special_tokens_mask", None)
batch["input_ids"], batch["labels"] = self.mask_tokens(
batch["input_ids"], special_tokens_mask=special_tokens_mask
)
return batch
def mask_tokens(
self, inputs: np.ndarray, special_tokens_mask: Optional[np.ndarray]
) -> Tuple[jnp.ndarray, jnp.ndarray]:
"""
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original.
"""
labels = inputs.copy()
# We sample a few tokens in each sequence for MLM training (with probability `self.mlm_probability`)
probability_matrix = np.full(labels.shape, self.mlm_probability)
special_tokens_mask = special_tokens_mask.astype("bool")
probability_matrix[special_tokens_mask] = 0.0
masked_indices = np.random.binomial(1, probability_matrix).astype("bool")
labels[~masked_indices] = -100 # We only compute loss on masked tokens
# 80% of the time, we replace masked input tokens with tokenizer.mask_token ([MASK])
indices_replaced = np.random.binomial(1, np.full(labels.shape, 0.8)).astype("bool") & masked_indices
inputs[indices_replaced] = self.tokenizer.convert_tokens_to_ids(self.tokenizer.mask_token)
# 10% of the time, we replace masked input tokens with random word
indices_random = np.random.binomial(1, np.full(labels.shape, 0.5)).astype("bool")
indices_random &= masked_indices & ~indices_replaced
random_words = np.random.randint(self.tokenizer.vocab_size, size=labels.shape, dtype="i4")
inputs[indices_random] = random_words[indices_random]
# The rest of the time (10% of the time) we keep the masked input tokens unchanged
return inputs, labels
def generate_batch_splits(samples_idx: np.ndarray, batch_size: int) -> np.ndarray:
num_samples = len(samples_idx)
samples_to_remove = num_samples % batch_size
if samples_to_remove != 0:
samples_idx = samples_idx[:-samples_to_remove]
sections_split = num_samples // batch_size
batch_idx = np.split(samples_idx, sections_split)
return batch_idx
def advance_iter_and_group_samples(train_iterator, num_samples, max_seq_length):
"""
The training iterator is advanced so that after groupifying the samples,
`num_samples` of length `max_seq_length` are returned.
"""
num_total_tokens = max_seq_length * num_samples
samples = defaultdict(list)
i = 0
while i < num_total_tokens:
tokenized_samples = next(train_iterator)
i += len(tokenized_samples["input_ids"])
# concatenate tokenized samples to list (excluding "id" and "text")
samples = {
k: samples[k] + tokenized_samples[k] for k in ["input_ids", "attention_mask", "special_tokens_mask"]
}
# Concatenated tokens are split to lists of length `max_seq_length`.
# Note that remainedr of % max_seq_length are thrown away.
def group_texts(examples):
result = {
k: [t[i : i + max_seq_length] for i in range(0, num_total_tokens, max_seq_length)]
for k, t in examples.items()
}
return result
grouped_samples = group_texts(samples)
return grouped_samples
def write_train_metric(summary_writer, train_metrics, train_time, step):
summary_writer.scalar("train_time", train_time, step)
train_metrics = get_metrics(train_metrics)
for key, vals in train_metrics.items():
tag = f"train_{key}"
for i, val in enumerate(vals):
summary_writer.scalar(tag, val, step - len(vals) + i + 1)
def write_eval_metric(summary_writer, eval_metrics, step):
for metric_name, value in eval_metrics.items():
summary_writer.scalar(f"eval_{metric_name}", value, step)
if __name__ == "__main__":
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
if (
os.path.exists(training_args.output_dir)
and os.listdir(training_args.output_dir)
and training_args.do_train
and not training_args.overwrite_output_dir
):
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty."
"Use --overwrite_output_dir to overcome."
)
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
level="INFO",
datefmt="[%X]",
)
# Log on each process the small summary:
logger = logging.getLogger(__name__)
logger.warning(
f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
+ f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
)
# Set the verbosity to info of the Transformers logger (on main process only):
logger.info(f"Training/evaluation parameters {training_args}")
# Set seed before initializing model.
set_seed(training_args.seed)
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
dataset = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
cache_dir=model_args.cache_dir,
streaming=True,
split="train",
)
if model_args.config_name:
config = AutoConfig.from_pretrained(model_args.config_name, cache_dir=model_args.cache_dir)
elif model_args.model_name_or_path:
config = AutoConfig.from_pretrained(model_args.model_name_or_path, cache_dir=model_args.cache_dir)
else:
config = CONFIG_MAPPING[model_args.model_type]()
logger.warning("You are instantiating a new config instance from scratch.")
if model_args.tokenizer_name:
tokenizer = AutoTokenizer.from_pretrained(
model_args.tokenizer_name, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
)
elif model_args.model_name_or_path:
tokenizer = AutoTokenizer.from_pretrained(
model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
)
else:
raise ValueError(
"You are instantiating a new tokenizer from scratch. This is not supported by this script."
"You can do it from another script, save it, and load it from here, using --tokenizer_name."
)
# Otherwise, we tokenize every text, then concatenate them together before splitting them in smaller parts.
# We use `return_special_tokens_mask=True` because DataCollatorForLanguageModeling (see below) is more
# efficient when it receives the `special_tokens_mask`.
def tokenize_function(examples):
return tokenizer(examples[data_args.text_column_name], return_special_tokens_mask=True)
tokenized_datasets = dataset.map(tokenize_function, batched=True, remove_columns=list(dataset.features.keys()))
shuffle_seed = training_args.seed
tokenized_datasets = tokenized_datasets.shuffle(buffer_size=data_args.shuffle_buffer_size, seed=shuffle_seed)
has_tensorboard = is_tensorboard_available()
if has_tensorboard and jax.process_index() == 0:
try:
from flax.metrics.tensorboard import SummaryWriter
except ImportError as ie:
has_tensorboard = False
logger.warning(
f"Unable to display metrics through TensorBoard because some package are not installed: {ie}"
)
summary_writer = SummaryWriter(log_dir=Path(training_args.output_dir))
# Data collator
# This one will take care of randomly masking the tokens.
data_collator = FlaxDataCollatorForLanguageModeling(tokenizer=tokenizer, mlm_probability=data_args.mlm_probability)
# Initialize our training
rng = jax.random.PRNGKey(training_args.seed)
dropout_rngs = jax.random.split(rng, jax.local_device_count())
if model_args.model_name_or_path:
model = FlaxAutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, config=config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
)
else:
model = FlaxAutoModelForMaskedLM.from_config(
config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
)
# Store some constant
num_epochs = int(training_args.num_train_epochs)
train_batch_size = int(training_args.per_device_train_batch_size) * jax.device_count()
eval_batch_size = int(training_args.per_device_eval_batch_size) * jax.device_count()
# define number steps per stream epoch
num_train_steps = data_args.num_train_steps
# Create learning rate schedule
warmup_fn = optax.linear_schedule(
init_value=0.0, end_value=training_args.learning_rate, transition_steps=training_args.warmup_steps
)
decay_fn = optax.linear_schedule(
init_value=training_args.learning_rate,
end_value=0,
transition_steps=num_train_steps - training_args.warmup_steps,
)
linear_decay_lr_schedule_fn = optax.join_schedules(
schedules=[warmup_fn, decay_fn], boundaries=[training_args.warmup_steps]
)
# We use Optax's "masking" functionality to not apply weight decay
# to bias and LayerNorm scale parameters. decay_mask_fn returns a
# mask boolean with the same structure as the parameters.
# The mask is True for parameters that should be decayed.
# Note that this mask is specifically adapted for FlaxBERT-like models.
# For other models, one should correct the layer norm parameter naming
# accordingly.
def decay_mask_fn(params):
flat_params = traverse_util.flatten_dict(params)
flat_mask = {path: (path[-1] != "bias" and path[-2:] != ("LayerNorm", "scale")) for path in flat_params}
return traverse_util.unflatten_dict(flat_mask)
# create adam optimizer
adamw = optax.adamw(
learning_rate=linear_decay_lr_schedule_fn,
b1=training_args.adam_beta1,
b2=training_args.adam_beta2,
eps=training_args.adam_epsilon,
weight_decay=training_args.weight_decay,
mask=decay_mask_fn,
)
# Setup train state
state = train_state.TrainState.create(apply_fn=model.__call__, params=model.params, tx=adamw)
# Define gradient update step fn
def train_step(state, batch, dropout_rng):
dropout_rng, new_dropout_rng = jax.random.split(dropout_rng)
def loss_fn(params):
labels = batch.pop("labels")
logits = state.apply_fn(**batch, params=params, dropout_rng=dropout_rng, train=True)[0]
# compute loss, ignore padded input tokens
label_mask = jnp.where(labels > 0, 1.0, 0.0)
loss = optax.softmax_cross_entropy(logits, onehot(labels, logits.shape[-1])) * label_mask
# take average
loss = loss.sum() / label_mask.sum()
return loss
grad_fn = jax.value_and_grad(loss_fn)
loss, grad = grad_fn(state.params)
grad = jax.lax.pmean(grad, "batch")
new_state = state.apply_gradients(grads=grad)
metrics = jax.lax.pmean(
{"loss": loss, "learning_rate": linear_decay_lr_schedule_fn(state.step)}, axis_name="batch"
)
return new_state, metrics, new_dropout_rng
# Create parallel version of the train step
p_train_step = jax.pmap(train_step, "batch", donate_argnums=(0,))
# Define eval fn
def eval_step(params, batch):
labels = batch.pop("labels")
logits = model(**batch, params=params, train=False)[0]
# compute loss, ignore padded input tokens
label_mask = jnp.where(labels > 0, 1.0, 0.0)
loss = optax.softmax_cross_entropy(logits, onehot(labels, logits.shape[-1])) * label_mask
# compute accuracy
accuracy = jnp.equal(jnp.argmax(logits, axis=-1), labels) * label_mask
# summarize metrics
metrics = {"loss": loss.sum(), "accuracy": accuracy.sum(), "normalizer": label_mask.sum()}
metrics = jax.lax.psum(metrics, axis_name="batch")
return metrics
p_eval_step = jax.pmap(eval_step, "batch", donate_argnums=(0,))
# Replicate the train state on each device
state = jax_utils.replicate(state)
train_time = 0
train_start = time.time()
train_metrics = []
eval_metrics = []
training_iter = iter(tokenized_datasets)
max_seq_length = min(data_args.max_seq_length, tokenizer.model_max_length)
eval_samples = advance_iter_and_group_samples(training_iter, data_args.num_eval_samples, max_seq_length)
steps = tqdm(range(num_train_steps), desc="Training...", position=0)
for step in range(num_train_steps):
# ======================== Training ================================
try:
samples = advance_iter_and_group_samples(training_iter, train_batch_size, max_seq_length)
except StopIteration:
# Once the end of the dataset stream is reached, the training iterator
# is reinitialized and reshuffled and a new eval dataset is randomly chosen.
shuffle_seed += 1
tokenized_datasets.set_epoch(shuffle_seed)
training_iter = iter(tokenized_datasets)
eval_dataset = advance_iter_and_group_samples(training_iter, data_args.num_eval_samples, max_seq_length)
samples = advance_iter_and_group_samples(training_iter, train_batch_size, max_seq_length)
# process input samples
model_inputs = data_collator(samples)
# Model forward
model_inputs = shard(model_inputs.data)
state, train_metric, dropout_rngs = p_train_step(state, model_inputs, dropout_rngs)
train_metrics.append(train_metric)
if step % training_args.logging_steps == 0 and step > 0:
steps.write(
f"Step... ({step} | Loss: {train_metric['loss'].mean()}, Learning Rate:"
f" {train_metric['learning_rate'].mean()})"
)
train_time += time.time() - train_start
if has_tensorboard and jax.process_index() == 0:
write_train_metric(summary_writer, train_metrics, train_time, step)
train_metrics = []
# ======================== Evaluating ==============================
if step % training_args.eval_steps == 0 and step > 0:
# Avoid using jax.numpy here in case of TPU training
eval_samples_idx = np.arange(data_args.num_eval_samples)
eval_batch_idx = generate_batch_splits(eval_samples_idx, eval_batch_size)
for i, batch_idx in enumerate(tqdm(eval_batch_idx, desc="Evaluating ...", position=1)):
# process input samples
batch_eval_samples = {k: [v[idx] for idx in batch_idx] for k, v in eval_samples.items()}
model_inputs = data_collator(batch_eval_samples)
# Model forward
model_inputs = shard(model_inputs.data)
metrics = p_eval_step(state.params, model_inputs)
eval_metrics.append(metrics)
# normalize eval metrics
eval_metrics = get_metrics(eval_metrics)
eval_metrics = jax.tree_util.tree_map(jnp.sum, eval_metrics)
eval_normalizer = eval_metrics.pop("normalizer")
eval_metrics = jax.tree_util.tree_map(lambda x: x / eval_normalizer, eval_metrics)
# Update progress bar
steps.desc = (
f"Step... ({step + 1}/{num_train_steps} | Loss: {eval_metrics['loss']}, Acc:"
f" {eval_metrics['accuracy']})"
)
if has_tensorboard and jax.process_index() == 0:
write_eval_metric(summary_writer, eval_metrics, step)
eval_metrics = []
# save checkpoint after each epoch and push checkpoint to the hub
if jax.process_index() == 0:
params = jax.device_get(jax.tree_util.tree_map(lambda x: x[0], state.params))
model.save_pretrained(
training_args.output_dir,
params=params,
push_to_hub=training_args.push_to_hub,
commit_message=f"Saving weights and logs of step {step+1}",
)
# update tqdm bar
steps.update(1)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./.git/objects/42/4caa4447aac94fb44979a1fa0748a2d67a9106 | x+)JMU057e040031Q,+dXD[,V Wj{V2,)517>9?(13|SUd+X.TU/u@V^XRWܦ'f{wGD\Terb^JfJbIj|zj^jQbI~H A Sewveϸӑ^
T_Zvĩ?NztZJK`K۲P i99%Eɩ` 1;C,~B
OF䑥:F,RSɴox4>BLSGl{o6u{U; yGpAd'ee&|*Quw懏1+gvnk#\}0S֜Vfݾ T]I>sjˏ~}^gd,́~3٥Оjds&ngl 5:w[.gvIsꔜ X | x+)JMU057e040031Q,+dXD[,V Wj{V2,)517>9?(13|SUd+X.TU/u@V^XRWܦ'f{wGD\Terb^JfJbIj|zj^jQbI~H A Sewveϸӑ^
T_Zvĩ?NztZJK`K۲P i99%Eɩ` 1;C,~B
OF䑥:F,RSɴox4>BLSGl{o6u{U; yGpAd'ee&|*Quw懏1+gvnk#\}0S֜Vfݾ T]I>sjˏ~}^gd,́~3٥Оjds&ngl 5:w[.gvIsꔜ X | -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/whisper/test_feature_extraction_whisper.py | # coding=utf-8
# Copyright 2022 HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import itertools
import os
import random
import tempfile
import unittest
import numpy as np
from transformers import is_speech_available
from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_speech_available():
from transformers import WhisperFeatureExtractor
if is_torch_available():
import torch
global_rng = random.Random()
def floats_list(shape, scale=1.0, rng=None, name=None):
"""Creates a random float32 tensor"""
if rng is None:
rng = global_rng
values = []
for batch_idx in range(shape[0]):
values.append([])
for _ in range(shape[1]):
values[-1].append(rng.random() * scale)
return values
@require_torch
@require_torchaudio
class WhisperFeatureExtractionTester(unittest.TestCase):
def __init__(
self,
parent,
batch_size=7,
min_seq_length=400,
max_seq_length=2000,
feature_size=10,
hop_length=160,
chunk_length=8,
padding_value=0.0,
sampling_rate=4_000,
return_attention_mask=False,
do_normalize=True,
):
self.parent = parent
self.batch_size = batch_size
self.min_seq_length = min_seq_length
self.max_seq_length = max_seq_length
self.seq_length_diff = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
self.padding_value = padding_value
self.sampling_rate = sampling_rate
self.return_attention_mask = return_attention_mask
self.do_normalize = do_normalize
self.feature_size = feature_size
self.chunk_length = chunk_length
self.hop_length = hop_length
def prepare_feat_extract_dict(self):
return {
"feature_size": self.feature_size,
"hop_length": self.hop_length,
"chunk_length": self.chunk_length,
"padding_value": self.padding_value,
"sampling_rate": self.sampling_rate,
"return_attention_mask": self.return_attention_mask,
"do_normalize": self.do_normalize,
}
def prepare_inputs_for_common(self, equal_length=False, numpify=False):
def _flatten(list_of_lists):
return list(itertools.chain(*list_of_lists))
if equal_length:
speech_inputs = [floats_list((self.max_seq_length, self.feature_size)) for _ in range(self.batch_size)]
else:
# make sure that inputs increase in size
speech_inputs = [
floats_list((x, self.feature_size))
for x in range(self.min_seq_length, self.max_seq_length, self.seq_length_diff)
]
if numpify:
speech_inputs = [np.asarray(x) for x in speech_inputs]
return speech_inputs
@require_torch
@require_torchaudio
class WhisperFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.TestCase):
feature_extraction_class = WhisperFeatureExtractor if is_speech_available() else None
def setUp(self):
self.feat_extract_tester = WhisperFeatureExtractionTester(self)
def test_feat_extract_from_and_save_pretrained(self):
feat_extract_first = self.feature_extraction_class(**self.feat_extract_dict)
with tempfile.TemporaryDirectory() as tmpdirname:
saved_file = feat_extract_first.save_pretrained(tmpdirname)[0]
check_json_file_has_correct_format(saved_file)
feat_extract_second = self.feature_extraction_class.from_pretrained(tmpdirname)
dict_first = feat_extract_first.to_dict()
dict_second = feat_extract_second.to_dict()
mel_1 = feat_extract_first.mel_filters
mel_2 = feat_extract_second.mel_filters
self.assertTrue(np.allclose(mel_1, mel_2))
self.assertEqual(dict_first, dict_second)
def test_feat_extract_to_json_file(self):
feat_extract_first = self.feature_extraction_class(**self.feat_extract_dict)
with tempfile.TemporaryDirectory() as tmpdirname:
json_file_path = os.path.join(tmpdirname, "feat_extract.json")
feat_extract_first.to_json_file(json_file_path)
feat_extract_second = self.feature_extraction_class.from_json_file(json_file_path)
dict_first = feat_extract_first.to_dict()
dict_second = feat_extract_second.to_dict()
mel_1 = feat_extract_first.mel_filters
mel_2 = feat_extract_second.mel_filters
self.assertTrue(np.allclose(mel_1, mel_2))
self.assertEqual(dict_first, dict_second)
def test_call(self):
# Tests that all call wrap to encode_plus and batch_encode_plus
feature_extractor = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict())
# create three inputs of length 800, 1000, and 1200
speech_inputs = [floats_list((1, x))[0] for x in range(800, 1400, 200)]
np_speech_inputs = [np.asarray(speech_input) for speech_input in speech_inputs]
# Test feature size
input_features = feature_extractor(np_speech_inputs, padding="max_length", return_tensors="np").input_features
self.assertTrue(input_features.ndim == 3)
self.assertTrue(input_features.shape[-1] == feature_extractor.nb_max_frames)
self.assertTrue(input_features.shape[-2] == feature_extractor.feature_size)
# Test not batched input
encoded_sequences_1 = feature_extractor(speech_inputs[0], return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs[0], return_tensors="np").input_features
self.assertTrue(np.allclose(encoded_sequences_1, encoded_sequences_2, atol=1e-3))
# Test batched
encoded_sequences_1 = feature_extractor(speech_inputs, return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs, return_tensors="np").input_features
for enc_seq_1, enc_seq_2 in zip(encoded_sequences_1, encoded_sequences_2):
self.assertTrue(np.allclose(enc_seq_1, enc_seq_2, atol=1e-3))
# Test truncation required
speech_inputs = [floats_list((1, x))[0] for x in range(200, (feature_extractor.n_samples + 500), 200)]
np_speech_inputs = [np.asarray(speech_input) for speech_input in speech_inputs]
speech_inputs_truncated = [x[: feature_extractor.n_samples] for x in speech_inputs]
np_speech_inputs_truncated = [np.asarray(speech_input) for speech_input in speech_inputs_truncated]
encoded_sequences_1 = feature_extractor(np_speech_inputs, return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs_truncated, return_tensors="np").input_features
for enc_seq_1, enc_seq_2 in zip(encoded_sequences_1, encoded_sequences_2):
self.assertTrue(np.allclose(enc_seq_1, enc_seq_2, atol=1e-3))
def test_double_precision_pad(self):
import torch
feature_extractor = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict())
np_speech_inputs = np.random.rand(100, 32).astype(np.float64)
py_speech_inputs = np_speech_inputs.tolist()
for inputs in [py_speech_inputs, np_speech_inputs]:
np_processed = feature_extractor.pad([{"input_features": inputs}], return_tensors="np")
self.assertTrue(np_processed.input_features.dtype == np.float32)
pt_processed = feature_extractor.pad([{"input_features": inputs}], return_tensors="pt")
self.assertTrue(pt_processed.input_features.dtype == torch.float32)
def _load_datasamples(self, num_samples):
from datasets import load_dataset
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
return [x["array"] for x in speech_samples]
def test_integration(self):
# fmt: off
EXPECTED_INPUT_FEATURES = torch.tensor(
[
0.1193, -0.0946, -0.1098, -0.0196, 0.0225, -0.0690, -0.1736, 0.0951,
0.0971, -0.0817, -0.0702, 0.0162, 0.0260, 0.0017, -0.0192, -0.1678,
0.0709, -0.1867, -0.0655, -0.0274, -0.0234, -0.1884, -0.0516, -0.0554,
-0.0274, -0.1425, -0.1423, 0.0837, 0.0377, -0.0854
]
)
# fmt: on
input_speech = self._load_datasamples(1)
feaure_extractor = WhisperFeatureExtractor()
input_features = feaure_extractor(input_speech, return_tensors="pt").input_features
self.assertTrue(torch.allclose(input_features[0, 0, :30], EXPECTED_INPUT_FEATURES, atol=1e-4))
| # coding=utf-8
# Copyright 2022 HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import itertools
import os
import random
import tempfile
import unittest
import numpy as np
from transformers import is_speech_available
from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_speech_available():
from transformers import WhisperFeatureExtractor
if is_torch_available():
import torch
global_rng = random.Random()
def floats_list(shape, scale=1.0, rng=None, name=None):
"""Creates a random float32 tensor"""
if rng is None:
rng = global_rng
values = []
for batch_idx in range(shape[0]):
values.append([])
for _ in range(shape[1]):
values[-1].append(rng.random() * scale)
return values
@require_torch
@require_torchaudio
class WhisperFeatureExtractionTester(unittest.TestCase):
def __init__(
self,
parent,
batch_size=7,
min_seq_length=400,
max_seq_length=2000,
feature_size=10,
hop_length=160,
chunk_length=8,
padding_value=0.0,
sampling_rate=4_000,
return_attention_mask=False,
do_normalize=True,
):
self.parent = parent
self.batch_size = batch_size
self.min_seq_length = min_seq_length
self.max_seq_length = max_seq_length
self.seq_length_diff = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
self.padding_value = padding_value
self.sampling_rate = sampling_rate
self.return_attention_mask = return_attention_mask
self.do_normalize = do_normalize
self.feature_size = feature_size
self.chunk_length = chunk_length
self.hop_length = hop_length
def prepare_feat_extract_dict(self):
return {
"feature_size": self.feature_size,
"hop_length": self.hop_length,
"chunk_length": self.chunk_length,
"padding_value": self.padding_value,
"sampling_rate": self.sampling_rate,
"return_attention_mask": self.return_attention_mask,
"do_normalize": self.do_normalize,
}
def prepare_inputs_for_common(self, equal_length=False, numpify=False):
def _flatten(list_of_lists):
return list(itertools.chain(*list_of_lists))
if equal_length:
speech_inputs = [floats_list((self.max_seq_length, self.feature_size)) for _ in range(self.batch_size)]
else:
# make sure that inputs increase in size
speech_inputs = [
floats_list((x, self.feature_size))
for x in range(self.min_seq_length, self.max_seq_length, self.seq_length_diff)
]
if numpify:
speech_inputs = [np.asarray(x) for x in speech_inputs]
return speech_inputs
@require_torch
@require_torchaudio
class WhisperFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.TestCase):
feature_extraction_class = WhisperFeatureExtractor if is_speech_available() else None
def setUp(self):
self.feat_extract_tester = WhisperFeatureExtractionTester(self)
def test_feat_extract_from_and_save_pretrained(self):
feat_extract_first = self.feature_extraction_class(**self.feat_extract_dict)
with tempfile.TemporaryDirectory() as tmpdirname:
saved_file = feat_extract_first.save_pretrained(tmpdirname)[0]
check_json_file_has_correct_format(saved_file)
feat_extract_second = self.feature_extraction_class.from_pretrained(tmpdirname)
dict_first = feat_extract_first.to_dict()
dict_second = feat_extract_second.to_dict()
mel_1 = feat_extract_first.mel_filters
mel_2 = feat_extract_second.mel_filters
self.assertTrue(np.allclose(mel_1, mel_2))
self.assertEqual(dict_first, dict_second)
def test_feat_extract_to_json_file(self):
feat_extract_first = self.feature_extraction_class(**self.feat_extract_dict)
with tempfile.TemporaryDirectory() as tmpdirname:
json_file_path = os.path.join(tmpdirname, "feat_extract.json")
feat_extract_first.to_json_file(json_file_path)
feat_extract_second = self.feature_extraction_class.from_json_file(json_file_path)
dict_first = feat_extract_first.to_dict()
dict_second = feat_extract_second.to_dict()
mel_1 = feat_extract_first.mel_filters
mel_2 = feat_extract_second.mel_filters
self.assertTrue(np.allclose(mel_1, mel_2))
self.assertEqual(dict_first, dict_second)
def test_call(self):
# Tests that all call wrap to encode_plus and batch_encode_plus
feature_extractor = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict())
# create three inputs of length 800, 1000, and 1200
speech_inputs = [floats_list((1, x))[0] for x in range(800, 1400, 200)]
np_speech_inputs = [np.asarray(speech_input) for speech_input in speech_inputs]
# Test feature size
input_features = feature_extractor(np_speech_inputs, padding="max_length", return_tensors="np").input_features
self.assertTrue(input_features.ndim == 3)
self.assertTrue(input_features.shape[-1] == feature_extractor.nb_max_frames)
self.assertTrue(input_features.shape[-2] == feature_extractor.feature_size)
# Test not batched input
encoded_sequences_1 = feature_extractor(speech_inputs[0], return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs[0], return_tensors="np").input_features
self.assertTrue(np.allclose(encoded_sequences_1, encoded_sequences_2, atol=1e-3))
# Test batched
encoded_sequences_1 = feature_extractor(speech_inputs, return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs, return_tensors="np").input_features
for enc_seq_1, enc_seq_2 in zip(encoded_sequences_1, encoded_sequences_2):
self.assertTrue(np.allclose(enc_seq_1, enc_seq_2, atol=1e-3))
# Test truncation required
speech_inputs = [floats_list((1, x))[0] for x in range(200, (feature_extractor.n_samples + 500), 200)]
np_speech_inputs = [np.asarray(speech_input) for speech_input in speech_inputs]
speech_inputs_truncated = [x[: feature_extractor.n_samples] for x in speech_inputs]
np_speech_inputs_truncated = [np.asarray(speech_input) for speech_input in speech_inputs_truncated]
encoded_sequences_1 = feature_extractor(np_speech_inputs, return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs_truncated, return_tensors="np").input_features
for enc_seq_1, enc_seq_2 in zip(encoded_sequences_1, encoded_sequences_2):
self.assertTrue(np.allclose(enc_seq_1, enc_seq_2, atol=1e-3))
def test_double_precision_pad(self):
import torch
feature_extractor = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict())
np_speech_inputs = np.random.rand(100, 32).astype(np.float64)
py_speech_inputs = np_speech_inputs.tolist()
for inputs in [py_speech_inputs, np_speech_inputs]:
np_processed = feature_extractor.pad([{"input_features": inputs}], return_tensors="np")
self.assertTrue(np_processed.input_features.dtype == np.float32)
pt_processed = feature_extractor.pad([{"input_features": inputs}], return_tensors="pt")
self.assertTrue(pt_processed.input_features.dtype == torch.float32)
def _load_datasamples(self, num_samples):
from datasets import load_dataset
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
return [x["array"] for x in speech_samples]
def test_integration(self):
# fmt: off
EXPECTED_INPUT_FEATURES = torch.tensor(
[
0.1193, -0.0946, -0.1098, -0.0196, 0.0225, -0.0690, -0.1736, 0.0951,
0.0971, -0.0817, -0.0702, 0.0162, 0.0260, 0.0017, -0.0192, -0.1678,
0.0709, -0.1867, -0.0655, -0.0274, -0.0234, -0.1884, -0.0516, -0.0554,
-0.0274, -0.1425, -0.1423, 0.0837, 0.0377, -0.0854
]
)
# fmt: on
input_speech = self._load_datasamples(1)
feaure_extractor = WhisperFeatureExtractor()
input_features = feaure_extractor(input_speech, return_tensors="pt").input_features
self.assertTrue(torch.allclose(input_features[0, 0, :30], EXPECTED_INPUT_FEATURES, atol=1e-4))
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/speech_encoder_decoder/test_modeling_flax_speech_encoder_decoder.py | # coding=utf-8
# Copyright 2022 HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import tempfile
import unittest
import numpy as np
from transformers import is_flax_available, is_torch_available
from transformers.testing_utils import is_pt_flax_cross_test, require_flax, slow, torch_device
from ...test_modeling_flax_common import floats_tensor, ids_tensor, random_attention_mask
from ..bart.test_modeling_flax_bart import FlaxBartStandaloneDecoderModelTester
from ..bert.test_modeling_flax_bert import FlaxBertModelTester
from ..gpt2.test_modeling_flax_gpt2 import FlaxGPT2ModelTester
from ..wav2vec2.test_modeling_flax_wav2vec2 import FlaxWav2Vec2ModelTester
if is_flax_available():
import jax
import jax.numpy as jnp
from flax.training.common_utils import onehot
from flax.traverse_util import flatten_dict
from transformers import (
FlaxBartForCausalLM,
FlaxBertForCausalLM,
FlaxGPT2LMHeadModel,
FlaxSpeechEncoderDecoderModel,
FlaxWav2Vec2Model,
SpeechEncoderDecoderConfig,
)
from transformers.modeling_flax_outputs import FlaxBaseModelOutput
from transformers.modeling_flax_pytorch_utils import (
convert_pytorch_state_dict_to_flax,
load_flax_weights_in_pytorch_model,
)
if is_torch_available():
import torch
from transformers import SpeechEncoderDecoderModel
@require_flax
class FlaxEncoderDecoderMixin:
def get_encoder_decoder_model(self, config, decoder_config):
raise NotImplementedError
def prepare_config_and_inputs(self):
raise NotImplementedError
def get_pretrained_model(self):
raise NotImplementedError
def check_encoder_decoder_model_from_pretrained_configs(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
self.assertTrue(encoder_decoder_config.decoder.is_decoder)
enc_dec_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
self.assertTrue(enc_dec_model.config.is_encoder_decoder)
self.assertFalse(enc_dec_model.config.tie_word_embeddings)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_encoder_decoder_model(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
enc_dec_model = SpeechEncoderDecoderModel(encoder=encoder_model, decoder=decoder_model)
self.assertTrue(enc_dec_model.config.decoder.is_decoder)
self.assertTrue(enc_dec_model.config.decoder.add_cross_attention)
self.assertTrue(enc_dec_model.config.is_encoder_decoder)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
encoder_outputs = FlaxBaseModelOutput(last_hidden_state=outputs_encoder_decoder.encoder_hidden_states[-1])
outputs_encoder_decoder = enc_dec_model(
attention_mask, decoder_input_ids, decoder_attention_mask, encoder_outputs=encoder_outputs
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_encoder_decoder_model_from_pretrained(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
return_dict,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model, "return_dict": return_dict}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
output_hidden_states=True,
return_dict=True,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_save_and_load(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
outputs = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
out_2 = np.array(outputs[0])
out_2[np.isnan(out_2)] = 0
with tempfile.TemporaryDirectory() as tmpdirname:
enc_dec_model.save_pretrained(tmpdirname)
FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname)
after_outputs = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
out_1 = np.array(after_outputs[0])
out_1[np.isnan(out_1)] = 0
max_diff = np.amax(np.abs(out_1 - out_2))
self.assertLessEqual(max_diff, 4e-2)
def check_encoder_decoder_model_from_encoder_decoder_pretrained(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
# assert that loading encoder and decoder models from configs has been correctly executed
self.assertEqual(config.add_adapter, encoder_model.config.add_adapter)
self.assertEqual(decoder_config.use_cache, decoder_model.config.use_cache)
with tempfile.TemporaryDirectory() as enc_tmpdir:
with tempfile.TemporaryDirectory() as dec_tmpdir:
encoder_model.save_pretrained(enc_tmpdir)
decoder_model.save_pretrained(dec_tmpdir)
# load a model from pretrained encoder and decoder checkpoints, setting one encoder and one decoder kwarg opposite to that specified in their respective configs
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
encoder_pretrained_model_name_or_path=enc_tmpdir,
decoder_pretrained_model_name_or_path=dec_tmpdir,
encoder_add_adapter=not config.add_adapter,
decoder_use_cache=not decoder_config.use_cache,
)
# assert that setting encoder and decoder kwargs opposite to those in the configs has correctly been applied
self.assertNotEqual(config.add_adapter, enc_dec_model.config.encoder.add_adapter)
self.assertNotEqual(decoder_config.use_cache, enc_dec_model.config.decoder.use_cache)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
output_hidden_states=True,
return_dict=True,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_encoder_decoder_model_output_attentions(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
# make the decoder inputs a different shape from the encoder inputs to harden the test
decoder_input_ids = decoder_input_ids[:, :-1]
decoder_attention_mask = decoder_attention_mask[:, :-1]
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
output_attentions=True,
)
encoder_attentions = outputs_encoder_decoder["encoder_attentions"]
self.assertEqual(len(encoder_attentions), config.num_hidden_layers)
seq_len = enc_dec_model._get_feat_extract_output_lengths(inputs.shape[1])
self.assertEqual(encoder_attentions[0].shape[-3:], (config.num_attention_heads, seq_len, seq_len))
decoder_attentions = outputs_encoder_decoder["decoder_attentions"]
num_decoder_layers = (
decoder_config.num_decoder_layers
if hasattr(decoder_config, "num_decoder_layers")
else decoder_config.num_hidden_layers
)
self.assertEqual(len(decoder_attentions), num_decoder_layers)
self.assertEqual(
decoder_attentions[0].shape[-3:],
(decoder_config.num_attention_heads, decoder_input_ids.shape[-1], decoder_input_ids.shape[-1]),
)
cross_attentions = outputs_encoder_decoder["cross_attentions"]
self.assertEqual(len(cross_attentions), num_decoder_layers)
cross_attention_input_seq_len = decoder_input_ids.shape[-1]
self.assertEqual(
cross_attentions[0].shape[-3:],
(decoder_config.num_attention_heads, cross_attention_input_seq_len, seq_len),
)
def check_encoder_decoder_model_generate(self, inputs, config, decoder_config, **kwargs):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
pad_token_id = enc_dec_model.config.decoder.pad_token_id
eos_token_id = enc_dec_model.config.decoder.eos_token_id
decoder_start_token_id = enc_dec_model.config.decoder.decoder_start_token_id
# Copied from generation.utils (GPT2 doesn't have `pad_token_id`)
if pad_token_id is None and eos_token_id is not None:
pad_token_id = eos_token_id
if decoder_start_token_id is None:
decoder_start_token_id = enc_dec_model.config.decoder.bos_token_id
# Bert does not have a bos token id, so use pad_token_id instead
# Copied from `test_modeling_encoder_decoder.py`
if decoder_start_token_id is None:
decoder_start_token_id = pad_token_id
generated_output = enc_dec_model.generate(
inputs,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
decoder_start_token_id=decoder_start_token_id,
)
generated_sequences = generated_output.sequences
self.assertEqual(generated_sequences.shape, (inputs.shape[0],) + (decoder_config.max_length,))
def check_freeze_feature_encoder(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
enc_dec_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
params = enc_dec_model.params
def cross_entropy(logits, labels):
return -jnp.sum(labels * jax.nn.log_softmax(logits, axis=-1), axis=-1)
# define a dummy loss function for computing the loss over a forward pass
def compute_loss(
params,
inputs,
attention_mask,
decoder_input_ids,
freeze_feature_encoder: bool = False,
):
outputs_enc_dec = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
freeze_feature_encoder=freeze_feature_encoder,
params=params,
)
logits = outputs_enc_dec.logits
vocab_size = logits.shape[-1]
loss = cross_entropy(logits, onehot(labels=decoder_input_ids, num_classes=vocab_size)).sum()
return (loss, logits)
# transform the loss function to get the gradients
grad_fn = jax.value_and_grad(compute_loss, has_aux=True)
# compute the loss, logits, and gradients for the unfrozen model
(loss, logits), grads = grad_fn(
params, inputs, attention_mask, decoder_input_ids, freeze_feature_encoder=False
)
# compare to the loss, logits and gradients for the frozen model
(loss_frozen, logits_frozen), grads_frozen = grad_fn(
params, inputs, attention_mask, decoder_input_ids, freeze_feature_encoder=True
)
# ensure that the logits and losses remain precisely equal
self.assertTrue((logits == logits_frozen).all())
self.assertEqual(loss, loss_frozen)
grads = flatten_dict(grads)
grads_frozen = flatten_dict(grads_frozen)
# ensure that the dicts of gradients contain the same keys
self.assertEqual(grads.keys(), grads_frozen.keys())
# ensure that the gradients of the feature extractor layers are precisely zero when frozen and contain non-zero entries when unfrozen
feature_extractor_grads = tuple(grads[k] for k in grads if "feature_extractor" in k)
feature_extractor_grads_frozen = tuple(grads_frozen[k] for k in grads_frozen if "feature_extractor" in k)
for feature_extractor_grad, feature_extractor_grad_frozen in zip(
feature_extractor_grads, feature_extractor_grads_frozen
):
self.assertTrue((feature_extractor_grad_frozen == 0.0).all())
self.assertTrue((feature_extractor_grad > 0.0).any())
# ensure that the gradients of all unfrozen layers remain precisely equal, i.e. all layers excluding the frozen 'feature_extractor'
grads = tuple(grads[k] for k in grads if "feature_extractor" not in k)
grads_frozen = tuple(grads_frozen[k] for k in grads_frozen if "feature_extractor" not in k)
for grad, grad_frozen in zip(grads, grads_frozen):
self.assertTrue((grad == grad_frozen).all())
def check_pt_flax_equivalence(self, pt_model, fx_model, inputs_dict):
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs).to_tuple()
fx_outputs = fx_model(**inputs_dict).to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
for fx_output, pt_output in zip(fx_outputs, pt_outputs):
self.assert_almost_equals(fx_output, pt_output.numpy(), 1e-5)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict).to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
for fx_output_loaded, pt_output in zip(fx_outputs_loaded, pt_outputs):
self.assert_almost_equals(fx_output_loaded, pt_output.numpy(), 1e-5)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs).to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
for fx_output, pt_output_loaded in zip(fx_outputs, pt_outputs_loaded):
self.assert_almost_equals(fx_output, pt_output_loaded.numpy(), 1e-5)
def check_equivalence_pt_to_flax(self, config, decoder_config, inputs_dict):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
pt_model = SpeechEncoderDecoderModel(encoder_decoder_config)
fx_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
fx_state = convert_pytorch_state_dict_to_flax(pt_model.state_dict(), fx_model)
fx_model.params = fx_state
self.check_pt_flax_equivalence(pt_model, fx_model, inputs_dict)
def check_equivalence_flax_to_pt(self, config, decoder_config, inputs_dict):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
pt_model = SpeechEncoderDecoderModel(encoder_decoder_config)
fx_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
pt_model = load_flax_weights_in_pytorch_model(pt_model, fx_model.params)
self.check_pt_flax_equivalence(pt_model, fx_model, inputs_dict)
def test_encoder_decoder_model_from_pretrained_configs(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_pretrained_configs(**input_ids_dict)
def test_encoder_decoder_model_from_pretrained(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_pretrained(**input_ids_dict, return_dict=False)
def test_encoder_decoder_model_from_pretrained_return_dict(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_pretrained(**input_ids_dict, return_dict=True)
def test_save_and_load_from_pretrained(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_save_and_load(**input_ids_dict)
def test_encoder_decoder_model_from_encoder_decoder_pretrained(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_encoder_decoder_pretrained(**input_ids_dict)
def test_encoder_decoder_model_output_attentions(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_output_attentions(**input_ids_dict)
def test_freeze_feature_encoder(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_freeze_feature_encoder(**input_ids_dict)
def test_encoder_decoder_model_generate(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_generate(**input_ids_dict)
def assert_almost_equals(self, a: np.ndarray, b: np.ndarray, tol: float):
diff = np.abs((a - b)).max()
self.assertLessEqual(diff, tol, f"Difference between torch and flax is {diff} (>= {tol}).")
@is_pt_flax_cross_test
def test_pt_flax_equivalence(self):
config_inputs_dict = self.prepare_config_and_inputs()
config = config_inputs_dict.pop("config")
decoder_config = config_inputs_dict.pop("decoder_config")
inputs_dict = config_inputs_dict
# `encoder_hidden_states` is not used in model call/forward
del inputs_dict["encoder_hidden_states"]
# Avoid the case where a sequence has no place to attend (after combined with the causal attention mask)
batch_size = inputs_dict["decoder_attention_mask"].shape[0]
inputs_dict["decoder_attention_mask"] = np.concatenate(
[np.ones(shape=(batch_size, 1)), inputs_dict["decoder_attention_mask"][:, 1:]], axis=1
)
# Flax models don't use the `use_cache` option and cache is not returned as a default.
# So we disable `use_cache` here for PyTorch model.
decoder_config.use_cache = False
self.assertTrue(decoder_config.cross_attention_hidden_size is None)
# check without `enc_to_dec_proj` projection
decoder_config.hidden_size = config.hidden_size
self.assertTrue(config.hidden_size == decoder_config.hidden_size)
self.check_equivalence_pt_to_flax(config, decoder_config, inputs_dict)
self.check_equivalence_flax_to_pt(config, decoder_config, inputs_dict)
# check `enc_to_dec_proj` work as expected
decoder_config.hidden_size = decoder_config.hidden_size * 2
self.assertTrue(config.hidden_size != decoder_config.hidden_size)
self.check_equivalence_pt_to_flax(config, decoder_config, inputs_dict)
self.check_equivalence_flax_to_pt(config, decoder_config, inputs_dict)
# check `add_adapter` works as expected
config.add_adapter = True
self.assertTrue(config.add_adapter)
self.check_equivalence_pt_to_flax(config, decoder_config, inputs_dict)
self.check_equivalence_flax_to_pt(config, decoder_config, inputs_dict)
@slow
def test_real_model_save_load_from_pretrained(self):
model_2 = self.get_pretrained_model()
inputs = ids_tensor([13, 5], model_2.config.encoder.vocab_size)
decoder_input_ids = ids_tensor([13, 1], model_2.config.decoder.vocab_size)
attention_mask = ids_tensor([13, 5], vocab_size=2)
outputs = model_2(
inputs=inputs,
decoder_input_ids=decoder_input_ids,
attention_mask=attention_mask,
)
out_2 = np.array(outputs[0])
out_2[np.isnan(out_2)] = 0
with tempfile.TemporaryDirectory() as tmp_dirname:
model_2.save_pretrained(tmp_dirname)
model_1 = FlaxSpeechEncoderDecoderModel.from_pretrained(tmp_dirname)
after_outputs = model_1(
inputs=inputs,
decoder_input_ids=decoder_input_ids,
attention_mask=attention_mask,
)
out_1 = np.array(after_outputs[0])
out_1[np.isnan(out_1)] = 0
max_diff = np.amax(np.abs(out_1 - out_2))
self.assertLessEqual(max_diff, 4e-2)
@require_flax
class FlaxWav2Vec2GPT2ModelTest(FlaxEncoderDecoderMixin, unittest.TestCase):
def get_pretrained_model_and_inputs(self):
model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
"facebook/wav2vec2-large-lv60", "gpt2-medium"
)
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
return model, inputs
def get_encoder_decoder_model(self, config, decoder_config):
encoder_model = FlaxWav2Vec2Model(config)
decoder_model = FlaxGPT2LMHeadModel(decoder_config)
return encoder_model, decoder_model
def prepare_config_and_inputs(self):
model_tester_encoder = FlaxWav2Vec2ModelTester(self, batch_size=13)
model_tester_decoder = FlaxGPT2ModelTester(self, batch_size=13)
encoder_config_and_inputs = model_tester_encoder.prepare_config_and_inputs()
decoder_config_and_inputs = model_tester_decoder.prepare_config_and_inputs_for_decoder()
(config, inputs, attention_mask) = encoder_config_and_inputs
(
decoder_config,
decoder_input_ids,
decoder_attention_mask,
encoder_hidden_states,
encoder_attention_mask,
) = decoder_config_and_inputs
# make sure that cross attention layers are added
decoder_config.add_cross_attention = True
return {
"config": config,
"inputs": inputs,
"attention_mask": attention_mask,
"decoder_config": decoder_config,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
"encoder_hidden_states": encoder_hidden_states,
}
@slow
def test_flaxwav2vec2gpt2_pt_flax_equivalence(self):
pt_model = SpeechEncoderDecoderModel.from_pretrained("jsnfly/wav2vec2-large-xlsr-53-german-gpt2")
fx_model = FlaxSpeechEncoderDecoderModel.from_pretrained(
"jsnfly/wav2vec2-large-xlsr-53-german-gpt2", from_pt=True
)
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], fx_model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs_dict = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs)
pt_logits = pt_outputs.logits
pt_outputs = pt_outputs.to_tuple()
fx_outputs = fx_model(**inputs_dict)
fx_logits = fx_outputs.logits
fx_outputs = fx_outputs.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits.numpy(), 4e-2)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict)
fx_logits_loaded = fx_outputs_loaded.logits
fx_outputs_loaded = fx_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits_loaded, pt_logits.numpy(), 4e-2)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs)
pt_logits_loaded = pt_outputs_loaded.logits
pt_outputs_loaded = pt_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits_loaded.numpy(), 4e-2)
@require_flax
class FlaxWav2Vec2BartModelTest(FlaxEncoderDecoderMixin, unittest.TestCase):
def get_pretrained_model_and_inputs(self):
model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
"facebook/wav2vec2-large-lv60", "bart-large"
)
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
return model, inputs
def get_encoder_decoder_model(self, config, decoder_config):
encoder_model = FlaxWav2Vec2Model(config)
decoder_model = FlaxBartForCausalLM(decoder_config)
return encoder_model, decoder_model
def prepare_config_and_inputs(self):
model_tester_encoder = FlaxWav2Vec2ModelTester(self, batch_size=13)
model_tester_decoder = FlaxBartStandaloneDecoderModelTester(self, batch_size=13)
encoder_config_and_inputs = model_tester_encoder.prepare_config_and_inputs()
decoder_config_and_inputs = model_tester_decoder.prepare_config_and_inputs_for_decoder()
(config, inputs, attention_mask) = encoder_config_and_inputs
(
decoder_config,
decoder_input_ids,
decoder_attention_mask,
encoder_hidden_states,
encoder_attention_mask,
) = decoder_config_and_inputs
# make sure that cross attention layers are added
decoder_config.add_cross_attention = True
return {
"config": config,
"inputs": inputs,
"attention_mask": attention_mask,
"decoder_config": decoder_config,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
"encoder_hidden_states": encoder_hidden_states,
}
@slow
def test_flaxwav2vec2bart_pt_flax_equivalence(self):
pt_model = SpeechEncoderDecoderModel.from_pretrained("patrickvonplaten/wav2vec2-2-bart-large")
fx_model = FlaxSpeechEncoderDecoderModel.from_pretrained(
"patrickvonplaten/wav2vec2-2-bart-large", from_pt=True
)
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], fx_model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs_dict = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs)
pt_logits = pt_outputs.logits
pt_outputs = pt_outputs.to_tuple()
fx_outputs = fx_model(**inputs_dict)
fx_logits = fx_outputs.logits
fx_outputs = fx_outputs.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits.numpy(), 4e-2)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict)
fx_logits_loaded = fx_outputs_loaded.logits
fx_outputs_loaded = fx_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits_loaded, pt_logits.numpy(), 4e-2)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs)
pt_logits_loaded = pt_outputs_loaded.logits
pt_outputs_loaded = pt_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits_loaded.numpy(), 4e-2)
@require_flax
class FlaxWav2Vec2BertModelTest(FlaxEncoderDecoderMixin, unittest.TestCase):
def get_pretrained_model_and_inputs(self):
model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
"facebook/wav2vec2-large-lv60", "bert-large-uncased"
)
batch_size = 13
input_values = floats_tensor([batch_size, 512], model.config.encoder.vocab_size)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
return model, inputs
def get_encoder_decoder_model(self, config, decoder_config):
encoder_model = FlaxWav2Vec2Model(config)
decoder_model = FlaxBertForCausalLM(decoder_config)
return encoder_model, decoder_model
def prepare_config_and_inputs(self):
model_tester_encoder = FlaxWav2Vec2ModelTester(self, batch_size=13)
model_tester_decoder = FlaxBertModelTester(self, batch_size=13)
encoder_config_and_inputs = model_tester_encoder.prepare_config_and_inputs()
decoder_config_and_inputs = model_tester_decoder.prepare_config_and_inputs_for_decoder()
(config, inputs, attention_mask) = encoder_config_and_inputs
(
decoder_config,
decoder_input_ids,
decoder_attention_mask,
encoder_hidden_states,
encoder_attention_mask,
) = decoder_config_and_inputs
# make sure that cross attention layers are added
decoder_config.add_cross_attention = True
return {
"config": config,
"inputs": inputs,
"attention_mask": attention_mask,
"decoder_config": decoder_config,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
"encoder_hidden_states": encoder_hidden_states,
}
@slow
def test_flaxwav2vec2bert_pt_flax_equivalence(self):
pt_model = SpeechEncoderDecoderModel.from_pretrained("speech-seq2seq/wav2vec2-2-bert-large")
fx_model = FlaxSpeechEncoderDecoderModel.from_pretrained("speech-seq2seq/wav2vec2-2-bert-large", from_pt=True)
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
batch_size = 13
input_values = floats_tensor([batch_size, 512], fx_model.config.encoder.vocab_size)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], fx_model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs_dict = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs)
pt_logits = pt_outputs.logits
pt_outputs = pt_outputs.to_tuple()
fx_outputs = fx_model(**inputs_dict)
fx_logits = fx_outputs.logits
fx_outputs = fx_outputs.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits.numpy(), 4e-2)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict)
fx_logits_loaded = fx_outputs_loaded.logits
fx_outputs_loaded = fx_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits_loaded, pt_logits.numpy(), 4e-2)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs)
pt_logits_loaded = pt_outputs_loaded.logits
pt_outputs_loaded = pt_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits_loaded.numpy(), 4e-2)
| # coding=utf-8
# Copyright 2022 HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import tempfile
import unittest
import numpy as np
from transformers import is_flax_available, is_torch_available
from transformers.testing_utils import is_pt_flax_cross_test, require_flax, slow, torch_device
from ...test_modeling_flax_common import floats_tensor, ids_tensor, random_attention_mask
from ..bart.test_modeling_flax_bart import FlaxBartStandaloneDecoderModelTester
from ..bert.test_modeling_flax_bert import FlaxBertModelTester
from ..gpt2.test_modeling_flax_gpt2 import FlaxGPT2ModelTester
from ..wav2vec2.test_modeling_flax_wav2vec2 import FlaxWav2Vec2ModelTester
if is_flax_available():
import jax
import jax.numpy as jnp
from flax.training.common_utils import onehot
from flax.traverse_util import flatten_dict
from transformers import (
FlaxBartForCausalLM,
FlaxBertForCausalLM,
FlaxGPT2LMHeadModel,
FlaxSpeechEncoderDecoderModel,
FlaxWav2Vec2Model,
SpeechEncoderDecoderConfig,
)
from transformers.modeling_flax_outputs import FlaxBaseModelOutput
from transformers.modeling_flax_pytorch_utils import (
convert_pytorch_state_dict_to_flax,
load_flax_weights_in_pytorch_model,
)
if is_torch_available():
import torch
from transformers import SpeechEncoderDecoderModel
@require_flax
class FlaxEncoderDecoderMixin:
def get_encoder_decoder_model(self, config, decoder_config):
raise NotImplementedError
def prepare_config_and_inputs(self):
raise NotImplementedError
def get_pretrained_model(self):
raise NotImplementedError
def check_encoder_decoder_model_from_pretrained_configs(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
self.assertTrue(encoder_decoder_config.decoder.is_decoder)
enc_dec_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
self.assertTrue(enc_dec_model.config.is_encoder_decoder)
self.assertFalse(enc_dec_model.config.tie_word_embeddings)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_encoder_decoder_model(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
enc_dec_model = SpeechEncoderDecoderModel(encoder=encoder_model, decoder=decoder_model)
self.assertTrue(enc_dec_model.config.decoder.is_decoder)
self.assertTrue(enc_dec_model.config.decoder.add_cross_attention)
self.assertTrue(enc_dec_model.config.is_encoder_decoder)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
encoder_outputs = FlaxBaseModelOutput(last_hidden_state=outputs_encoder_decoder.encoder_hidden_states[-1])
outputs_encoder_decoder = enc_dec_model(
attention_mask, decoder_input_ids, decoder_attention_mask, encoder_outputs=encoder_outputs
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_encoder_decoder_model_from_pretrained(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
return_dict,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model, "return_dict": return_dict}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
output_hidden_states=True,
return_dict=True,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_save_and_load(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
outputs = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
out_2 = np.array(outputs[0])
out_2[np.isnan(out_2)] = 0
with tempfile.TemporaryDirectory() as tmpdirname:
enc_dec_model.save_pretrained(tmpdirname)
FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname)
after_outputs = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
)
out_1 = np.array(after_outputs[0])
out_1[np.isnan(out_1)] = 0
max_diff = np.amax(np.abs(out_1 - out_2))
self.assertLessEqual(max_diff, 4e-2)
def check_encoder_decoder_model_from_encoder_decoder_pretrained(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
# assert that loading encoder and decoder models from configs has been correctly executed
self.assertEqual(config.add_adapter, encoder_model.config.add_adapter)
self.assertEqual(decoder_config.use_cache, decoder_model.config.use_cache)
with tempfile.TemporaryDirectory() as enc_tmpdir:
with tempfile.TemporaryDirectory() as dec_tmpdir:
encoder_model.save_pretrained(enc_tmpdir)
decoder_model.save_pretrained(dec_tmpdir)
# load a model from pretrained encoder and decoder checkpoints, setting one encoder and one decoder kwarg opposite to that specified in their respective configs
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
encoder_pretrained_model_name_or_path=enc_tmpdir,
decoder_pretrained_model_name_or_path=dec_tmpdir,
encoder_add_adapter=not config.add_adapter,
decoder_use_cache=not decoder_config.use_cache,
)
# assert that setting encoder and decoder kwargs opposite to those in the configs has correctly been applied
self.assertNotEqual(config.add_adapter, enc_dec_model.config.encoder.add_adapter)
self.assertNotEqual(decoder_config.use_cache, enc_dec_model.config.decoder.use_cache)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
output_hidden_states=True,
return_dict=True,
)
self.assertEqual(
outputs_encoder_decoder["logits"].shape, (decoder_input_ids.shape + (decoder_config.vocab_size,))
)
def check_encoder_decoder_model_output_attentions(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
# make the decoder inputs a different shape from the encoder inputs to harden the test
decoder_input_ids = decoder_input_ids[:, :-1]
decoder_attention_mask = decoder_attention_mask[:, :-1]
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
outputs_encoder_decoder = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
output_attentions=True,
)
encoder_attentions = outputs_encoder_decoder["encoder_attentions"]
self.assertEqual(len(encoder_attentions), config.num_hidden_layers)
seq_len = enc_dec_model._get_feat_extract_output_lengths(inputs.shape[1])
self.assertEqual(encoder_attentions[0].shape[-3:], (config.num_attention_heads, seq_len, seq_len))
decoder_attentions = outputs_encoder_decoder["decoder_attentions"]
num_decoder_layers = (
decoder_config.num_decoder_layers
if hasattr(decoder_config, "num_decoder_layers")
else decoder_config.num_hidden_layers
)
self.assertEqual(len(decoder_attentions), num_decoder_layers)
self.assertEqual(
decoder_attentions[0].shape[-3:],
(decoder_config.num_attention_heads, decoder_input_ids.shape[-1], decoder_input_ids.shape[-1]),
)
cross_attentions = outputs_encoder_decoder["cross_attentions"]
self.assertEqual(len(cross_attentions), num_decoder_layers)
cross_attention_input_seq_len = decoder_input_ids.shape[-1]
self.assertEqual(
cross_attentions[0].shape[-3:],
(decoder_config.num_attention_heads, cross_attention_input_seq_len, seq_len),
)
def check_encoder_decoder_model_generate(self, inputs, config, decoder_config, **kwargs):
encoder_model, decoder_model = self.get_encoder_decoder_model(config, decoder_config)
kwargs = {"encoder_model": encoder_model, "decoder_model": decoder_model}
enc_dec_model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(**kwargs)
pad_token_id = enc_dec_model.config.decoder.pad_token_id
eos_token_id = enc_dec_model.config.decoder.eos_token_id
decoder_start_token_id = enc_dec_model.config.decoder.decoder_start_token_id
# Copied from generation.utils (GPT2 doesn't have `pad_token_id`)
if pad_token_id is None and eos_token_id is not None:
pad_token_id = eos_token_id
if decoder_start_token_id is None:
decoder_start_token_id = enc_dec_model.config.decoder.bos_token_id
# Bert does not have a bos token id, so use pad_token_id instead
# Copied from `test_modeling_encoder_decoder.py`
if decoder_start_token_id is None:
decoder_start_token_id = pad_token_id
generated_output = enc_dec_model.generate(
inputs,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
decoder_start_token_id=decoder_start_token_id,
)
generated_sequences = generated_output.sequences
self.assertEqual(generated_sequences.shape, (inputs.shape[0],) + (decoder_config.max_length,))
def check_freeze_feature_encoder(
self,
config,
inputs,
attention_mask,
encoder_hidden_states,
decoder_config,
decoder_input_ids,
decoder_attention_mask,
**kwargs,
):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
enc_dec_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
params = enc_dec_model.params
def cross_entropy(logits, labels):
return -jnp.sum(labels * jax.nn.log_softmax(logits, axis=-1), axis=-1)
# define a dummy loss function for computing the loss over a forward pass
def compute_loss(
params,
inputs,
attention_mask,
decoder_input_ids,
freeze_feature_encoder: bool = False,
):
outputs_enc_dec = enc_dec_model(
inputs=inputs,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
freeze_feature_encoder=freeze_feature_encoder,
params=params,
)
logits = outputs_enc_dec.logits
vocab_size = logits.shape[-1]
loss = cross_entropy(logits, onehot(labels=decoder_input_ids, num_classes=vocab_size)).sum()
return (loss, logits)
# transform the loss function to get the gradients
grad_fn = jax.value_and_grad(compute_loss, has_aux=True)
# compute the loss, logits, and gradients for the unfrozen model
(loss, logits), grads = grad_fn(
params, inputs, attention_mask, decoder_input_ids, freeze_feature_encoder=False
)
# compare to the loss, logits and gradients for the frozen model
(loss_frozen, logits_frozen), grads_frozen = grad_fn(
params, inputs, attention_mask, decoder_input_ids, freeze_feature_encoder=True
)
# ensure that the logits and losses remain precisely equal
self.assertTrue((logits == logits_frozen).all())
self.assertEqual(loss, loss_frozen)
grads = flatten_dict(grads)
grads_frozen = flatten_dict(grads_frozen)
# ensure that the dicts of gradients contain the same keys
self.assertEqual(grads.keys(), grads_frozen.keys())
# ensure that the gradients of the feature extractor layers are precisely zero when frozen and contain non-zero entries when unfrozen
feature_extractor_grads = tuple(grads[k] for k in grads if "feature_extractor" in k)
feature_extractor_grads_frozen = tuple(grads_frozen[k] for k in grads_frozen if "feature_extractor" in k)
for feature_extractor_grad, feature_extractor_grad_frozen in zip(
feature_extractor_grads, feature_extractor_grads_frozen
):
self.assertTrue((feature_extractor_grad_frozen == 0.0).all())
self.assertTrue((feature_extractor_grad > 0.0).any())
# ensure that the gradients of all unfrozen layers remain precisely equal, i.e. all layers excluding the frozen 'feature_extractor'
grads = tuple(grads[k] for k in grads if "feature_extractor" not in k)
grads_frozen = tuple(grads_frozen[k] for k in grads_frozen if "feature_extractor" not in k)
for grad, grad_frozen in zip(grads, grads_frozen):
self.assertTrue((grad == grad_frozen).all())
def check_pt_flax_equivalence(self, pt_model, fx_model, inputs_dict):
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs).to_tuple()
fx_outputs = fx_model(**inputs_dict).to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
for fx_output, pt_output in zip(fx_outputs, pt_outputs):
self.assert_almost_equals(fx_output, pt_output.numpy(), 1e-5)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict).to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
for fx_output_loaded, pt_output in zip(fx_outputs_loaded, pt_outputs):
self.assert_almost_equals(fx_output_loaded, pt_output.numpy(), 1e-5)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs).to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
for fx_output, pt_output_loaded in zip(fx_outputs, pt_outputs_loaded):
self.assert_almost_equals(fx_output, pt_output_loaded.numpy(), 1e-5)
def check_equivalence_pt_to_flax(self, config, decoder_config, inputs_dict):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
pt_model = SpeechEncoderDecoderModel(encoder_decoder_config)
fx_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
fx_state = convert_pytorch_state_dict_to_flax(pt_model.state_dict(), fx_model)
fx_model.params = fx_state
self.check_pt_flax_equivalence(pt_model, fx_model, inputs_dict)
def check_equivalence_flax_to_pt(self, config, decoder_config, inputs_dict):
encoder_decoder_config = SpeechEncoderDecoderConfig.from_encoder_decoder_configs(config, decoder_config)
pt_model = SpeechEncoderDecoderModel(encoder_decoder_config)
fx_model = FlaxSpeechEncoderDecoderModel(encoder_decoder_config)
pt_model = load_flax_weights_in_pytorch_model(pt_model, fx_model.params)
self.check_pt_flax_equivalence(pt_model, fx_model, inputs_dict)
def test_encoder_decoder_model_from_pretrained_configs(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_pretrained_configs(**input_ids_dict)
def test_encoder_decoder_model_from_pretrained(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_pretrained(**input_ids_dict, return_dict=False)
def test_encoder_decoder_model_from_pretrained_return_dict(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_pretrained(**input_ids_dict, return_dict=True)
def test_save_and_load_from_pretrained(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_save_and_load(**input_ids_dict)
def test_encoder_decoder_model_from_encoder_decoder_pretrained(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_from_encoder_decoder_pretrained(**input_ids_dict)
def test_encoder_decoder_model_output_attentions(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_output_attentions(**input_ids_dict)
def test_freeze_feature_encoder(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_freeze_feature_encoder(**input_ids_dict)
def test_encoder_decoder_model_generate(self):
input_ids_dict = self.prepare_config_and_inputs()
self.check_encoder_decoder_model_generate(**input_ids_dict)
def assert_almost_equals(self, a: np.ndarray, b: np.ndarray, tol: float):
diff = np.abs((a - b)).max()
self.assertLessEqual(diff, tol, f"Difference between torch and flax is {diff} (>= {tol}).")
@is_pt_flax_cross_test
def test_pt_flax_equivalence(self):
config_inputs_dict = self.prepare_config_and_inputs()
config = config_inputs_dict.pop("config")
decoder_config = config_inputs_dict.pop("decoder_config")
inputs_dict = config_inputs_dict
# `encoder_hidden_states` is not used in model call/forward
del inputs_dict["encoder_hidden_states"]
# Avoid the case where a sequence has no place to attend (after combined with the causal attention mask)
batch_size = inputs_dict["decoder_attention_mask"].shape[0]
inputs_dict["decoder_attention_mask"] = np.concatenate(
[np.ones(shape=(batch_size, 1)), inputs_dict["decoder_attention_mask"][:, 1:]], axis=1
)
# Flax models don't use the `use_cache` option and cache is not returned as a default.
# So we disable `use_cache` here for PyTorch model.
decoder_config.use_cache = False
self.assertTrue(decoder_config.cross_attention_hidden_size is None)
# check without `enc_to_dec_proj` projection
decoder_config.hidden_size = config.hidden_size
self.assertTrue(config.hidden_size == decoder_config.hidden_size)
self.check_equivalence_pt_to_flax(config, decoder_config, inputs_dict)
self.check_equivalence_flax_to_pt(config, decoder_config, inputs_dict)
# check `enc_to_dec_proj` work as expected
decoder_config.hidden_size = decoder_config.hidden_size * 2
self.assertTrue(config.hidden_size != decoder_config.hidden_size)
self.check_equivalence_pt_to_flax(config, decoder_config, inputs_dict)
self.check_equivalence_flax_to_pt(config, decoder_config, inputs_dict)
# check `add_adapter` works as expected
config.add_adapter = True
self.assertTrue(config.add_adapter)
self.check_equivalence_pt_to_flax(config, decoder_config, inputs_dict)
self.check_equivalence_flax_to_pt(config, decoder_config, inputs_dict)
@slow
def test_real_model_save_load_from_pretrained(self):
model_2 = self.get_pretrained_model()
inputs = ids_tensor([13, 5], model_2.config.encoder.vocab_size)
decoder_input_ids = ids_tensor([13, 1], model_2.config.decoder.vocab_size)
attention_mask = ids_tensor([13, 5], vocab_size=2)
outputs = model_2(
inputs=inputs,
decoder_input_ids=decoder_input_ids,
attention_mask=attention_mask,
)
out_2 = np.array(outputs[0])
out_2[np.isnan(out_2)] = 0
with tempfile.TemporaryDirectory() as tmp_dirname:
model_2.save_pretrained(tmp_dirname)
model_1 = FlaxSpeechEncoderDecoderModel.from_pretrained(tmp_dirname)
after_outputs = model_1(
inputs=inputs,
decoder_input_ids=decoder_input_ids,
attention_mask=attention_mask,
)
out_1 = np.array(after_outputs[0])
out_1[np.isnan(out_1)] = 0
max_diff = np.amax(np.abs(out_1 - out_2))
self.assertLessEqual(max_diff, 4e-2)
@require_flax
class FlaxWav2Vec2GPT2ModelTest(FlaxEncoderDecoderMixin, unittest.TestCase):
def get_pretrained_model_and_inputs(self):
model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
"facebook/wav2vec2-large-lv60", "gpt2-medium"
)
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
return model, inputs
def get_encoder_decoder_model(self, config, decoder_config):
encoder_model = FlaxWav2Vec2Model(config)
decoder_model = FlaxGPT2LMHeadModel(decoder_config)
return encoder_model, decoder_model
def prepare_config_and_inputs(self):
model_tester_encoder = FlaxWav2Vec2ModelTester(self, batch_size=13)
model_tester_decoder = FlaxGPT2ModelTester(self, batch_size=13)
encoder_config_and_inputs = model_tester_encoder.prepare_config_and_inputs()
decoder_config_and_inputs = model_tester_decoder.prepare_config_and_inputs_for_decoder()
(config, inputs, attention_mask) = encoder_config_and_inputs
(
decoder_config,
decoder_input_ids,
decoder_attention_mask,
encoder_hidden_states,
encoder_attention_mask,
) = decoder_config_and_inputs
# make sure that cross attention layers are added
decoder_config.add_cross_attention = True
return {
"config": config,
"inputs": inputs,
"attention_mask": attention_mask,
"decoder_config": decoder_config,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
"encoder_hidden_states": encoder_hidden_states,
}
@slow
def test_flaxwav2vec2gpt2_pt_flax_equivalence(self):
pt_model = SpeechEncoderDecoderModel.from_pretrained("jsnfly/wav2vec2-large-xlsr-53-german-gpt2")
fx_model = FlaxSpeechEncoderDecoderModel.from_pretrained(
"jsnfly/wav2vec2-large-xlsr-53-german-gpt2", from_pt=True
)
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], fx_model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs_dict = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs)
pt_logits = pt_outputs.logits
pt_outputs = pt_outputs.to_tuple()
fx_outputs = fx_model(**inputs_dict)
fx_logits = fx_outputs.logits
fx_outputs = fx_outputs.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits.numpy(), 4e-2)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict)
fx_logits_loaded = fx_outputs_loaded.logits
fx_outputs_loaded = fx_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits_loaded, pt_logits.numpy(), 4e-2)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs)
pt_logits_loaded = pt_outputs_loaded.logits
pt_outputs_loaded = pt_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits_loaded.numpy(), 4e-2)
@require_flax
class FlaxWav2Vec2BartModelTest(FlaxEncoderDecoderMixin, unittest.TestCase):
def get_pretrained_model_and_inputs(self):
model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
"facebook/wav2vec2-large-lv60", "bart-large"
)
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
return model, inputs
def get_encoder_decoder_model(self, config, decoder_config):
encoder_model = FlaxWav2Vec2Model(config)
decoder_model = FlaxBartForCausalLM(decoder_config)
return encoder_model, decoder_model
def prepare_config_and_inputs(self):
model_tester_encoder = FlaxWav2Vec2ModelTester(self, batch_size=13)
model_tester_decoder = FlaxBartStandaloneDecoderModelTester(self, batch_size=13)
encoder_config_and_inputs = model_tester_encoder.prepare_config_and_inputs()
decoder_config_and_inputs = model_tester_decoder.prepare_config_and_inputs_for_decoder()
(config, inputs, attention_mask) = encoder_config_and_inputs
(
decoder_config,
decoder_input_ids,
decoder_attention_mask,
encoder_hidden_states,
encoder_attention_mask,
) = decoder_config_and_inputs
# make sure that cross attention layers are added
decoder_config.add_cross_attention = True
return {
"config": config,
"inputs": inputs,
"attention_mask": attention_mask,
"decoder_config": decoder_config,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
"encoder_hidden_states": encoder_hidden_states,
}
@slow
def test_flaxwav2vec2bart_pt_flax_equivalence(self):
pt_model = SpeechEncoderDecoderModel.from_pretrained("patrickvonplaten/wav2vec2-2-bart-large")
fx_model = FlaxSpeechEncoderDecoderModel.from_pretrained(
"patrickvonplaten/wav2vec2-2-bart-large", from_pt=True
)
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
batch_size = 13
input_values = floats_tensor([batch_size, 512], scale=1.0)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], fx_model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs_dict = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs)
pt_logits = pt_outputs.logits
pt_outputs = pt_outputs.to_tuple()
fx_outputs = fx_model(**inputs_dict)
fx_logits = fx_outputs.logits
fx_outputs = fx_outputs.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits.numpy(), 4e-2)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict)
fx_logits_loaded = fx_outputs_loaded.logits
fx_outputs_loaded = fx_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits_loaded, pt_logits.numpy(), 4e-2)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs)
pt_logits_loaded = pt_outputs_loaded.logits
pt_outputs_loaded = pt_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits_loaded.numpy(), 4e-2)
@require_flax
class FlaxWav2Vec2BertModelTest(FlaxEncoderDecoderMixin, unittest.TestCase):
def get_pretrained_model_and_inputs(self):
model = FlaxSpeechEncoderDecoderModel.from_encoder_decoder_pretrained(
"facebook/wav2vec2-large-lv60", "bert-large-uncased"
)
batch_size = 13
input_values = floats_tensor([batch_size, 512], model.config.encoder.vocab_size)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
return model, inputs
def get_encoder_decoder_model(self, config, decoder_config):
encoder_model = FlaxWav2Vec2Model(config)
decoder_model = FlaxBertForCausalLM(decoder_config)
return encoder_model, decoder_model
def prepare_config_and_inputs(self):
model_tester_encoder = FlaxWav2Vec2ModelTester(self, batch_size=13)
model_tester_decoder = FlaxBertModelTester(self, batch_size=13)
encoder_config_and_inputs = model_tester_encoder.prepare_config_and_inputs()
decoder_config_and_inputs = model_tester_decoder.prepare_config_and_inputs_for_decoder()
(config, inputs, attention_mask) = encoder_config_and_inputs
(
decoder_config,
decoder_input_ids,
decoder_attention_mask,
encoder_hidden_states,
encoder_attention_mask,
) = decoder_config_and_inputs
# make sure that cross attention layers are added
decoder_config.add_cross_attention = True
return {
"config": config,
"inputs": inputs,
"attention_mask": attention_mask,
"decoder_config": decoder_config,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
"encoder_hidden_states": encoder_hidden_states,
}
@slow
def test_flaxwav2vec2bert_pt_flax_equivalence(self):
pt_model = SpeechEncoderDecoderModel.from_pretrained("speech-seq2seq/wav2vec2-2-bert-large")
fx_model = FlaxSpeechEncoderDecoderModel.from_pretrained("speech-seq2seq/wav2vec2-2-bert-large", from_pt=True)
pt_model.to(torch_device)
pt_model.eval()
# prepare inputs
batch_size = 13
input_values = floats_tensor([batch_size, 512], fx_model.config.encoder.vocab_size)
attention_mask = random_attention_mask([batch_size, 512])
decoder_input_ids = ids_tensor([batch_size, 4], fx_model.config.decoder.vocab_size)
decoder_attention_mask = random_attention_mask([batch_size, 4])
inputs_dict = {
"inputs": input_values,
"attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"decoder_attention_mask": decoder_attention_mask,
}
flax_inputs = inputs_dict
pt_inputs = {k: torch.tensor(v.tolist()) for k, v in flax_inputs.items()}
with torch.no_grad():
pt_outputs = pt_model(**pt_inputs)
pt_logits = pt_outputs.logits
pt_outputs = pt_outputs.to_tuple()
fx_outputs = fx_model(**inputs_dict)
fx_logits = fx_outputs.logits
fx_outputs = fx_outputs.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits.numpy(), 4e-2)
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(tmpdirname)
fx_model_loaded = FlaxSpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_pt=True)
fx_outputs_loaded = fx_model_loaded(**inputs_dict)
fx_logits_loaded = fx_outputs_loaded.logits
fx_outputs_loaded = fx_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs_loaded), len(pt_outputs), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits_loaded, pt_logits.numpy(), 4e-2)
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(tmpdirname)
pt_model_loaded = SpeechEncoderDecoderModel.from_pretrained(tmpdirname, from_flax=True)
pt_model_loaded.to(torch_device)
pt_model_loaded.eval()
with torch.no_grad():
pt_outputs_loaded = pt_model_loaded(**pt_inputs)
pt_logits_loaded = pt_outputs_loaded.logits
pt_outputs_loaded = pt_outputs_loaded.to_tuple()
self.assertEqual(len(fx_outputs), len(pt_outputs_loaded), "Output lengths differ between Flax and PyTorch")
self.assert_almost_equals(fx_logits, pt_logits_loaded.numpy(), 4e-2)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./.circleci/TROUBLESHOOT.md | # Troubleshooting
This is a document explaining how to deal with various issues on Circle-CI. The entries may include actually solutions or pointers to Issues that cover those.
## Circle CI
* pytest worker runs out of resident RAM and gets killed by `cgroups`: https://github.com/huggingface/transformers/issues/11408
| # Troubleshooting
This is a document explaining how to deal with various issues on Circle-CI. The entries may include actually solutions or pointers to Issues that cover those.
## Circle CI
* pytest worker runs out of resident RAM and gets killed by `cgroups`: https://github.com/huggingface/transformers/issues/11408
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/clipseg/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/clipseg/convert_clipseg_original_pytorch_to_hf.py | # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Convert CLIPSeg checkpoints from the original repository. URL: https://github.com/timojl/clipseg."""
import argparse
import requests
import torch
from PIL import Image
from transformers import (
CLIPSegConfig,
CLIPSegForImageSegmentation,
CLIPSegProcessor,
CLIPSegTextConfig,
CLIPSegVisionConfig,
CLIPTokenizer,
ViTFeatureExtractor,
)
def get_clipseg_config(model_name):
text_config = CLIPSegTextConfig()
vision_config = CLIPSegVisionConfig(patch_size=16)
use_complex_transposed_convolution = True if "refined" in model_name else False
reduce_dim = 16 if "rd16" in model_name else 64
config = CLIPSegConfig.from_text_vision_configs(
text_config,
vision_config,
use_complex_transposed_convolution=use_complex_transposed_convolution,
reduce_dim=reduce_dim,
)
return config
def rename_key(name):
# update prefixes
if "clip_model" in name:
name = name.replace("clip_model", "clip")
if "transformer" in name:
if "visual" in name:
name = name.replace("visual.transformer", "vision_model")
else:
name = name.replace("transformer", "text_model")
if "resblocks" in name:
name = name.replace("resblocks", "encoder.layers")
if "ln_1" in name:
name = name.replace("ln_1", "layer_norm1")
if "ln_2" in name:
name = name.replace("ln_2", "layer_norm2")
if "c_fc" in name:
name = name.replace("c_fc", "fc1")
if "c_proj" in name:
name = name.replace("c_proj", "fc2")
if "attn" in name and "self" not in name:
name = name.replace("attn", "self_attn")
# text encoder
if "token_embedding" in name:
name = name.replace("token_embedding", "text_model.embeddings.token_embedding")
if "positional_embedding" in name and "visual" not in name:
name = name.replace("positional_embedding", "text_model.embeddings.position_embedding.weight")
if "ln_final" in name:
name = name.replace("ln_final", "text_model.final_layer_norm")
# vision encoder
if "visual.class_embedding" in name:
name = name.replace("visual.class_embedding", "vision_model.embeddings.class_embedding")
if "visual.conv1" in name:
name = name.replace("visual.conv1", "vision_model.embeddings.patch_embedding")
if "visual.positional_embedding" in name:
name = name.replace("visual.positional_embedding", "vision_model.embeddings.position_embedding.weight")
if "visual.ln_pre" in name:
name = name.replace("visual.ln_pre", "vision_model.pre_layrnorm")
if "visual.ln_post" in name:
name = name.replace("visual.ln_post", "vision_model.post_layernorm")
# projection layers
if "visual.proj" in name:
name = name.replace("visual.proj", "visual_projection.weight")
if "text_projection" in name:
name = name.replace("text_projection", "text_projection.weight")
# decoder
if "trans_conv" in name:
name = name.replace("trans_conv", "transposed_convolution")
if "film_mul" in name or "film_add" in name or "reduce" in name or "transposed_convolution" in name:
name = "decoder." + name
if "blocks" in name:
name = name.replace("blocks", "decoder.layers")
if "linear1" in name:
name = name.replace("linear1", "mlp.fc1")
if "linear2" in name:
name = name.replace("linear2", "mlp.fc2")
if "norm1" in name and "layer_" not in name:
name = name.replace("norm1", "layer_norm1")
if "norm2" in name and "layer_" not in name:
name = name.replace("norm2", "layer_norm2")
return name
def convert_state_dict(orig_state_dict, config):
for key in orig_state_dict.copy().keys():
val = orig_state_dict.pop(key)
if key.startswith("clip_model") and "attn.in_proj" in key:
key_split = key.split(".")
if "visual" in key:
layer_num = int(key_split[4])
dim = config.vision_config.hidden_size
prefix = "vision_model"
else:
layer_num = int(key_split[3])
dim = config.text_config.hidden_size
prefix = "text_model"
if "weight" in key:
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.q_proj.weight"] = val[:dim, :]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.k_proj.weight"] = val[
dim : dim * 2, :
]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.v_proj.weight"] = val[-dim:, :]
else:
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.q_proj.bias"] = val[:dim]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.k_proj.bias"] = val[dim : dim * 2]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.v_proj.bias"] = val[-dim:]
elif "self_attn" in key and "out_proj" not in key:
key_split = key.split(".")
layer_num = int(key_split[1])
dim = config.reduce_dim
if "weight" in key:
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.q_proj.weight"] = val[:dim, :]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.k_proj.weight"] = val[dim : dim * 2, :]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.v_proj.weight"] = val[-dim:, :]
else:
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.q_proj.bias"] = val[:dim]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.k_proj.bias"] = val[dim : dim * 2]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.v_proj.bias"] = val[-dim:]
else:
new_name = rename_key(key)
if "visual_projection" in new_name or "text_projection" in new_name:
val = val.T
orig_state_dict[new_name] = val
return orig_state_dict
# We will verify our results on an image of cute cats
def prepare_img():
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
image = Image.open(requests.get(url, stream=True).raw)
return image
def convert_clipseg_checkpoint(model_name, checkpoint_path, pytorch_dump_folder_path, push_to_hub):
config = get_clipseg_config(model_name)
model = CLIPSegForImageSegmentation(config)
model.eval()
state_dict = torch.load(checkpoint_path, map_location="cpu")
# remove some keys
for key in state_dict.copy().keys():
if key.startswith("model"):
state_dict.pop(key, None)
# rename some keys
state_dict = convert_state_dict(state_dict, config)
missing_keys, unexpected_keys = model.load_state_dict(state_dict, strict=False)
if missing_keys != ["clip.text_model.embeddings.position_ids", "clip.vision_model.embeddings.position_ids"]:
raise ValueError("Missing keys that are not expected: {}".format(missing_keys))
if unexpected_keys != ["decoder.reduce.weight", "decoder.reduce.bias"]:
raise ValueError(f"Unexpected keys: {unexpected_keys}")
feature_extractor = ViTFeatureExtractor(size=352)
tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-base-patch32")
processor = CLIPSegProcessor(feature_extractor=feature_extractor, tokenizer=tokenizer)
image = prepare_img()
text = ["a glass", "something to fill", "wood", "a jar"]
inputs = processor(text=text, images=[image] * len(text), padding="max_length", return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
# verify values
expected_conditional = torch.tensor([0.1110, -0.1882, 0.1645])
expected_pooled_output = torch.tensor([0.2692, -0.7197, -0.1328])
if model_name == "clipseg-rd64-refined":
expected_masks_slice = torch.tensor(
[[-10.0407, -9.9431, -10.2646], [-9.9751, -9.7064, -9.9586], [-9.6891, -9.5645, -9.9618]]
)
elif model_name == "clipseg-rd64":
expected_masks_slice = torch.tensor(
[[-7.2877, -7.2711, -7.2463], [-7.2652, -7.2780, -7.2520], [-7.2239, -7.2204, -7.2001]]
)
elif model_name == "clipseg-rd16":
expected_masks_slice = torch.tensor(
[[-6.3955, -6.4055, -6.4151], [-6.3911, -6.4033, -6.4100], [-6.3474, -6.3702, -6.3762]]
)
else:
raise ValueError(f"Model name {model_name} not supported.")
assert torch.allclose(outputs.logits[0, :3, :3], expected_masks_slice, atol=1e-3)
assert torch.allclose(outputs.conditional_embeddings[0, :3], expected_conditional, atol=1e-3)
assert torch.allclose(outputs.pooled_output[0, :3], expected_pooled_output, atol=1e-3)
print("Looks ok!")
if pytorch_dump_folder_path is not None:
print(f"Saving model and processor to {pytorch_dump_folder_path}")
model.save_pretrained(pytorch_dump_folder_path)
processor.save_pretrained(pytorch_dump_folder_path)
if push_to_hub:
print(f"Pushing model and processor for {model_name} to the hub")
model.push_to_hub(f"CIDAS/{model_name}")
processor.push_to_hub(f"CIDAS/{model_name}")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--model_name",
default="clipseg-rd64",
type=str,
choices=["clipseg-rd16", "clipseg-rd64", "clipseg-rd64-refined"],
help=(
"Name of the model. Supported models are: clipseg-rd64, clipseg-rd16 and clipseg-rd64-refined (rd meaning"
" reduce dimension)"
),
)
parser.add_argument(
"--checkpoint_path",
default="/Users/nielsrogge/Documents/CLIPSeg/clip_plus_rd64-uni.pth",
type=str,
help=(
"Path to the original checkpoint. Note that the script assumes that the checkpoint includes both CLIP and"
" the decoder weights."
),
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory."
)
parser.add_argument(
"--push_to_hub", action="store_true", help="Whether or not to push the converted model to the 🤗 hub."
)
args = parser.parse_args()
convert_clipseg_checkpoint(args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub)
| # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Convert CLIPSeg checkpoints from the original repository. URL: https://github.com/timojl/clipseg."""
import argparse
import requests
import torch
from PIL import Image
from transformers import (
CLIPSegConfig,
CLIPSegForImageSegmentation,
CLIPSegProcessor,
CLIPSegTextConfig,
CLIPSegVisionConfig,
CLIPTokenizer,
ViTFeatureExtractor,
)
def get_clipseg_config(model_name):
text_config = CLIPSegTextConfig()
vision_config = CLIPSegVisionConfig(patch_size=16)
use_complex_transposed_convolution = True if "refined" in model_name else False
reduce_dim = 16 if "rd16" in model_name else 64
config = CLIPSegConfig.from_text_vision_configs(
text_config,
vision_config,
use_complex_transposed_convolution=use_complex_transposed_convolution,
reduce_dim=reduce_dim,
)
return config
def rename_key(name):
# update prefixes
if "clip_model" in name:
name = name.replace("clip_model", "clip")
if "transformer" in name:
if "visual" in name:
name = name.replace("visual.transformer", "vision_model")
else:
name = name.replace("transformer", "text_model")
if "resblocks" in name:
name = name.replace("resblocks", "encoder.layers")
if "ln_1" in name:
name = name.replace("ln_1", "layer_norm1")
if "ln_2" in name:
name = name.replace("ln_2", "layer_norm2")
if "c_fc" in name:
name = name.replace("c_fc", "fc1")
if "c_proj" in name:
name = name.replace("c_proj", "fc2")
if "attn" in name and "self" not in name:
name = name.replace("attn", "self_attn")
# text encoder
if "token_embedding" in name:
name = name.replace("token_embedding", "text_model.embeddings.token_embedding")
if "positional_embedding" in name and "visual" not in name:
name = name.replace("positional_embedding", "text_model.embeddings.position_embedding.weight")
if "ln_final" in name:
name = name.replace("ln_final", "text_model.final_layer_norm")
# vision encoder
if "visual.class_embedding" in name:
name = name.replace("visual.class_embedding", "vision_model.embeddings.class_embedding")
if "visual.conv1" in name:
name = name.replace("visual.conv1", "vision_model.embeddings.patch_embedding")
if "visual.positional_embedding" in name:
name = name.replace("visual.positional_embedding", "vision_model.embeddings.position_embedding.weight")
if "visual.ln_pre" in name:
name = name.replace("visual.ln_pre", "vision_model.pre_layrnorm")
if "visual.ln_post" in name:
name = name.replace("visual.ln_post", "vision_model.post_layernorm")
# projection layers
if "visual.proj" in name:
name = name.replace("visual.proj", "visual_projection.weight")
if "text_projection" in name:
name = name.replace("text_projection", "text_projection.weight")
# decoder
if "trans_conv" in name:
name = name.replace("trans_conv", "transposed_convolution")
if "film_mul" in name or "film_add" in name or "reduce" in name or "transposed_convolution" in name:
name = "decoder." + name
if "blocks" in name:
name = name.replace("blocks", "decoder.layers")
if "linear1" in name:
name = name.replace("linear1", "mlp.fc1")
if "linear2" in name:
name = name.replace("linear2", "mlp.fc2")
if "norm1" in name and "layer_" not in name:
name = name.replace("norm1", "layer_norm1")
if "norm2" in name and "layer_" not in name:
name = name.replace("norm2", "layer_norm2")
return name
def convert_state_dict(orig_state_dict, config):
for key in orig_state_dict.copy().keys():
val = orig_state_dict.pop(key)
if key.startswith("clip_model") and "attn.in_proj" in key:
key_split = key.split(".")
if "visual" in key:
layer_num = int(key_split[4])
dim = config.vision_config.hidden_size
prefix = "vision_model"
else:
layer_num = int(key_split[3])
dim = config.text_config.hidden_size
prefix = "text_model"
if "weight" in key:
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.q_proj.weight"] = val[:dim, :]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.k_proj.weight"] = val[
dim : dim * 2, :
]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.v_proj.weight"] = val[-dim:, :]
else:
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.q_proj.bias"] = val[:dim]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.k_proj.bias"] = val[dim : dim * 2]
orig_state_dict[f"clip.{prefix}.encoder.layers.{layer_num}.self_attn.v_proj.bias"] = val[-dim:]
elif "self_attn" in key and "out_proj" not in key:
key_split = key.split(".")
layer_num = int(key_split[1])
dim = config.reduce_dim
if "weight" in key:
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.q_proj.weight"] = val[:dim, :]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.k_proj.weight"] = val[dim : dim * 2, :]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.v_proj.weight"] = val[-dim:, :]
else:
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.q_proj.bias"] = val[:dim]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.k_proj.bias"] = val[dim : dim * 2]
orig_state_dict[f"decoder.layers.{layer_num}.self_attn.v_proj.bias"] = val[-dim:]
else:
new_name = rename_key(key)
if "visual_projection" in new_name or "text_projection" in new_name:
val = val.T
orig_state_dict[new_name] = val
return orig_state_dict
# We will verify our results on an image of cute cats
def prepare_img():
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
image = Image.open(requests.get(url, stream=True).raw)
return image
def convert_clipseg_checkpoint(model_name, checkpoint_path, pytorch_dump_folder_path, push_to_hub):
config = get_clipseg_config(model_name)
model = CLIPSegForImageSegmentation(config)
model.eval()
state_dict = torch.load(checkpoint_path, map_location="cpu")
# remove some keys
for key in state_dict.copy().keys():
if key.startswith("model"):
state_dict.pop(key, None)
# rename some keys
state_dict = convert_state_dict(state_dict, config)
missing_keys, unexpected_keys = model.load_state_dict(state_dict, strict=False)
if missing_keys != ["clip.text_model.embeddings.position_ids", "clip.vision_model.embeddings.position_ids"]:
raise ValueError("Missing keys that are not expected: {}".format(missing_keys))
if unexpected_keys != ["decoder.reduce.weight", "decoder.reduce.bias"]:
raise ValueError(f"Unexpected keys: {unexpected_keys}")
feature_extractor = ViTFeatureExtractor(size=352)
tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-base-patch32")
processor = CLIPSegProcessor(feature_extractor=feature_extractor, tokenizer=tokenizer)
image = prepare_img()
text = ["a glass", "something to fill", "wood", "a jar"]
inputs = processor(text=text, images=[image] * len(text), padding="max_length", return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
# verify values
expected_conditional = torch.tensor([0.1110, -0.1882, 0.1645])
expected_pooled_output = torch.tensor([0.2692, -0.7197, -0.1328])
if model_name == "clipseg-rd64-refined":
expected_masks_slice = torch.tensor(
[[-10.0407, -9.9431, -10.2646], [-9.9751, -9.7064, -9.9586], [-9.6891, -9.5645, -9.9618]]
)
elif model_name == "clipseg-rd64":
expected_masks_slice = torch.tensor(
[[-7.2877, -7.2711, -7.2463], [-7.2652, -7.2780, -7.2520], [-7.2239, -7.2204, -7.2001]]
)
elif model_name == "clipseg-rd16":
expected_masks_slice = torch.tensor(
[[-6.3955, -6.4055, -6.4151], [-6.3911, -6.4033, -6.4100], [-6.3474, -6.3702, -6.3762]]
)
else:
raise ValueError(f"Model name {model_name} not supported.")
assert torch.allclose(outputs.logits[0, :3, :3], expected_masks_slice, atol=1e-3)
assert torch.allclose(outputs.conditional_embeddings[0, :3], expected_conditional, atol=1e-3)
assert torch.allclose(outputs.pooled_output[0, :3], expected_pooled_output, atol=1e-3)
print("Looks ok!")
if pytorch_dump_folder_path is not None:
print(f"Saving model and processor to {pytorch_dump_folder_path}")
model.save_pretrained(pytorch_dump_folder_path)
processor.save_pretrained(pytorch_dump_folder_path)
if push_to_hub:
print(f"Pushing model and processor for {model_name} to the hub")
model.push_to_hub(f"CIDAS/{model_name}")
processor.push_to_hub(f"CIDAS/{model_name}")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--model_name",
default="clipseg-rd64",
type=str,
choices=["clipseg-rd16", "clipseg-rd64", "clipseg-rd64-refined"],
help=(
"Name of the model. Supported models are: clipseg-rd64, clipseg-rd16 and clipseg-rd64-refined (rd meaning"
" reduce dimension)"
),
)
parser.add_argument(
"--checkpoint_path",
default="/Users/nielsrogge/Documents/CLIPSeg/clip_plus_rd64-uni.pth",
type=str,
help=(
"Path to the original checkpoint. Note that the script assumes that the checkpoint includes both CLIP and"
" the decoder weights."
),
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory."
)
parser.add_argument(
"--push_to_hub", action="store_true", help="Whether or not to push the converted model to the 🤗 hub."
)
args = parser.parse_args()
convert_clipseg_checkpoint(args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/xlnet/test_modeling_xlnet.py | # coding=utf-8
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import random
import unittest
from transformers import XLNetConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
XLNetForMultipleChoice,
XLNetForQuestionAnswering,
XLNetForQuestionAnsweringSimple,
XLNetForSequenceClassification,
XLNetForTokenClassification,
XLNetLMHeadModel,
XLNetModel,
)
from transformers.models.xlnet.modeling_xlnet import XLNET_PRETRAINED_MODEL_ARCHIVE_LIST
class XLNetModelTester:
def __init__(
self,
parent,
batch_size=14,
seq_length=7,
mem_len=10,
clamp_len=-1,
reuse_len=15,
is_training=True,
use_labels=True,
vocab_size=99,
cutoffs=[10, 50, 80],
hidden_size=32,
num_attention_heads=4,
d_inner=128,
num_hidden_layers=5,
type_sequence_label_size=2,
untie_r=True,
bi_data=False,
same_length=False,
initializer_range=0.05,
seed=1,
type_vocab_size=2,
bos_token_id=1,
eos_token_id=2,
pad_token_id=5,
num_choices=4,
):
self.parent = parent
self.batch_size = 14
self.seq_length = 7
self.mem_len = 10
# self.key_len = seq_length + mem_len
self.clamp_len = -1
self.reuse_len = 15
self.is_training = True
self.use_labels = True
self.vocab_size = 99
self.cutoffs = [10, 50, 80]
self.hidden_size = 32
self.num_attention_heads = 4
self.d_inner = 128
self.num_hidden_layers = 5
self.type_sequence_label_size = 2
self.untie_r = True
self.bi_data = False
self.same_length = False
self.initializer_range = 0.05
self.seed = 1
self.type_vocab_size = 2
self.bos_token_id = 1
self.eos_token_id = 2
self.pad_token_id = 5
self.num_choices = 4
def prepare_config_and_inputs(self):
input_ids_1 = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
input_ids_2 = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
segment_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
input_mask = random_attention_mask([self.batch_size, self.seq_length])
input_ids_q = ids_tensor([self.batch_size, self.seq_length + 1], self.vocab_size)
perm_mask = torch.zeros(
self.batch_size,
self.seq_length + 1,
self.seq_length + 1,
dtype=torch.float,
device=torch_device,
)
perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token
target_mapping = torch.zeros(
self.batch_size,
1,
self.seq_length + 1,
dtype=torch.float,
device=torch_device,
)
target_mapping[:, 0, -1] = 1.0 # predict last token
sequence_labels = None
lm_labels = None
is_impossible_labels = None
token_labels = None
if self.use_labels:
lm_labels = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
is_impossible_labels = ids_tensor([self.batch_size], 2).float()
token_labels = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
config = self.get_config()
return (
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
)
def get_config(self):
return XLNetConfig(
vocab_size=self.vocab_size,
d_model=self.hidden_size,
n_head=self.num_attention_heads,
d_inner=self.d_inner,
n_layer=self.num_hidden_layers,
untie_r=self.untie_r,
mem_len=self.mem_len,
clamp_len=self.clamp_len,
same_length=self.same_length,
reuse_len=self.reuse_len,
bi_data=self.bi_data,
initializer_range=self.initializer_range,
num_labels=self.type_sequence_label_size,
bos_token_id=self.bos_token_id,
pad_token_id=self.pad_token_id,
eos_token_id=self.eos_token_id,
)
def set_seed(self):
random.seed(self.seed)
torch.manual_seed(self.seed)
def create_and_check_xlnet_base_model(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetModel(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1, input_mask=input_mask)
result = model(input_ids_1, attention_mask=input_mask)
result = model(input_ids_1, token_type_ids=segment_ids)
result = model(input_ids_1)
config.mem_len = 0
model = XLNetModel(config)
model.to(torch_device)
model.eval()
base_model_output = model(input_ids_1)
self.parent.assertEqual(len(base_model_output), 2)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_use_mems_train(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForSequenceClassification(config)
model.to(torch_device)
model.train()
train_size = input_ids_1.shape[0]
batch_size = 4
for i in range(train_size // batch_size + 1):
input_ids = input_ids_1[i : (i + 1) * batch_size]
labels = sequence_labels[i : (i + 1) * batch_size]
outputs = model(input_ids=input_ids, labels=labels, return_dict=True)
self.parent.assertIsNone(outputs.mems)
self.parent.assertIsNotNone(outputs.loss)
def create_and_check_xlnet_model_use_mems(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetModel(config=config)
model.to(torch_device)
model.eval()
# first forward pass
causal_mask = torch.ones(
input_ids_1.shape[0],
input_ids_1.shape[1],
input_ids_1.shape[1],
dtype=torch.float,
device=torch_device,
)
causal_mask = torch.triu(causal_mask, diagonal=0)
outputs_cache = model(input_ids_1, use_mems=True, perm_mask=causal_mask)
outputs_no_cache = model(input_ids_1, use_mems=False, perm_mask=causal_mask)
outputs_conf = model(input_ids_1)
self.parent.assertTrue(len(outputs_cache) == len(outputs_conf))
self.parent.assertTrue(len(outputs_cache) == len(outputs_no_cache) + 1)
output, mems = outputs_cache.to_tuple()
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)
# append to next input_ids and token_type_ids
next_input_ids = torch.cat([input_ids_1, next_tokens], dim=-1)
# causal mask
causal_mask = torch.ones(
input_ids_1.shape[0],
input_ids_1.shape[1] + 1,
input_ids_1.shape[1] + 1,
dtype=torch.float,
device=torch_device,
)
causal_mask = torch.triu(causal_mask, diagonal=0)
single_mask = torch.ones(input_ids_1.shape[0], 1, 1, dtype=torch.float, device=torch_device)
# second forward pass
output_from_no_past = model(next_input_ids, perm_mask=causal_mask)["last_hidden_state"]
output_from_past = model(next_tokens, mems=mems, perm_mask=single_mask)["last_hidden_state"]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, -1, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))
def create_and_check_xlnet_base_model_with_att_output(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetModel(config)
model.to(torch_device)
model.eval()
attentions = model(input_ids_1, target_mapping=target_mapping, output_attentions=True)["attentions"]
self.parent.assertEqual(len(attentions), config.n_layer)
self.parent.assertIsInstance(attentions[0], tuple)
self.parent.assertEqual(len(attentions[0]), 2)
self.parent.assertTrue(attentions[0][0].shape, attentions[0][0].shape)
def create_and_check_xlnet_lm_head(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetLMHeadModel(config)
model.to(torch_device)
model.eval()
result1 = model(input_ids_1, token_type_ids=segment_ids, labels=lm_labels)
result2 = model(input_ids_2, token_type_ids=segment_ids, labels=lm_labels, mems=result1.mems)
_ = model(input_ids_q, perm_mask=perm_mask, target_mapping=target_mapping)
self.parent.assertEqual(result1.loss.shape, ())
self.parent.assertEqual(result1.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
self.parent.assertListEqual(
[mem.shape for mem in result1.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
self.parent.assertEqual(result2.loss.shape, ())
self.parent.assertEqual(result2.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
self.parent.assertListEqual(
[mem.shape for mem in result2.mems],
[(self.mem_len, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_xlnet_qa(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForQuestionAnswering(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1)
result_with_labels = model(
input_ids_1,
start_positions=sequence_labels,
end_positions=sequence_labels,
cls_index=sequence_labels,
is_impossible=is_impossible_labels,
p_mask=input_mask,
)
result_with_labels = model(
input_ids_1,
start_positions=sequence_labels,
end_positions=sequence_labels,
cls_index=sequence_labels,
is_impossible=is_impossible_labels,
)
total_loss, mems = result_with_labels.to_tuple()
result_with_labels = model(
input_ids_1,
start_positions=sequence_labels,
end_positions=sequence_labels,
)
total_loss, mems = result_with_labels.to_tuple()
self.parent.assertEqual(result_with_labels.loss.shape, ())
self.parent.assertEqual(result.start_top_log_probs.shape, (self.batch_size, model.config.start_n_top))
self.parent.assertEqual(result.start_top_index.shape, (self.batch_size, model.config.start_n_top))
self.parent.assertEqual(
result.end_top_log_probs.shape, (self.batch_size, model.config.start_n_top * model.config.end_n_top)
)
self.parent.assertEqual(
result.end_top_index.shape, (self.batch_size, model.config.start_n_top * model.config.end_n_top)
)
self.parent.assertEqual(result.cls_logits.shape, (self.batch_size,))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_xlnet_token_classif(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForTokenClassification(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1)
result = model(input_ids_1, labels=token_labels)
self.parent.assertEqual(result.loss.shape, ())
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.type_sequence_label_size))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_xlnet_sequence_classif(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForSequenceClassification(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1)
result = model(input_ids_1, labels=sequence_labels)
self.parent.assertEqual(result.loss.shape, ())
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.type_sequence_label_size))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
(
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
) = config_and_inputs
inputs_dict = {"input_ids": input_ids_1}
return config, inputs_dict
@require_torch
class XLNetModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(
XLNetModel,
XLNetLMHeadModel,
XLNetForTokenClassification,
XLNetForSequenceClassification,
XLNetForQuestionAnswering,
XLNetForQuestionAnsweringSimple,
XLNetForMultipleChoice,
)
if is_torch_available()
else ()
)
all_generative_model_classes = (
(XLNetLMHeadModel,) if is_torch_available() else ()
) # TODO (PVP): Check other models whether language generation is also applicable
pipeline_model_mapping = (
{
"feature-extraction": XLNetModel,
"question-answering": XLNetForQuestionAnsweringSimple,
"text-classification": XLNetForSequenceClassification,
"text-generation": XLNetLMHeadModel,
"token-classification": XLNetForTokenClassification,
"zero-shot": XLNetForSequenceClassification,
}
if is_torch_available()
else {}
)
fx_compatible = False
test_pruning = False
# XLNet has 2 QA models -> need to manually set the correct labels for one of them here
def _prepare_for_class(self, inputs_dict, model_class, return_labels=False):
inputs_dict = super()._prepare_for_class(inputs_dict, model_class, return_labels=return_labels)
if return_labels:
if model_class.__name__ == "XLNetForQuestionAnswering":
inputs_dict["start_positions"] = torch.zeros(
self.model_tester.batch_size, dtype=torch.long, device=torch_device
)
inputs_dict["end_positions"] = torch.zeros(
self.model_tester.batch_size, dtype=torch.long, device=torch_device
)
return inputs_dict
def setUp(self):
self.model_tester = XLNetModelTester(self)
self.config_tester = ConfigTester(self, config_class=XLNetConfig, d_inner=37)
def test_config(self):
self.config_tester.run_common_tests()
def test_xlnet_base_model(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_base_model(*config_and_inputs)
def test_xlnet_base_model_use_mems(self):
# checking that in auto-regressive mode, `use_mems` gives the same results
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_model_use_mems(*config_and_inputs)
def test_seq_classification_use_mems_train(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_use_mems_train(*config_and_inputs)
def test_xlnet_base_model_with_att_output(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_base_model_with_att_output(*config_and_inputs)
def test_xlnet_lm_head(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_lm_head(*config_and_inputs)
def test_xlnet_sequence_classif(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_sequence_classif(*config_and_inputs)
def test_xlnet_token_classif(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_token_classif(*config_and_inputs)
def test_xlnet_qa(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_qa(*config_and_inputs)
def test_retain_grad_hidden_states_attentions(self):
# xlnet cannot keep gradients in attentions or hidden states
return
# overwrite from test_modeling_common
def _mock_init_weights(self, module):
if hasattr(module, "weight") and module.weight is not None:
module.weight.data.fill_(3)
if hasattr(module, "bias") and module.bias is not None:
module.bias.data.fill_(3)
for param in ["q", "k", "v", "o", "r", "r_r_bias", "r_s_bias", "r_w_bias", "seg_embed", "mask_emb"]:
if hasattr(module, param) and getattr(module, param) is not None:
weight = getattr(module, param)
weight.data.fill_(3)
def _check_hidden_states_for_generate(
self, batch_size, hidden_states, min_length, max_length, config, use_cache=False, num_beam_groups=1
):
self.assertIsInstance(hidden_states, tuple)
self.assertListEqual(
[isinstance(iter_hidden_states, tuple) for iter_hidden_states in hidden_states],
[True] * len(hidden_states),
)
self.assertEqual(len(hidden_states), (max_length - min_length) * num_beam_groups)
for idx, iter_hidden_states in enumerate(hidden_states):
# check hidden size
for i, layer_hidden_states in enumerate(iter_hidden_states):
# every 2nd tensor is from extra stream
if i % 2 != 0:
seq_len = 1
else:
# for first item dummy PAD token is appended so need one more
seq_len = (min_length + 1) if idx == 0 else min_length
expected_shape = (batch_size * num_beam_groups, seq_len, config.hidden_size)
self.assertEqual(layer_hidden_states.shape, expected_shape)
def _check_attentions_for_generate(
self, batch_size, attentions, min_length, max_length, config, use_cache=False, num_beam_groups=1
):
self.assertIsInstance(attentions, tuple)
self.assertListEqual(
[isinstance(iter_attentions, tuple) for iter_attentions in attentions], [True] * len(attentions)
)
self.assertEqual(len(attentions), (max_length - min_length) * num_beam_groups)
for idx, attentions_item in enumerate(attentions):
for iter_attentions in attentions_item:
tgt_len = min_length
# for first item dummy PAD token is appended so need one more
if idx == 0:
tgt_len += 1
src_len = min_length + idx + 1
expected_shape = (
batch_size * num_beam_groups,
config.num_attention_heads,
tgt_len,
src_len,
)
# check attn size
self.assertListEqual(
[layer_attention.shape for layer_attention in iter_attentions],
[expected_shape] * len(iter_attentions),
)
@slow
def test_model_from_pretrained(self):
for model_name in XLNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
model = XLNetModel.from_pretrained(model_name)
self.assertIsNotNone(model)
@require_torch
class XLNetModelLanguageGenerationTest(unittest.TestCase):
@slow
def test_lm_generate_xlnet_base_cased(self):
model = XLNetLMHeadModel.from_pretrained("xlnet-base-cased")
model.to(torch_device)
# fmt: off
input_ids = torch.tensor(
[
[
67, 2840, 19, 18, 1484, 20, 965, 29077, 8719, 1273, 21, 45, 273, 17, 10, 15048, 28, 27511, 21, 4185, 11, 41, 2444, 9, 32, 1025, 20, 8719, 26, 23, 673, 966, 19, 29077, 20643, 27511, 20822, 20643, 19, 17, 6616, 17511, 18, 8978, 20, 18, 777, 9, 19233, 1527, 17669, 19, 24, 673, 17, 28756, 150, 12943, 4354, 153, 27, 442, 37, 45, 668, 21, 24, 256, 20, 416, 22, 2771, 4901, 9, 12943, 4354, 153, 51, 24, 3004, 21, 28142, 23, 65, 20, 18, 416, 34, 24, 2958, 22947, 9, 1177, 45, 668, 3097, 13768, 23, 103, 28, 441, 148, 48, 20522, 19, 12943, 4354, 153, 12860, 34, 18, 326, 27, 17492, 684, 21, 6709, 9, 8585, 123, 266, 19, 12943, 4354, 153, 6872, 24, 3004, 20, 18, 9225, 2198, 19, 12717, 103, 22, 401, 24, 6348, 9, 12943, 4354, 153, 1068, 2768, 2286, 19, 33, 104, 19, 176, 24, 9313, 19, 20086, 28, 45, 10292, 9, 4, 3,
]
],
dtype=torch.long,
device=torch_device,
)
# fmt: on
# In 1991, the remains of Russian Tsar Nicholas II and his family
# (except for Alexei and Maria) are discovered.
# The voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the
# remainder of the story. 1883 Western Siberia,
# a young Grigori Rasputin is asked by his father and a group of men to perform magic.
# Rasputin has a vision and denounces one of the men as a horse thief. Although his
# father initially slaps him for making such an accusation, Rasputin watches as the
# man is chased outside and beaten. Twenty years later, Rasputin sees a vision of
# the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous,
# with people, even a bishop, begging for his blessing. """
# fmt: off
expected_output_ids = [
67, 2840, 19, 18, 1484, 20, 965, 29077, 8719, 1273, 21, 45, 273, 17, 10, 15048, 28, 27511, 21, 4185, 11, 41, 2444, 9, 32, 1025, 20, 8719, 26, 23, 673, 966, 19, 29077, 20643, 27511, 20822, 20643, 19, 17, 6616, 17511, 18, 8978, 20, 18, 777, 9, 19233, 1527, 17669, 19, 24, 673, 17, 28756, 150, 12943, 4354, 153, 27, 442, 37, 45, 668, 21, 24, 256, 20, 416, 22, 2771, 4901, 9, 12943, 4354, 153, 51, 24, 3004, 21, 28142, 23, 65, 20, 18, 416, 34, 24, 2958, 22947, 9, 1177, 45, 668, 3097, 13768, 23, 103, 28, 441, 148, 48, 20522, 19, 12943, 4354, 153, 12860, 34, 18, 326, 27, 17492, 684, 21, 6709, 9, 8585, 123, 266, 19, 12943, 4354, 153, 6872, 24, 3004, 20, 18, 9225, 2198, 19, 12717, 103, 22, 401, 24, 6348, 9, 12943, 4354, 153, 1068, 2768, 2286, 19, 33, 104, 19, 176, 24, 9313, 19, 20086, 28, 45, 10292, 9, 4, 3, 19, 12943, 4354, 153, 27, 442, 22, 2771, 4901, 9, 69, 27, 442, 22, 2771, 24, 11335, 20, 18, 9225, 2198, 9, 69, 27, 442, 22, 2771, 24, 11335, 20, 18, 9225, 2198, 9, 69, 27, 442, 22, 2771,
]
# fmt: on
# In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria)
# are discovered. The voice of Nicholas's young son, Tsarevich Alexei Nikolaevich,
# narrates the remainder of the story. 1883 Western Siberia, a young Grigori Rasputin
# is asked by his father and a group of men to perform magic. Rasputin has a vision and
# denounces one of the men as a horse thief. Although his father initially slaps
# him for making such an accusation, Rasputin watches as the man is chased outside and beaten.
# Twenty years later, Rasputin sees a vision of the Virgin Mary, prompting him to become a priest.
# Rasputin quickly becomes famous, with people, even a bishop, begging for his blessing.
# <sep><cls>, Rasputin is asked to perform magic. He is asked to perform a ritual of the Virgin Mary.
# He is asked to perform a ritual of the Virgin Mary. He is asked to perform
output_ids = model.generate(input_ids, max_length=200, do_sample=False)
self.assertListEqual(output_ids[0].tolist(), expected_output_ids)
| # coding=utf-8
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import random
import unittest
from transformers import XLNetConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
XLNetForMultipleChoice,
XLNetForQuestionAnswering,
XLNetForQuestionAnsweringSimple,
XLNetForSequenceClassification,
XLNetForTokenClassification,
XLNetLMHeadModel,
XLNetModel,
)
from transformers.models.xlnet.modeling_xlnet import XLNET_PRETRAINED_MODEL_ARCHIVE_LIST
class XLNetModelTester:
def __init__(
self,
parent,
batch_size=14,
seq_length=7,
mem_len=10,
clamp_len=-1,
reuse_len=15,
is_training=True,
use_labels=True,
vocab_size=99,
cutoffs=[10, 50, 80],
hidden_size=32,
num_attention_heads=4,
d_inner=128,
num_hidden_layers=5,
type_sequence_label_size=2,
untie_r=True,
bi_data=False,
same_length=False,
initializer_range=0.05,
seed=1,
type_vocab_size=2,
bos_token_id=1,
eos_token_id=2,
pad_token_id=5,
num_choices=4,
):
self.parent = parent
self.batch_size = 14
self.seq_length = 7
self.mem_len = 10
# self.key_len = seq_length + mem_len
self.clamp_len = -1
self.reuse_len = 15
self.is_training = True
self.use_labels = True
self.vocab_size = 99
self.cutoffs = [10, 50, 80]
self.hidden_size = 32
self.num_attention_heads = 4
self.d_inner = 128
self.num_hidden_layers = 5
self.type_sequence_label_size = 2
self.untie_r = True
self.bi_data = False
self.same_length = False
self.initializer_range = 0.05
self.seed = 1
self.type_vocab_size = 2
self.bos_token_id = 1
self.eos_token_id = 2
self.pad_token_id = 5
self.num_choices = 4
def prepare_config_and_inputs(self):
input_ids_1 = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
input_ids_2 = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
segment_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
input_mask = random_attention_mask([self.batch_size, self.seq_length])
input_ids_q = ids_tensor([self.batch_size, self.seq_length + 1], self.vocab_size)
perm_mask = torch.zeros(
self.batch_size,
self.seq_length + 1,
self.seq_length + 1,
dtype=torch.float,
device=torch_device,
)
perm_mask[:, :, -1] = 1.0 # Previous tokens don't see last token
target_mapping = torch.zeros(
self.batch_size,
1,
self.seq_length + 1,
dtype=torch.float,
device=torch_device,
)
target_mapping[:, 0, -1] = 1.0 # predict last token
sequence_labels = None
lm_labels = None
is_impossible_labels = None
token_labels = None
if self.use_labels:
lm_labels = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
is_impossible_labels = ids_tensor([self.batch_size], 2).float()
token_labels = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
config = self.get_config()
return (
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
)
def get_config(self):
return XLNetConfig(
vocab_size=self.vocab_size,
d_model=self.hidden_size,
n_head=self.num_attention_heads,
d_inner=self.d_inner,
n_layer=self.num_hidden_layers,
untie_r=self.untie_r,
mem_len=self.mem_len,
clamp_len=self.clamp_len,
same_length=self.same_length,
reuse_len=self.reuse_len,
bi_data=self.bi_data,
initializer_range=self.initializer_range,
num_labels=self.type_sequence_label_size,
bos_token_id=self.bos_token_id,
pad_token_id=self.pad_token_id,
eos_token_id=self.eos_token_id,
)
def set_seed(self):
random.seed(self.seed)
torch.manual_seed(self.seed)
def create_and_check_xlnet_base_model(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetModel(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1, input_mask=input_mask)
result = model(input_ids_1, attention_mask=input_mask)
result = model(input_ids_1, token_type_ids=segment_ids)
result = model(input_ids_1)
config.mem_len = 0
model = XLNetModel(config)
model.to(torch_device)
model.eval()
base_model_output = model(input_ids_1)
self.parent.assertEqual(len(base_model_output), 2)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_use_mems_train(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForSequenceClassification(config)
model.to(torch_device)
model.train()
train_size = input_ids_1.shape[0]
batch_size = 4
for i in range(train_size // batch_size + 1):
input_ids = input_ids_1[i : (i + 1) * batch_size]
labels = sequence_labels[i : (i + 1) * batch_size]
outputs = model(input_ids=input_ids, labels=labels, return_dict=True)
self.parent.assertIsNone(outputs.mems)
self.parent.assertIsNotNone(outputs.loss)
def create_and_check_xlnet_model_use_mems(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetModel(config=config)
model.to(torch_device)
model.eval()
# first forward pass
causal_mask = torch.ones(
input_ids_1.shape[0],
input_ids_1.shape[1],
input_ids_1.shape[1],
dtype=torch.float,
device=torch_device,
)
causal_mask = torch.triu(causal_mask, diagonal=0)
outputs_cache = model(input_ids_1, use_mems=True, perm_mask=causal_mask)
outputs_no_cache = model(input_ids_1, use_mems=False, perm_mask=causal_mask)
outputs_conf = model(input_ids_1)
self.parent.assertTrue(len(outputs_cache) == len(outputs_conf))
self.parent.assertTrue(len(outputs_cache) == len(outputs_no_cache) + 1)
output, mems = outputs_cache.to_tuple()
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)
# append to next input_ids and token_type_ids
next_input_ids = torch.cat([input_ids_1, next_tokens], dim=-1)
# causal mask
causal_mask = torch.ones(
input_ids_1.shape[0],
input_ids_1.shape[1] + 1,
input_ids_1.shape[1] + 1,
dtype=torch.float,
device=torch_device,
)
causal_mask = torch.triu(causal_mask, diagonal=0)
single_mask = torch.ones(input_ids_1.shape[0], 1, 1, dtype=torch.float, device=torch_device)
# second forward pass
output_from_no_past = model(next_input_ids, perm_mask=causal_mask)["last_hidden_state"]
output_from_past = model(next_tokens, mems=mems, perm_mask=single_mask)["last_hidden_state"]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, -1, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))
def create_and_check_xlnet_base_model_with_att_output(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetModel(config)
model.to(torch_device)
model.eval()
attentions = model(input_ids_1, target_mapping=target_mapping, output_attentions=True)["attentions"]
self.parent.assertEqual(len(attentions), config.n_layer)
self.parent.assertIsInstance(attentions[0], tuple)
self.parent.assertEqual(len(attentions[0]), 2)
self.parent.assertTrue(attentions[0][0].shape, attentions[0][0].shape)
def create_and_check_xlnet_lm_head(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetLMHeadModel(config)
model.to(torch_device)
model.eval()
result1 = model(input_ids_1, token_type_ids=segment_ids, labels=lm_labels)
result2 = model(input_ids_2, token_type_ids=segment_ids, labels=lm_labels, mems=result1.mems)
_ = model(input_ids_q, perm_mask=perm_mask, target_mapping=target_mapping)
self.parent.assertEqual(result1.loss.shape, ())
self.parent.assertEqual(result1.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
self.parent.assertListEqual(
[mem.shape for mem in result1.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
self.parent.assertEqual(result2.loss.shape, ())
self.parent.assertEqual(result2.logits.shape, (self.batch_size, self.seq_length, self.vocab_size))
self.parent.assertListEqual(
[mem.shape for mem in result2.mems],
[(self.mem_len, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_xlnet_qa(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForQuestionAnswering(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1)
result_with_labels = model(
input_ids_1,
start_positions=sequence_labels,
end_positions=sequence_labels,
cls_index=sequence_labels,
is_impossible=is_impossible_labels,
p_mask=input_mask,
)
result_with_labels = model(
input_ids_1,
start_positions=sequence_labels,
end_positions=sequence_labels,
cls_index=sequence_labels,
is_impossible=is_impossible_labels,
)
total_loss, mems = result_with_labels.to_tuple()
result_with_labels = model(
input_ids_1,
start_positions=sequence_labels,
end_positions=sequence_labels,
)
total_loss, mems = result_with_labels.to_tuple()
self.parent.assertEqual(result_with_labels.loss.shape, ())
self.parent.assertEqual(result.start_top_log_probs.shape, (self.batch_size, model.config.start_n_top))
self.parent.assertEqual(result.start_top_index.shape, (self.batch_size, model.config.start_n_top))
self.parent.assertEqual(
result.end_top_log_probs.shape, (self.batch_size, model.config.start_n_top * model.config.end_n_top)
)
self.parent.assertEqual(
result.end_top_index.shape, (self.batch_size, model.config.start_n_top * model.config.end_n_top)
)
self.parent.assertEqual(result.cls_logits.shape, (self.batch_size,))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_xlnet_token_classif(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForTokenClassification(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1)
result = model(input_ids_1, labels=token_labels)
self.parent.assertEqual(result.loss.shape, ())
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.seq_length, self.type_sequence_label_size))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def create_and_check_xlnet_sequence_classif(
self,
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
):
model = XLNetForSequenceClassification(config)
model.to(torch_device)
model.eval()
result = model(input_ids_1)
result = model(input_ids_1, labels=sequence_labels)
self.parent.assertEqual(result.loss.shape, ())
self.parent.assertEqual(result.logits.shape, (self.batch_size, self.type_sequence_label_size))
self.parent.assertListEqual(
[mem.shape for mem in result.mems],
[(self.seq_length, self.batch_size, self.hidden_size)] * self.num_hidden_layers,
)
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
(
config,
input_ids_1,
input_ids_2,
input_ids_q,
perm_mask,
input_mask,
target_mapping,
segment_ids,
lm_labels,
sequence_labels,
is_impossible_labels,
token_labels,
) = config_and_inputs
inputs_dict = {"input_ids": input_ids_1}
return config, inputs_dict
@require_torch
class XLNetModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(
XLNetModel,
XLNetLMHeadModel,
XLNetForTokenClassification,
XLNetForSequenceClassification,
XLNetForQuestionAnswering,
XLNetForQuestionAnsweringSimple,
XLNetForMultipleChoice,
)
if is_torch_available()
else ()
)
all_generative_model_classes = (
(XLNetLMHeadModel,) if is_torch_available() else ()
) # TODO (PVP): Check other models whether language generation is also applicable
pipeline_model_mapping = (
{
"feature-extraction": XLNetModel,
"question-answering": XLNetForQuestionAnsweringSimple,
"text-classification": XLNetForSequenceClassification,
"text-generation": XLNetLMHeadModel,
"token-classification": XLNetForTokenClassification,
"zero-shot": XLNetForSequenceClassification,
}
if is_torch_available()
else {}
)
fx_compatible = False
test_pruning = False
# XLNet has 2 QA models -> need to manually set the correct labels for one of them here
def _prepare_for_class(self, inputs_dict, model_class, return_labels=False):
inputs_dict = super()._prepare_for_class(inputs_dict, model_class, return_labels=return_labels)
if return_labels:
if model_class.__name__ == "XLNetForQuestionAnswering":
inputs_dict["start_positions"] = torch.zeros(
self.model_tester.batch_size, dtype=torch.long, device=torch_device
)
inputs_dict["end_positions"] = torch.zeros(
self.model_tester.batch_size, dtype=torch.long, device=torch_device
)
return inputs_dict
def setUp(self):
self.model_tester = XLNetModelTester(self)
self.config_tester = ConfigTester(self, config_class=XLNetConfig, d_inner=37)
def test_config(self):
self.config_tester.run_common_tests()
def test_xlnet_base_model(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_base_model(*config_and_inputs)
def test_xlnet_base_model_use_mems(self):
# checking that in auto-regressive mode, `use_mems` gives the same results
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_model_use_mems(*config_and_inputs)
def test_seq_classification_use_mems_train(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_use_mems_train(*config_and_inputs)
def test_xlnet_base_model_with_att_output(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_base_model_with_att_output(*config_and_inputs)
def test_xlnet_lm_head(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_lm_head(*config_and_inputs)
def test_xlnet_sequence_classif(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_sequence_classif(*config_and_inputs)
def test_xlnet_token_classif(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_token_classif(*config_and_inputs)
def test_xlnet_qa(self):
self.model_tester.set_seed()
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_xlnet_qa(*config_and_inputs)
def test_retain_grad_hidden_states_attentions(self):
# xlnet cannot keep gradients in attentions or hidden states
return
# overwrite from test_modeling_common
def _mock_init_weights(self, module):
if hasattr(module, "weight") and module.weight is not None:
module.weight.data.fill_(3)
if hasattr(module, "bias") and module.bias is not None:
module.bias.data.fill_(3)
for param in ["q", "k", "v", "o", "r", "r_r_bias", "r_s_bias", "r_w_bias", "seg_embed", "mask_emb"]:
if hasattr(module, param) and getattr(module, param) is not None:
weight = getattr(module, param)
weight.data.fill_(3)
def _check_hidden_states_for_generate(
self, batch_size, hidden_states, min_length, max_length, config, use_cache=False, num_beam_groups=1
):
self.assertIsInstance(hidden_states, tuple)
self.assertListEqual(
[isinstance(iter_hidden_states, tuple) for iter_hidden_states in hidden_states],
[True] * len(hidden_states),
)
self.assertEqual(len(hidden_states), (max_length - min_length) * num_beam_groups)
for idx, iter_hidden_states in enumerate(hidden_states):
# check hidden size
for i, layer_hidden_states in enumerate(iter_hidden_states):
# every 2nd tensor is from extra stream
if i % 2 != 0:
seq_len = 1
else:
# for first item dummy PAD token is appended so need one more
seq_len = (min_length + 1) if idx == 0 else min_length
expected_shape = (batch_size * num_beam_groups, seq_len, config.hidden_size)
self.assertEqual(layer_hidden_states.shape, expected_shape)
def _check_attentions_for_generate(
self, batch_size, attentions, min_length, max_length, config, use_cache=False, num_beam_groups=1
):
self.assertIsInstance(attentions, tuple)
self.assertListEqual(
[isinstance(iter_attentions, tuple) for iter_attentions in attentions], [True] * len(attentions)
)
self.assertEqual(len(attentions), (max_length - min_length) * num_beam_groups)
for idx, attentions_item in enumerate(attentions):
for iter_attentions in attentions_item:
tgt_len = min_length
# for first item dummy PAD token is appended so need one more
if idx == 0:
tgt_len += 1
src_len = min_length + idx + 1
expected_shape = (
batch_size * num_beam_groups,
config.num_attention_heads,
tgt_len,
src_len,
)
# check attn size
self.assertListEqual(
[layer_attention.shape for layer_attention in iter_attentions],
[expected_shape] * len(iter_attentions),
)
@slow
def test_model_from_pretrained(self):
for model_name in XLNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
model = XLNetModel.from_pretrained(model_name)
self.assertIsNotNone(model)
@require_torch
class XLNetModelLanguageGenerationTest(unittest.TestCase):
@slow
def test_lm_generate_xlnet_base_cased(self):
model = XLNetLMHeadModel.from_pretrained("xlnet-base-cased")
model.to(torch_device)
# fmt: off
input_ids = torch.tensor(
[
[
67, 2840, 19, 18, 1484, 20, 965, 29077, 8719, 1273, 21, 45, 273, 17, 10, 15048, 28, 27511, 21, 4185, 11, 41, 2444, 9, 32, 1025, 20, 8719, 26, 23, 673, 966, 19, 29077, 20643, 27511, 20822, 20643, 19, 17, 6616, 17511, 18, 8978, 20, 18, 777, 9, 19233, 1527, 17669, 19, 24, 673, 17, 28756, 150, 12943, 4354, 153, 27, 442, 37, 45, 668, 21, 24, 256, 20, 416, 22, 2771, 4901, 9, 12943, 4354, 153, 51, 24, 3004, 21, 28142, 23, 65, 20, 18, 416, 34, 24, 2958, 22947, 9, 1177, 45, 668, 3097, 13768, 23, 103, 28, 441, 148, 48, 20522, 19, 12943, 4354, 153, 12860, 34, 18, 326, 27, 17492, 684, 21, 6709, 9, 8585, 123, 266, 19, 12943, 4354, 153, 6872, 24, 3004, 20, 18, 9225, 2198, 19, 12717, 103, 22, 401, 24, 6348, 9, 12943, 4354, 153, 1068, 2768, 2286, 19, 33, 104, 19, 176, 24, 9313, 19, 20086, 28, 45, 10292, 9, 4, 3,
]
],
dtype=torch.long,
device=torch_device,
)
# fmt: on
# In 1991, the remains of Russian Tsar Nicholas II and his family
# (except for Alexei and Maria) are discovered.
# The voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the
# remainder of the story. 1883 Western Siberia,
# a young Grigori Rasputin is asked by his father and a group of men to perform magic.
# Rasputin has a vision and denounces one of the men as a horse thief. Although his
# father initially slaps him for making such an accusation, Rasputin watches as the
# man is chased outside and beaten. Twenty years later, Rasputin sees a vision of
# the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous,
# with people, even a bishop, begging for his blessing. """
# fmt: off
expected_output_ids = [
67, 2840, 19, 18, 1484, 20, 965, 29077, 8719, 1273, 21, 45, 273, 17, 10, 15048, 28, 27511, 21, 4185, 11, 41, 2444, 9, 32, 1025, 20, 8719, 26, 23, 673, 966, 19, 29077, 20643, 27511, 20822, 20643, 19, 17, 6616, 17511, 18, 8978, 20, 18, 777, 9, 19233, 1527, 17669, 19, 24, 673, 17, 28756, 150, 12943, 4354, 153, 27, 442, 37, 45, 668, 21, 24, 256, 20, 416, 22, 2771, 4901, 9, 12943, 4354, 153, 51, 24, 3004, 21, 28142, 23, 65, 20, 18, 416, 34, 24, 2958, 22947, 9, 1177, 45, 668, 3097, 13768, 23, 103, 28, 441, 148, 48, 20522, 19, 12943, 4354, 153, 12860, 34, 18, 326, 27, 17492, 684, 21, 6709, 9, 8585, 123, 266, 19, 12943, 4354, 153, 6872, 24, 3004, 20, 18, 9225, 2198, 19, 12717, 103, 22, 401, 24, 6348, 9, 12943, 4354, 153, 1068, 2768, 2286, 19, 33, 104, 19, 176, 24, 9313, 19, 20086, 28, 45, 10292, 9, 4, 3, 19, 12943, 4354, 153, 27, 442, 22, 2771, 4901, 9, 69, 27, 442, 22, 2771, 24, 11335, 20, 18, 9225, 2198, 9, 69, 27, 442, 22, 2771, 24, 11335, 20, 18, 9225, 2198, 9, 69, 27, 442, 22, 2771,
]
# fmt: on
# In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria)
# are discovered. The voice of Nicholas's young son, Tsarevich Alexei Nikolaevich,
# narrates the remainder of the story. 1883 Western Siberia, a young Grigori Rasputin
# is asked by his father and a group of men to perform magic. Rasputin has a vision and
# denounces one of the men as a horse thief. Although his father initially slaps
# him for making such an accusation, Rasputin watches as the man is chased outside and beaten.
# Twenty years later, Rasputin sees a vision of the Virgin Mary, prompting him to become a priest.
# Rasputin quickly becomes famous, with people, even a bishop, begging for his blessing.
# <sep><cls>, Rasputin is asked to perform magic. He is asked to perform a ritual of the Virgin Mary.
# He is asked to perform a ritual of the Virgin Mary. He is asked to perform
output_ids = model.generate(input_ids, max_length=200, do_sample=False)
self.assertListEqual(output_ids[0].tolist(), expected_output_ids)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/bart/test_modeling_tf_bart.py | # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import copy
import tempfile
import unittest
import numpy as np
from transformers import BartConfig, BartTokenizer, is_tf_available
from transformers.testing_utils import require_tf, slow, tooslow
from transformers.utils import cached_property
from ...test_configuration_common import ConfigTester
from ...test_modeling_tf_common import TFModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
from ...utils.test_modeling_tf_core import TFCoreModelTesterMixin
if is_tf_available():
import tensorflow as tf
from transformers import TFBartForConditionalGeneration, TFBartForSequenceClassification, TFBartModel
@require_tf
class TFBartModelTester:
config_cls = BartConfig
config_updates = {}
hidden_act = "gelu"
def __init__(
self,
parent,
batch_size=13,
seq_length=7,
is_training=True,
use_labels=False,
vocab_size=99,
hidden_size=32,
num_hidden_layers=5,
num_attention_heads=4,
intermediate_size=37,
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=20,
eos_token_id=2,
pad_token_id=1,
bos_token_id=0,
):
self.parent = parent
self.batch_size = batch_size
self.seq_length = seq_length
self.is_training = is_training
self.use_labels = use_labels
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.eos_token_id = eos_token_id
self.pad_token_id = pad_token_id
self.bos_token_id = bos_token_id
def prepare_config_and_inputs_for_common(self):
# Ids are clipped to avoid "beginng of sequence", "end of sequence", and "pad" tokens
input_ids = tf.clip_by_value(
ids_tensor([self.batch_size, self.seq_length - 1], self.vocab_size),
clip_value_min=self.eos_token_id + 1,
clip_value_max=self.vocab_size + 1,
)
# Explicity add "end of sequence" to the inputs
eos_tensor = tf.expand_dims(tf.constant([self.eos_token_id] * self.batch_size), 1)
input_ids = tf.concat([input_ids, eos_tensor], axis=1)
decoder_input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
config = self.config_cls(
vocab_size=self.vocab_size,
d_model=self.hidden_size,
encoder_layers=self.num_hidden_layers,
decoder_layers=self.num_hidden_layers,
encoder_attention_heads=self.num_attention_heads,
decoder_attention_heads=self.num_attention_heads,
encoder_ffn_dim=self.intermediate_size,
decoder_ffn_dim=self.intermediate_size,
dropout=self.hidden_dropout_prob,
attention_dropout=self.attention_probs_dropout_prob,
max_position_embeddings=self.max_position_embeddings,
eos_token_ids=[2],
bos_token_id=self.bos_token_id,
pad_token_id=self.pad_token_id,
decoder_start_token_id=self.pad_token_id,
**self.config_updates,
)
inputs_dict = prepare_bart_inputs_dict(config, input_ids, decoder_input_ids)
return config, inputs_dict
def check_decoder_model_past_large_inputs(self, config, inputs_dict):
model = TFBartModel(config=config).get_decoder()
input_ids = inputs_dict["input_ids"]
input_ids = input_ids[:1, :]
attention_mask = inputs_dict["attention_mask"][:1, :]
head_mask = inputs_dict["head_mask"]
self.batch_size = 1
# first forward pass
outputs = model(input_ids, attention_mask=attention_mask, head_mask=head_mask, use_cache=True)
output, past_key_values = outputs.to_tuple()
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
next_attn_mask = tf.cast(ids_tensor((self.batch_size, 3), 2), tf.int8)
# append to next input_ids and
next_input_ids = tf.concat([input_ids, next_tokens], axis=-1)
next_attention_mask = tf.concat([attention_mask, next_attn_mask], axis=-1)
output_from_no_past = model(next_input_ids, attention_mask=next_attention_mask)
output_from_no_past = output_from_no_past[0]
output_from_past = model(next_tokens, attention_mask=next_attention_mask, past_key_values=past_key_values)
output_from_past = output_from_past[0]
self.parent.assertEqual(next_tokens.shape[1], output_from_past.shape[1])
# select random slice
random_slice_idx = int(ids_tensor((1,), output_from_past.shape[-1]))
output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx]
output_from_past_slice = output_from_past[:, :, random_slice_idx]
# test that outputs are equal for slice
tf.debugging.assert_near(output_from_past_slice, output_from_no_past_slice, rtol=1e-3)
def prepare_bart_inputs_dict(
config,
input_ids,
decoder_input_ids,
attention_mask=None,
decoder_attention_mask=None,
head_mask=None,
decoder_head_mask=None,
cross_attn_head_mask=None,
):
if attention_mask is None:
attention_mask = tf.cast(tf.math.not_equal(input_ids, config.pad_token_id), tf.int8)
if decoder_attention_mask is None:
decoder_attention_mask = tf.concat(
[
tf.ones(decoder_input_ids[:, :1].shape, dtype=tf.int8),
tf.cast(tf.math.not_equal(decoder_input_ids[:, 1:], config.pad_token_id), tf.int8),
],
axis=-1,
)
if head_mask is None:
head_mask = tf.ones((config.encoder_layers, config.encoder_attention_heads))
if decoder_head_mask is None:
decoder_head_mask = tf.ones((config.decoder_layers, config.decoder_attention_heads))
if cross_attn_head_mask is None:
cross_attn_head_mask = tf.ones((config.decoder_layers, config.decoder_attention_heads))
return {
"input_ids": input_ids,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"decoder_attention_mask": decoder_attention_mask,
"head_mask": head_mask,
"decoder_head_mask": decoder_head_mask,
"cross_attn_head_mask": cross_attn_head_mask,
}
@require_tf
class TFBartModelTest(TFModelTesterMixin, TFCoreModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(TFBartForConditionalGeneration, TFBartForSequenceClassification, TFBartModel) if is_tf_available() else ()
)
all_generative_model_classes = (TFBartForConditionalGeneration,) if is_tf_available() else ()
pipeline_model_mapping = (
{
"conversational": TFBartForConditionalGeneration,
"feature-extraction": TFBartModel,
"summarization": TFBartForConditionalGeneration,
"text2text-generation": TFBartForConditionalGeneration,
"text-classification": TFBartForSequenceClassification,
"zero-shot": TFBartForSequenceClassification,
}
if is_tf_available()
else {}
)
is_encoder_decoder = True
test_pruning = False
test_onnx = True
onnx_min_opset = 10
def setUp(self):
self.model_tester = TFBartModelTester(self)
self.config_tester = ConfigTester(self, config_class=BartConfig)
def test_config(self):
self.config_tester.run_common_tests()
def test_decoder_model_past_large_inputs(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs_for_common()
self.model_tester.check_decoder_model_past_large_inputs(*config_and_inputs)
def test_model_common_attributes(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
assert isinstance(model.get_input_embeddings(), tf.keras.layers.Layer)
if model_class in self.all_generative_model_classes:
x = model.get_output_embeddings()
assert isinstance(x, tf.keras.layers.Layer)
name = model.get_bias()
assert isinstance(name, dict)
for k, v in name.items():
assert isinstance(v, tf.Variable)
else:
x = model.get_output_embeddings()
assert x is None
name = model.get_bias()
assert name is None
@tooslow
def test_saved_model_creation(self):
pass
# TODO (Joao): fix me
@unittest.skip("Onnx compliancy broke with TF 2.10")
def test_onnx_compliancy(self):
pass
# TFBartForSequenceClassification does not support inputs_embeds
def test_inputs_embeds(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in (TFBartForConditionalGeneration, TFBartModel):
model = model_class(config)
inputs = copy.deepcopy(inputs_dict)
if not self.is_encoder_decoder:
input_ids = inputs["input_ids"]
del inputs["input_ids"]
else:
encoder_input_ids = inputs["input_ids"]
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
del inputs["input_ids"]
inputs.pop("decoder_input_ids", None)
if not self.is_encoder_decoder:
inputs["inputs_embeds"] = model.get_input_embeddings()(input_ids)
else:
inputs["inputs_embeds"] = model.get_input_embeddings()(encoder_input_ids)
inputs["decoder_inputs_embeds"] = model.get_input_embeddings()(decoder_input_ids)
inputs = self._prepare_for_class(inputs, model_class)
model(inputs)
# TFBartForSequenceClassification does not support inputs_embeds
@slow
def test_graph_mode_with_inputs_embeds(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in (TFBartForConditionalGeneration, TFBartModel):
model = model_class(config)
inputs = copy.deepcopy(inputs_dict)
if not self.is_encoder_decoder:
input_ids = inputs["input_ids"]
del inputs["input_ids"]
else:
encoder_input_ids = inputs["input_ids"]
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
del inputs["input_ids"]
inputs.pop("decoder_input_ids", None)
if not self.is_encoder_decoder:
inputs["inputs_embeds"] = model.get_input_embeddings()(input_ids)
else:
inputs["inputs_embeds"] = model.get_input_embeddings()(encoder_input_ids)
inputs["decoder_inputs_embeds"] = model.get_input_embeddings()(decoder_input_ids)
inputs = self._prepare_for_class(inputs, model_class)
@tf.function
def run_in_graph_mode():
return model(inputs)
outputs = run_in_graph_mode()
self.assertIsNotNone(outputs)
@slow
def test_save_load_after_resize_token_embeddings(self):
# Custom version of this test to ensure "end of sequence" tokens are present throughout
if not self.test_resize_embeddings:
return
config, original_inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
# create a model with resized (expended) embeddings
new_tokens_size = 10
old_total_size = config.vocab_size
new_total_size = old_total_size + new_tokens_size
model = model_class(config=copy.deepcopy(config)) # `resize_token_embeddings` mutates `config`
model(model.dummy_inputs) # builds the embeddings layer
model.resize_token_embeddings(new_total_size)
# fetch the output for an input exclusively made of new members of the vocabulary
inputs_dict = copy.deepcopy(original_inputs_dict)
ids_feat_name = None
if "input_ids" in inputs_dict:
ids_feat_name = "input_ids"
elif "decoder_input_ids" in inputs_dict:
ids_feat_name = "decoder_input_ids"
else:
assert False, "No input ids feature found in the inputs dict"
new_vocab_input_ids = ids_tensor(inputs_dict[ids_feat_name].shape, new_tokens_size)
new_vocab_input_ids += old_total_size
# Replace last id with EOS token
new_vocab_input_ids = new_vocab_input_ids[:, :-1]
new_vocab_input_ids = tf.concat(
[new_vocab_input_ids, tf.ones((tf.shape(new_vocab_input_ids)[0], 1), dtype=tf.int32) * 2], axis=1
)
inputs_dict[ids_feat_name] = new_vocab_input_ids
if "input_ids" in inputs_dict:
inputs_dict["input_ids"] = new_vocab_input_ids
if "decoder_input_ids" in inputs_dict:
inputs_dict["decoder_input_ids"] = new_vocab_input_ids
prepared_inputs = self._prepare_for_class(inputs_dict, model_class)
outputs = model(**prepared_inputs)
# save and load the model
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(tmpdirname, saved_model=False)
model = model_class.from_pretrained(tmpdirname)
restored_model_outputs = model(**prepared_inputs)
# check that the output for the restored model is the same
self.assert_outputs_same(restored_model_outputs, outputs)
def _long_tensor(tok_lst):
return tf.constant(tok_lst, dtype=tf.int32)
@require_tf
class TFBartHeadTests(unittest.TestCase):
vocab_size = 99
def _get_config_and_data(self):
eos_column_vector = tf.ones((4, 1), dtype=tf.int32) * 2
input_ids = tf.concat([ids_tensor((4, 6), self.vocab_size - 3) + 3, eos_column_vector], axis=1)
batch_size = input_ids.shape[0]
config = BartConfig(
vocab_size=self.vocab_size,
d_model=24,
encoder_layers=2,
decoder_layers=2,
encoder_attention_heads=2,
decoder_attention_heads=2,
encoder_ffn_dim=32,
decoder_ffn_dim=32,
max_position_embeddings=48,
eos_token_id=2,
pad_token_id=1,
bos_token_id=0,
decoder_start_token_id=2,
)
return config, input_ids, batch_size
def test_lm_forward(self):
config, input_ids, batch_size = self._get_config_and_data()
decoder_lm_labels = ids_tensor([batch_size, input_ids.shape[1]], self.vocab_size)
lm_model = TFBartForConditionalGeneration(config)
outputs = lm_model(input_ids=input_ids, labels=decoder_lm_labels, decoder_input_ids=input_ids, use_cache=False)
expected_shape = (batch_size, input_ids.shape[1], config.vocab_size)
self.assertEqual(outputs.logits.shape, expected_shape)
def test_lm_uneven_forward(self):
config = BartConfig(
vocab_size=10,
d_model=24,
encoder_layers=2,
decoder_layers=2,
encoder_attention_heads=2,
decoder_attention_heads=2,
encoder_ffn_dim=32,
decoder_ffn_dim=32,
max_position_embeddings=48,
)
lm_model = TFBartForConditionalGeneration(config)
context = tf.fill((7, 2), 4)
summary = tf.fill((7, 7), 6)
outputs = lm_model(input_ids=context, decoder_input_ids=summary, use_cache=False)
expected_shape = (*summary.shape, config.vocab_size)
self.assertEqual(outputs.logits.shape, expected_shape)
@require_tf
class TFBartForSequenceClassificationTest(unittest.TestCase):
def test_model_fails_for_uneven_eos_tokens(self):
config = BartConfig(eos_token_id=2)
model = TFBartForSequenceClassification(config)
inputs = {
"input_ids": tf.constant([[1, 2, 2, 2], [1, 3, 2, 2], [2, 2, 3, 3]]),
"attention_mask": tf.constant([[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]),
}
with self.assertRaises(tf.errors.InvalidArgumentError):
model(inputs)
@slow
@require_tf
class TFBartModelIntegrationTest(unittest.TestCase):
def test_inference_no_head(self):
model = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large").model
input_ids = _long_tensor([[0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2]])
attention_mask = tf.cast(tf.math.not_equal(input_ids, model.config.pad_token_id), tf.int8)
output = model(input_ids=input_ids, attention_mask=attention_mask)[0]
expected_shape = (1, 11, 1024)
self.assertEqual(output.shape, expected_shape)
expected_slice = tf.convert_to_tensor(
[[0.7144, 0.8143, -1.2813], [0.7144, 0.8143, -1.2813], [-0.0467, 2.5911, -2.1845]],
)
tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=1e-3)
def test_cnn_summarization_same_as_fairseq_hard(self):
hf = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
tok = self.tok
FRANCE_ARTICLE = ( # @noqa
" Marseille, France (CNN)The French prosecutor leading an investigation into the crash of Germanwings"
" Flight 9525 insisted Wednesday that he was not aware of any video footage from on board the plane."
' Marseille prosecutor Brice Robin told CNN that "so far no videos were used in the crash investigation."'
' He added, "A person who has such a video needs to immediately give it to the investigators." Robin\'s'
" comments follow claims by two magazines, German daily Bild and French Paris Match, of a cell phone video"
" showing the harrowing final seconds from on board Germanwings Flight 9525 as it crashed into the French"
" Alps. All 150 on board were killed. Paris Match and Bild reported that the video was recovered from a"
" phone at the wreckage site. The two publications described the supposed video, but did not post it on"
" their websites. The publications said that they watched the video, which was found by a source close to"
" the investigation. \"One can hear cries of 'My God' in several languages,\" Paris Match reported."
' "Metallic banging can also be heard more than three times, perhaps of the pilot trying to open the'
" cockpit door with a heavy object. Towards the end, after a heavy shake, stronger than the others, the"
' screaming intensifies. Then nothing." "It is a very disturbing scene," said Julian Reichelt,'
" editor-in-chief of Bild online. An official with France's accident investigation agency, the BEA, said"
" the agency is not aware of any such video. Lt. Col. Jean-Marc Menichini, a French Gendarmerie spokesman"
" in charge of communications on rescue efforts around the Germanwings crash site, told CNN that the"
' reports were "completely wrong" and "unwarranted." Cell phones have been collected at the site, he said,'
' but that they "hadn\'t been exploited yet." Menichini said he believed the cell phones would need to be'
" sent to the Criminal Research Institute in Rosny sous-Bois, near Paris, in order to be analyzed by"
" specialized technicians working hand-in-hand with investigators. But none of the cell phones found so"
" far have been sent to the institute, Menichini said. Asked whether staff involved in the search could"
' have leaked a memory card to the media, Menichini answered with a categorical "no." Reichelt told "Erin'
' Burnett: Outfront" that he had watched the video and stood by the report, saying Bild and Paris Match'
' are "very confident" that the clip is real. He noted that investigators only revealed they\'d recovered'
' cell phones from the crash site after Bild and Paris Match published their reports. "That is something'
" we did not know before. ... Overall we can say many things of the investigation weren't revealed by the"
' investigation at the beginning," he said. What was mental state of Germanwings co-pilot? German airline'
" Lufthansa confirmed Tuesday that co-pilot Andreas Lubitz had battled depression years before he took the"
" controls of Germanwings Flight 9525, which he's accused of deliberately crashing last week in the"
' French Alps. Lubitz told his Lufthansa flight training school in 2009 that he had a "previous episode of'
' severe depression," the airline said Tuesday. Email correspondence between Lubitz and the school'
" discovered in an internal investigation, Lufthansa said, included medical documents he submitted in"
" connection with resuming his flight training. The announcement indicates that Lufthansa, the parent"
" company of Germanwings, knew of Lubitz's battle with depression, allowed him to continue training and"
" ultimately put him in the cockpit. Lufthansa, whose CEO Carsten Spohr previously said Lubitz was 100%"
' fit to fly, described its statement Tuesday as a "swift and seamless clarification" and said it was'
" sharing the information and documents -- including training and medical records -- with public"
" prosecutors. Spohr traveled to the crash site Wednesday, where recovery teams have been working for the"
" past week to recover human remains and plane debris scattered across a steep mountainside. He saw the"
" crisis center set up in Seyne-les-Alpes, laid a wreath in the village of Le Vernet, closer to the crash"
" site, where grieving families have left flowers at a simple stone memorial. Menichini told CNN late"
" Tuesday that no visible human remains were left at the site but recovery teams would keep searching."
" French President Francois Hollande, speaking Tuesday, said that it should be possible to identify all"
" the victims using DNA analysis by the end of the week, sooner than authorities had previously suggested."
" In the meantime, the recovery of the victims' personal belongings will start Wednesday, Menichini said."
" Among those personal belongings could be more cell phones belonging to the 144 passengers and six crew"
" on board. Check out the latest from our correspondents . The details about Lubitz's correspondence with"
" the flight school during his training were among several developments as investigators continued to"
" delve into what caused the crash and Lubitz's possible motive for downing the jet. A Lufthansa"
" spokesperson told CNN on Tuesday that Lubitz had a valid medical certificate, had passed all his"
' examinations and "held all the licenses required." Earlier, a spokesman for the prosecutor\'s office in'
" Dusseldorf, Christoph Kumpa, said medical records reveal Lubitz suffered from suicidal tendencies at"
" some point before his aviation career and underwent psychotherapy before he got his pilot's license."
" Kumpa emphasized there's no evidence suggesting Lubitz was suicidal or acting aggressively before the"
" crash. Investigators are looking into whether Lubitz feared his medical condition would cause him to"
" lose his pilot's license, a European government official briefed on the investigation told CNN on"
' Tuesday. While flying was "a big part of his life," the source said, it\'s only one theory being'
" considered. Another source, a law enforcement official briefed on the investigation, also told CNN that"
" authorities believe the primary motive for Lubitz to bring down the plane was that he feared he would"
" not be allowed to fly because of his medical problems. Lubitz's girlfriend told investigators he had"
" seen an eye doctor and a neuropsychologist, both of whom deemed him unfit to work recently and concluded"
" he had psychological issues, the European government official said. But no matter what details emerge"
" about his previous mental health struggles, there's more to the story, said Brian Russell, a forensic"
' psychologist. "Psychology can explain why somebody would turn rage inward on themselves about the fact'
" that maybe they weren't going to keep doing their job and they're upset about that and so they're"
' suicidal," he said. "But there is no mental illness that explains why somebody then feels entitled to'
" also take that rage and turn it outward on 149 other people who had nothing to do with the person's"
' problems." Germanwings crash compensation: What we know . Who was the captain of Germanwings Flight'
" 9525? CNN's Margot Haddad reported from Marseille and Pamela Brown from Dusseldorf, while Laura"
" Smith-Spark wrote from London. CNN's Frederik Pleitgen, Pamela Boykoff, Antonia Mortensen, Sandrine"
" Amiel and Anna-Maja Rappard contributed to this report."
)
EXPECTED_SUMMARY_FRANCE = (
"French prosecutor says he's not aware of any video footage from on board the plane. German daily Bild"
" and French Paris Match claim to have found a cell phone video of the crash. A French Gendarmerie"
' spokesman calls the reports "completely wrong" and "unwarranted" German airline Lufthansa confirms'
" co-pilot Andreas Lubitz had battled depression."
)
SHORTER_ARTICLE = (
" (CNN)The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes. CNN's Vasco Cotovio, Kareem Khadder"
" and Faith Karimi contributed to this report."
)
EXPECTED_SUMMARY_SHORTER = (
"The Palestinian Authority becomes the 123rd member of the International Criminal Court. The move gives"
" the court jurisdiction over alleged crimes in Palestinian territories. Israel and the United States"
" opposed the Palestinians' efforts to join the body. But Palestinian Foreign Minister Riad al-Malki said"
" it was a move toward greater justice."
)
# The below article tests that we don't add any hypotheses outside of the top n_beams
IRAN_ARTICLE = (
" (CNN)The United States and its negotiating partners reached a very strong framework agreement with Iran"
" in Lausanne, Switzerland, on Thursday that limits Iran's nuclear program in such a way as to effectively"
" block it from building a nuclear weapon. Expect pushback anyway, if the recent past is any harbinger."
" Just last month, in an attempt to head off such an agreement, House Speaker John Boehner invited Israeli"
" Prime Minister Benjamin Netanyahu to preemptively blast it before Congress, and 47 senators sent a"
" letter to the Iranian leadership warning them away from a deal. The debate that has already begun since"
" the announcement of the new framework will likely result in more heat than light. It will not be helped"
" by the gathering swirl of dubious assumptions and doubtful assertions. Let us address some of these: ."
" The most misleading assertion, despite universal rejection by experts, is that the negotiations'"
" objective at the outset was the total elimination of any nuclear program in Iran. That is the position"
" of Netanyahu and his acolytes in the U.S. Congress. But that is not and never was the objective. If it"
" had been, there would have been no Iranian team at the negotiating table. Rather, the objective has"
" always been to structure an agreement or series of agreements so that Iran could not covertly develop a"
" nuclear arsenal before the United States and its allies could respond. The new framework has exceeded"
" expectations in achieving that goal. It would reduce Iran's low-enriched uranium stockpile, cut by"
" two-thirds its number of installed centrifuges and implement a rigorous inspection regime. Another"
" dubious assumption of opponents is that the Iranian nuclear program is a covert weapons program. Despite"
" sharp accusations by some in the United States and its allies, Iran denies having such a program, and"
" U.S. intelligence contends that Iran has not yet made the decision to build a nuclear weapon. Iran's"
" continued cooperation with International Atomic Energy Agency inspections is further evidence on this"
" point, and we'll know even more about Iran's program in the coming months and years because of the deal."
" In fact, the inspections provisions that are part of this agreement are designed to protect against any"
" covert action by the Iranians. What's more, the rhetoric of some members of Congress has implied that"
" the negotiations have been between only the United States and Iran (i.e., the 47 senators' letter"
" warning that a deal might be killed by Congress or a future president). This of course is not the case."
" The talks were between Iran and the five permanent members of the U.N. Security Council (United States,"
" United Kingdom, France, China and Russia) plus Germany, dubbed the P5+1. While the United States has"
" played a leading role in the effort, it negotiated the terms alongside its partners. If the agreement"
" reached by the P5+1 is rejected by Congress, it could result in an unraveling of the sanctions on Iran"
" and threaten NATO cohesion in other areas. Another questionable assertion is that this agreement"
" contains a sunset clause, after which Iran will be free to do as it pleases. Again, this is not the"
" case. Some of the restrictions on Iran's nuclear activities, such as uranium enrichment, will be eased"
" or eliminated over time, as long as 15 years. But most importantly, the framework agreement includes"
" Iran's ratification of the Additional Protocol, which allows IAEA inspectors expanded access to nuclear"
" sites both declared and nondeclared. This provision will be permanent. It does not sunset. Thus, going"
" forward, if Iran decides to enrich uranium to weapons-grade levels, monitors will be able to detect such"
" a move in a matter of days and alert the U.N. Security Council. Many in Congress have said that the"
' agreement should be a formal treaty requiring the Senate to "advise and consent." But the issue is not'
" suited for a treaty. Treaties impose equivalent obligations on all signatories. For example, the New"
" START treaty limits Russia and the United States to 1,550 deployed strategic warheads. But any agreement"
" with Iran will not be so balanced. The restrictions and obligations in the final framework agreement"
" will be imposed almost exclusively on Iran. The P5+1 are obligated only to ease and eventually remove"
" most but not all economic sanctions, which were imposed as leverage to gain this final deal. Finally"
" some insist that any agreement must address Iranian missile programs, human rights violations or support"
" for Hamas or Hezbollah. As important as these issues are, and they must indeed be addressed, they are"
" unrelated to the most important aim of a nuclear deal: preventing a nuclear Iran. To include them in"
" the negotiations would be a poison pill. This agreement should be judged on its merits and on how it"
" affects the security of our negotiating partners and allies, including Israel. Those judgments should be"
" fact-based, not based on questionable assertions or dubious assumptions."
)
EXPECTED_SUMMARY_IRAN = (
"The U.S. and its negotiating partners reached a very strong framework agreement with Iran. Peter Bergen:"
" The debate that has already begun will likely result in more heat than light. He says the agreement"
" limits Iran's nuclear program in such a way as to effectively block it from building a nuclear weapon."
" Bergen says the most important aim of a nuclear deal is preventing a nuclear Iran."
)
ARTICLE_SUBWAY = (
" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York. A"
" year later, she got married again in Westchester County, but to a different man and without divorcing"
" her first husband. Only 18 days after that marriage, she got hitched yet again. Then, Barrientos"
' declared "I do" five more times, sometimes only within two weeks of each other. In 2010, she married'
" once more, this time in the Bronx. In an application for a marriage license, she stated it was her"
' "first and only" marriage. Barrientos, now 39, is facing two criminal counts of "offering a false'
' instrument for filing in the first degree," referring to her false statements on the 2010 marriage'
" license application, according to court documents. Prosecutors said the marriages were part of an"
" immigration scam. On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to"
" her attorney, Christopher Wright, who declined to comment further. After leaving court, Barrientos was"
" arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New"
" York subway through an emergency exit, said Detective Annette Markowski, a police spokeswoman. In total,"
" Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002. All"
" occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be"
" married to four men, and at one time, she was married to eight men at once, prosecutors say. Prosecutors"
" said the immigration scam involved some of her husbands, who filed for permanent residence status"
" shortly after the marriages. Any divorces happened only after such filings were approved. It was"
" unclear whether any of the men will be prosecuted. The case was referred to the Bronx District"
" Attorney's Office by Immigration and Customs Enforcement and the Department of Homeland Security's"
' Investigation Division. Seven of the men are from so-called "red-flagged" countries, including Egypt,'
" Turkey, Georgia, Pakistan and Mali. Her eighth husband, Rashid Rajput, was deported in 2006 to his"
" native Pakistan after an investigation by the Joint Terrorism Task Force. If convicted, Barrientos faces"
" up to four years in prison. Her next court appearance is scheduled for May 18."
)
EXPECTED_SUMMARY_SUBWAY = (
"Liana Barrientos has been married 10 times, sometimes within two weeks of each other. Prosecutors say the"
" marriages were part of an immigration scam. On Friday, she pleaded not guilty at State Supreme Court in"
" the Bronx. She was arrested and charged with theft of service and criminal trespass for allegedly"
" sneaking into the subway."
)
dct = tok(
[FRANCE_ARTICLE, SHORTER_ARTICLE, IRAN_ARTICLE, ARTICLE_SUBWAY],
max_length=1024,
truncation_strategy="only_first",
padding="longest",
truncation=True,
return_tensors="tf",
)
self.assertEqual(1024, dct["input_ids"].shape[1])
hypotheses_batch = hf.generate(
input_ids=dct["input_ids"],
attention_mask=dct["attention_mask"],
)
assert hypotheses_batch[:, 1].numpy().tolist() == [0, 0, 0, 0] # test force_bos_token_to_be_generated
decoded = tok.batch_decode(hypotheses_batch, skip_special_tokens=True, clean_up_tokenization_spaces=False)
expected_batch = [
EXPECTED_SUMMARY_FRANCE,
EXPECTED_SUMMARY_SHORTER,
EXPECTED_SUMMARY_IRAN,
EXPECTED_SUMMARY_SUBWAY,
]
assert decoded == expected_batch
@cached_property
def tok(self):
return BartTokenizer.from_pretrained("facebook/bart-large")
@slow
def test_contrastive_search_bart(self):
article = (
" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York. A"
" year later, she got married again in Westchester County, but to a different man and without divorcing"
" her first husband. Only 18 days after that marriage, she got hitched yet again. Then, Barrientos"
' declared "I do" five more times, sometimes only within two weeks of each other. In 2010, she married'
" once more, this time in the Bronx. In an application for a marriage license, she stated it was her"
' "first and only" marriage. Barrientos, now 39, is facing two criminal counts of "offering a false'
' instrument for filing in the first degree," referring to her false statements on the 2010 marriage'
" license application, according to court documents. Prosecutors said the marriages were part of an"
" immigration scam. On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to"
" her attorney, Christopher Wright, who declined to comment further. After leaving court, Barrientos was"
" arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New"
" York subway through an emergency exit, said Detective Annette Markowski, a police spokeswoman. In total,"
" Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002. All"
" occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be"
" married to four men, and at one time, she was married to eight men at once, prosecutors say. Prosecutors"
" said the immigration scam involved some of her husbands, who filed for permanent residence status"
" shortly after the marriages. Any divorces happened only after such filings were approved. It was"
" unclear whether any of the men will be prosecuted. The case was referred to the Bronx District"
" Attorney's Office by Immigration and Customs Enforcement and the Department of Homeland Security's"
' Investigation Division. Seven of the men are from so-called "red-flagged" countries, including Egypt,'
" Turkey, Georgia, Pakistan and Mali. Her eighth husband, Rashid Rajput, was deported in 2006 to his"
" native Pakistan after an investigation by the Joint Terrorism Task Force. If convicted, Barrientos faces"
" up to four years in prison. Her next court appearance is scheduled for May 18."
)
bart_tokenizer = BartTokenizer.from_pretrained("facebook/bart-large-cnn")
bart_model = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
input_ids = bart_tokenizer(
article, add_special_tokens=False, truncation=True, max_length=512, return_tensors="tf"
).input_ids
outputs = bart_model.generate(input_ids, penalty_alpha=0.5, top_k=5, max_length=64)
generated_text = bart_tokenizer.batch_decode(outputs, skip_special_tokens=True)
self.assertListEqual(
generated_text,
[
"Liana Barrientos, 39, pleaded not guilty to charges related to false marriage statements. "
"Prosecutors say she married at least 10 times, sometimes within two weeks of each other. She is "
"accused of being part of an immigration scam to get permanent residency. If convicted, she faces up "
"to four years in"
],
)
@slow
def test_contrastive_search_bart_xla(self):
article = (
" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York. A"
" year later, she got married again in Westchester County, but to a different man and without divorcing"
" her first husband. Only 18 days after that marriage, she got hitched yet again. Then, Barrientos"
' declared "I do" five more times, sometimes only within two weeks of each other. In 2010, she married'
" once more, this time in the Bronx. In an application for a marriage license, she stated it was her"
' "first and only" marriage. Barrientos, now 39, is facing two criminal counts of "offering a false'
' instrument for filing in the first degree," referring to her false statements on the 2010 marriage'
" license application, according to court documents. Prosecutors said the marriages were part of an"
" immigration scam. On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to"
" her attorney, Christopher Wright, who declined to comment further. After leaving court, Barrientos was"
" arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New"
" York subway through an emergency exit, said Detective Annette Markowski, a police spokeswoman. In total,"
" Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002. All"
" occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be"
" married to four men, and at one time, she was married to eight men at once, prosecutors say. Prosecutors"
" said the immigration scam involved some of her husbands, who filed for permanent residence status"
" shortly after the marriages. Any divorces happened only after such filings were approved. It was"
" unclear whether any of the men will be prosecuted. The case was referred to the Bronx District"
" Attorney's Office by Immigration and Customs Enforcement and the Department of Homeland Security's"
' Investigation Division. Seven of the men are from so-called "red-flagged" countries, including Egypt,'
" Turkey, Georgia, Pakistan and Mali. Her eighth husband, Rashid Rajput, was deported in 2006 to his"
" native Pakistan after an investigation by the Joint Terrorism Task Force. If convicted, Barrientos faces"
" up to four years in prison. Her next court appearance is scheduled for May 18."
)
bart_tokenizer = BartTokenizer.from_pretrained("facebook/bart-large-cnn")
bart_model = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
input_ids = bart_tokenizer(
article, add_special_tokens=False, truncation=True, max_length=512, return_tensors="tf"
).input_ids
xla_generate = tf.function(bart_model.generate, jit_compile=True)
# no_repeat_ngram_size set to 0 because it isn't compatible with XLA, but doesn't change the original output
outputs = xla_generate(input_ids, penalty_alpha=0.5, top_k=5, max_length=64, no_repeat_ngram_size=0)
generated_text = bart_tokenizer.batch_decode(outputs, skip_special_tokens=True)
self.assertListEqual(
generated_text,
[
"Liana Barrientos, 39, pleaded not guilty to charges related to false marriage statements. "
"Prosecutors say she married at least 10 times, sometimes within two weeks of each other. She is "
"accused of being part of an immigration scam to get permanent residency. If convicted, she faces up "
"to four years in"
],
)
@slow
@require_tf
class FasterTFBartModelIntegrationTests(unittest.TestCase):
"""These tests are useful for debugging since they operate on a model with 1 encoder layer and 1 decoder layer."""
@cached_property
def tok(self):
return BartTokenizer.from_pretrained("facebook/bart-large")
@cached_property
def xsum_1_1_model(self):
return TFBartForConditionalGeneration.from_pretrained("sshleifer/distilbart-xsum-1-1")
def test_xsum_1_1_generation(self):
model = self.xsum_1_1_model
assert model.model.decoder.embed_tokens == model.model.shared
ARTICLE = (
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes."
)
EXPECTED = (
" The International Criminal Court (ICC) has announced that it has been announced by the International"
" Criminal court."
)
dct = self.tok(ARTICLE, return_tensors="tf")
generated_ids = model.generate(**dct, num_beams=4)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)[0]
assert result == EXPECTED
def test_xsum_1_1_xla_generation(self):
# same test as above, but with `no_repeat_ngram_size=0` (not compatible with XLA) and XLA comparison enabled
model = self.xsum_1_1_model
assert model.model.decoder.embed_tokens == model.model.shared
ARTICLE = (
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes."
)
EXPECTED = (
" The International Criminal Court (ICC) has announced that it is to be investigated by the International"
" Criminal Court (ICC) over allegations of war crimes."
)
dct = self.tok(ARTICLE, return_tensors="tf")
generated_ids = model.generate(**dct, num_beams=4, no_repeat_ngram_size=0)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)[0]
assert result == EXPECTED
xla_generate = tf.function(model.generate, jit_compile=True)
generated_ids = xla_generate(**dct, num_beams=4, no_repeat_ngram_size=0)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)[0]
assert result == EXPECTED
def test_xsum_1_1_batch_generation(self):
batch = self.tok(
[
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories."
" The formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is"
" based. The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted"
' its jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including'
' East Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination'
" into the situation in Palestinian territories, paving the way for possible war crimes investigations"
" against Israelis. As members of the court, Palestinians may be subject to counter-charges as well."
" Israel and the United States, neither of which is an ICC member, opposed the Palestinians' efforts"
" to join the body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony,"
' said it was a move toward greater justice. "As Palestine formally becomes a State Party to the Rome'
' Statute today, the world is also a step closer to ending a long era of impunity and injustice," he'
' said, according to an ICC news release. "Indeed, today brings us closer to our shared goals of'
' justice and peace." Judge Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was'
' just the first step for the Palestinians. "As the Rome Statute today enters into force for the State'
" of Palestine, Palestine acquires all the rights as well as responsibilities that come with being a"
' State Party to the Statute. These are substantive commitments, which cannot be taken lightly," she'
' said. Rights group Human Rights Watch welcomed the development. "Governments seeking to penalize'
" Palestine for joining the ICC should immediately end their pressure, and countries that support"
" universal acceptance of the court's treaty should speak out to welcome its membership,\" said"
" Balkees Jarrah, international justice counsel for the group. \"What's objectionable is the attempts"
" to undermine international justice, not Palestine's decision to join a treaty to which over 100"
' countries around the world are members." In January, when the preliminary ICC examination was'
" opened, Israeli Prime Minister Benjamin Netanyahu described it as an outrage, saying the court was"
' overstepping its boundaries. The United States also said it "strongly" disagreed with the court\'s'
' decision. "As we have said repeatedly, we do not believe that Palestine is a state and therefore we'
' do not believe that it is eligible to join the ICC," the State Department said in a statement. It'
' urged the warring sides to resolve their differences through direct negotiations. "We will continue'
' to oppose actions against Israel at the ICC as counterproductive to the cause of peace," it said.'
" But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows'
" the court to review evidence and determine whether to investigate suspects on both sides. Prosecutor"
' Fatou Bensouda said her office would "conduct its analysis in full independence and impartiality."'
" The war between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The"
" inquiry will include alleged war crimes committed since June. The International Criminal Court was"
" set up in 2002 to prosecute genocide, crimes against humanity and war crimes.",
"The French prosecutor leading an investigation into the crash of Germanwings Flight 9525 insisted"
" Wednesday that he was not aware of any video footage from on board the plane. Marseille prosecutor"
' Brice Robin told CNN that "so far no videos were used in the crash investigation." He added, "A'
" person who has such a video needs to immediately give it to the investigators.\" Robin's comments"
" follow claims by two magazines, German daily Bild and French Paris Match, of a cell phone video"
" showing the harrowing final seconds from on board Germanwings Flight 9525 as it crashed into the"
" French Alps. All 150 on board were killed. Paris Match and Bild reported that the video was"
" recovered from a phone at the wreckage site. The two publications described the supposed video, but"
" did not post it on their websites. The publications said that they watched the video, which was"
" found by a source close to the investigation. \"One can hear cries of 'My God' in several"
' languages," Paris Match reported. "Metallic banging can also be heard more than three times, perhaps'
" of the pilot trying to open the cockpit door with a heavy object. Towards the end, after a heavy"
' shake, stronger than the others, the screaming intensifies. Then nothing." "It is a very disturbing'
" scene,\" said Julian Reichelt, editor-in-chief of Bild online. An official with France's accident"
" investigation agency, the BEA, said the agency is not aware of any such video. Lt. Col. Jean-Marc"
" Menichini, a French Gendarmerie spokesman in charge of communications on rescue efforts around the"
' Germanwings crash site, told CNN that the reports were "completely wrong" and "unwarranted." Cell'
' phones have been collected at the site, he said, but that they "hadn\'t been exploited yet."'
" Menichini said he believed the cell phones would need to be sent to the Criminal Research Institute"
" in Rosny sous-Bois, near Paris, in order to be analyzed by specialized technicians working"
" hand-in-hand with investigators. But none of the cell phones found so far have been sent to the"
" institute, Menichini said. Asked whether staff involved in the search could have leaked a memory"
' card to the media, Menichini answered with a categorical "no." Reichelt told "Erin Burnett:'
' Outfront" that he had watched the video and stood by the report, saying Bild and Paris Match are'
' "very confident" that the clip is real. He noted that investigators only revealed they\'d recovered'
' cell phones from the crash site after Bild and Paris Match published their reports. "That is'
" something we did not know before. ... Overall we can say many things of the investigation weren't"
' revealed by the investigation at the beginning," he said. What was mental state of Germanwings'
" co-pilot? German airline Lufthansa confirmed Tuesday that co-pilot Andreas Lubitz had battled"
" depression years before he took the controls of Germanwings Flight 9525, which he's accused of"
" deliberately crashing last week in the French Alps. Lubitz told his Lufthansa flight training school"
' in 2009 that he had a "previous episode of severe depression," the airline said Tuesday. Email'
" correspondence between Lubitz and the school discovered in an internal investigation, Lufthansa"
" said, included medical documents he submitted in connection with resuming his flight training. The"
" announcement indicates that Lufthansa, the parent company of Germanwings, knew of Lubitz's battle"
" with depression, allowed him to continue training and ultimately put him in the cockpit. Lufthansa,"
" whose CEO Carsten Spohr previously said Lubitz was 100% fit to fly, described its statement Tuesday"
' as a "swift and seamless clarification" and said it was sharing the information and documents --'
" including training and medical records -- with public prosecutors. Spohr traveled to the crash site"
" Wednesday, where recovery teams have been working for the past week to recover human remains and"
" plane debris scattered across a steep mountainside. He saw the crisis center set up in"
" Seyne-les-Alpes, laid a wreath in the village of Le Vernet, closer to the crash site, where grieving"
" families have left flowers at a simple stone memorial. Menichini told CNN late Tuesday that no"
" visible human remains were left at the site but recovery teams would keep searching. French"
" President Francois Hollande, speaking Tuesday, said that it should be possible to identify all the"
" victims using DNA analysis by the end of the week, sooner than authorities had previously suggested."
" In the meantime, the recovery of the victims' personal belongings will start Wednesday, Menichini"
" said. Among those personal belongings could be more cell phones belonging to the 144 passengers and"
" six crew on board. Check out the latest from our correspondents . The details about Lubitz's"
" correspondence with the flight school during his training were among several developments as"
" investigators continued to delve into what caused the crash and Lubitz's possible motive for"
" downing the jet. A Lufthansa spokesperson told CNN on Tuesday that Lubitz had a valid medical"
' certificate, had passed all his examinations and "held all the licenses required." Earlier, a'
" spokesman for the prosecutor's office in Dusseldorf, Christoph Kumpa, said medical records reveal"
" Lubitz suffered from suicidal tendencies at some point before his aviation career and underwent"
" psychotherapy before he got his pilot's license. Kumpa emphasized there's no evidence suggesting"
" Lubitz was suicidal or acting aggressively before the crash. Investigators are looking into whether"
" Lubitz feared his medical condition would cause him to lose his pilot's license, a European"
' government official briefed on the investigation told CNN on Tuesday. While flying was "a big part'
" of his life,\" the source said, it's only one theory being considered. Another source, a law"
" enforcement official briefed on the investigation, also told CNN that authorities believe the"
" primary motive for Lubitz to bring down the plane was that he feared he would not be allowed to fly"
" because of his medical problems. Lubitz's girlfriend told investigators he had seen an eye doctor"
" and a neuropsychologist, both of whom deemed him unfit to work recently and concluded he had"
" psychological issues, the European government official said. But no matter what details emerge about"
" his previous mental health struggles, there's more to the story, said Brian Russell, a forensic"
' psychologist. "Psychology can explain why somebody would turn rage inward on themselves about the'
" fact that maybe they weren't going to keep doing their job and they're upset about that and so"
' they\'re suicidal," he said. "But there is no mental illness that explains why somebody then feels'
" entitled to also take that rage and turn it outward on 149 other people who had nothing to do with"
" the person's problems.\" Germanwings crash compensation: What we know . Who was the captain of"
" Germanwings Flight 9525? CNN's Margot Haddad reported from Marseille and Pamela Brown from"
" Dusseldorf, while Laura Smith-Spark wrote from London. CNN's Frederik Pleitgen, Pamela Boykoff,"
" Antonia Mortensen, Sandrine Amiel and Anna-Maja Rappard contributed to this report.",
],
return_tensors="tf",
padding="longest",
truncation=True,
)
generated_ids = self.xsum_1_1_model.generate(**batch, num_beams=4)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)
assert (
result[0]
== " The International Criminal Court (ICC) has announced that it has been announced by the International"
" Criminal court."
)
assert (
result[1]
== " An investigation into the crash that killed at least 10 people in the French capital has been"
" released by the French police investigating the crash."
)
def test_encoder_equiv(self):
batch = self.tok(
[
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories."
" The formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is"
" based. The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted"
' its jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including'
' East Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination'
" into the situation in Palestinian territories, paving the way for possible war crimes investigations"
" against Israelis. As members of the court, Palestinians may be subject to counter-charges as well."
" Israel and the United States, neither of which is an ICC member, opposed the Palestinians' efforts"
" to join the body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony,"
' said it was a move toward greater justice. "As Palestine formally becomes a State Party to the Rome'
' Statute today, the world is also a step closer to ending a long era of impunity and injustice," he'
' said, according to an ICC news release. "Indeed, today brings us closer to our shared goals of'
' justice and peace." Judge Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was'
' just the first step for the Palestinians. "As the Rome Statute today enters into force for the State'
" of Palestine, Palestine acquires all the rights as well as responsibilities that come with being a"
' State Party to the Statute. These are substantive commitments, which cannot be taken lightly," she'
' said. Rights group Human Rights Watch welcomed the development. "Governments seeking to penalize'
" Palestine for joining the ICC should immediately end their pressure, and countries that support"
" universal acceptance of the court's treaty should speak out to welcome its membership,\" said"
" Balkees Jarrah, international justice counsel for the group. \"What's objectionable is the attempts"
" to undermine international justice, not Palestine's decision to join a treaty to which over 100"
' countries around the world are members." In January, when the preliminary ICC examination was'
" opened, Israeli Prime Minister Benjamin Netanyahu described it as an outrage, saying the court was"
' overstepping its boundaries. The United States also said it "strongly" disagreed with the court\'s'
' decision. "As we have said repeatedly, we do not believe that Palestine is a state and therefore we'
' do not believe that it is eligible to join the ICC," the State Department said in a statement. It'
' urged the warring sides to resolve their differences through direct negotiations. "We will continue'
' to oppose actions against Israel at the ICC as counterproductive to the cause of peace," it said.'
" But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows'
" the court to review evidence and determine whether to investigate suspects on both sides. Prosecutor"
' Fatou Bensouda said her office would "conduct its analysis in full independence and impartiality."'
" The war between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The"
" inquiry will include alleged war crimes committed since June. The International Criminal Court was"
" set up in 2002 to prosecute genocide, crimes against humanity and war crimes.",
"The French prosecutor leading an investigation into the crash of Germanwings Flight 9525 insisted"
" Wednesday that he was not aware of any video footage from on board the plane. Marseille prosecutor"
' Brice Robin told CNN that "so far no videos were used in the crash investigation." He added, "A'
" person who has such a video needs to immediately give it to the investigators.\" Robin's comments"
" follow claims by two magazines, German daily Bild and French Paris Match, of a cell phone video"
" showing the harrowing final seconds from on board Germanwings Flight 9525 as it crashed into the"
" French Alps. All 150 on board were killed. Paris Match and Bild reported that the video was"
" recovered from a phone at the wreckage site. The two publications described the supposed video, but"
" did not post it on their websites. The publications said that they watched the video, which was"
" found by a source close to the investigation. \"One can hear cries of 'My God' in several"
' languages," Paris Match reported. "Metallic banging can also be heard more than three times, perhaps'
" of the pilot trying to open the cockpit door with a heavy object. Towards the end, after a heavy"
' shake, stronger than the others, the screaming intensifies. Then nothing." "It is a very disturbing'
" scene,\" said Julian Reichelt, editor-in-chief of Bild online. An official with France's accident"
" investigation agency, the BEA, said the agency is not aware of any such video. Lt. Col. Jean-Marc"
" Menichini, a French Gendarmerie spokesman in charge of communications on rescue efforts around the"
' Germanwings crash site, told CNN that the reports were "completely wrong" and "unwarranted." Cell'
' phones have been collected at the site, he said, but that they "hadn\'t been exploited yet."'
" Menichini said he believed the cell phones would need to be sent to the Criminal Research Institute"
" in Rosny sous-Bois, near Paris, in order to be analyzed by specialized technicians working"
" hand-in-hand with investigators. But none of the cell phones found so far have been sent to the"
" institute, Menichini said. Asked whether staff involved in the search could have leaked a memory"
' card to the media, Menichini answered with a categorical "no." Reichelt told "Erin Burnett:'
' Outfront" that he had watched the video and stood by the report, saying Bild and Paris Match are'
' "very confident" that the clip is real. He noted that investigators only revealed they\'d recovered'
' cell phones from the crash site after Bild and Paris Match published their reports. "That is'
" something we did not know before. ... Overall we can say many things of the investigation weren't"
' revealed by the investigation at the beginning," he said. What was mental state of Germanwings'
" co-pilot? German airline Lufthansa confirmed Tuesday that co-pilot Andreas Lubitz had battled"
" depression years before he took the controls of Germanwings Flight 9525, which he's accused of"
" deliberately crashing last week in the French Alps. Lubitz told his Lufthansa flight training school"
' in 2009 that he had a "previous episode of severe depression," the airline said Tuesday. Email'
" correspondence between Lubitz and the school discovered in an internal investigation, Lufthansa"
" said, included medical documents he submitted in connection with resuming his flight training. The"
" announcement indicates that Lufthansa, the parent company of Germanwings, knew of Lubitz's battle"
" with depression, allowed him to continue training and ultimately put him in the cockpit. Lufthansa,"
" whose CEO Carsten Spohr previously said Lubitz was 100% fit to fly, described its statement Tuesday"
' as a "swift and seamless clarification" and said it was sharing the information and documents --'
" including training and medical records -- with public prosecutors. Spohr traveled to the crash site"
" Wednesday, where recovery teams have been working for the past week to recover human remains and"
" plane debris scattered across a steep mountainside. He saw the crisis center set up in"
" Seyne-les-Alpes, laid a wreath in the village of Le Vernet, closer to the crash site, where grieving"
" families have left flowers at a simple stone memorial. Menichini told CNN late Tuesday that no"
" visible human remains were left at the site but recovery teams would keep searching. French"
" President Francois Hollande, speaking Tuesday, said that it should be possible to identify all the"
" victims using DNA analysis by the end of the week, sooner than authorities had previously suggested."
" In the meantime, the recovery of the victims' personal belongings will start Wednesday, Menichini"
" said. Among those personal belongings could be more cell phones belonging to the 144 passengers and"
" six crew on board. Check out the latest from our correspondents . The details about Lubitz's"
" correspondence with the flight school during his training were among several developments as"
" investigators continued to delve into what caused the crash and Lubitz's possible motive for"
" downing the jet. A Lufthansa spokesperson told CNN on Tuesday that Lubitz had a valid medical"
' certificate, had passed all his examinations and "held all the licenses required." Earlier, a'
" spokesman for the prosecutor's office in Dusseldorf, Christoph Kumpa, said medical records reveal"
" Lubitz suffered from suicidal tendencies at some point before his aviation career and underwent"
" psychotherapy before he got his pilot's license. Kumpa emphasized there's no evidence suggesting"
" Lubitz was suicidal or acting aggressively before the crash. Investigators are looking into whether"
" Lubitz feared his medical condition would cause him to lose his pilot's license, a European"
' government official briefed on the investigation told CNN on Tuesday. While flying was "a big part'
" of his life,\" the source said, it's only one theory being considered. Another source, a law"
" enforcement official briefed on the investigation, also told CNN that authorities believe the"
" primary motive for Lubitz to bring down the plane was that he feared he would not be allowed to fly"
" because of his medical problems. Lubitz's girlfriend told investigators he had seen an eye doctor"
" and a neuropsychologist, both of whom deemed him unfit to work recently and concluded he had"
" psychological issues, the European government official said. But no matter what details emerge about"
" his previous mental health struggles, there's more to the story, said Brian Russell, a forensic"
' psychologist. "Psychology can explain why somebody would turn rage inward on themselves about the'
" fact that maybe they weren't going to keep doing their job and they're upset about that and so"
' they\'re suicidal," he said. "But there is no mental illness that explains why somebody then feels'
" entitled to also take that rage and turn it outward on 149 other people who had nothing to do with"
" the person's problems.\" Germanwings crash compensation: What we know . Who was the captain of"
" Germanwings Flight 9525? CNN's Margot Haddad reported from Marseille and Pamela Brown from"
" Dusseldorf, while Laura Smith-Spark wrote from London. CNN's Frederik Pleitgen, Pamela Boykoff,"
" Antonia Mortensen, Sandrine Amiel and Anna-Maja Rappard contributed to this report.",
],
return_tensors="tf",
padding="longest",
truncation=True,
)
features = self.xsum_1_1_model.get_encoder()(**batch).last_hidden_state
expected = np.array([[-0.0828, -0.0251, -0.0674], [0.1277, 0.3311, -0.0255], [0.2613, -0.0840, -0.2763]])
assert np.allclose(features[0, :3, :3].numpy(), expected, atol=1e-3)
| # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import copy
import tempfile
import unittest
import numpy as np
from transformers import BartConfig, BartTokenizer, is_tf_available
from transformers.testing_utils import require_tf, slow, tooslow
from transformers.utils import cached_property
from ...test_configuration_common import ConfigTester
from ...test_modeling_tf_common import TFModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
from ...utils.test_modeling_tf_core import TFCoreModelTesterMixin
if is_tf_available():
import tensorflow as tf
from transformers import TFBartForConditionalGeneration, TFBartForSequenceClassification, TFBartModel
@require_tf
class TFBartModelTester:
config_cls = BartConfig
config_updates = {}
hidden_act = "gelu"
def __init__(
self,
parent,
batch_size=13,
seq_length=7,
is_training=True,
use_labels=False,
vocab_size=99,
hidden_size=32,
num_hidden_layers=5,
num_attention_heads=4,
intermediate_size=37,
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=20,
eos_token_id=2,
pad_token_id=1,
bos_token_id=0,
):
self.parent = parent
self.batch_size = batch_size
self.seq_length = seq_length
self.is_training = is_training
self.use_labels = use_labels
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.eos_token_id = eos_token_id
self.pad_token_id = pad_token_id
self.bos_token_id = bos_token_id
def prepare_config_and_inputs_for_common(self):
# Ids are clipped to avoid "beginng of sequence", "end of sequence", and "pad" tokens
input_ids = tf.clip_by_value(
ids_tensor([self.batch_size, self.seq_length - 1], self.vocab_size),
clip_value_min=self.eos_token_id + 1,
clip_value_max=self.vocab_size + 1,
)
# Explicity add "end of sequence" to the inputs
eos_tensor = tf.expand_dims(tf.constant([self.eos_token_id] * self.batch_size), 1)
input_ids = tf.concat([input_ids, eos_tensor], axis=1)
decoder_input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
config = self.config_cls(
vocab_size=self.vocab_size,
d_model=self.hidden_size,
encoder_layers=self.num_hidden_layers,
decoder_layers=self.num_hidden_layers,
encoder_attention_heads=self.num_attention_heads,
decoder_attention_heads=self.num_attention_heads,
encoder_ffn_dim=self.intermediate_size,
decoder_ffn_dim=self.intermediate_size,
dropout=self.hidden_dropout_prob,
attention_dropout=self.attention_probs_dropout_prob,
max_position_embeddings=self.max_position_embeddings,
eos_token_ids=[2],
bos_token_id=self.bos_token_id,
pad_token_id=self.pad_token_id,
decoder_start_token_id=self.pad_token_id,
**self.config_updates,
)
inputs_dict = prepare_bart_inputs_dict(config, input_ids, decoder_input_ids)
return config, inputs_dict
def check_decoder_model_past_large_inputs(self, config, inputs_dict):
model = TFBartModel(config=config).get_decoder()
input_ids = inputs_dict["input_ids"]
input_ids = input_ids[:1, :]
attention_mask = inputs_dict["attention_mask"][:1, :]
head_mask = inputs_dict["head_mask"]
self.batch_size = 1
# first forward pass
outputs = model(input_ids, attention_mask=attention_mask, head_mask=head_mask, use_cache=True)
output, past_key_values = outputs.to_tuple()
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
next_attn_mask = tf.cast(ids_tensor((self.batch_size, 3), 2), tf.int8)
# append to next input_ids and
next_input_ids = tf.concat([input_ids, next_tokens], axis=-1)
next_attention_mask = tf.concat([attention_mask, next_attn_mask], axis=-1)
output_from_no_past = model(next_input_ids, attention_mask=next_attention_mask)
output_from_no_past = output_from_no_past[0]
output_from_past = model(next_tokens, attention_mask=next_attention_mask, past_key_values=past_key_values)
output_from_past = output_from_past[0]
self.parent.assertEqual(next_tokens.shape[1], output_from_past.shape[1])
# select random slice
random_slice_idx = int(ids_tensor((1,), output_from_past.shape[-1]))
output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx]
output_from_past_slice = output_from_past[:, :, random_slice_idx]
# test that outputs are equal for slice
tf.debugging.assert_near(output_from_past_slice, output_from_no_past_slice, rtol=1e-3)
def prepare_bart_inputs_dict(
config,
input_ids,
decoder_input_ids,
attention_mask=None,
decoder_attention_mask=None,
head_mask=None,
decoder_head_mask=None,
cross_attn_head_mask=None,
):
if attention_mask is None:
attention_mask = tf.cast(tf.math.not_equal(input_ids, config.pad_token_id), tf.int8)
if decoder_attention_mask is None:
decoder_attention_mask = tf.concat(
[
tf.ones(decoder_input_ids[:, :1].shape, dtype=tf.int8),
tf.cast(tf.math.not_equal(decoder_input_ids[:, 1:], config.pad_token_id), tf.int8),
],
axis=-1,
)
if head_mask is None:
head_mask = tf.ones((config.encoder_layers, config.encoder_attention_heads))
if decoder_head_mask is None:
decoder_head_mask = tf.ones((config.decoder_layers, config.decoder_attention_heads))
if cross_attn_head_mask is None:
cross_attn_head_mask = tf.ones((config.decoder_layers, config.decoder_attention_heads))
return {
"input_ids": input_ids,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"decoder_attention_mask": decoder_attention_mask,
"head_mask": head_mask,
"decoder_head_mask": decoder_head_mask,
"cross_attn_head_mask": cross_attn_head_mask,
}
@require_tf
class TFBartModelTest(TFModelTesterMixin, TFCoreModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(TFBartForConditionalGeneration, TFBartForSequenceClassification, TFBartModel) if is_tf_available() else ()
)
all_generative_model_classes = (TFBartForConditionalGeneration,) if is_tf_available() else ()
pipeline_model_mapping = (
{
"conversational": TFBartForConditionalGeneration,
"feature-extraction": TFBartModel,
"summarization": TFBartForConditionalGeneration,
"text2text-generation": TFBartForConditionalGeneration,
"text-classification": TFBartForSequenceClassification,
"zero-shot": TFBartForSequenceClassification,
}
if is_tf_available()
else {}
)
is_encoder_decoder = True
test_pruning = False
test_onnx = True
onnx_min_opset = 10
def setUp(self):
self.model_tester = TFBartModelTester(self)
self.config_tester = ConfigTester(self, config_class=BartConfig)
def test_config(self):
self.config_tester.run_common_tests()
def test_decoder_model_past_large_inputs(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs_for_common()
self.model_tester.check_decoder_model_past_large_inputs(*config_and_inputs)
def test_model_common_attributes(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
assert isinstance(model.get_input_embeddings(), tf.keras.layers.Layer)
if model_class in self.all_generative_model_classes:
x = model.get_output_embeddings()
assert isinstance(x, tf.keras.layers.Layer)
name = model.get_bias()
assert isinstance(name, dict)
for k, v in name.items():
assert isinstance(v, tf.Variable)
else:
x = model.get_output_embeddings()
assert x is None
name = model.get_bias()
assert name is None
@tooslow
def test_saved_model_creation(self):
pass
# TODO (Joao): fix me
@unittest.skip("Onnx compliancy broke with TF 2.10")
def test_onnx_compliancy(self):
pass
# TFBartForSequenceClassification does not support inputs_embeds
def test_inputs_embeds(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in (TFBartForConditionalGeneration, TFBartModel):
model = model_class(config)
inputs = copy.deepcopy(inputs_dict)
if not self.is_encoder_decoder:
input_ids = inputs["input_ids"]
del inputs["input_ids"]
else:
encoder_input_ids = inputs["input_ids"]
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
del inputs["input_ids"]
inputs.pop("decoder_input_ids", None)
if not self.is_encoder_decoder:
inputs["inputs_embeds"] = model.get_input_embeddings()(input_ids)
else:
inputs["inputs_embeds"] = model.get_input_embeddings()(encoder_input_ids)
inputs["decoder_inputs_embeds"] = model.get_input_embeddings()(decoder_input_ids)
inputs = self._prepare_for_class(inputs, model_class)
model(inputs)
# TFBartForSequenceClassification does not support inputs_embeds
@slow
def test_graph_mode_with_inputs_embeds(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in (TFBartForConditionalGeneration, TFBartModel):
model = model_class(config)
inputs = copy.deepcopy(inputs_dict)
if not self.is_encoder_decoder:
input_ids = inputs["input_ids"]
del inputs["input_ids"]
else:
encoder_input_ids = inputs["input_ids"]
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
del inputs["input_ids"]
inputs.pop("decoder_input_ids", None)
if not self.is_encoder_decoder:
inputs["inputs_embeds"] = model.get_input_embeddings()(input_ids)
else:
inputs["inputs_embeds"] = model.get_input_embeddings()(encoder_input_ids)
inputs["decoder_inputs_embeds"] = model.get_input_embeddings()(decoder_input_ids)
inputs = self._prepare_for_class(inputs, model_class)
@tf.function
def run_in_graph_mode():
return model(inputs)
outputs = run_in_graph_mode()
self.assertIsNotNone(outputs)
@slow
def test_save_load_after_resize_token_embeddings(self):
# Custom version of this test to ensure "end of sequence" tokens are present throughout
if not self.test_resize_embeddings:
return
config, original_inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
# create a model with resized (expended) embeddings
new_tokens_size = 10
old_total_size = config.vocab_size
new_total_size = old_total_size + new_tokens_size
model = model_class(config=copy.deepcopy(config)) # `resize_token_embeddings` mutates `config`
model(model.dummy_inputs) # builds the embeddings layer
model.resize_token_embeddings(new_total_size)
# fetch the output for an input exclusively made of new members of the vocabulary
inputs_dict = copy.deepcopy(original_inputs_dict)
ids_feat_name = None
if "input_ids" in inputs_dict:
ids_feat_name = "input_ids"
elif "decoder_input_ids" in inputs_dict:
ids_feat_name = "decoder_input_ids"
else:
assert False, "No input ids feature found in the inputs dict"
new_vocab_input_ids = ids_tensor(inputs_dict[ids_feat_name].shape, new_tokens_size)
new_vocab_input_ids += old_total_size
# Replace last id with EOS token
new_vocab_input_ids = new_vocab_input_ids[:, :-1]
new_vocab_input_ids = tf.concat(
[new_vocab_input_ids, tf.ones((tf.shape(new_vocab_input_ids)[0], 1), dtype=tf.int32) * 2], axis=1
)
inputs_dict[ids_feat_name] = new_vocab_input_ids
if "input_ids" in inputs_dict:
inputs_dict["input_ids"] = new_vocab_input_ids
if "decoder_input_ids" in inputs_dict:
inputs_dict["decoder_input_ids"] = new_vocab_input_ids
prepared_inputs = self._prepare_for_class(inputs_dict, model_class)
outputs = model(**prepared_inputs)
# save and load the model
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(tmpdirname, saved_model=False)
model = model_class.from_pretrained(tmpdirname)
restored_model_outputs = model(**prepared_inputs)
# check that the output for the restored model is the same
self.assert_outputs_same(restored_model_outputs, outputs)
def _long_tensor(tok_lst):
return tf.constant(tok_lst, dtype=tf.int32)
@require_tf
class TFBartHeadTests(unittest.TestCase):
vocab_size = 99
def _get_config_and_data(self):
eos_column_vector = tf.ones((4, 1), dtype=tf.int32) * 2
input_ids = tf.concat([ids_tensor((4, 6), self.vocab_size - 3) + 3, eos_column_vector], axis=1)
batch_size = input_ids.shape[0]
config = BartConfig(
vocab_size=self.vocab_size,
d_model=24,
encoder_layers=2,
decoder_layers=2,
encoder_attention_heads=2,
decoder_attention_heads=2,
encoder_ffn_dim=32,
decoder_ffn_dim=32,
max_position_embeddings=48,
eos_token_id=2,
pad_token_id=1,
bos_token_id=0,
decoder_start_token_id=2,
)
return config, input_ids, batch_size
def test_lm_forward(self):
config, input_ids, batch_size = self._get_config_and_data()
decoder_lm_labels = ids_tensor([batch_size, input_ids.shape[1]], self.vocab_size)
lm_model = TFBartForConditionalGeneration(config)
outputs = lm_model(input_ids=input_ids, labels=decoder_lm_labels, decoder_input_ids=input_ids, use_cache=False)
expected_shape = (batch_size, input_ids.shape[1], config.vocab_size)
self.assertEqual(outputs.logits.shape, expected_shape)
def test_lm_uneven_forward(self):
config = BartConfig(
vocab_size=10,
d_model=24,
encoder_layers=2,
decoder_layers=2,
encoder_attention_heads=2,
decoder_attention_heads=2,
encoder_ffn_dim=32,
decoder_ffn_dim=32,
max_position_embeddings=48,
)
lm_model = TFBartForConditionalGeneration(config)
context = tf.fill((7, 2), 4)
summary = tf.fill((7, 7), 6)
outputs = lm_model(input_ids=context, decoder_input_ids=summary, use_cache=False)
expected_shape = (*summary.shape, config.vocab_size)
self.assertEqual(outputs.logits.shape, expected_shape)
@require_tf
class TFBartForSequenceClassificationTest(unittest.TestCase):
def test_model_fails_for_uneven_eos_tokens(self):
config = BartConfig(eos_token_id=2)
model = TFBartForSequenceClassification(config)
inputs = {
"input_ids": tf.constant([[1, 2, 2, 2], [1, 3, 2, 2], [2, 2, 3, 3]]),
"attention_mask": tf.constant([[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]),
}
with self.assertRaises(tf.errors.InvalidArgumentError):
model(inputs)
@slow
@require_tf
class TFBartModelIntegrationTest(unittest.TestCase):
def test_inference_no_head(self):
model = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large").model
input_ids = _long_tensor([[0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2]])
attention_mask = tf.cast(tf.math.not_equal(input_ids, model.config.pad_token_id), tf.int8)
output = model(input_ids=input_ids, attention_mask=attention_mask)[0]
expected_shape = (1, 11, 1024)
self.assertEqual(output.shape, expected_shape)
expected_slice = tf.convert_to_tensor(
[[0.7144, 0.8143, -1.2813], [0.7144, 0.8143, -1.2813], [-0.0467, 2.5911, -2.1845]],
)
tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=1e-3)
def test_cnn_summarization_same_as_fairseq_hard(self):
hf = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
tok = self.tok
FRANCE_ARTICLE = ( # @noqa
" Marseille, France (CNN)The French prosecutor leading an investigation into the crash of Germanwings"
" Flight 9525 insisted Wednesday that he was not aware of any video footage from on board the plane."
' Marseille prosecutor Brice Robin told CNN that "so far no videos were used in the crash investigation."'
' He added, "A person who has such a video needs to immediately give it to the investigators." Robin\'s'
" comments follow claims by two magazines, German daily Bild and French Paris Match, of a cell phone video"
" showing the harrowing final seconds from on board Germanwings Flight 9525 as it crashed into the French"
" Alps. All 150 on board were killed. Paris Match and Bild reported that the video was recovered from a"
" phone at the wreckage site. The two publications described the supposed video, but did not post it on"
" their websites. The publications said that they watched the video, which was found by a source close to"
" the investigation. \"One can hear cries of 'My God' in several languages,\" Paris Match reported."
' "Metallic banging can also be heard more than three times, perhaps of the pilot trying to open the'
" cockpit door with a heavy object. Towards the end, after a heavy shake, stronger than the others, the"
' screaming intensifies. Then nothing." "It is a very disturbing scene," said Julian Reichelt,'
" editor-in-chief of Bild online. An official with France's accident investigation agency, the BEA, said"
" the agency is not aware of any such video. Lt. Col. Jean-Marc Menichini, a French Gendarmerie spokesman"
" in charge of communications on rescue efforts around the Germanwings crash site, told CNN that the"
' reports were "completely wrong" and "unwarranted." Cell phones have been collected at the site, he said,'
' but that they "hadn\'t been exploited yet." Menichini said he believed the cell phones would need to be'
" sent to the Criminal Research Institute in Rosny sous-Bois, near Paris, in order to be analyzed by"
" specialized technicians working hand-in-hand with investigators. But none of the cell phones found so"
" far have been sent to the institute, Menichini said. Asked whether staff involved in the search could"
' have leaked a memory card to the media, Menichini answered with a categorical "no." Reichelt told "Erin'
' Burnett: Outfront" that he had watched the video and stood by the report, saying Bild and Paris Match'
' are "very confident" that the clip is real. He noted that investigators only revealed they\'d recovered'
' cell phones from the crash site after Bild and Paris Match published their reports. "That is something'
" we did not know before. ... Overall we can say many things of the investigation weren't revealed by the"
' investigation at the beginning," he said. What was mental state of Germanwings co-pilot? German airline'
" Lufthansa confirmed Tuesday that co-pilot Andreas Lubitz had battled depression years before he took the"
" controls of Germanwings Flight 9525, which he's accused of deliberately crashing last week in the"
' French Alps. Lubitz told his Lufthansa flight training school in 2009 that he had a "previous episode of'
' severe depression," the airline said Tuesday. Email correspondence between Lubitz and the school'
" discovered in an internal investigation, Lufthansa said, included medical documents he submitted in"
" connection with resuming his flight training. The announcement indicates that Lufthansa, the parent"
" company of Germanwings, knew of Lubitz's battle with depression, allowed him to continue training and"
" ultimately put him in the cockpit. Lufthansa, whose CEO Carsten Spohr previously said Lubitz was 100%"
' fit to fly, described its statement Tuesday as a "swift and seamless clarification" and said it was'
" sharing the information and documents -- including training and medical records -- with public"
" prosecutors. Spohr traveled to the crash site Wednesday, where recovery teams have been working for the"
" past week to recover human remains and plane debris scattered across a steep mountainside. He saw the"
" crisis center set up in Seyne-les-Alpes, laid a wreath in the village of Le Vernet, closer to the crash"
" site, where grieving families have left flowers at a simple stone memorial. Menichini told CNN late"
" Tuesday that no visible human remains were left at the site but recovery teams would keep searching."
" French President Francois Hollande, speaking Tuesday, said that it should be possible to identify all"
" the victims using DNA analysis by the end of the week, sooner than authorities had previously suggested."
" In the meantime, the recovery of the victims' personal belongings will start Wednesday, Menichini said."
" Among those personal belongings could be more cell phones belonging to the 144 passengers and six crew"
" on board. Check out the latest from our correspondents . The details about Lubitz's correspondence with"
" the flight school during his training were among several developments as investigators continued to"
" delve into what caused the crash and Lubitz's possible motive for downing the jet. A Lufthansa"
" spokesperson told CNN on Tuesday that Lubitz had a valid medical certificate, had passed all his"
' examinations and "held all the licenses required." Earlier, a spokesman for the prosecutor\'s office in'
" Dusseldorf, Christoph Kumpa, said medical records reveal Lubitz suffered from suicidal tendencies at"
" some point before his aviation career and underwent psychotherapy before he got his pilot's license."
" Kumpa emphasized there's no evidence suggesting Lubitz was suicidal or acting aggressively before the"
" crash. Investigators are looking into whether Lubitz feared his medical condition would cause him to"
" lose his pilot's license, a European government official briefed on the investigation told CNN on"
' Tuesday. While flying was "a big part of his life," the source said, it\'s only one theory being'
" considered. Another source, a law enforcement official briefed on the investigation, also told CNN that"
" authorities believe the primary motive for Lubitz to bring down the plane was that he feared he would"
" not be allowed to fly because of his medical problems. Lubitz's girlfriend told investigators he had"
" seen an eye doctor and a neuropsychologist, both of whom deemed him unfit to work recently and concluded"
" he had psychological issues, the European government official said. But no matter what details emerge"
" about his previous mental health struggles, there's more to the story, said Brian Russell, a forensic"
' psychologist. "Psychology can explain why somebody would turn rage inward on themselves about the fact'
" that maybe they weren't going to keep doing their job and they're upset about that and so they're"
' suicidal," he said. "But there is no mental illness that explains why somebody then feels entitled to'
" also take that rage and turn it outward on 149 other people who had nothing to do with the person's"
' problems." Germanwings crash compensation: What we know . Who was the captain of Germanwings Flight'
" 9525? CNN's Margot Haddad reported from Marseille and Pamela Brown from Dusseldorf, while Laura"
" Smith-Spark wrote from London. CNN's Frederik Pleitgen, Pamela Boykoff, Antonia Mortensen, Sandrine"
" Amiel and Anna-Maja Rappard contributed to this report."
)
EXPECTED_SUMMARY_FRANCE = (
"French prosecutor says he's not aware of any video footage from on board the plane. German daily Bild"
" and French Paris Match claim to have found a cell phone video of the crash. A French Gendarmerie"
' spokesman calls the reports "completely wrong" and "unwarranted" German airline Lufthansa confirms'
" co-pilot Andreas Lubitz had battled depression."
)
SHORTER_ARTICLE = (
" (CNN)The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes. CNN's Vasco Cotovio, Kareem Khadder"
" and Faith Karimi contributed to this report."
)
EXPECTED_SUMMARY_SHORTER = (
"The Palestinian Authority becomes the 123rd member of the International Criminal Court. The move gives"
" the court jurisdiction over alleged crimes in Palestinian territories. Israel and the United States"
" opposed the Palestinians' efforts to join the body. But Palestinian Foreign Minister Riad al-Malki said"
" it was a move toward greater justice."
)
# The below article tests that we don't add any hypotheses outside of the top n_beams
IRAN_ARTICLE = (
" (CNN)The United States and its negotiating partners reached a very strong framework agreement with Iran"
" in Lausanne, Switzerland, on Thursday that limits Iran's nuclear program in such a way as to effectively"
" block it from building a nuclear weapon. Expect pushback anyway, if the recent past is any harbinger."
" Just last month, in an attempt to head off such an agreement, House Speaker John Boehner invited Israeli"
" Prime Minister Benjamin Netanyahu to preemptively blast it before Congress, and 47 senators sent a"
" letter to the Iranian leadership warning them away from a deal. The debate that has already begun since"
" the announcement of the new framework will likely result in more heat than light. It will not be helped"
" by the gathering swirl of dubious assumptions and doubtful assertions. Let us address some of these: ."
" The most misleading assertion, despite universal rejection by experts, is that the negotiations'"
" objective at the outset was the total elimination of any nuclear program in Iran. That is the position"
" of Netanyahu and his acolytes in the U.S. Congress. But that is not and never was the objective. If it"
" had been, there would have been no Iranian team at the negotiating table. Rather, the objective has"
" always been to structure an agreement or series of agreements so that Iran could not covertly develop a"
" nuclear arsenal before the United States and its allies could respond. The new framework has exceeded"
" expectations in achieving that goal. It would reduce Iran's low-enriched uranium stockpile, cut by"
" two-thirds its number of installed centrifuges and implement a rigorous inspection regime. Another"
" dubious assumption of opponents is that the Iranian nuclear program is a covert weapons program. Despite"
" sharp accusations by some in the United States and its allies, Iran denies having such a program, and"
" U.S. intelligence contends that Iran has not yet made the decision to build a nuclear weapon. Iran's"
" continued cooperation with International Atomic Energy Agency inspections is further evidence on this"
" point, and we'll know even more about Iran's program in the coming months and years because of the deal."
" In fact, the inspections provisions that are part of this agreement are designed to protect against any"
" covert action by the Iranians. What's more, the rhetoric of some members of Congress has implied that"
" the negotiations have been between only the United States and Iran (i.e., the 47 senators' letter"
" warning that a deal might be killed by Congress or a future president). This of course is not the case."
" The talks were between Iran and the five permanent members of the U.N. Security Council (United States,"
" United Kingdom, France, China and Russia) plus Germany, dubbed the P5+1. While the United States has"
" played a leading role in the effort, it negotiated the terms alongside its partners. If the agreement"
" reached by the P5+1 is rejected by Congress, it could result in an unraveling of the sanctions on Iran"
" and threaten NATO cohesion in other areas. Another questionable assertion is that this agreement"
" contains a sunset clause, after which Iran will be free to do as it pleases. Again, this is not the"
" case. Some of the restrictions on Iran's nuclear activities, such as uranium enrichment, will be eased"
" or eliminated over time, as long as 15 years. But most importantly, the framework agreement includes"
" Iran's ratification of the Additional Protocol, which allows IAEA inspectors expanded access to nuclear"
" sites both declared and nondeclared. This provision will be permanent. It does not sunset. Thus, going"
" forward, if Iran decides to enrich uranium to weapons-grade levels, monitors will be able to detect such"
" a move in a matter of days and alert the U.N. Security Council. Many in Congress have said that the"
' agreement should be a formal treaty requiring the Senate to "advise and consent." But the issue is not'
" suited for a treaty. Treaties impose equivalent obligations on all signatories. For example, the New"
" START treaty limits Russia and the United States to 1,550 deployed strategic warheads. But any agreement"
" with Iran will not be so balanced. The restrictions and obligations in the final framework agreement"
" will be imposed almost exclusively on Iran. The P5+1 are obligated only to ease and eventually remove"
" most but not all economic sanctions, which were imposed as leverage to gain this final deal. Finally"
" some insist that any agreement must address Iranian missile programs, human rights violations or support"
" for Hamas or Hezbollah. As important as these issues are, and they must indeed be addressed, they are"
" unrelated to the most important aim of a nuclear deal: preventing a nuclear Iran. To include them in"
" the negotiations would be a poison pill. This agreement should be judged on its merits and on how it"
" affects the security of our negotiating partners and allies, including Israel. Those judgments should be"
" fact-based, not based on questionable assertions or dubious assumptions."
)
EXPECTED_SUMMARY_IRAN = (
"The U.S. and its negotiating partners reached a very strong framework agreement with Iran. Peter Bergen:"
" The debate that has already begun will likely result in more heat than light. He says the agreement"
" limits Iran's nuclear program in such a way as to effectively block it from building a nuclear weapon."
" Bergen says the most important aim of a nuclear deal is preventing a nuclear Iran."
)
ARTICLE_SUBWAY = (
" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York. A"
" year later, she got married again in Westchester County, but to a different man and without divorcing"
" her first husband. Only 18 days after that marriage, she got hitched yet again. Then, Barrientos"
' declared "I do" five more times, sometimes only within two weeks of each other. In 2010, she married'
" once more, this time in the Bronx. In an application for a marriage license, she stated it was her"
' "first and only" marriage. Barrientos, now 39, is facing two criminal counts of "offering a false'
' instrument for filing in the first degree," referring to her false statements on the 2010 marriage'
" license application, according to court documents. Prosecutors said the marriages were part of an"
" immigration scam. On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to"
" her attorney, Christopher Wright, who declined to comment further. After leaving court, Barrientos was"
" arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New"
" York subway through an emergency exit, said Detective Annette Markowski, a police spokeswoman. In total,"
" Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002. All"
" occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be"
" married to four men, and at one time, she was married to eight men at once, prosecutors say. Prosecutors"
" said the immigration scam involved some of her husbands, who filed for permanent residence status"
" shortly after the marriages. Any divorces happened only after such filings were approved. It was"
" unclear whether any of the men will be prosecuted. The case was referred to the Bronx District"
" Attorney's Office by Immigration and Customs Enforcement and the Department of Homeland Security's"
' Investigation Division. Seven of the men are from so-called "red-flagged" countries, including Egypt,'
" Turkey, Georgia, Pakistan and Mali. Her eighth husband, Rashid Rajput, was deported in 2006 to his"
" native Pakistan after an investigation by the Joint Terrorism Task Force. If convicted, Barrientos faces"
" up to four years in prison. Her next court appearance is scheduled for May 18."
)
EXPECTED_SUMMARY_SUBWAY = (
"Liana Barrientos has been married 10 times, sometimes within two weeks of each other. Prosecutors say the"
" marriages were part of an immigration scam. On Friday, she pleaded not guilty at State Supreme Court in"
" the Bronx. She was arrested and charged with theft of service and criminal trespass for allegedly"
" sneaking into the subway."
)
dct = tok(
[FRANCE_ARTICLE, SHORTER_ARTICLE, IRAN_ARTICLE, ARTICLE_SUBWAY],
max_length=1024,
truncation_strategy="only_first",
padding="longest",
truncation=True,
return_tensors="tf",
)
self.assertEqual(1024, dct["input_ids"].shape[1])
hypotheses_batch = hf.generate(
input_ids=dct["input_ids"],
attention_mask=dct["attention_mask"],
)
assert hypotheses_batch[:, 1].numpy().tolist() == [0, 0, 0, 0] # test force_bos_token_to_be_generated
decoded = tok.batch_decode(hypotheses_batch, skip_special_tokens=True, clean_up_tokenization_spaces=False)
expected_batch = [
EXPECTED_SUMMARY_FRANCE,
EXPECTED_SUMMARY_SHORTER,
EXPECTED_SUMMARY_IRAN,
EXPECTED_SUMMARY_SUBWAY,
]
assert decoded == expected_batch
@cached_property
def tok(self):
return BartTokenizer.from_pretrained("facebook/bart-large")
@slow
def test_contrastive_search_bart(self):
article = (
" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York. A"
" year later, she got married again in Westchester County, but to a different man and without divorcing"
" her first husband. Only 18 days after that marriage, she got hitched yet again. Then, Barrientos"
' declared "I do" five more times, sometimes only within two weeks of each other. In 2010, she married'
" once more, this time in the Bronx. In an application for a marriage license, she stated it was her"
' "first and only" marriage. Barrientos, now 39, is facing two criminal counts of "offering a false'
' instrument for filing in the first degree," referring to her false statements on the 2010 marriage'
" license application, according to court documents. Prosecutors said the marriages were part of an"
" immigration scam. On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to"
" her attorney, Christopher Wright, who declined to comment further. After leaving court, Barrientos was"
" arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New"
" York subway through an emergency exit, said Detective Annette Markowski, a police spokeswoman. In total,"
" Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002. All"
" occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be"
" married to four men, and at one time, she was married to eight men at once, prosecutors say. Prosecutors"
" said the immigration scam involved some of her husbands, who filed for permanent residence status"
" shortly after the marriages. Any divorces happened only after such filings were approved. It was"
" unclear whether any of the men will be prosecuted. The case was referred to the Bronx District"
" Attorney's Office by Immigration and Customs Enforcement and the Department of Homeland Security's"
' Investigation Division. Seven of the men are from so-called "red-flagged" countries, including Egypt,'
" Turkey, Georgia, Pakistan and Mali. Her eighth husband, Rashid Rajput, was deported in 2006 to his"
" native Pakistan after an investigation by the Joint Terrorism Task Force. If convicted, Barrientos faces"
" up to four years in prison. Her next court appearance is scheduled for May 18."
)
bart_tokenizer = BartTokenizer.from_pretrained("facebook/bart-large-cnn")
bart_model = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
input_ids = bart_tokenizer(
article, add_special_tokens=False, truncation=True, max_length=512, return_tensors="tf"
).input_ids
outputs = bart_model.generate(input_ids, penalty_alpha=0.5, top_k=5, max_length=64)
generated_text = bart_tokenizer.batch_decode(outputs, skip_special_tokens=True)
self.assertListEqual(
generated_text,
[
"Liana Barrientos, 39, pleaded not guilty to charges related to false marriage statements. "
"Prosecutors say she married at least 10 times, sometimes within two weeks of each other. She is "
"accused of being part of an immigration scam to get permanent residency. If convicted, she faces up "
"to four years in"
],
)
@slow
def test_contrastive_search_bart_xla(self):
article = (
" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York. A"
" year later, she got married again in Westchester County, but to a different man and without divorcing"
" her first husband. Only 18 days after that marriage, she got hitched yet again. Then, Barrientos"
' declared "I do" five more times, sometimes only within two weeks of each other. In 2010, she married'
" once more, this time in the Bronx. In an application for a marriage license, she stated it was her"
' "first and only" marriage. Barrientos, now 39, is facing two criminal counts of "offering a false'
' instrument for filing in the first degree," referring to her false statements on the 2010 marriage'
" license application, according to court documents. Prosecutors said the marriages were part of an"
" immigration scam. On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to"
" her attorney, Christopher Wright, who declined to comment further. After leaving court, Barrientos was"
" arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New"
" York subway through an emergency exit, said Detective Annette Markowski, a police spokeswoman. In total,"
" Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002. All"
" occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be"
" married to four men, and at one time, she was married to eight men at once, prosecutors say. Prosecutors"
" said the immigration scam involved some of her husbands, who filed for permanent residence status"
" shortly after the marriages. Any divorces happened only after such filings were approved. It was"
" unclear whether any of the men will be prosecuted. The case was referred to the Bronx District"
" Attorney's Office by Immigration and Customs Enforcement and the Department of Homeland Security's"
' Investigation Division. Seven of the men are from so-called "red-flagged" countries, including Egypt,'
" Turkey, Georgia, Pakistan and Mali. Her eighth husband, Rashid Rajput, was deported in 2006 to his"
" native Pakistan after an investigation by the Joint Terrorism Task Force. If convicted, Barrientos faces"
" up to four years in prison. Her next court appearance is scheduled for May 18."
)
bart_tokenizer = BartTokenizer.from_pretrained("facebook/bart-large-cnn")
bart_model = TFBartForConditionalGeneration.from_pretrained("facebook/bart-large-cnn")
input_ids = bart_tokenizer(
article, add_special_tokens=False, truncation=True, max_length=512, return_tensors="tf"
).input_ids
xla_generate = tf.function(bart_model.generate, jit_compile=True)
# no_repeat_ngram_size set to 0 because it isn't compatible with XLA, but doesn't change the original output
outputs = xla_generate(input_ids, penalty_alpha=0.5, top_k=5, max_length=64, no_repeat_ngram_size=0)
generated_text = bart_tokenizer.batch_decode(outputs, skip_special_tokens=True)
self.assertListEqual(
generated_text,
[
"Liana Barrientos, 39, pleaded not guilty to charges related to false marriage statements. "
"Prosecutors say she married at least 10 times, sometimes within two weeks of each other. She is "
"accused of being part of an immigration scam to get permanent residency. If convicted, she faces up "
"to four years in"
],
)
@slow
@require_tf
class FasterTFBartModelIntegrationTests(unittest.TestCase):
"""These tests are useful for debugging since they operate on a model with 1 encoder layer and 1 decoder layer."""
@cached_property
def tok(self):
return BartTokenizer.from_pretrained("facebook/bart-large")
@cached_property
def xsum_1_1_model(self):
return TFBartForConditionalGeneration.from_pretrained("sshleifer/distilbart-xsum-1-1")
def test_xsum_1_1_generation(self):
model = self.xsum_1_1_model
assert model.model.decoder.embed_tokens == model.model.shared
ARTICLE = (
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes."
)
EXPECTED = (
" The International Criminal Court (ICC) has announced that it has been announced by the International"
" Criminal court."
)
dct = self.tok(ARTICLE, return_tensors="tf")
generated_ids = model.generate(**dct, num_beams=4)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)[0]
assert result == EXPECTED
def test_xsum_1_1_xla_generation(self):
# same test as above, but with `no_repeat_ngram_size=0` (not compatible with XLA) and XLA comparison enabled
model = self.xsum_1_1_model
assert model.model.decoder.embed_tokens == model.model.shared
ARTICLE = (
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories. The"
" formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is based."
" The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted its"
' jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including East'
' Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination into the'
" situation in Palestinian territories, paving the way for possible war crimes investigations against"
" Israelis. As members of the court, Palestinians may be subject to counter-charges as well. Israel and"
" the United States, neither of which is an ICC member, opposed the Palestinians' efforts to join the"
" body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony, said it was a"
' move toward greater justice. "As Palestine formally becomes a State Party to the Rome Statute today, the'
' world is also a step closer to ending a long era of impunity and injustice," he said, according to an'
' ICC news release. "Indeed, today brings us closer to our shared goals of justice and peace." Judge'
" Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was just the first step for the"
' Palestinians. "As the Rome Statute today enters into force for the State of Palestine, Palestine'
" acquires all the rights as well as responsibilities that come with being a State Party to the Statute."
' These are substantive commitments, which cannot be taken lightly," she said. Rights group Human Rights'
' Watch welcomed the development. "Governments seeking to penalize Palestine for joining the ICC should'
" immediately end their pressure, and countries that support universal acceptance of the court's treaty"
' should speak out to welcome its membership," said Balkees Jarrah, international justice counsel for the'
" group. \"What's objectionable is the attempts to undermine international justice, not Palestine's"
' decision to join a treaty to which over 100 countries around the world are members." In January, when'
" the preliminary ICC examination was opened, Israeli Prime Minister Benjamin Netanyahu described it as an"
' outrage, saying the court was overstepping its boundaries. The United States also said it "strongly"'
" disagreed with the court's decision. \"As we have said repeatedly, we do not believe that Palestine is a"
' state and therefore we do not believe that it is eligible to join the ICC," the State Department said in'
' a statement. It urged the warring sides to resolve their differences through direct negotiations. "We'
' will continue to oppose actions against Israel at the ICC as counterproductive to the cause of peace,"'
" it said. But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows the'
" court to review evidence and determine whether to investigate suspects on both sides. Prosecutor Fatou"
' Bensouda said her office would "conduct its analysis in full independence and impartiality." The war'
" between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The inquiry"
" will include alleged war crimes committed since June. The International Criminal Court was set up in"
" 2002 to prosecute genocide, crimes against humanity and war crimes."
)
EXPECTED = (
" The International Criminal Court (ICC) has announced that it is to be investigated by the International"
" Criminal Court (ICC) over allegations of war crimes."
)
dct = self.tok(ARTICLE, return_tensors="tf")
generated_ids = model.generate(**dct, num_beams=4, no_repeat_ngram_size=0)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)[0]
assert result == EXPECTED
xla_generate = tf.function(model.generate, jit_compile=True)
generated_ids = xla_generate(**dct, num_beams=4, no_repeat_ngram_size=0)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)[0]
assert result == EXPECTED
def test_xsum_1_1_batch_generation(self):
batch = self.tok(
[
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories."
" The formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is"
" based. The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted"
' its jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including'
' East Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination'
" into the situation in Palestinian territories, paving the way for possible war crimes investigations"
" against Israelis. As members of the court, Palestinians may be subject to counter-charges as well."
" Israel and the United States, neither of which is an ICC member, opposed the Palestinians' efforts"
" to join the body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony,"
' said it was a move toward greater justice. "As Palestine formally becomes a State Party to the Rome'
' Statute today, the world is also a step closer to ending a long era of impunity and injustice," he'
' said, according to an ICC news release. "Indeed, today brings us closer to our shared goals of'
' justice and peace." Judge Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was'
' just the first step for the Palestinians. "As the Rome Statute today enters into force for the State'
" of Palestine, Palestine acquires all the rights as well as responsibilities that come with being a"
' State Party to the Statute. These are substantive commitments, which cannot be taken lightly," she'
' said. Rights group Human Rights Watch welcomed the development. "Governments seeking to penalize'
" Palestine for joining the ICC should immediately end their pressure, and countries that support"
" universal acceptance of the court's treaty should speak out to welcome its membership,\" said"
" Balkees Jarrah, international justice counsel for the group. \"What's objectionable is the attempts"
" to undermine international justice, not Palestine's decision to join a treaty to which over 100"
' countries around the world are members." In January, when the preliminary ICC examination was'
" opened, Israeli Prime Minister Benjamin Netanyahu described it as an outrage, saying the court was"
' overstepping its boundaries. The United States also said it "strongly" disagreed with the court\'s'
' decision. "As we have said repeatedly, we do not believe that Palestine is a state and therefore we'
' do not believe that it is eligible to join the ICC," the State Department said in a statement. It'
' urged the warring sides to resolve their differences through direct negotiations. "We will continue'
' to oppose actions against Israel at the ICC as counterproductive to the cause of peace," it said.'
" But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows'
" the court to review evidence and determine whether to investigate suspects on both sides. Prosecutor"
' Fatou Bensouda said her office would "conduct its analysis in full independence and impartiality."'
" The war between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The"
" inquiry will include alleged war crimes committed since June. The International Criminal Court was"
" set up in 2002 to prosecute genocide, crimes against humanity and war crimes.",
"The French prosecutor leading an investigation into the crash of Germanwings Flight 9525 insisted"
" Wednesday that he was not aware of any video footage from on board the plane. Marseille prosecutor"
' Brice Robin told CNN that "so far no videos were used in the crash investigation." He added, "A'
" person who has such a video needs to immediately give it to the investigators.\" Robin's comments"
" follow claims by two magazines, German daily Bild and French Paris Match, of a cell phone video"
" showing the harrowing final seconds from on board Germanwings Flight 9525 as it crashed into the"
" French Alps. All 150 on board were killed. Paris Match and Bild reported that the video was"
" recovered from a phone at the wreckage site. The two publications described the supposed video, but"
" did not post it on their websites. The publications said that they watched the video, which was"
" found by a source close to the investigation. \"One can hear cries of 'My God' in several"
' languages," Paris Match reported. "Metallic banging can also be heard more than three times, perhaps'
" of the pilot trying to open the cockpit door with a heavy object. Towards the end, after a heavy"
' shake, stronger than the others, the screaming intensifies. Then nothing." "It is a very disturbing'
" scene,\" said Julian Reichelt, editor-in-chief of Bild online. An official with France's accident"
" investigation agency, the BEA, said the agency is not aware of any such video. Lt. Col. Jean-Marc"
" Menichini, a French Gendarmerie spokesman in charge of communications on rescue efforts around the"
' Germanwings crash site, told CNN that the reports were "completely wrong" and "unwarranted." Cell'
' phones have been collected at the site, he said, but that they "hadn\'t been exploited yet."'
" Menichini said he believed the cell phones would need to be sent to the Criminal Research Institute"
" in Rosny sous-Bois, near Paris, in order to be analyzed by specialized technicians working"
" hand-in-hand with investigators. But none of the cell phones found so far have been sent to the"
" institute, Menichini said. Asked whether staff involved in the search could have leaked a memory"
' card to the media, Menichini answered with a categorical "no." Reichelt told "Erin Burnett:'
' Outfront" that he had watched the video and stood by the report, saying Bild and Paris Match are'
' "very confident" that the clip is real. He noted that investigators only revealed they\'d recovered'
' cell phones from the crash site after Bild and Paris Match published their reports. "That is'
" something we did not know before. ... Overall we can say many things of the investigation weren't"
' revealed by the investigation at the beginning," he said. What was mental state of Germanwings'
" co-pilot? German airline Lufthansa confirmed Tuesday that co-pilot Andreas Lubitz had battled"
" depression years before he took the controls of Germanwings Flight 9525, which he's accused of"
" deliberately crashing last week in the French Alps. Lubitz told his Lufthansa flight training school"
' in 2009 that he had a "previous episode of severe depression," the airline said Tuesday. Email'
" correspondence between Lubitz and the school discovered in an internal investigation, Lufthansa"
" said, included medical documents he submitted in connection with resuming his flight training. The"
" announcement indicates that Lufthansa, the parent company of Germanwings, knew of Lubitz's battle"
" with depression, allowed him to continue training and ultimately put him in the cockpit. Lufthansa,"
" whose CEO Carsten Spohr previously said Lubitz was 100% fit to fly, described its statement Tuesday"
' as a "swift and seamless clarification" and said it was sharing the information and documents --'
" including training and medical records -- with public prosecutors. Spohr traveled to the crash site"
" Wednesday, where recovery teams have been working for the past week to recover human remains and"
" plane debris scattered across a steep mountainside. He saw the crisis center set up in"
" Seyne-les-Alpes, laid a wreath in the village of Le Vernet, closer to the crash site, where grieving"
" families have left flowers at a simple stone memorial. Menichini told CNN late Tuesday that no"
" visible human remains were left at the site but recovery teams would keep searching. French"
" President Francois Hollande, speaking Tuesday, said that it should be possible to identify all the"
" victims using DNA analysis by the end of the week, sooner than authorities had previously suggested."
" In the meantime, the recovery of the victims' personal belongings will start Wednesday, Menichini"
" said. Among those personal belongings could be more cell phones belonging to the 144 passengers and"
" six crew on board. Check out the latest from our correspondents . The details about Lubitz's"
" correspondence with the flight school during his training were among several developments as"
" investigators continued to delve into what caused the crash and Lubitz's possible motive for"
" downing the jet. A Lufthansa spokesperson told CNN on Tuesday that Lubitz had a valid medical"
' certificate, had passed all his examinations and "held all the licenses required." Earlier, a'
" spokesman for the prosecutor's office in Dusseldorf, Christoph Kumpa, said medical records reveal"
" Lubitz suffered from suicidal tendencies at some point before his aviation career and underwent"
" psychotherapy before he got his pilot's license. Kumpa emphasized there's no evidence suggesting"
" Lubitz was suicidal or acting aggressively before the crash. Investigators are looking into whether"
" Lubitz feared his medical condition would cause him to lose his pilot's license, a European"
' government official briefed on the investigation told CNN on Tuesday. While flying was "a big part'
" of his life,\" the source said, it's only one theory being considered. Another source, a law"
" enforcement official briefed on the investigation, also told CNN that authorities believe the"
" primary motive for Lubitz to bring down the plane was that he feared he would not be allowed to fly"
" because of his medical problems. Lubitz's girlfriend told investigators he had seen an eye doctor"
" and a neuropsychologist, both of whom deemed him unfit to work recently and concluded he had"
" psychological issues, the European government official said. But no matter what details emerge about"
" his previous mental health struggles, there's more to the story, said Brian Russell, a forensic"
' psychologist. "Psychology can explain why somebody would turn rage inward on themselves about the'
" fact that maybe they weren't going to keep doing their job and they're upset about that and so"
' they\'re suicidal," he said. "But there is no mental illness that explains why somebody then feels'
" entitled to also take that rage and turn it outward on 149 other people who had nothing to do with"
" the person's problems.\" Germanwings crash compensation: What we know . Who was the captain of"
" Germanwings Flight 9525? CNN's Margot Haddad reported from Marseille and Pamela Brown from"
" Dusseldorf, while Laura Smith-Spark wrote from London. CNN's Frederik Pleitgen, Pamela Boykoff,"
" Antonia Mortensen, Sandrine Amiel and Anna-Maja Rappard contributed to this report.",
],
return_tensors="tf",
padding="longest",
truncation=True,
)
generated_ids = self.xsum_1_1_model.generate(**batch, num_beams=4)
result = self.tok.batch_decode(generated_ids, skip_special_tokens=True)
assert (
result[0]
== " The International Criminal Court (ICC) has announced that it has been announced by the International"
" Criminal court."
)
assert (
result[1]
== " An investigation into the crash that killed at least 10 people in the French capital has been"
" released by the French police investigating the crash."
)
def test_encoder_equiv(self):
batch = self.tok(
[
"The Palestinian Authority officially became the 123rd member of the International Criminal Court on"
" Wednesday, a step that gives the court jurisdiction over alleged crimes in Palestinian territories."
" The formal accession was marked with a ceremony at The Hague, in the Netherlands, where the court is"
" based. The Palestinians signed the ICC's founding Rome Statute in January, when they also accepted"
' its jurisdiction over alleged crimes committed "in the occupied Palestinian territory, including'
' East Jerusalem, since June 13, 2014." Later that month, the ICC opened a preliminary examination'
" into the situation in Palestinian territories, paving the way for possible war crimes investigations"
" against Israelis. As members of the court, Palestinians may be subject to counter-charges as well."
" Israel and the United States, neither of which is an ICC member, opposed the Palestinians' efforts"
" to join the body. But Palestinian Foreign Minister Riad al-Malki, speaking at Wednesday's ceremony,"
' said it was a move toward greater justice. "As Palestine formally becomes a State Party to the Rome'
' Statute today, the world is also a step closer to ending a long era of impunity and injustice," he'
' said, according to an ICC news release. "Indeed, today brings us closer to our shared goals of'
' justice and peace." Judge Kuniko Ozaki, a vice president of the ICC, said acceding to the treaty was'
' just the first step for the Palestinians. "As the Rome Statute today enters into force for the State'
" of Palestine, Palestine acquires all the rights as well as responsibilities that come with being a"
' State Party to the Statute. These are substantive commitments, which cannot be taken lightly," she'
' said. Rights group Human Rights Watch welcomed the development. "Governments seeking to penalize'
" Palestine for joining the ICC should immediately end their pressure, and countries that support"
" universal acceptance of the court's treaty should speak out to welcome its membership,\" said"
" Balkees Jarrah, international justice counsel for the group. \"What's objectionable is the attempts"
" to undermine international justice, not Palestine's decision to join a treaty to which over 100"
' countries around the world are members." In January, when the preliminary ICC examination was'
" opened, Israeli Prime Minister Benjamin Netanyahu described it as an outrage, saying the court was"
' overstepping its boundaries. The United States also said it "strongly" disagreed with the court\'s'
' decision. "As we have said repeatedly, we do not believe that Palestine is a state and therefore we'
' do not believe that it is eligible to join the ICC," the State Department said in a statement. It'
' urged the warring sides to resolve their differences through direct negotiations. "We will continue'
' to oppose actions against Israel at the ICC as counterproductive to the cause of peace," it said.'
" But the ICC begs to differ with the definition of a state for its purposes and refers to the"
' territories as "Palestine." While a preliminary examination is not a formal investigation, it allows'
" the court to review evidence and determine whether to investigate suspects on both sides. Prosecutor"
' Fatou Bensouda said her office would "conduct its analysis in full independence and impartiality."'
" The war between Israel and Hamas militants in Gaza last summer left more than 2,000 people dead. The"
" inquiry will include alleged war crimes committed since June. The International Criminal Court was"
" set up in 2002 to prosecute genocide, crimes against humanity and war crimes.",
"The French prosecutor leading an investigation into the crash of Germanwings Flight 9525 insisted"
" Wednesday that he was not aware of any video footage from on board the plane. Marseille prosecutor"
' Brice Robin told CNN that "so far no videos were used in the crash investigation." He added, "A'
" person who has such a video needs to immediately give it to the investigators.\" Robin's comments"
" follow claims by two magazines, German daily Bild and French Paris Match, of a cell phone video"
" showing the harrowing final seconds from on board Germanwings Flight 9525 as it crashed into the"
" French Alps. All 150 on board were killed. Paris Match and Bild reported that the video was"
" recovered from a phone at the wreckage site. The two publications described the supposed video, but"
" did not post it on their websites. The publications said that they watched the video, which was"
" found by a source close to the investigation. \"One can hear cries of 'My God' in several"
' languages," Paris Match reported. "Metallic banging can also be heard more than three times, perhaps'
" of the pilot trying to open the cockpit door with a heavy object. Towards the end, after a heavy"
' shake, stronger than the others, the screaming intensifies. Then nothing." "It is a very disturbing'
" scene,\" said Julian Reichelt, editor-in-chief of Bild online. An official with France's accident"
" investigation agency, the BEA, said the agency is not aware of any such video. Lt. Col. Jean-Marc"
" Menichini, a French Gendarmerie spokesman in charge of communications on rescue efforts around the"
' Germanwings crash site, told CNN that the reports were "completely wrong" and "unwarranted." Cell'
' phones have been collected at the site, he said, but that they "hadn\'t been exploited yet."'
" Menichini said he believed the cell phones would need to be sent to the Criminal Research Institute"
" in Rosny sous-Bois, near Paris, in order to be analyzed by specialized technicians working"
" hand-in-hand with investigators. But none of the cell phones found so far have been sent to the"
" institute, Menichini said. Asked whether staff involved in the search could have leaked a memory"
' card to the media, Menichini answered with a categorical "no." Reichelt told "Erin Burnett:'
' Outfront" that he had watched the video and stood by the report, saying Bild and Paris Match are'
' "very confident" that the clip is real. He noted that investigators only revealed they\'d recovered'
' cell phones from the crash site after Bild and Paris Match published their reports. "That is'
" something we did not know before. ... Overall we can say many things of the investigation weren't"
' revealed by the investigation at the beginning," he said. What was mental state of Germanwings'
" co-pilot? German airline Lufthansa confirmed Tuesday that co-pilot Andreas Lubitz had battled"
" depression years before he took the controls of Germanwings Flight 9525, which he's accused of"
" deliberately crashing last week in the French Alps. Lubitz told his Lufthansa flight training school"
' in 2009 that he had a "previous episode of severe depression," the airline said Tuesday. Email'
" correspondence between Lubitz and the school discovered in an internal investigation, Lufthansa"
" said, included medical documents he submitted in connection with resuming his flight training. The"
" announcement indicates that Lufthansa, the parent company of Germanwings, knew of Lubitz's battle"
" with depression, allowed him to continue training and ultimately put him in the cockpit. Lufthansa,"
" whose CEO Carsten Spohr previously said Lubitz was 100% fit to fly, described its statement Tuesday"
' as a "swift and seamless clarification" and said it was sharing the information and documents --'
" including training and medical records -- with public prosecutors. Spohr traveled to the crash site"
" Wednesday, where recovery teams have been working for the past week to recover human remains and"
" plane debris scattered across a steep mountainside. He saw the crisis center set up in"
" Seyne-les-Alpes, laid a wreath in the village of Le Vernet, closer to the crash site, where grieving"
" families have left flowers at a simple stone memorial. Menichini told CNN late Tuesday that no"
" visible human remains were left at the site but recovery teams would keep searching. French"
" President Francois Hollande, speaking Tuesday, said that it should be possible to identify all the"
" victims using DNA analysis by the end of the week, sooner than authorities had previously suggested."
" In the meantime, the recovery of the victims' personal belongings will start Wednesday, Menichini"
" said. Among those personal belongings could be more cell phones belonging to the 144 passengers and"
" six crew on board. Check out the latest from our correspondents . The details about Lubitz's"
" correspondence with the flight school during his training were among several developments as"
" investigators continued to delve into what caused the crash and Lubitz's possible motive for"
" downing the jet. A Lufthansa spokesperson told CNN on Tuesday that Lubitz had a valid medical"
' certificate, had passed all his examinations and "held all the licenses required." Earlier, a'
" spokesman for the prosecutor's office in Dusseldorf, Christoph Kumpa, said medical records reveal"
" Lubitz suffered from suicidal tendencies at some point before his aviation career and underwent"
" psychotherapy before he got his pilot's license. Kumpa emphasized there's no evidence suggesting"
" Lubitz was suicidal or acting aggressively before the crash. Investigators are looking into whether"
" Lubitz feared his medical condition would cause him to lose his pilot's license, a European"
' government official briefed on the investigation told CNN on Tuesday. While flying was "a big part'
" of his life,\" the source said, it's only one theory being considered. Another source, a law"
" enforcement official briefed on the investigation, also told CNN that authorities believe the"
" primary motive for Lubitz to bring down the plane was that he feared he would not be allowed to fly"
" because of his medical problems. Lubitz's girlfriend told investigators he had seen an eye doctor"
" and a neuropsychologist, both of whom deemed him unfit to work recently and concluded he had"
" psychological issues, the European government official said. But no matter what details emerge about"
" his previous mental health struggles, there's more to the story, said Brian Russell, a forensic"
' psychologist. "Psychology can explain why somebody would turn rage inward on themselves about the'
" fact that maybe they weren't going to keep doing their job and they're upset about that and so"
' they\'re suicidal," he said. "But there is no mental illness that explains why somebody then feels'
" entitled to also take that rage and turn it outward on 149 other people who had nothing to do with"
" the person's problems.\" Germanwings crash compensation: What we know . Who was the captain of"
" Germanwings Flight 9525? CNN's Margot Haddad reported from Marseille and Pamela Brown from"
" Dusseldorf, while Laura Smith-Spark wrote from London. CNN's Frederik Pleitgen, Pamela Boykoff,"
" Antonia Mortensen, Sandrine Amiel and Anna-Maja Rappard contributed to this report.",
],
return_tensors="tf",
padding="longest",
truncation=True,
)
features = self.xsum_1_1_model.get_encoder()(**batch).last_hidden_state
expected = np.array([[-0.0828, -0.0251, -0.0674], [0.1277, 0.3311, -0.0255], [0.2613, -0.0840, -0.2763]])
assert np.allclose(features[0, :3, :3].numpy(), expected, atol=1e-3)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/megatron_gpt2/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/xlm_prophetnet/configuration_xlm_prophetnet.py | # coding=utf-8
# Copyright 2020 The Microsoft Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" XLM-ProphetNet model configuration"""
from typing import Callable, Optional, Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"microsoft/xprophetnet-large-wiki100-cased": (
"https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json"
),
}
class XLMProphetNetConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`XLMProphetNetModel`]. It is used to instantiate a
XLMProphetNet model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the XLMProphetNet
[microsoft/xprophetnet-large-wiki100-cased](https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased)
architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
activation_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for activations inside the fully connected layer.
activation_function (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
vocab_size (`int`, *optional*, defaults to 30522):
Vocabulary size of the ProphetNET model. Defines the number of different tokens that can be represented by
the `inputs_ids` passed when calling [`XLMProphetNetModel`].
hidden_size (`int`, *optional*, defaults to 1024):
Dimensionality of the layers and the pooler layer.
encoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimensionality of the "intermediate" (often named feed-forward) layer in decoder.
num_encoder_layers (`int`, *optional*, defaults to 12):
Number of encoder layers.
num_encoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
decoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimensionality of the `intermediate` (often named feed-forward) layer in decoder.
num_decoder_layers (`int`, *optional*, defaults to 12):
Number of decoder layers.
num_decoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer decoder.
attention_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
max_position_embeddings (`int`, *optional*, defaults to 512):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
init_std (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
add_cross_attention (`bool`, *optional*, defaults to `True`):
Whether cross-attention layers should be added to the model.
is_encoder_decoder (`bool`, *optional*, defaults to `True`):
Whether this is an encoder/decoder model.
pad_token_id (`int`, *optional*, defaults to 1)
Padding token id.
bos_token_id (`int`, *optional*, defaults to 0)
Beginning of stream token id.
eos_token_id (`int`, *optional*, defaults to 2)
End of stream token id.
ngram (`int`, *optional*, defaults to 2)
Number of future tokens to predict. Set to 1 to be same as traditional Language model to predict next first
token.
num_buckets (`int`, *optional*, defaults to 32)
The number of buckets to use for each attention layer. This is for relative position calculation. See the
[T5 paper](see https://arxiv.org/abs/1910.10683) for more details.
relative_max_distance (`int`, *optional*, defaults to 128)
Relative distances greater than this number will be put into the last same bucket. This is for relative
position calculation. See the [T5 paper](see https://arxiv.org/abs/1910.10683) for more details.
disable_ngram_loss (`bool`, *optional*, defaults to `False`):
Whether be trained predicting only the next first token.
eps (`float`, *optional*, defaults to 0.0):
Controls the `epsilon` parameter value for label smoothing in the loss calculation. If set to 0, no label
smoothing is performed.
use_cache (`bool`, *optional*, defaults to `True`):
Whether or not the model should return the last key/values attentions (not used by all models).
"""
model_type = "xlm-prophetnet"
keys_to_ignore_at_inference = ["past_key_values"]
attribute_map = {
"num_attention_heads": "num_encoder_attention_heads",
}
def __init__(
self,
activation_dropout: Optional[float] = 0.1,
activation_function: Optional[Union[str, Callable]] = "gelu",
vocab_size: Optional[int] = 30522,
hidden_size: Optional[int] = 1024,
encoder_ffn_dim: Optional[int] = 4096,
num_encoder_layers: Optional[int] = 12,
num_encoder_attention_heads: Optional[int] = 16,
decoder_ffn_dim: Optional[int] = 4096,
num_decoder_layers: Optional[int] = 12,
num_decoder_attention_heads: Optional[int] = 16,
attention_dropout: Optional[float] = 0.1,
dropout: Optional[float] = 0.1,
max_position_embeddings: Optional[int] = 512,
init_std: Optional[float] = 0.02,
is_encoder_decoder: Optional[bool] = True,
add_cross_attention: Optional[bool] = True,
decoder_start_token_id: Optional[int] = 0,
ngram: Optional[int] = 2,
num_buckets: Optional[int] = 32,
relative_max_distance: Optional[int] = 128,
disable_ngram_loss: Optional[bool] = False,
eps: Optional[float] = 0.0,
use_cache: Optional[bool] = True,
pad_token_id: Optional[int] = 0,
bos_token_id: Optional[int] = 1,
eos_token_id: Optional[int] = 2,
**kwargs,
):
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.encoder_ffn_dim = encoder_ffn_dim
self.num_encoder_layers = num_encoder_layers
self.num_encoder_attention_heads = num_encoder_attention_heads
self.decoder_ffn_dim = decoder_ffn_dim
self.num_decoder_layers = num_decoder_layers
self.num_decoder_attention_heads = num_decoder_attention_heads
self.max_position_embeddings = max_position_embeddings
self.init_std = init_std # Normal(0, this parameter)
self.activation_function = activation_function
# parameters for xlmprophetnet
self.ngram = ngram
self.num_buckets = num_buckets
self.relative_max_distance = relative_max_distance
self.disable_ngram_loss = disable_ngram_loss
self.eps = eps
# 3 Types of Dropout
self.attention_dropout = attention_dropout
self.activation_dropout = activation_dropout
self.dropout = dropout
self.use_cache = use_cache
super().__init__(
pad_token_id=pad_token_id,
bos_token_id=bos_token_id,
eos_token_id=eos_token_id,
is_encoder_decoder=is_encoder_decoder,
add_cross_attention=add_cross_attention,
decoder_start_token_id=decoder_start_token_id,
**kwargs,
)
@property
def num_hidden_layers(self) -> int:
return self.num_encoder_layers + self.num_decoder_layers
@num_hidden_layers.setter
def num_hidden_layers(self, value):
raise NotImplementedError(
"This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and"
" `num_decoder_layers`."
)
| # coding=utf-8
# Copyright 2020 The Microsoft Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" XLM-ProphetNet model configuration"""
from typing import Callable, Optional, Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"microsoft/xprophetnet-large-wiki100-cased": (
"https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json"
),
}
class XLMProphetNetConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`XLMProphetNetModel`]. It is used to instantiate a
XLMProphetNet model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the XLMProphetNet
[microsoft/xprophetnet-large-wiki100-cased](https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased)
architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
activation_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for activations inside the fully connected layer.
activation_function (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
vocab_size (`int`, *optional*, defaults to 30522):
Vocabulary size of the ProphetNET model. Defines the number of different tokens that can be represented by
the `inputs_ids` passed when calling [`XLMProphetNetModel`].
hidden_size (`int`, *optional*, defaults to 1024):
Dimensionality of the layers and the pooler layer.
encoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimensionality of the "intermediate" (often named feed-forward) layer in decoder.
num_encoder_layers (`int`, *optional*, defaults to 12):
Number of encoder layers.
num_encoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
decoder_ffn_dim (`int`, *optional*, defaults to 4096):
Dimensionality of the `intermediate` (often named feed-forward) layer in decoder.
num_decoder_layers (`int`, *optional*, defaults to 12):
Number of decoder layers.
num_decoder_attention_heads (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer decoder.
attention_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
max_position_embeddings (`int`, *optional*, defaults to 512):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 512 or 1024 or 2048).
init_std (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
add_cross_attention (`bool`, *optional*, defaults to `True`):
Whether cross-attention layers should be added to the model.
is_encoder_decoder (`bool`, *optional*, defaults to `True`):
Whether this is an encoder/decoder model.
pad_token_id (`int`, *optional*, defaults to 1)
Padding token id.
bos_token_id (`int`, *optional*, defaults to 0)
Beginning of stream token id.
eos_token_id (`int`, *optional*, defaults to 2)
End of stream token id.
ngram (`int`, *optional*, defaults to 2)
Number of future tokens to predict. Set to 1 to be same as traditional Language model to predict next first
token.
num_buckets (`int`, *optional*, defaults to 32)
The number of buckets to use for each attention layer. This is for relative position calculation. See the
[T5 paper](see https://arxiv.org/abs/1910.10683) for more details.
relative_max_distance (`int`, *optional*, defaults to 128)
Relative distances greater than this number will be put into the last same bucket. This is for relative
position calculation. See the [T5 paper](see https://arxiv.org/abs/1910.10683) for more details.
disable_ngram_loss (`bool`, *optional*, defaults to `False`):
Whether be trained predicting only the next first token.
eps (`float`, *optional*, defaults to 0.0):
Controls the `epsilon` parameter value for label smoothing in the loss calculation. If set to 0, no label
smoothing is performed.
use_cache (`bool`, *optional*, defaults to `True`):
Whether or not the model should return the last key/values attentions (not used by all models).
"""
model_type = "xlm-prophetnet"
keys_to_ignore_at_inference = ["past_key_values"]
attribute_map = {
"num_attention_heads": "num_encoder_attention_heads",
}
def __init__(
self,
activation_dropout: Optional[float] = 0.1,
activation_function: Optional[Union[str, Callable]] = "gelu",
vocab_size: Optional[int] = 30522,
hidden_size: Optional[int] = 1024,
encoder_ffn_dim: Optional[int] = 4096,
num_encoder_layers: Optional[int] = 12,
num_encoder_attention_heads: Optional[int] = 16,
decoder_ffn_dim: Optional[int] = 4096,
num_decoder_layers: Optional[int] = 12,
num_decoder_attention_heads: Optional[int] = 16,
attention_dropout: Optional[float] = 0.1,
dropout: Optional[float] = 0.1,
max_position_embeddings: Optional[int] = 512,
init_std: Optional[float] = 0.02,
is_encoder_decoder: Optional[bool] = True,
add_cross_attention: Optional[bool] = True,
decoder_start_token_id: Optional[int] = 0,
ngram: Optional[int] = 2,
num_buckets: Optional[int] = 32,
relative_max_distance: Optional[int] = 128,
disable_ngram_loss: Optional[bool] = False,
eps: Optional[float] = 0.0,
use_cache: Optional[bool] = True,
pad_token_id: Optional[int] = 0,
bos_token_id: Optional[int] = 1,
eos_token_id: Optional[int] = 2,
**kwargs,
):
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.encoder_ffn_dim = encoder_ffn_dim
self.num_encoder_layers = num_encoder_layers
self.num_encoder_attention_heads = num_encoder_attention_heads
self.decoder_ffn_dim = decoder_ffn_dim
self.num_decoder_layers = num_decoder_layers
self.num_decoder_attention_heads = num_decoder_attention_heads
self.max_position_embeddings = max_position_embeddings
self.init_std = init_std # Normal(0, this parameter)
self.activation_function = activation_function
# parameters for xlmprophetnet
self.ngram = ngram
self.num_buckets = num_buckets
self.relative_max_distance = relative_max_distance
self.disable_ngram_loss = disable_ngram_loss
self.eps = eps
# 3 Types of Dropout
self.attention_dropout = attention_dropout
self.activation_dropout = activation_dropout
self.dropout = dropout
self.use_cache = use_cache
super().__init__(
pad_token_id=pad_token_id,
bos_token_id=bos_token_id,
eos_token_id=eos_token_id,
is_encoder_decoder=is_encoder_decoder,
add_cross_attention=add_cross_attention,
decoder_start_token_id=decoder_start_token_id,
**kwargs,
)
@property
def num_hidden_layers(self) -> int:
return self.num_encoder_layers + self.num_decoder_layers
@num_hidden_layers.setter
def num_hidden_layers(self, value):
raise NotImplementedError(
"This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and"
" `num_decoder_layers`."
)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./.github/workflows/add-model-like.yml | name: Add model like runner
on:
push:
branches:
- main
pull_request:
paths:
- "src/**"
- "tests/**"
- ".github/**"
types: [opened, synchronize, reopened]
jobs:
run_tests_templates_like:
name: "Add new model like template tests"
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Install dependencies
run: |
sudo apt -y update && sudo apt install -y libsndfile1-dev
- name: Load cached virtual environment
uses: actions/cache@v2
id: cache
with:
path: ~/venv/
key: v4-tests_model_like-${{ hashFiles('setup.py') }}
- name: Create virtual environment on cache miss
if: steps.cache.outputs.cache-hit != 'true'
run: |
python -m venv ~/venv && . ~/venv/bin/activate
pip install --upgrade pip!=21.3
pip install -e .[dev]
- name: Check transformers location
# make `transformers` available as package (required since we use `-e` flag) and check it's indeed from the repo.
run: |
. ~/venv/bin/activate
python setup.py develop
transformers_install=$(pip list -e | grep transformers)
transformers_install_array=($transformers_install)
transformers_loc=${transformers_install_array[-1]}
transformers_repo_loc=$(pwd .)
if [ "$transformers_loc" != "$transformers_repo_loc" ]; then
echo "transformers is from $transformers_loc but it shoud be from $transformers_repo_loc/src."
echo "A fix is required. Stop testing."
exit 1
fi
- name: Create model files
run: |
. ~/venv/bin/activate
transformers-cli add-new-model-like --config_file tests/fixtures/add_distilbert_like_config.json --path_to_repo .
make style
make fix-copies
- name: Run all PyTorch modeling test
run: |
. ~/venv/bin/activate
python -m pytest -n 2 --dist=loadfile -s --make-reports=tests_new_models tests/bert_new/test_modeling_bert_new.py
- name: Run style changes
run: |
. ~/venv/bin/activate
make style && make quality && make repo-consistency
- name: Failure short reports
if: ${{ always() }}
run: cat reports/tests_new_models/failures_short.txt
- name: Test suite reports artifacts
if: ${{ always() }}
uses: actions/upload-artifact@v3
with:
name: run_all_tests_new_models_test_reports
path: reports/tests_new_models
| name: Add model like runner
on:
push:
branches:
- main
pull_request:
paths:
- "src/**"
- "tests/**"
- ".github/**"
types: [opened, synchronize, reopened]
jobs:
run_tests_templates_like:
name: "Add new model like template tests"
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Install dependencies
run: |
sudo apt -y update && sudo apt install -y libsndfile1-dev
- name: Load cached virtual environment
uses: actions/cache@v2
id: cache
with:
path: ~/venv/
key: v4-tests_model_like-${{ hashFiles('setup.py') }}
- name: Create virtual environment on cache miss
if: steps.cache.outputs.cache-hit != 'true'
run: |
python -m venv ~/venv && . ~/venv/bin/activate
pip install --upgrade pip!=21.3
pip install -e .[dev]
- name: Check transformers location
# make `transformers` available as package (required since we use `-e` flag) and check it's indeed from the repo.
run: |
. ~/venv/bin/activate
python setup.py develop
transformers_install=$(pip list -e | grep transformers)
transformers_install_array=($transformers_install)
transformers_loc=${transformers_install_array[-1]}
transformers_repo_loc=$(pwd .)
if [ "$transformers_loc" != "$transformers_repo_loc" ]; then
echo "transformers is from $transformers_loc but it shoud be from $transformers_repo_loc/src."
echo "A fix is required. Stop testing."
exit 1
fi
- name: Create model files
run: |
. ~/venv/bin/activate
transformers-cli add-new-model-like --config_file tests/fixtures/add_distilbert_like_config.json --path_to_repo .
make style
make fix-copies
- name: Run all PyTorch modeling test
run: |
. ~/venv/bin/activate
python -m pytest -n 2 --dist=loadfile -s --make-reports=tests_new_models tests/bert_new/test_modeling_bert_new.py
- name: Run style changes
run: |
. ~/venv/bin/activate
make style && make quality && make repo-consistency
- name: Failure short reports
if: ${{ always() }}
run: cat reports/tests_new_models/failures_short.txt
- name: Test suite reports artifacts
if: ${{ always() }}
uses: actions/upload-artifact@v3
with:
name: run_all_tests_new_models_test_reports
path: reports/tests_new_models
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/pytorch/audio-classification/run_audio_classification.py | #!/usr/bin/env python
# coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
import os
import sys
import warnings
from dataclasses import dataclass, field
from random import randint
from typing import Optional
import datasets
import evaluate
import numpy as np
from datasets import DatasetDict, load_dataset
import transformers
from transformers import (
AutoConfig,
AutoFeatureExtractor,
AutoModelForAudioClassification,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version, send_example_telemetry
from transformers.utils.versions import require_version
logger = logging.getLogger(__name__)
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.27.0.dev0")
require_version("datasets>=1.14.0", "To fix: pip install -r examples/pytorch/audio-classification/requirements.txt")
def random_subsample(wav: np.ndarray, max_length: float, sample_rate: int = 16000):
"""Randomly sample chunks of `max_length` seconds from the input audio"""
sample_length = int(round(sample_rate * max_length))
if len(wav) <= sample_length:
return wav
random_offset = randint(0, len(wav) - sample_length - 1)
return wav[random_offset : random_offset + sample_length]
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
Using `HfArgumentParser` we can turn this class
into argparse arguments to be able to specify them on
the command line.
"""
dataset_name: Optional[str] = field(default=None, metadata={"help": "Name of a dataset from the datasets package"})
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
train_file: Optional[str] = field(
default=None, metadata={"help": "A file containing the training audio paths and labels."}
)
eval_file: Optional[str] = field(
default=None, metadata={"help": "A file containing the validation audio paths and labels."}
)
train_split_name: str = field(
default="train",
metadata={
"help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
},
)
eval_split_name: str = field(
default="validation",
metadata={
"help": (
"The name of the training data set split to use (via the datasets library). Defaults to 'validation'"
)
},
)
audio_column_name: str = field(
default="audio",
metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
)
label_column_name: str = field(
default="label", metadata={"help": "The name of the dataset column containing the labels. Defaults to 'label'"}
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_length_seconds: float = field(
default=20,
metadata={"help": "Audio clips will be randomly cut to this length during training if the value is set."},
)
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
"""
model_name_or_path: str = field(
default="facebook/wav2vec2-base",
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"},
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None, metadata={"help": "Where do you want to store the pretrained models downloaded from the Hub"}
)
model_revision: str = field(
default="main",
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
)
feature_extractor_name: Optional[str] = field(
default=None, metadata={"help": "Name or path of preprocessor config."}
)
freeze_feature_encoder: bool = field(
default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
)
attention_mask: bool = field(
default=True, metadata={"help": "Whether to generate an attention mask in the feature extractor."}
)
use_auth_token: bool = field(
default=False,
metadata={
"help": (
"Will use the token generated when running `huggingface-cli login` (necessary to use this script "
"with private models)."
)
},
)
freeze_feature_extractor: Optional[bool] = field(
default=None, metadata={"help": "Whether to freeze the feature extractor layers of the model."}
)
ignore_mismatched_sizes: bool = field(
default=False,
metadata={"help": "Will enable to load a pretrained model whose head dimensions are different."},
)
def __post_init__(self):
if not self.freeze_feature_extractor and self.freeze_feature_encoder:
warnings.warn(
"The argument `--freeze_feature_extractor` is deprecated and "
"will be removed in a future version. Use `--freeze_feature_encoder`"
"instead. Setting `freeze_feature_encoder==True`.",
FutureWarning,
)
if self.freeze_feature_extractor and not self.freeze_feature_encoder:
raise ValueError(
"The argument `--freeze_feature_extractor` is deprecated and "
"should not be used in combination with `--freeze_feature_encoder`."
"Only make use of `--freeze_feature_encoder`."
)
def main():
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
# information sent is the one passed as arguments along with your Python/PyTorch versions.
send_example_telemetry("run_audio_classification", model_args, data_args)
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
handlers=[logging.StreamHandler(sys.stdout)],
)
if training_args.should_log:
# The default of training_args.log_level is passive, so we set log level at info here to have that default.
transformers.utils.logging.set_verbosity_info()
log_level = training_args.get_process_log_level()
logger.setLevel(log_level)
transformers.utils.logging.set_verbosity(log_level)
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Log on each process the small summary:
logger.warning(
f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu} "
+ f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
)
logger.info(f"Training/evaluation parameters {training_args}")
# Set seed before initializing model.
set_seed(training_args.seed)
# Detecting last checkpoint.
last_checkpoint = None
if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
last_checkpoint = get_last_checkpoint(training_args.output_dir)
if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty. "
"Use --overwrite_output_dir to train from scratch."
)
elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
logger.info(
f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
"the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
)
# Initialize our dataset and prepare it for the audio classification task.
raw_datasets = DatasetDict()
raw_datasets["train"] = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
split=data_args.train_split_name,
use_auth_token=True if model_args.use_auth_token else None,
)
raw_datasets["eval"] = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
split=data_args.eval_split_name,
use_auth_token=True if model_args.use_auth_token else None,
)
if data_args.audio_column_name not in raw_datasets["train"].column_names:
raise ValueError(
f"--audio_column_name {data_args.audio_column_name} not found in dataset '{data_args.dataset_name}'. "
"Make sure to set `--audio_column_name` to the correct audio column - one of "
f"{', '.join(raw_datasets['train'].column_names)}."
)
if data_args.label_column_name not in raw_datasets["train"].column_names:
raise ValueError(
f"--label_column_name {data_args.label_column_name} not found in dataset '{data_args.dataset_name}'. "
"Make sure to set `--label_column_name` to the correct text column - one of "
f"{', '.join(raw_datasets['train'].column_names)}."
)
# Setting `return_attention_mask=True` is the way to get a correctly masked mean-pooling over
# transformer outputs in the classifier, but it doesn't always lead to better accuracy
feature_extractor = AutoFeatureExtractor.from_pretrained(
model_args.feature_extractor_name or model_args.model_name_or_path,
return_attention_mask=model_args.attention_mask,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
# `datasets` takes care of automatically loading and resampling the audio,
# so we just need to set the correct target sampling rate.
raw_datasets = raw_datasets.cast_column(
data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
)
model_input_name = feature_extractor.model_input_names[0]
def train_transforms(batch):
"""Apply train_transforms across a batch."""
subsampled_wavs = []
for audio in batch[data_args.audio_column_name]:
wav = random_subsample(
audio["array"], max_length=data_args.max_length_seconds, sample_rate=feature_extractor.sampling_rate
)
subsampled_wavs.append(wav)
inputs = feature_extractor(subsampled_wavs, sampling_rate=feature_extractor.sampling_rate)
output_batch = {model_input_name: inputs.get(model_input_name)}
output_batch["labels"] = list(batch[data_args.label_column_name])
return output_batch
def val_transforms(batch):
"""Apply val_transforms across a batch."""
wavs = [audio["array"] for audio in batch[data_args.audio_column_name]]
inputs = feature_extractor(wavs, sampling_rate=feature_extractor.sampling_rate)
output_batch = {model_input_name: inputs.get(model_input_name)}
output_batch["labels"] = list(batch[data_args.label_column_name])
return output_batch
# Prepare label mappings.
# We'll include these in the model's config to get human readable labels in the Inference API.
labels = raw_datasets["train"].features[data_args.label_column_name].names
label2id, id2label = {}, {}
for i, label in enumerate(labels):
label2id[label] = str(i)
id2label[str(i)] = label
# Load the accuracy metric from the datasets package
metric = evaluate.load("accuracy")
# Define our compute_metrics function. It takes an `EvalPrediction` object (a namedtuple with
# `predictions` and `label_ids` fields) and has to return a dictionary string to float.
def compute_metrics(eval_pred):
"""Computes accuracy on a batch of predictions"""
predictions = np.argmax(eval_pred.predictions, axis=1)
return metric.compute(predictions=predictions, references=eval_pred.label_ids)
config = AutoConfig.from_pretrained(
model_args.config_name or model_args.model_name_or_path,
num_labels=len(labels),
label2id=label2id,
id2label=id2label,
finetuning_task="audio-classification",
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
model = AutoModelForAudioClassification.from_pretrained(
model_args.model_name_or_path,
from_tf=bool(".ckpt" in model_args.model_name_or_path),
config=config,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
ignore_mismatched_sizes=model_args.ignore_mismatched_sizes,
)
# freeze the convolutional waveform encoder
if model_args.freeze_feature_encoder:
model.freeze_feature_encoder()
if training_args.do_train:
if data_args.max_train_samples is not None:
raw_datasets["train"] = (
raw_datasets["train"].shuffle(seed=training_args.seed).select(range(data_args.max_train_samples))
)
# Set the training transforms
raw_datasets["train"].set_transform(train_transforms, output_all_columns=False)
if training_args.do_eval:
if data_args.max_eval_samples is not None:
raw_datasets["eval"] = (
raw_datasets["eval"].shuffle(seed=training_args.seed).select(range(data_args.max_eval_samples))
)
# Set the validation transforms
raw_datasets["eval"].set_transform(val_transforms, output_all_columns=False)
# Initialize our trainer
trainer = Trainer(
model=model,
args=training_args,
train_dataset=raw_datasets["train"] if training_args.do_train else None,
eval_dataset=raw_datasets["eval"] if training_args.do_eval else None,
compute_metrics=compute_metrics,
tokenizer=feature_extractor,
)
# Training
if training_args.do_train:
checkpoint = None
if training_args.resume_from_checkpoint is not None:
checkpoint = training_args.resume_from_checkpoint
elif last_checkpoint is not None:
checkpoint = last_checkpoint
train_result = trainer.train(resume_from_checkpoint=checkpoint)
trainer.save_model()
trainer.log_metrics("train", train_result.metrics)
trainer.save_metrics("train", train_result.metrics)
trainer.save_state()
# Evaluation
if training_args.do_eval:
metrics = trainer.evaluate()
trainer.log_metrics("eval", metrics)
trainer.save_metrics("eval", metrics)
# Write model card and (optionally) push to hub
kwargs = {
"finetuned_from": model_args.model_name_or_path,
"tasks": "audio-classification",
"dataset": data_args.dataset_name,
"tags": ["audio-classification"],
}
if training_args.push_to_hub:
trainer.push_to_hub(**kwargs)
else:
trainer.create_model_card(**kwargs)
if __name__ == "__main__":
main()
| #!/usr/bin/env python
# coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
import os
import sys
import warnings
from dataclasses import dataclass, field
from random import randint
from typing import Optional
import datasets
import evaluate
import numpy as np
from datasets import DatasetDict, load_dataset
import transformers
from transformers import (
AutoConfig,
AutoFeatureExtractor,
AutoModelForAudioClassification,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version, send_example_telemetry
from transformers.utils.versions import require_version
logger = logging.getLogger(__name__)
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.27.0.dev0")
require_version("datasets>=1.14.0", "To fix: pip install -r examples/pytorch/audio-classification/requirements.txt")
def random_subsample(wav: np.ndarray, max_length: float, sample_rate: int = 16000):
"""Randomly sample chunks of `max_length` seconds from the input audio"""
sample_length = int(round(sample_rate * max_length))
if len(wav) <= sample_length:
return wav
random_offset = randint(0, len(wav) - sample_length - 1)
return wav[random_offset : random_offset + sample_length]
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
Using `HfArgumentParser` we can turn this class
into argparse arguments to be able to specify them on
the command line.
"""
dataset_name: Optional[str] = field(default=None, metadata={"help": "Name of a dataset from the datasets package"})
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
train_file: Optional[str] = field(
default=None, metadata={"help": "A file containing the training audio paths and labels."}
)
eval_file: Optional[str] = field(
default=None, metadata={"help": "A file containing the validation audio paths and labels."}
)
train_split_name: str = field(
default="train",
metadata={
"help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
},
)
eval_split_name: str = field(
default="validation",
metadata={
"help": (
"The name of the training data set split to use (via the datasets library). Defaults to 'validation'"
)
},
)
audio_column_name: str = field(
default="audio",
metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
)
label_column_name: str = field(
default="label", metadata={"help": "The name of the dataset column containing the labels. Defaults to 'label'"}
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_length_seconds: float = field(
default=20,
metadata={"help": "Audio clips will be randomly cut to this length during training if the value is set."},
)
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
"""
model_name_or_path: str = field(
default="facebook/wav2vec2-base",
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"},
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None, metadata={"help": "Where do you want to store the pretrained models downloaded from the Hub"}
)
model_revision: str = field(
default="main",
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
)
feature_extractor_name: Optional[str] = field(
default=None, metadata={"help": "Name or path of preprocessor config."}
)
freeze_feature_encoder: bool = field(
default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
)
attention_mask: bool = field(
default=True, metadata={"help": "Whether to generate an attention mask in the feature extractor."}
)
use_auth_token: bool = field(
default=False,
metadata={
"help": (
"Will use the token generated when running `huggingface-cli login` (necessary to use this script "
"with private models)."
)
},
)
freeze_feature_extractor: Optional[bool] = field(
default=None, metadata={"help": "Whether to freeze the feature extractor layers of the model."}
)
ignore_mismatched_sizes: bool = field(
default=False,
metadata={"help": "Will enable to load a pretrained model whose head dimensions are different."},
)
def __post_init__(self):
if not self.freeze_feature_extractor and self.freeze_feature_encoder:
warnings.warn(
"The argument `--freeze_feature_extractor` is deprecated and "
"will be removed in a future version. Use `--freeze_feature_encoder`"
"instead. Setting `freeze_feature_encoder==True`.",
FutureWarning,
)
if self.freeze_feature_extractor and not self.freeze_feature_encoder:
raise ValueError(
"The argument `--freeze_feature_extractor` is deprecated and "
"should not be used in combination with `--freeze_feature_encoder`."
"Only make use of `--freeze_feature_encoder`."
)
def main():
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
else:
model_args, data_args, training_args = parser.parse_args_into_dataclasses()
# Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
# information sent is the one passed as arguments along with your Python/PyTorch versions.
send_example_telemetry("run_audio_classification", model_args, data_args)
# Setup logging
logging.basicConfig(
format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
datefmt="%m/%d/%Y %H:%M:%S",
handlers=[logging.StreamHandler(sys.stdout)],
)
if training_args.should_log:
# The default of training_args.log_level is passive, so we set log level at info here to have that default.
transformers.utils.logging.set_verbosity_info()
log_level = training_args.get_process_log_level()
logger.setLevel(log_level)
transformers.utils.logging.set_verbosity(log_level)
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Log on each process the small summary:
logger.warning(
f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu} "
+ f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
)
logger.info(f"Training/evaluation parameters {training_args}")
# Set seed before initializing model.
set_seed(training_args.seed)
# Detecting last checkpoint.
last_checkpoint = None
if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
last_checkpoint = get_last_checkpoint(training_args.output_dir)
if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty. "
"Use --overwrite_output_dir to train from scratch."
)
elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
logger.info(
f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
"the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
)
# Initialize our dataset and prepare it for the audio classification task.
raw_datasets = DatasetDict()
raw_datasets["train"] = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
split=data_args.train_split_name,
use_auth_token=True if model_args.use_auth_token else None,
)
raw_datasets["eval"] = load_dataset(
data_args.dataset_name,
data_args.dataset_config_name,
split=data_args.eval_split_name,
use_auth_token=True if model_args.use_auth_token else None,
)
if data_args.audio_column_name not in raw_datasets["train"].column_names:
raise ValueError(
f"--audio_column_name {data_args.audio_column_name} not found in dataset '{data_args.dataset_name}'. "
"Make sure to set `--audio_column_name` to the correct audio column - one of "
f"{', '.join(raw_datasets['train'].column_names)}."
)
if data_args.label_column_name not in raw_datasets["train"].column_names:
raise ValueError(
f"--label_column_name {data_args.label_column_name} not found in dataset '{data_args.dataset_name}'. "
"Make sure to set `--label_column_name` to the correct text column - one of "
f"{', '.join(raw_datasets['train'].column_names)}."
)
# Setting `return_attention_mask=True` is the way to get a correctly masked mean-pooling over
# transformer outputs in the classifier, but it doesn't always lead to better accuracy
feature_extractor = AutoFeatureExtractor.from_pretrained(
model_args.feature_extractor_name or model_args.model_name_or_path,
return_attention_mask=model_args.attention_mask,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
# `datasets` takes care of automatically loading and resampling the audio,
# so we just need to set the correct target sampling rate.
raw_datasets = raw_datasets.cast_column(
data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
)
model_input_name = feature_extractor.model_input_names[0]
def train_transforms(batch):
"""Apply train_transforms across a batch."""
subsampled_wavs = []
for audio in batch[data_args.audio_column_name]:
wav = random_subsample(
audio["array"], max_length=data_args.max_length_seconds, sample_rate=feature_extractor.sampling_rate
)
subsampled_wavs.append(wav)
inputs = feature_extractor(subsampled_wavs, sampling_rate=feature_extractor.sampling_rate)
output_batch = {model_input_name: inputs.get(model_input_name)}
output_batch["labels"] = list(batch[data_args.label_column_name])
return output_batch
def val_transforms(batch):
"""Apply val_transforms across a batch."""
wavs = [audio["array"] for audio in batch[data_args.audio_column_name]]
inputs = feature_extractor(wavs, sampling_rate=feature_extractor.sampling_rate)
output_batch = {model_input_name: inputs.get(model_input_name)}
output_batch["labels"] = list(batch[data_args.label_column_name])
return output_batch
# Prepare label mappings.
# We'll include these in the model's config to get human readable labels in the Inference API.
labels = raw_datasets["train"].features[data_args.label_column_name].names
label2id, id2label = {}, {}
for i, label in enumerate(labels):
label2id[label] = str(i)
id2label[str(i)] = label
# Load the accuracy metric from the datasets package
metric = evaluate.load("accuracy")
# Define our compute_metrics function. It takes an `EvalPrediction` object (a namedtuple with
# `predictions` and `label_ids` fields) and has to return a dictionary string to float.
def compute_metrics(eval_pred):
"""Computes accuracy on a batch of predictions"""
predictions = np.argmax(eval_pred.predictions, axis=1)
return metric.compute(predictions=predictions, references=eval_pred.label_ids)
config = AutoConfig.from_pretrained(
model_args.config_name or model_args.model_name_or_path,
num_labels=len(labels),
label2id=label2id,
id2label=id2label,
finetuning_task="audio-classification",
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
)
model = AutoModelForAudioClassification.from_pretrained(
model_args.model_name_or_path,
from_tf=bool(".ckpt" in model_args.model_name_or_path),
config=config,
cache_dir=model_args.cache_dir,
revision=model_args.model_revision,
use_auth_token=True if model_args.use_auth_token else None,
ignore_mismatched_sizes=model_args.ignore_mismatched_sizes,
)
# freeze the convolutional waveform encoder
if model_args.freeze_feature_encoder:
model.freeze_feature_encoder()
if training_args.do_train:
if data_args.max_train_samples is not None:
raw_datasets["train"] = (
raw_datasets["train"].shuffle(seed=training_args.seed).select(range(data_args.max_train_samples))
)
# Set the training transforms
raw_datasets["train"].set_transform(train_transforms, output_all_columns=False)
if training_args.do_eval:
if data_args.max_eval_samples is not None:
raw_datasets["eval"] = (
raw_datasets["eval"].shuffle(seed=training_args.seed).select(range(data_args.max_eval_samples))
)
# Set the validation transforms
raw_datasets["eval"].set_transform(val_transforms, output_all_columns=False)
# Initialize our trainer
trainer = Trainer(
model=model,
args=training_args,
train_dataset=raw_datasets["train"] if training_args.do_train else None,
eval_dataset=raw_datasets["eval"] if training_args.do_eval else None,
compute_metrics=compute_metrics,
tokenizer=feature_extractor,
)
# Training
if training_args.do_train:
checkpoint = None
if training_args.resume_from_checkpoint is not None:
checkpoint = training_args.resume_from_checkpoint
elif last_checkpoint is not None:
checkpoint = last_checkpoint
train_result = trainer.train(resume_from_checkpoint=checkpoint)
trainer.save_model()
trainer.log_metrics("train", train_result.metrics)
trainer.save_metrics("train", train_result.metrics)
trainer.save_state()
# Evaluation
if training_args.do_eval:
metrics = trainer.evaluate()
trainer.log_metrics("eval", metrics)
trainer.save_metrics("eval", metrics)
# Write model card and (optionally) push to hub
kwargs = {
"finetuned_from": model_args.model_name_or_path,
"tasks": "audio-classification",
"dataset": data_args.dataset_name,
"tags": ["audio-classification"],
}
if training_args.push_to_hub:
trainer.push_to_hub(**kwargs)
else:
trainer.create_model_card(**kwargs)
if __name__ == "__main__":
main()
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/maskformer/__init__.py | # Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
_import_structure = {
"configuration_maskformer": ["MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "MaskFormerConfig"],
"configuration_maskformer_swin": ["MaskFormerSwinConfig"],
}
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["feature_extraction_maskformer"] = ["MaskFormerFeatureExtractor"]
_import_structure["image_processing_maskformer"] = ["MaskFormerImageProcessor"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_maskformer"] = [
"MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"MaskFormerForInstanceSegmentation",
"MaskFormerModel",
"MaskFormerPreTrainedModel",
]
_import_structure["modeling_maskformer_swin"] = [
"MaskFormerSwinBackbone",
"MaskFormerSwinModel",
"MaskFormerSwinPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_maskformer import MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskFormerConfig
from .configuration_maskformer_swin import MaskFormerSwinConfig
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .feature_extraction_maskformer import MaskFormerFeatureExtractor
from .image_processing_maskformer import MaskFormerImageProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_maskformer import (
MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
MaskFormerForInstanceSegmentation,
MaskFormerModel,
MaskFormerPreTrainedModel,
)
from .modeling_maskformer_swin import (
MaskFormerSwinBackbone,
MaskFormerSwinModel,
MaskFormerSwinPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure)
| # Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
_import_structure = {
"configuration_maskformer": ["MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "MaskFormerConfig"],
"configuration_maskformer_swin": ["MaskFormerSwinConfig"],
}
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["feature_extraction_maskformer"] = ["MaskFormerFeatureExtractor"]
_import_structure["image_processing_maskformer"] = ["MaskFormerImageProcessor"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_maskformer"] = [
"MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"MaskFormerForInstanceSegmentation",
"MaskFormerModel",
"MaskFormerPreTrainedModel",
]
_import_structure["modeling_maskformer_swin"] = [
"MaskFormerSwinBackbone",
"MaskFormerSwinModel",
"MaskFormerSwinPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_maskformer import MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskFormerConfig
from .configuration_maskformer_swin import MaskFormerSwinConfig
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .feature_extraction_maskformer import MaskFormerFeatureExtractor
from .image_processing_maskformer import MaskFormerImageProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_maskformer import (
MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
MaskFormerForInstanceSegmentation,
MaskFormerModel,
MaskFormerPreTrainedModel,
)
from .modeling_maskformer_swin import (
MaskFormerSwinBackbone,
MaskFormerSwinModel,
MaskFormerSwinPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/blenderbot_small/__init__.py | # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
_import_structure = {
"configuration_blenderbot_small": [
"BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BlenderbotSmallConfig",
"BlenderbotSmallOnnxConfig",
],
"tokenization_blenderbot_small": ["BlenderbotSmallTokenizer"],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["tokenization_blenderbot_small_fast"] = ["BlenderbotSmallTokenizerFast"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_blenderbot_small"] = [
"BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST",
"BlenderbotSmallForCausalLM",
"BlenderbotSmallForConditionalGeneration",
"BlenderbotSmallModel",
"BlenderbotSmallPreTrainedModel",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tf_blenderbot_small"] = [
"TFBlenderbotSmallForConditionalGeneration",
"TFBlenderbotSmallModel",
"TFBlenderbotSmallPreTrainedModel",
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_flax_blenderbot_small"] = [
"FlaxBlenderbotSmallForConditionalGeneration",
"FlaxBlenderbotSmallModel",
"FlaxBlenderbotSmallPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
BlenderbotSmallConfig,
BlenderbotSmallOnnxConfig,
)
from .tokenization_blenderbot_small import BlenderbotSmallTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_blenderbot_small_fast import BlenderbotSmallTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotSmallForCausalLM,
BlenderbotSmallForConditionalGeneration,
BlenderbotSmallModel,
BlenderbotSmallPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_blenderbot_small import (
TFBlenderbotSmallForConditionalGeneration,
TFBlenderbotSmallModel,
TFBlenderbotSmallPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_blenderbot_small import (
FlaxBlenderbotSmallForConditionalGeneration,
FlaxBlenderbotSmallModel,
FlaxBlenderbotSmallPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
_import_structure = {
"configuration_blenderbot_small": [
"BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BlenderbotSmallConfig",
"BlenderbotSmallOnnxConfig",
],
"tokenization_blenderbot_small": ["BlenderbotSmallTokenizer"],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["tokenization_blenderbot_small_fast"] = ["BlenderbotSmallTokenizerFast"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_blenderbot_small"] = [
"BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST",
"BlenderbotSmallForCausalLM",
"BlenderbotSmallForConditionalGeneration",
"BlenderbotSmallModel",
"BlenderbotSmallPreTrainedModel",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tf_blenderbot_small"] = [
"TFBlenderbotSmallForConditionalGeneration",
"TFBlenderbotSmallModel",
"TFBlenderbotSmallPreTrainedModel",
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_flax_blenderbot_small"] = [
"FlaxBlenderbotSmallForConditionalGeneration",
"FlaxBlenderbotSmallModel",
"FlaxBlenderbotSmallPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
BlenderbotSmallConfig,
BlenderbotSmallOnnxConfig,
)
from .tokenization_blenderbot_small import BlenderbotSmallTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_blenderbot_small_fast import BlenderbotSmallTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotSmallForCausalLM,
BlenderbotSmallForConditionalGeneration,
BlenderbotSmallModel,
BlenderbotSmallPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_blenderbot_small import (
TFBlenderbotSmallForConditionalGeneration,
TFBlenderbotSmallModel,
TFBlenderbotSmallPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_blenderbot_small import (
FlaxBlenderbotSmallForConditionalGeneration,
FlaxBlenderbotSmallModel,
FlaxBlenderbotSmallPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/en/model_doc/xlm-roberta-xl.mdx | <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# XLM-RoBERTa-XL
## Overview
The XLM-RoBERTa-XL model was proposed in [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
The abstract from the paper is the following:
*Recent work has demonstrated the effectiveness of cross-lingual language model pretraining for cross-lingual understanding. In this study, we present the results of two larger multilingual masked language models, with 3.5B and 10.7B parameters. Our two new models dubbed XLM-R XL and XLM-R XXL outperform XLM-R by 1.8% and 2.4% average accuracy on XNLI. Our model also outperforms the RoBERTa-Large model on several English tasks of the GLUE benchmark by 0.3% on average while handling 99 more languages. This suggests pretrained models with larger capacity may obtain both strong performance on high-resource languages while greatly improving low-resource languages. We make our code and models publicly available.*
Tips:
- XLM-RoBERTa-XL is a multilingual model trained on 100 different languages. Unlike some XLM multilingual models, it does
not require `lang` tensors to understand which language is used, and should be able to determine the correct
language from the input ids.
This model was contributed by [Soonhwan-Kwon](https://github.com/Soonhwan-Kwon) and [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/xlmr).
## Documentation resources
- [Text classification task guide](./tasks/sequence_classification)
- [Token classification task guide](./tasks/token_classification)
- [Question answering task guide](./tasks/question_answering)
- [Causal language modeling task guide](./tasks/language_modeling)
- [Masked language modeling task guide](./tasks/masked_language_modeling)
- [Multiple choice task guide](./tasks/multiple_choice)
## XLMRobertaXLConfig
[[autodoc]] XLMRobertaXLConfig
## XLMRobertaXLModel
[[autodoc]] XLMRobertaXLModel
- forward
## XLMRobertaXLForCausalLM
[[autodoc]] XLMRobertaXLForCausalLM
- forward
## XLMRobertaXLForMaskedLM
[[autodoc]] XLMRobertaXLForMaskedLM
- forward
## XLMRobertaXLForSequenceClassification
[[autodoc]] XLMRobertaXLForSequenceClassification
- forward
## XLMRobertaXLForMultipleChoice
[[autodoc]] XLMRobertaXLForMultipleChoice
- forward
## XLMRobertaXLForTokenClassification
[[autodoc]] XLMRobertaXLForTokenClassification
- forward
## XLMRobertaXLForQuestionAnswering
[[autodoc]] XLMRobertaXLForQuestionAnswering
- forward
| <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# XLM-RoBERTa-XL
## Overview
The XLM-RoBERTa-XL model was proposed in [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
The abstract from the paper is the following:
*Recent work has demonstrated the effectiveness of cross-lingual language model pretraining for cross-lingual understanding. In this study, we present the results of two larger multilingual masked language models, with 3.5B and 10.7B parameters. Our two new models dubbed XLM-R XL and XLM-R XXL outperform XLM-R by 1.8% and 2.4% average accuracy on XNLI. Our model also outperforms the RoBERTa-Large model on several English tasks of the GLUE benchmark by 0.3% on average while handling 99 more languages. This suggests pretrained models with larger capacity may obtain both strong performance on high-resource languages while greatly improving low-resource languages. We make our code and models publicly available.*
Tips:
- XLM-RoBERTa-XL is a multilingual model trained on 100 different languages. Unlike some XLM multilingual models, it does
not require `lang` tensors to understand which language is used, and should be able to determine the correct
language from the input ids.
This model was contributed by [Soonhwan-Kwon](https://github.com/Soonhwan-Kwon) and [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/xlmr).
## Documentation resources
- [Text classification task guide](./tasks/sequence_classification)
- [Token classification task guide](./tasks/token_classification)
- [Question answering task guide](./tasks/question_answering)
- [Causal language modeling task guide](./tasks/language_modeling)
- [Masked language modeling task guide](./tasks/masked_language_modeling)
- [Multiple choice task guide](./tasks/multiple_choice)
## XLMRobertaXLConfig
[[autodoc]] XLMRobertaXLConfig
## XLMRobertaXLModel
[[autodoc]] XLMRobertaXLModel
- forward
## XLMRobertaXLForCausalLM
[[autodoc]] XLMRobertaXLForCausalLM
- forward
## XLMRobertaXLForMaskedLM
[[autodoc]] XLMRobertaXLForMaskedLM
- forward
## XLMRobertaXLForSequenceClassification
[[autodoc]] XLMRobertaXLForSequenceClassification
- forward
## XLMRobertaXLForMultipleChoice
[[autodoc]] XLMRobertaXLForMultipleChoice
- forward
## XLMRobertaXLForTokenClassification
[[autodoc]] XLMRobertaXLForTokenClassification
- forward
## XLMRobertaXLForQuestionAnswering
[[autodoc]] XLMRobertaXLForQuestionAnswering
- forward
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/layoutlmv3/test_image_processing_layoutlmv3.py | # coding=utf-8
# Copyright 2022 HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import unittest
import numpy as np
from transformers.testing_utils import require_pytesseract, require_torch
from transformers.utils import is_pytesseract_available, is_torch_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
if is_pytesseract_available():
from PIL import Image
from transformers import LayoutLMv3ImageProcessor
class LayoutLMv3ImageProcessingTester(unittest.TestCase):
def __init__(
self,
parent,
batch_size=7,
num_channels=3,
image_size=18,
min_resolution=30,
max_resolution=400,
do_resize=True,
size=None,
apply_ocr=True,
):
size = size if size is not None else {"height": 18, "width": 18}
self.parent = parent
self.batch_size = batch_size
self.num_channels = num_channels
self.image_size = image_size
self.min_resolution = min_resolution
self.max_resolution = max_resolution
self.do_resize = do_resize
self.size = size
self.apply_ocr = apply_ocr
def prepare_image_processor_dict(self):
return {"do_resize": self.do_resize, "size": self.size, "apply_ocr": self.apply_ocr}
@require_torch
@require_pytesseract
class LayoutLMv3ImageProcessingTest(ImageProcessingSavingTestMixin, unittest.TestCase):
image_processing_class = LayoutLMv3ImageProcessor if is_pytesseract_available() else None
def setUp(self):
self.image_processor_tester = LayoutLMv3ImageProcessingTester(self)
@property
def image_processor_dict(self):
return self.image_processor_tester.prepare_image_processor_dict()
def test_image_processor_properties(self):
image_processing = self.image_processing_class(**self.image_processor_dict)
self.assertTrue(hasattr(image_processing, "do_resize"))
self.assertTrue(hasattr(image_processing, "size"))
self.assertTrue(hasattr(image_processing, "apply_ocr"))
def test_image_processor_from_dict_with_kwargs(self):
image_processor = self.image_processing_class.from_dict(self.image_processor_dict)
self.assertEqual(image_processor.size, {"height": 18, "width": 18})
image_processor = self.image_processing_class.from_dict(self.image_processor_dict, size=42)
self.assertEqual(image_processor.size, {"height": 42, "width": 42})
def test_batch_feature(self):
pass
def test_call_pil(self):
# Initialize image_processing
image_processing = self.image_processing_class(**self.image_processor_dict)
# create random PIL images
image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False)
for image in image_inputs:
self.assertIsInstance(image, Image.Image)
# Test not batched input
encoding = image_processing(image_inputs[0], return_tensors="pt")
self.assertEqual(
encoding.pixel_values.shape,
(
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
self.assertIsInstance(encoding.words, list)
self.assertIsInstance(encoding.boxes, list)
# Test batched
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
def test_call_numpy(self):
# Initialize image_processing
image_processing = self.image_processing_class(**self.image_processor_dict)
# create random numpy tensors
image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, numpify=True)
for image in image_inputs:
self.assertIsInstance(image, np.ndarray)
# Test not batched input
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
# Test batched
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
def test_call_pytorch(self):
# Initialize image_processing
image_processing = self.image_processing_class(**self.image_processor_dict)
# create random PyTorch tensors
image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, torchify=True)
for image in image_inputs:
self.assertIsInstance(image, torch.Tensor)
# Test not batched input
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
# Test batched
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
def test_LayoutLMv3_integration_test(self):
# with apply_OCR = True
image_processing = LayoutLMv3ImageProcessor()
from datasets import load_dataset
ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
image = Image.open(ds[0]["file"]).convert("RGB")
encoding = image_processing(image, return_tensors="pt")
self.assertEqual(encoding.pixel_values.shape, (1, 3, 224, 224))
self.assertEqual(len(encoding.words), len(encoding.boxes))
# fmt: off
# the words and boxes were obtained with Tesseract 4.1.1
expected_words = [['11:14', 'to', '11:39', 'a.m', '11:39', 'to', '11:44', 'a.m.', '11:44', 'a.m.', 'to', '12:25', 'p.m.', '12:25', 'to', '12:58', 'p.m.', '12:58', 'to', '4:00', 'p.m.', '2:00', 'to', '5:00', 'p.m.', 'Coffee', 'Break', 'Coffee', 'will', 'be', 'served', 'for', 'men', 'and', 'women', 'in', 'the', 'lobby', 'adjacent', 'to', 'exhibit', 'area.', 'Please', 'move', 'into', 'exhibit', 'area.', '(Exhibits', 'Open)', 'TRRF', 'GENERAL', 'SESSION', '(PART', '|)', 'Presiding:', 'Lee', 'A.', 'Waller', 'TRRF', 'Vice', 'President', '“Introductory', 'Remarks”', 'Lee', 'A.', 'Waller,', 'TRRF', 'Vice', 'Presi-', 'dent', 'Individual', 'Interviews', 'with', 'TRRF', 'Public', 'Board', 'Members', 'and', 'Sci-', 'entific', 'Advisory', 'Council', 'Mem-', 'bers', 'Conducted', 'by', 'TRRF', 'Treasurer', 'Philip', 'G.', 'Kuehn', 'to', 'get', 'answers', 'which', 'the', 'public', 'refrigerated', 'warehousing', 'industry', 'is', 'looking', 'for.', 'Plus', 'questions', 'from', 'the', 'floor.', 'Dr.', 'Emil', 'M.', 'Mrak,', 'University', 'of', 'Cal-', 'ifornia,', 'Chairman,', 'TRRF', 'Board;', 'Sam', 'R.', 'Cecil,', 'University', 'of', 'Georgia', 'College', 'of', 'Agriculture;', 'Dr.', 'Stanley', 'Charm,', 'Tufts', 'University', 'School', 'of', 'Medicine;', 'Dr.', 'Robert', 'H.', 'Cotton,', 'ITT', 'Continental', 'Baking', 'Company;', 'Dr.', 'Owen', 'Fennema,', 'University', 'of', 'Wis-', 'consin;', 'Dr.', 'Robert', 'E.', 'Hardenburg,', 'USDA.', 'Questions', 'and', 'Answers', 'Exhibits', 'Open', 'Capt.', 'Jack', 'Stoney', 'Room', 'TRRF', 'Scientific', 'Advisory', 'Council', 'Meeting', 'Ballroom', 'Foyer']] # noqa: E231
expected_boxes = [[[141, 57, 214, 69], [228, 58, 252, 69], [141, 75, 216, 88], [230, 79, 280, 88], [142, 260, 218, 273], [230, 261, 255, 273], [143, 279, 218, 290], [231, 282, 290, 291], [143, 342, 218, 354], [231, 345, 289, 355], [202, 362, 227, 373], [143, 379, 220, 392], [231, 382, 291, 394], [144, 714, 220, 726], [231, 715, 256, 726], [144, 732, 220, 745], [232, 736, 291, 747], [144, 769, 218, 782], [231, 770, 256, 782], [141, 788, 202, 801], [215, 791, 274, 804], [143, 826, 204, 838], [215, 826, 240, 838], [142, 844, 202, 857], [215, 847, 274, 859], [334, 57, 427, 69], [440, 57, 522, 69], [369, 75, 461, 88], [469, 75, 516, 88], [528, 76, 562, 88], [570, 76, 667, 88], [675, 75, 711, 87], [721, 79, 778, 88], [789, 75, 840, 88], [369, 97, 470, 107], [484, 94, 507, 106], [518, 94, 562, 107], [576, 94, 655, 110], [668, 94, 792, 109], [804, 95, 829, 107], [369, 113, 465, 125], [477, 116, 547, 125], [562, 113, 658, 125], [671, 116, 748, 125], [761, 113, 811, 125], [369, 131, 465, 143], [477, 133, 548, 143], [563, 130, 698, 145], [710, 130, 802, 146], [336, 171, 412, 183], [423, 171, 572, 183], [582, 170, 716, 184], [728, 171, 817, 187], [829, 171, 844, 186], [338, 197, 482, 212], [507, 196, 557, 209], [569, 196, 595, 208], [610, 196, 702, 209], [505, 214, 583, 226], [595, 214, 656, 227], [670, 215, 807, 227], [335, 259, 543, 274], [556, 259, 708, 272], [372, 279, 422, 291], [435, 279, 460, 291], [474, 279, 574, 292], [587, 278, 664, 291], [676, 278, 738, 291], [751, 279, 834, 291], [372, 298, 434, 310], [335, 341, 483, 354], [497, 341, 655, 354], [667, 341, 728, 354], [740, 341, 825, 354], [335, 360, 430, 372], [442, 360, 534, 372], [545, 359, 687, 372], [697, 360, 754, 372], [765, 360, 823, 373], [334, 378, 428, 391], [440, 378, 577, 394], [590, 378, 705, 391], [720, 378, 801, 391], [334, 397, 400, 409], [370, 416, 529, 429], [544, 416, 576, 432], [587, 416, 665, 428], [677, 416, 814, 429], [372, 435, 452, 450], [465, 434, 495, 447], [511, 434, 600, 447], [611, 436, 637, 447], [649, 436, 694, 451], [705, 438, 824, 447], [369, 453, 452, 466], [464, 454, 509, 466], [522, 453, 611, 469], [625, 453, 792, 469], [370, 472, 556, 488], [570, 472, 684, 487], [697, 472, 718, 485], [732, 472, 835, 488], [369, 490, 411, 503], [425, 490, 484, 503], [496, 490, 635, 506], [645, 490, 707, 503], [718, 491, 761, 503], [771, 490, 840, 503], [336, 510, 374, 521], [388, 510, 447, 522], [460, 510, 489, 521], [503, 510, 580, 522], [592, 509, 736, 525], [745, 509, 770, 522], [781, 509, 840, 522], [338, 528, 434, 541], [448, 528, 596, 541], [609, 527, 687, 540], [700, 528, 792, 541], [336, 546, 397, 559], [407, 546, 431, 559], [443, 546, 525, 560], [537, 546, 680, 562], [688, 546, 714, 559], [722, 546, 837, 562], [336, 565, 449, 581], [461, 565, 485, 577], [497, 565, 665, 581], [681, 565, 718, 577], [732, 565, 837, 580], [337, 584, 438, 597], [452, 583, 521, 596], [535, 584, 677, 599], [690, 583, 787, 596], [801, 583, 825, 596], [338, 602, 478, 615], [492, 602, 530, 614], [543, 602, 638, 615], [650, 602, 676, 614], [688, 602, 788, 615], [802, 602, 843, 614], [337, 621, 502, 633], [516, 621, 615, 637], [629, 621, 774, 636], [789, 621, 827, 633], [337, 639, 418, 652], [432, 640, 571, 653], [587, 639, 731, 655], [743, 639, 769, 652], [780, 639, 841, 652], [338, 658, 440, 673], [455, 658, 491, 670], [508, 658, 602, 671], [616, 658, 638, 670], [654, 658, 835, 674], [337, 677, 429, 689], [337, 714, 482, 726], [495, 714, 548, 726], [561, 714, 683, 726], [338, 770, 461, 782], [474, 769, 554, 785], [489, 788, 562, 803], [576, 788, 643, 801], [656, 787, 751, 804], [764, 788, 844, 801], [334, 825, 421, 838], [430, 824, 574, 838], [584, 824, 723, 841], [335, 844, 450, 857], [464, 843, 583, 860], [628, 862, 755, 875], [769, 861, 848, 878]]] # noqa: E231
# fmt: on
self.assertListEqual(encoding.words, expected_words)
self.assertListEqual(encoding.boxes, expected_boxes)
# with apply_OCR = False
image_processing = LayoutLMv3ImageProcessor(apply_ocr=False)
encoding = image_processing(image, return_tensors="pt")
self.assertEqual(encoding.pixel_values.shape, (1, 3, 224, 224))
| # coding=utf-8
# Copyright 2022 HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import unittest
import numpy as np
from transformers.testing_utils import require_pytesseract, require_torch
from transformers.utils import is_pytesseract_available, is_torch_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
if is_pytesseract_available():
from PIL import Image
from transformers import LayoutLMv3ImageProcessor
class LayoutLMv3ImageProcessingTester(unittest.TestCase):
def __init__(
self,
parent,
batch_size=7,
num_channels=3,
image_size=18,
min_resolution=30,
max_resolution=400,
do_resize=True,
size=None,
apply_ocr=True,
):
size = size if size is not None else {"height": 18, "width": 18}
self.parent = parent
self.batch_size = batch_size
self.num_channels = num_channels
self.image_size = image_size
self.min_resolution = min_resolution
self.max_resolution = max_resolution
self.do_resize = do_resize
self.size = size
self.apply_ocr = apply_ocr
def prepare_image_processor_dict(self):
return {"do_resize": self.do_resize, "size": self.size, "apply_ocr": self.apply_ocr}
@require_torch
@require_pytesseract
class LayoutLMv3ImageProcessingTest(ImageProcessingSavingTestMixin, unittest.TestCase):
image_processing_class = LayoutLMv3ImageProcessor if is_pytesseract_available() else None
def setUp(self):
self.image_processor_tester = LayoutLMv3ImageProcessingTester(self)
@property
def image_processor_dict(self):
return self.image_processor_tester.prepare_image_processor_dict()
def test_image_processor_properties(self):
image_processing = self.image_processing_class(**self.image_processor_dict)
self.assertTrue(hasattr(image_processing, "do_resize"))
self.assertTrue(hasattr(image_processing, "size"))
self.assertTrue(hasattr(image_processing, "apply_ocr"))
def test_image_processor_from_dict_with_kwargs(self):
image_processor = self.image_processing_class.from_dict(self.image_processor_dict)
self.assertEqual(image_processor.size, {"height": 18, "width": 18})
image_processor = self.image_processing_class.from_dict(self.image_processor_dict, size=42)
self.assertEqual(image_processor.size, {"height": 42, "width": 42})
def test_batch_feature(self):
pass
def test_call_pil(self):
# Initialize image_processing
image_processing = self.image_processing_class(**self.image_processor_dict)
# create random PIL images
image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False)
for image in image_inputs:
self.assertIsInstance(image, Image.Image)
# Test not batched input
encoding = image_processing(image_inputs[0], return_tensors="pt")
self.assertEqual(
encoding.pixel_values.shape,
(
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
self.assertIsInstance(encoding.words, list)
self.assertIsInstance(encoding.boxes, list)
# Test batched
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
def test_call_numpy(self):
# Initialize image_processing
image_processing = self.image_processing_class(**self.image_processor_dict)
# create random numpy tensors
image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, numpify=True)
for image in image_inputs:
self.assertIsInstance(image, np.ndarray)
# Test not batched input
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
# Test batched
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
def test_call_pytorch(self):
# Initialize image_processing
image_processing = self.image_processing_class(**self.image_processor_dict)
# create random PyTorch tensors
image_inputs = prepare_image_inputs(self.image_processor_tester, equal_resolution=False, torchify=True)
for image in image_inputs:
self.assertIsInstance(image, torch.Tensor)
# Test not batched input
encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
# Test batched
encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
self.assertEqual(
encoded_images.shape,
(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size["height"],
self.image_processor_tester.size["width"],
),
)
def test_LayoutLMv3_integration_test(self):
# with apply_OCR = True
image_processing = LayoutLMv3ImageProcessor()
from datasets import load_dataset
ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
image = Image.open(ds[0]["file"]).convert("RGB")
encoding = image_processing(image, return_tensors="pt")
self.assertEqual(encoding.pixel_values.shape, (1, 3, 224, 224))
self.assertEqual(len(encoding.words), len(encoding.boxes))
# fmt: off
# the words and boxes were obtained with Tesseract 4.1.1
expected_words = [['11:14', 'to', '11:39', 'a.m', '11:39', 'to', '11:44', 'a.m.', '11:44', 'a.m.', 'to', '12:25', 'p.m.', '12:25', 'to', '12:58', 'p.m.', '12:58', 'to', '4:00', 'p.m.', '2:00', 'to', '5:00', 'p.m.', 'Coffee', 'Break', 'Coffee', 'will', 'be', 'served', 'for', 'men', 'and', 'women', 'in', 'the', 'lobby', 'adjacent', 'to', 'exhibit', 'area.', 'Please', 'move', 'into', 'exhibit', 'area.', '(Exhibits', 'Open)', 'TRRF', 'GENERAL', 'SESSION', '(PART', '|)', 'Presiding:', 'Lee', 'A.', 'Waller', 'TRRF', 'Vice', 'President', '“Introductory', 'Remarks”', 'Lee', 'A.', 'Waller,', 'TRRF', 'Vice', 'Presi-', 'dent', 'Individual', 'Interviews', 'with', 'TRRF', 'Public', 'Board', 'Members', 'and', 'Sci-', 'entific', 'Advisory', 'Council', 'Mem-', 'bers', 'Conducted', 'by', 'TRRF', 'Treasurer', 'Philip', 'G.', 'Kuehn', 'to', 'get', 'answers', 'which', 'the', 'public', 'refrigerated', 'warehousing', 'industry', 'is', 'looking', 'for.', 'Plus', 'questions', 'from', 'the', 'floor.', 'Dr.', 'Emil', 'M.', 'Mrak,', 'University', 'of', 'Cal-', 'ifornia,', 'Chairman,', 'TRRF', 'Board;', 'Sam', 'R.', 'Cecil,', 'University', 'of', 'Georgia', 'College', 'of', 'Agriculture;', 'Dr.', 'Stanley', 'Charm,', 'Tufts', 'University', 'School', 'of', 'Medicine;', 'Dr.', 'Robert', 'H.', 'Cotton,', 'ITT', 'Continental', 'Baking', 'Company;', 'Dr.', 'Owen', 'Fennema,', 'University', 'of', 'Wis-', 'consin;', 'Dr.', 'Robert', 'E.', 'Hardenburg,', 'USDA.', 'Questions', 'and', 'Answers', 'Exhibits', 'Open', 'Capt.', 'Jack', 'Stoney', 'Room', 'TRRF', 'Scientific', 'Advisory', 'Council', 'Meeting', 'Ballroom', 'Foyer']] # noqa: E231
expected_boxes = [[[141, 57, 214, 69], [228, 58, 252, 69], [141, 75, 216, 88], [230, 79, 280, 88], [142, 260, 218, 273], [230, 261, 255, 273], [143, 279, 218, 290], [231, 282, 290, 291], [143, 342, 218, 354], [231, 345, 289, 355], [202, 362, 227, 373], [143, 379, 220, 392], [231, 382, 291, 394], [144, 714, 220, 726], [231, 715, 256, 726], [144, 732, 220, 745], [232, 736, 291, 747], [144, 769, 218, 782], [231, 770, 256, 782], [141, 788, 202, 801], [215, 791, 274, 804], [143, 826, 204, 838], [215, 826, 240, 838], [142, 844, 202, 857], [215, 847, 274, 859], [334, 57, 427, 69], [440, 57, 522, 69], [369, 75, 461, 88], [469, 75, 516, 88], [528, 76, 562, 88], [570, 76, 667, 88], [675, 75, 711, 87], [721, 79, 778, 88], [789, 75, 840, 88], [369, 97, 470, 107], [484, 94, 507, 106], [518, 94, 562, 107], [576, 94, 655, 110], [668, 94, 792, 109], [804, 95, 829, 107], [369, 113, 465, 125], [477, 116, 547, 125], [562, 113, 658, 125], [671, 116, 748, 125], [761, 113, 811, 125], [369, 131, 465, 143], [477, 133, 548, 143], [563, 130, 698, 145], [710, 130, 802, 146], [336, 171, 412, 183], [423, 171, 572, 183], [582, 170, 716, 184], [728, 171, 817, 187], [829, 171, 844, 186], [338, 197, 482, 212], [507, 196, 557, 209], [569, 196, 595, 208], [610, 196, 702, 209], [505, 214, 583, 226], [595, 214, 656, 227], [670, 215, 807, 227], [335, 259, 543, 274], [556, 259, 708, 272], [372, 279, 422, 291], [435, 279, 460, 291], [474, 279, 574, 292], [587, 278, 664, 291], [676, 278, 738, 291], [751, 279, 834, 291], [372, 298, 434, 310], [335, 341, 483, 354], [497, 341, 655, 354], [667, 341, 728, 354], [740, 341, 825, 354], [335, 360, 430, 372], [442, 360, 534, 372], [545, 359, 687, 372], [697, 360, 754, 372], [765, 360, 823, 373], [334, 378, 428, 391], [440, 378, 577, 394], [590, 378, 705, 391], [720, 378, 801, 391], [334, 397, 400, 409], [370, 416, 529, 429], [544, 416, 576, 432], [587, 416, 665, 428], [677, 416, 814, 429], [372, 435, 452, 450], [465, 434, 495, 447], [511, 434, 600, 447], [611, 436, 637, 447], [649, 436, 694, 451], [705, 438, 824, 447], [369, 453, 452, 466], [464, 454, 509, 466], [522, 453, 611, 469], [625, 453, 792, 469], [370, 472, 556, 488], [570, 472, 684, 487], [697, 472, 718, 485], [732, 472, 835, 488], [369, 490, 411, 503], [425, 490, 484, 503], [496, 490, 635, 506], [645, 490, 707, 503], [718, 491, 761, 503], [771, 490, 840, 503], [336, 510, 374, 521], [388, 510, 447, 522], [460, 510, 489, 521], [503, 510, 580, 522], [592, 509, 736, 525], [745, 509, 770, 522], [781, 509, 840, 522], [338, 528, 434, 541], [448, 528, 596, 541], [609, 527, 687, 540], [700, 528, 792, 541], [336, 546, 397, 559], [407, 546, 431, 559], [443, 546, 525, 560], [537, 546, 680, 562], [688, 546, 714, 559], [722, 546, 837, 562], [336, 565, 449, 581], [461, 565, 485, 577], [497, 565, 665, 581], [681, 565, 718, 577], [732, 565, 837, 580], [337, 584, 438, 597], [452, 583, 521, 596], [535, 584, 677, 599], [690, 583, 787, 596], [801, 583, 825, 596], [338, 602, 478, 615], [492, 602, 530, 614], [543, 602, 638, 615], [650, 602, 676, 614], [688, 602, 788, 615], [802, 602, 843, 614], [337, 621, 502, 633], [516, 621, 615, 637], [629, 621, 774, 636], [789, 621, 827, 633], [337, 639, 418, 652], [432, 640, 571, 653], [587, 639, 731, 655], [743, 639, 769, 652], [780, 639, 841, 652], [338, 658, 440, 673], [455, 658, 491, 670], [508, 658, 602, 671], [616, 658, 638, 670], [654, 658, 835, 674], [337, 677, 429, 689], [337, 714, 482, 726], [495, 714, 548, 726], [561, 714, 683, 726], [338, 770, 461, 782], [474, 769, 554, 785], [489, 788, 562, 803], [576, 788, 643, 801], [656, 787, 751, 804], [764, 788, 844, 801], [334, 825, 421, 838], [430, 824, 574, 838], [584, 824, 723, 841], [335, 844, 450, 857], [464, 843, 583, 860], [628, 862, 755, 875], [769, 861, 848, 878]]] # noqa: E231
# fmt: on
self.assertListEqual(encoding.words, expected_words)
self.assertListEqual(encoding.boxes, expected_boxes)
# with apply_OCR = False
image_processing = LayoutLMv3ImageProcessor(apply_ocr=False)
encoding = image_processing(image, return_tensors="pt")
self.assertEqual(encoding.pixel_values.shape, (1, 3, 224, 224))
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/bit/__init__.py | # Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
_import_structure = {"configuration_bit": ["BIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BitConfig", "BitOnnxConfig"]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_bit"] = [
"BIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BitForImageClassification",
"BitModel",
"BitPreTrainedModel",
"BitBackbone",
]
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["image_processing_bit"] = ["BitImageProcessor"]
if TYPE_CHECKING:
from .configuration_bit import BIT_PRETRAINED_CONFIG_ARCHIVE_MAP, BitConfig, BitOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_bit import (
BIT_PRETRAINED_MODEL_ARCHIVE_LIST,
BitBackbone,
BitForImageClassification,
BitModel,
BitPreTrainedModel,
)
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .image_processing_bit import BitImageProcessor
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure)
| # Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
_import_structure = {"configuration_bit": ["BIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BitConfig", "BitOnnxConfig"]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_bit"] = [
"BIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BitForImageClassification",
"BitModel",
"BitPreTrainedModel",
"BitBackbone",
]
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["image_processing_bit"] = ["BitImageProcessor"]
if TYPE_CHECKING:
from .configuration_bit import BIT_PRETRAINED_CONFIG_ARCHIVE_MAP, BitConfig, BitOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_bit import (
BIT_PRETRAINED_MODEL_ARCHIVE_LIST,
BitBackbone,
BitForImageClassification,
BitModel,
BitPreTrainedModel,
)
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .image_processing_bit import BitImageProcessor
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/plbart/__init__.py | # Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_sentencepiece_available,
is_tokenizers_available,
is_torch_available,
)
_import_structure = {"configuration_plbart": ["PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP", "PLBartConfig"]}
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["tokenization_plbart"] = ["PLBartTokenizer"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_plbart"] = [
"PLBART_PRETRAINED_MODEL_ARCHIVE_LIST",
"PLBartForCausalLM",
"PLBartForConditionalGeneration",
"PLBartForSequenceClassification",
"PLBartModel",
"PLBartPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_plbart import PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP, PLBartConfig
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_plbart import PLBartTokenizer
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_plbart import (
PLBART_PRETRAINED_MODEL_ARCHIVE_LIST,
PLBartForCausalLM,
PLBartForConditionalGeneration,
PLBartForSequenceClassification,
PLBartModel,
PLBartPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure)
| # Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_sentencepiece_available,
is_tokenizers_available,
is_torch_available,
)
_import_structure = {"configuration_plbart": ["PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP", "PLBartConfig"]}
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["tokenization_plbart"] = ["PLBartTokenizer"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_plbart"] = [
"PLBART_PRETRAINED_MODEL_ARCHIVE_LIST",
"PLBartForCausalLM",
"PLBartForConditionalGeneration",
"PLBartForSequenceClassification",
"PLBartModel",
"PLBartPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_plbart import PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP, PLBartConfig
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_plbart import PLBartTokenizer
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_plbart import (
PLBART_PRETRAINED_MODEL_ARCHIVE_LIST,
PLBartForCausalLM,
PLBartForConditionalGeneration,
PLBartForSequenceClassification,
PLBartModel,
PLBartPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/pipelines/image_classification.py | from typing import List, Union
from ..utils import (
add_end_docstrings,
is_tf_available,
is_torch_available,
is_vision_available,
logging,
requires_backends,
)
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_tf_available():
import tensorflow as tf
from ..models.auto.modeling_tf_auto import TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
from ..tf_utils import stable_softmax
if is_torch_available():
from ..models.auto.modeling_auto import MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
logger = logging.get_logger(__name__)
@add_end_docstrings(PIPELINE_INIT_ARGS)
class ImageClassificationPipeline(Pipeline):
"""
Image classification pipeline using any `AutoModelForImageClassification`. This pipeline predicts the class of an
image.
Example:
```python
>>> from transformers import pipeline
>>> classifier = pipeline(model="microsoft/beit-base-patch16-224-pt22k-ft22k")
>>> classifier("https://huggingface.co/datasets/Narsil/image_dummy/raw/main/parrots.png")
[{'score': 0.442, 'label': 'macaw'}, {'score': 0.088, 'label': 'popinjay'}, {'score': 0.075, 'label': 'parrot'}, {'score': 0.073, 'label': 'parodist, lampooner'}, {'score': 0.046, 'label': 'poll, poll_parrot'}]
```
Learn more about the basics of using a pipeline in the [pipeline tutorial](../pipeline_tutorial)
This image classification pipeline can currently be loaded from [`pipeline`] using the following task identifier:
`"image-classification"`.
See the list of available models on
[huggingface.co/models](https://huggingface.co/models?filter=image-classification).
"""
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
requires_backends(self, "vision")
self.check_model_type(
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
if self.framework == "tf"
else MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
)
def _sanitize_parameters(self, top_k=None):
postprocess_params = {}
if top_k is not None:
postprocess_params["top_k"] = top_k
return {}, {}, postprocess_params
def __call__(self, images: Union[str, List[str], "Image.Image", List["Image.Image"]], **kwargs):
"""
Assign labels to the image(s) passed as inputs.
Args:
images (`str`, `List[str]`, `PIL.Image` or `List[PIL.Image]`):
The pipeline handles three types of images:
- A string containing a http link pointing to an image
- A string containing a local path to an image
- An image loaded in PIL directly
The pipeline accepts either a single image or a batch of images, which must then be passed as a string.
Images in a batch must all be in the same format: all as http links, all as local paths, or all as PIL
images.
top_k (`int`, *optional*, defaults to 5):
The number of top labels that will be returned by the pipeline. If the provided number is higher than
the number of labels available in the model configuration, it will default to the number of labels.
Return:
A dictionary or a list of dictionaries containing result. If the input is a single image, will return a
dictionary, if the input is a list of several images, will return a list of dictionaries corresponding to
the images.
The dictionaries contain the following keys:
- **label** (`str`) -- The label identified by the model.
- **score** (`int`) -- The score attributed by the model for that label.
"""
return super().__call__(images, **kwargs)
def preprocess(self, image):
image = load_image(image)
model_inputs = self.image_processor(images=image, return_tensors=self.framework)
return model_inputs
def _forward(self, model_inputs):
model_outputs = self.model(**model_inputs)
return model_outputs
def postprocess(self, model_outputs, top_k=5):
if top_k > self.model.config.num_labels:
top_k = self.model.config.num_labels
if self.framework == "pt":
probs = model_outputs.logits.softmax(-1)[0]
scores, ids = probs.topk(top_k)
elif self.framework == "tf":
probs = stable_softmax(model_outputs.logits, axis=-1)[0]
topk = tf.math.top_k(probs, k=top_k)
scores, ids = topk.values.numpy(), topk.indices.numpy()
else:
raise ValueError(f"Unsupported framework: {self.framework}")
scores = scores.tolist()
ids = ids.tolist()
return [{"score": score, "label": self.model.config.id2label[_id]} for score, _id in zip(scores, ids)]
| from typing import List, Union
from ..utils import (
add_end_docstrings,
is_tf_available,
is_torch_available,
is_vision_available,
logging,
requires_backends,
)
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_tf_available():
import tensorflow as tf
from ..models.auto.modeling_tf_auto import TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
from ..tf_utils import stable_softmax
if is_torch_available():
from ..models.auto.modeling_auto import MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
logger = logging.get_logger(__name__)
@add_end_docstrings(PIPELINE_INIT_ARGS)
class ImageClassificationPipeline(Pipeline):
"""
Image classification pipeline using any `AutoModelForImageClassification`. This pipeline predicts the class of an
image.
Example:
```python
>>> from transformers import pipeline
>>> classifier = pipeline(model="microsoft/beit-base-patch16-224-pt22k-ft22k")
>>> classifier("https://huggingface.co/datasets/Narsil/image_dummy/raw/main/parrots.png")
[{'score': 0.442, 'label': 'macaw'}, {'score': 0.088, 'label': 'popinjay'}, {'score': 0.075, 'label': 'parrot'}, {'score': 0.073, 'label': 'parodist, lampooner'}, {'score': 0.046, 'label': 'poll, poll_parrot'}]
```
Learn more about the basics of using a pipeline in the [pipeline tutorial](../pipeline_tutorial)
This image classification pipeline can currently be loaded from [`pipeline`] using the following task identifier:
`"image-classification"`.
See the list of available models on
[huggingface.co/models](https://huggingface.co/models?filter=image-classification).
"""
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
requires_backends(self, "vision")
self.check_model_type(
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
if self.framework == "tf"
else MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
)
def _sanitize_parameters(self, top_k=None):
postprocess_params = {}
if top_k is not None:
postprocess_params["top_k"] = top_k
return {}, {}, postprocess_params
def __call__(self, images: Union[str, List[str], "Image.Image", List["Image.Image"]], **kwargs):
"""
Assign labels to the image(s) passed as inputs.
Args:
images (`str`, `List[str]`, `PIL.Image` or `List[PIL.Image]`):
The pipeline handles three types of images:
- A string containing a http link pointing to an image
- A string containing a local path to an image
- An image loaded in PIL directly
The pipeline accepts either a single image or a batch of images, which must then be passed as a string.
Images in a batch must all be in the same format: all as http links, all as local paths, or all as PIL
images.
top_k (`int`, *optional*, defaults to 5):
The number of top labels that will be returned by the pipeline. If the provided number is higher than
the number of labels available in the model configuration, it will default to the number of labels.
Return:
A dictionary or a list of dictionaries containing result. If the input is a single image, will return a
dictionary, if the input is a list of several images, will return a list of dictionaries corresponding to
the images.
The dictionaries contain the following keys:
- **label** (`str`) -- The label identified by the model.
- **score** (`int`) -- The score attributed by the model for that label.
"""
return super().__call__(images, **kwargs)
def preprocess(self, image):
image = load_image(image)
model_inputs = self.image_processor(images=image, return_tensors=self.framework)
return model_inputs
def _forward(self, model_inputs):
model_outputs = self.model(**model_inputs)
return model_outputs
def postprocess(self, model_outputs, top_k=5):
if top_k > self.model.config.num_labels:
top_k = self.model.config.num_labels
if self.framework == "pt":
probs = model_outputs.logits.softmax(-1)[0]
scores, ids = probs.topk(top_k)
elif self.framework == "tf":
probs = stable_softmax(model_outputs.logits, axis=-1)[0]
topk = tf.math.top_k(probs, k=top_k)
scores, ids = topk.values.numpy(), topk.indices.numpy()
else:
raise ValueError(f"Unsupported framework: {self.framework}")
scores = scores.tolist()
ids = ids.tolist()
return [{"score": score, "label": self.model.config.id2label[_id]} for score, _id in zip(scores, ids)]
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/cvt/modeling_cvt.py | # coding=utf-8
# Copyright 2022 Microsoft Research and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PyTorch CvT model."""
import collections.abc
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import torch
import torch.utils.checkpoint
from torch import nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
from ...file_utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward
from ...modeling_outputs import ImageClassifierOutputWithNoAttention, ModelOutput
from ...modeling_utils import PreTrainedModel, find_pruneable_heads_and_indices, prune_linear_layer
from ...utils import logging
from .configuration_cvt import CvtConfig
logger = logging.get_logger(__name__)
# General docstring
_CONFIG_FOR_DOC = "CvtConfig"
# Base docstring
_CHECKPOINT_FOR_DOC = "microsoft/cvt-13"
_EXPECTED_OUTPUT_SHAPE = [1, 384, 14, 14]
# Image classification docstring
_IMAGE_CLASS_CHECKPOINT = "microsoft/cvt-13"
_IMAGE_CLASS_EXPECTED_OUTPUT = "tabby, tabby cat"
CVT_PRETRAINED_MODEL_ARCHIVE_LIST = [
"microsoft/cvt-13",
"microsoft/cvt-13-384",
"microsoft/cvt-13-384-22k",
"microsoft/cvt-21",
"microsoft/cvt-21-384",
"microsoft/cvt-21-384-22k",
# See all Cvt models at https://huggingface.co/models?filter=cvt
]
@dataclass
class BaseModelOutputWithCLSToken(ModelOutput):
"""
Base class for model's outputs, with potential hidden states and attentions.
Args:
last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
cls_token_value (`torch.FloatTensor` of shape `(batch_size, 1, hidden_size)`):
Classification token at the output of the last layer of the model.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer
plus the initial embedding outputs.
"""
last_hidden_state: torch.FloatTensor = None
cls_token_value: torch.FloatTensor = None
hidden_states: Optional[Tuple[torch.FloatTensor]] = None
# Copied from transformers.models.beit.modeling_beit.drop_path
def drop_path(input, drop_prob: float = 0.0, training: bool = False):
"""
Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
Comment by Ross Wightman: This is the same as the DropConnect impl I created for EfficientNet, etc networks,
however, the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for changing the
layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use 'survival rate' as the
argument.
"""
if drop_prob == 0.0 or not training:
return input
keep_prob = 1 - drop_prob
shape = (input.shape[0],) + (1,) * (input.ndim - 1) # work with diff dim tensors, not just 2D ConvNets
random_tensor = keep_prob + torch.rand(shape, dtype=input.dtype, device=input.device)
random_tensor.floor_() # binarize
output = input.div(keep_prob) * random_tensor
return output
# Copied from transformers.models.beit.modeling_beit.BeitDropPath
class CvtDropPath(nn.Module):
"""Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks)."""
def __init__(self, drop_prob: Optional[float] = None) -> None:
super().__init__()
self.drop_prob = drop_prob
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
return drop_path(hidden_states, self.drop_prob, self.training)
def extra_repr(self) -> str:
return "p={}".format(self.drop_prob)
class CvtEmbeddings(nn.Module):
"""
Construct the CvT embeddings.
"""
def __init__(self, patch_size, num_channels, embed_dim, stride, padding, dropout_rate):
super().__init__()
self.convolution_embeddings = CvtConvEmbeddings(
patch_size=patch_size, num_channels=num_channels, embed_dim=embed_dim, stride=stride, padding=padding
)
self.dropout = nn.Dropout(dropout_rate)
def forward(self, pixel_values):
hidden_state = self.convolution_embeddings(pixel_values)
hidden_state = self.dropout(hidden_state)
return hidden_state
class CvtConvEmbeddings(nn.Module):
"""
Image to Conv Embedding.
"""
def __init__(self, patch_size, num_channels, embed_dim, stride, padding):
super().__init__()
patch_size = patch_size if isinstance(patch_size, collections.abc.Iterable) else (patch_size, patch_size)
self.patch_size = patch_size
self.projection = nn.Conv2d(num_channels, embed_dim, kernel_size=patch_size, stride=stride, padding=padding)
self.normalization = nn.LayerNorm(embed_dim)
def forward(self, pixel_values):
pixel_values = self.projection(pixel_values)
batch_size, num_channels, height, width = pixel_values.shape
hidden_size = height * width
# rearrange "b c h w -> b (h w) c"
pixel_values = pixel_values.view(batch_size, num_channels, hidden_size).permute(0, 2, 1)
if self.normalization:
pixel_values = self.normalization(pixel_values)
# rearrange "b (h w) c" -> b c h w"
pixel_values = pixel_values.permute(0, 2, 1).view(batch_size, num_channels, height, width)
return pixel_values
class CvtSelfAttentionConvProjection(nn.Module):
def __init__(self, embed_dim, kernel_size, padding, stride):
super().__init__()
self.convolution = nn.Conv2d(
embed_dim,
embed_dim,
kernel_size=kernel_size,
padding=padding,
stride=stride,
bias=False,
groups=embed_dim,
)
self.normalization = nn.BatchNorm2d(embed_dim)
def forward(self, hidden_state):
hidden_state = self.convolution(hidden_state)
hidden_state = self.normalization(hidden_state)
return hidden_state
class CvtSelfAttentionLinearProjection(nn.Module):
def forward(self, hidden_state):
batch_size, num_channels, height, width = hidden_state.shape
hidden_size = height * width
# rearrange " b c h w -> b (h w) c"
hidden_state = hidden_state.view(batch_size, num_channels, hidden_size).permute(0, 2, 1)
return hidden_state
class CvtSelfAttentionProjection(nn.Module):
def __init__(self, embed_dim, kernel_size, padding, stride, projection_method="dw_bn"):
super().__init__()
if projection_method == "dw_bn":
self.convolution_projection = CvtSelfAttentionConvProjection(embed_dim, kernel_size, padding, stride)
self.linear_projection = CvtSelfAttentionLinearProjection()
def forward(self, hidden_state):
hidden_state = self.convolution_projection(hidden_state)
hidden_state = self.linear_projection(hidden_state)
return hidden_state
class CvtSelfAttention(nn.Module):
def __init__(
self,
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
with_cls_token=True,
**kwargs,
):
super().__init__()
self.scale = embed_dim**-0.5
self.with_cls_token = with_cls_token
self.embed_dim = embed_dim
self.num_heads = num_heads
self.convolution_projection_query = CvtSelfAttentionProjection(
embed_dim,
kernel_size,
padding_q,
stride_q,
projection_method="linear" if qkv_projection_method == "avg" else qkv_projection_method,
)
self.convolution_projection_key = CvtSelfAttentionProjection(
embed_dim, kernel_size, padding_kv, stride_kv, projection_method=qkv_projection_method
)
self.convolution_projection_value = CvtSelfAttentionProjection(
embed_dim, kernel_size, padding_kv, stride_kv, projection_method=qkv_projection_method
)
self.projection_query = nn.Linear(embed_dim, embed_dim, bias=qkv_bias)
self.projection_key = nn.Linear(embed_dim, embed_dim, bias=qkv_bias)
self.projection_value = nn.Linear(embed_dim, embed_dim, bias=qkv_bias)
self.dropout = nn.Dropout(attention_drop_rate)
def rearrange_for_multi_head_attention(self, hidden_state):
batch_size, hidden_size, _ = hidden_state.shape
head_dim = self.embed_dim // self.num_heads
# rearrange 'b t (h d) -> b h t d'
return hidden_state.view(batch_size, hidden_size, self.num_heads, head_dim).permute(0, 2, 1, 3)
def forward(self, hidden_state, height, width):
if self.with_cls_token:
cls_token, hidden_state = torch.split(hidden_state, [1, height * width], 1)
batch_size, hidden_size, num_channels = hidden_state.shape
# rearrange "b (h w) c -> b c h w"
hidden_state = hidden_state.permute(0, 2, 1).view(batch_size, num_channels, height, width)
key = self.convolution_projection_key(hidden_state)
query = self.convolution_projection_query(hidden_state)
value = self.convolution_projection_value(hidden_state)
if self.with_cls_token:
query = torch.cat((cls_token, query), dim=1)
key = torch.cat((cls_token, key), dim=1)
value = torch.cat((cls_token, value), dim=1)
head_dim = self.embed_dim // self.num_heads
query = self.rearrange_for_multi_head_attention(self.projection_query(query))
key = self.rearrange_for_multi_head_attention(self.projection_key(key))
value = self.rearrange_for_multi_head_attention(self.projection_value(value))
attention_score = torch.einsum("bhlk,bhtk->bhlt", [query, key]) * self.scale
attention_probs = torch.nn.functional.softmax(attention_score, dim=-1)
attention_probs = self.dropout(attention_probs)
context = torch.einsum("bhlt,bhtv->bhlv", [attention_probs, value])
# rearrange"b h t d -> b t (h d)"
_, _, hidden_size, _ = context.shape
context = context.permute(0, 2, 1, 3).contiguous().view(batch_size, hidden_size, self.num_heads * head_dim)
return context
class CvtSelfOutput(nn.Module):
"""
The residual connection is defined in CvtLayer instead of here (as is the case with other models), due to the
layernorm applied before each block.
"""
def __init__(self, embed_dim, drop_rate):
super().__init__()
self.dense = nn.Linear(embed_dim, embed_dim)
self.dropout = nn.Dropout(drop_rate)
def forward(self, hidden_state, input_tensor):
hidden_state = self.dense(hidden_state)
hidden_state = self.dropout(hidden_state)
return hidden_state
class CvtAttention(nn.Module):
def __init__(
self,
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
with_cls_token=True,
):
super().__init__()
self.attention = CvtSelfAttention(
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
with_cls_token,
)
self.output = CvtSelfOutput(embed_dim, drop_rate)
self.pruned_heads = set()
def prune_heads(self, heads):
if len(heads) == 0:
return
heads, index = find_pruneable_heads_and_indices(
heads, self.attention.num_attention_heads, self.attention.attention_head_size, self.pruned_heads
)
# Prune linear layers
self.attention.query = prune_linear_layer(self.attention.query, index)
self.attention.key = prune_linear_layer(self.attention.key, index)
self.attention.value = prune_linear_layer(self.attention.value, index)
self.output.dense = prune_linear_layer(self.output.dense, index, dim=1)
# Update hyper params and store pruned heads
self.attention.num_attention_heads = self.attention.num_attention_heads - len(heads)
self.attention.all_head_size = self.attention.attention_head_size * self.attention.num_attention_heads
self.pruned_heads = self.pruned_heads.union(heads)
def forward(self, hidden_state, height, width):
self_output = self.attention(hidden_state, height, width)
attention_output = self.output(self_output, hidden_state)
return attention_output
class CvtIntermediate(nn.Module):
def __init__(self, embed_dim, mlp_ratio):
super().__init__()
self.dense = nn.Linear(embed_dim, int(embed_dim * mlp_ratio))
self.activation = nn.GELU()
def forward(self, hidden_state):
hidden_state = self.dense(hidden_state)
hidden_state = self.activation(hidden_state)
return hidden_state
class CvtOutput(nn.Module):
def __init__(self, embed_dim, mlp_ratio, drop_rate):
super().__init__()
self.dense = nn.Linear(int(embed_dim * mlp_ratio), embed_dim)
self.dropout = nn.Dropout(drop_rate)
def forward(self, hidden_state, input_tensor):
hidden_state = self.dense(hidden_state)
hidden_state = self.dropout(hidden_state)
hidden_state = hidden_state + input_tensor
return hidden_state
class CvtLayer(nn.Module):
"""
CvtLayer composed by attention layers, normalization and multi-layer perceptrons (mlps).
"""
def __init__(
self,
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
mlp_ratio,
drop_path_rate,
with_cls_token=True,
):
super().__init__()
self.attention = CvtAttention(
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
with_cls_token,
)
self.intermediate = CvtIntermediate(embed_dim, mlp_ratio)
self.output = CvtOutput(embed_dim, mlp_ratio, drop_rate)
self.drop_path = CvtDropPath(drop_prob=drop_path_rate) if drop_path_rate > 0.0 else nn.Identity()
self.layernorm_before = nn.LayerNorm(embed_dim)
self.layernorm_after = nn.LayerNorm(embed_dim)
def forward(self, hidden_state, height, width):
self_attention_output = self.attention(
self.layernorm_before(hidden_state), # in Cvt, layernorm is applied before self-attention
height,
width,
)
attention_output = self_attention_output
attention_output = self.drop_path(attention_output)
# first residual connection
hidden_state = attention_output + hidden_state
# in Cvt, layernorm is also applied after self-attention
layer_output = self.layernorm_after(hidden_state)
layer_output = self.intermediate(layer_output)
# second residual connection is done here
layer_output = self.output(layer_output, hidden_state)
layer_output = self.drop_path(layer_output)
return layer_output
class CvtStage(nn.Module):
def __init__(self, config, stage):
super().__init__()
self.config = config
self.stage = stage
if self.config.cls_token[self.stage]:
self.cls_token = nn.Parameter(torch.randn(1, 1, self.config.embed_dim[-1]))
self.embedding = CvtEmbeddings(
patch_size=config.patch_sizes[self.stage],
stride=config.patch_stride[self.stage],
num_channels=config.num_channels if self.stage == 0 else config.embed_dim[self.stage - 1],
embed_dim=config.embed_dim[self.stage],
padding=config.patch_padding[self.stage],
dropout_rate=config.drop_rate[self.stage],
)
drop_path_rates = [x.item() for x in torch.linspace(0, config.drop_path_rate[self.stage], config.depth[stage])]
self.layers = nn.Sequential(
*[
CvtLayer(
num_heads=config.num_heads[self.stage],
embed_dim=config.embed_dim[self.stage],
kernel_size=config.kernel_qkv[self.stage],
padding_q=config.padding_q[self.stage],
padding_kv=config.padding_kv[self.stage],
stride_kv=config.stride_kv[self.stage],
stride_q=config.stride_q[self.stage],
qkv_projection_method=config.qkv_projection_method[self.stage],
qkv_bias=config.qkv_bias[self.stage],
attention_drop_rate=config.attention_drop_rate[self.stage],
drop_rate=config.drop_rate[self.stage],
drop_path_rate=drop_path_rates[self.stage],
mlp_ratio=config.mlp_ratio[self.stage],
with_cls_token=config.cls_token[self.stage],
)
for _ in range(config.depth[self.stage])
]
)
def forward(self, hidden_state):
cls_token = None
hidden_state = self.embedding(hidden_state)
batch_size, num_channels, height, width = hidden_state.shape
# rearrange b c h w -> b (h w) c"
hidden_state = hidden_state.view(batch_size, num_channels, height * width).permute(0, 2, 1)
if self.config.cls_token[self.stage]:
cls_token = self.cls_token.expand(batch_size, -1, -1)
hidden_state = torch.cat((cls_token, hidden_state), dim=1)
for layer in self.layers:
layer_outputs = layer(hidden_state, height, width)
hidden_state = layer_outputs
if self.config.cls_token[self.stage]:
cls_token, hidden_state = torch.split(hidden_state, [1, height * width], 1)
hidden_state = hidden_state.permute(0, 2, 1).view(batch_size, num_channels, height, width)
return hidden_state, cls_token
class CvtEncoder(nn.Module):
def __init__(self, config):
super().__init__()
self.config = config
self.stages = nn.ModuleList([])
for stage_idx in range(len(config.depth)):
self.stages.append(CvtStage(config, stage_idx))
def forward(self, pixel_values, output_hidden_states=False, return_dict=True):
all_hidden_states = () if output_hidden_states else None
hidden_state = pixel_values
cls_token = None
for _, (stage_module) in enumerate(self.stages):
hidden_state, cls_token = stage_module(hidden_state)
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_state,)
if not return_dict:
return tuple(v for v in [hidden_state, cls_token, all_hidden_states] if v is not None)
return BaseModelOutputWithCLSToken(
last_hidden_state=hidden_state,
cls_token_value=cls_token,
hidden_states=all_hidden_states,
)
class CvtPreTrainedModel(PreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = CvtConfig
base_model_prefix = "cvt"
main_input_name = "pixel_values"
def _init_weights(self, module):
"""Initialize the weights"""
if isinstance(module, (nn.Linear, nn.Conv2d)):
module.weight.data = nn.init.trunc_normal_(module.weight.data, mean=0.0, std=self.config.initializer_range)
if module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.LayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
elif isinstance(module, CvtStage):
if self.config.cls_token[module.stage]:
module.cls_token.data = nn.init.trunc_normal_(
torch.zeros(1, 1, self.config.embed_dim[-1]), mean=0.0, std=self.config.initializer_range
)
CVT_START_DOCSTRING = r"""
This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass. Use it
as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
Parameters:
config ([`CvtConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""
CVT_INPUTS_DOCSTRING = r"""
Args:
pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See [`CvtImageProcessor.__call__`]
for details.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
@add_start_docstrings(
"The bare Cvt Model transformer outputting raw hidden-states without any specific head on top.",
CVT_START_DOCSTRING,
)
class CvtModel(CvtPreTrainedModel):
def __init__(self, config, add_pooling_layer=True):
super().__init__(config)
self.config = config
self.encoder = CvtEncoder(config)
self.post_init()
def _prune_heads(self, heads_to_prune):
"""
Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base
class PreTrainedModel
"""
for layer, heads in heads_to_prune.items():
self.encoder.layer[layer].attention.prune_heads(heads)
@add_start_docstrings_to_model_forward(CVT_INPUTS_DOCSTRING)
@add_code_sample_docstrings(
checkpoint=_CHECKPOINT_FOR_DOC,
output_type=BaseModelOutputWithCLSToken,
config_class=_CONFIG_FOR_DOC,
modality="vision",
expected_output=_EXPECTED_OUTPUT_SHAPE,
)
def forward(
self,
pixel_values: Optional[torch.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple, BaseModelOutputWithCLSToken]:
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if pixel_values is None:
raise ValueError("You have to specify pixel_values")
encoder_outputs = self.encoder(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = encoder_outputs[0]
if not return_dict:
return (sequence_output,) + encoder_outputs[1:]
return BaseModelOutputWithCLSToken(
last_hidden_state=sequence_output,
cls_token_value=encoder_outputs.cls_token_value,
hidden_states=encoder_outputs.hidden_states,
)
@add_start_docstrings(
"""
Cvt Model transformer with an image classification head on top (a linear layer on top of the final hidden state of
the [CLS] token) e.g. for ImageNet.
""",
CVT_START_DOCSTRING,
)
class CvtForImageClassification(CvtPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.cvt = CvtModel(config, add_pooling_layer=False)
self.layernorm = nn.LayerNorm(config.embed_dim[-1])
# Classifier head
self.classifier = (
nn.Linear(config.embed_dim[-1], config.num_labels) if config.num_labels > 0 else nn.Identity()
)
# Initialize weights and apply final processing
self.post_init()
@add_start_docstrings_to_model_forward(CVT_INPUTS_DOCSTRING)
@add_code_sample_docstrings(
checkpoint=_IMAGE_CLASS_CHECKPOINT,
output_type=ImageClassifierOutputWithNoAttention,
config_class=_CONFIG_FOR_DOC,
expected_output=_IMAGE_CLASS_EXPECTED_OUTPUT,
)
def forward(
self,
pixel_values: Optional[torch.Tensor] = None,
labels: Optional[torch.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple, ImageClassifierOutputWithNoAttention]:
r"""
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the image classification/regression loss. Indices should be in `[0, ...,
config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.cvt(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
cls_token = outputs[1]
if self.config.cls_token[-1]:
sequence_output = self.layernorm(cls_token)
else:
batch_size, num_channels, height, width = sequence_output.shape
# rearrange "b c h w -> b (h w) c"
sequence_output = sequence_output.view(batch_size, num_channels, height * width).permute(0, 2, 1)
sequence_output = self.layernorm(sequence_output)
sequence_output_mean = sequence_output.mean(dim=1)
logits = self.classifier(sequence_output_mean)
loss = None
if labels is not None:
if self.config.problem_type is None:
if self.config.num_labels == 1:
self.config.problem_type = "regression"
elif self.config.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
self.config.problem_type = "single_label_classification"
else:
self.config.problem_type = "multi_label_classification"
if self.config.problem_type == "regression":
loss_fct = MSELoss()
if self.config.num_labels == 1:
loss = loss_fct(logits.squeeze(), labels.squeeze())
else:
loss = loss_fct(logits, labels)
elif self.config.problem_type == "single_label_classification":
loss_fct = CrossEntropyLoss()
loss = loss_fct(logits.view(-1, self.config.num_labels), labels.view(-1))
elif self.config.problem_type == "multi_label_classification":
loss_fct = BCEWithLogitsLoss()
loss = loss_fct(logits, labels)
if not return_dict:
output = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return ImageClassifierOutputWithNoAttention(loss=loss, logits=logits, hidden_states=outputs.hidden_states)
| # coding=utf-8
# Copyright 2022 Microsoft Research and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PyTorch CvT model."""
import collections.abc
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import torch
import torch.utils.checkpoint
from torch import nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
from ...file_utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward
from ...modeling_outputs import ImageClassifierOutputWithNoAttention, ModelOutput
from ...modeling_utils import PreTrainedModel, find_pruneable_heads_and_indices, prune_linear_layer
from ...utils import logging
from .configuration_cvt import CvtConfig
logger = logging.get_logger(__name__)
# General docstring
_CONFIG_FOR_DOC = "CvtConfig"
# Base docstring
_CHECKPOINT_FOR_DOC = "microsoft/cvt-13"
_EXPECTED_OUTPUT_SHAPE = [1, 384, 14, 14]
# Image classification docstring
_IMAGE_CLASS_CHECKPOINT = "microsoft/cvt-13"
_IMAGE_CLASS_EXPECTED_OUTPUT = "tabby, tabby cat"
CVT_PRETRAINED_MODEL_ARCHIVE_LIST = [
"microsoft/cvt-13",
"microsoft/cvt-13-384",
"microsoft/cvt-13-384-22k",
"microsoft/cvt-21",
"microsoft/cvt-21-384",
"microsoft/cvt-21-384-22k",
# See all Cvt models at https://huggingface.co/models?filter=cvt
]
@dataclass
class BaseModelOutputWithCLSToken(ModelOutput):
"""
Base class for model's outputs, with potential hidden states and attentions.
Args:
last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
Sequence of hidden-states at the output of the last layer of the model.
cls_token_value (`torch.FloatTensor` of shape `(batch_size, 1, hidden_size)`):
Classification token at the output of the last layer of the model.
hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer
plus the initial embedding outputs.
"""
last_hidden_state: torch.FloatTensor = None
cls_token_value: torch.FloatTensor = None
hidden_states: Optional[Tuple[torch.FloatTensor]] = None
# Copied from transformers.models.beit.modeling_beit.drop_path
def drop_path(input, drop_prob: float = 0.0, training: bool = False):
"""
Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
Comment by Ross Wightman: This is the same as the DropConnect impl I created for EfficientNet, etc networks,
however, the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for changing the
layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use 'survival rate' as the
argument.
"""
if drop_prob == 0.0 or not training:
return input
keep_prob = 1 - drop_prob
shape = (input.shape[0],) + (1,) * (input.ndim - 1) # work with diff dim tensors, not just 2D ConvNets
random_tensor = keep_prob + torch.rand(shape, dtype=input.dtype, device=input.device)
random_tensor.floor_() # binarize
output = input.div(keep_prob) * random_tensor
return output
# Copied from transformers.models.beit.modeling_beit.BeitDropPath
class CvtDropPath(nn.Module):
"""Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks)."""
def __init__(self, drop_prob: Optional[float] = None) -> None:
super().__init__()
self.drop_prob = drop_prob
def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
return drop_path(hidden_states, self.drop_prob, self.training)
def extra_repr(self) -> str:
return "p={}".format(self.drop_prob)
class CvtEmbeddings(nn.Module):
"""
Construct the CvT embeddings.
"""
def __init__(self, patch_size, num_channels, embed_dim, stride, padding, dropout_rate):
super().__init__()
self.convolution_embeddings = CvtConvEmbeddings(
patch_size=patch_size, num_channels=num_channels, embed_dim=embed_dim, stride=stride, padding=padding
)
self.dropout = nn.Dropout(dropout_rate)
def forward(self, pixel_values):
hidden_state = self.convolution_embeddings(pixel_values)
hidden_state = self.dropout(hidden_state)
return hidden_state
class CvtConvEmbeddings(nn.Module):
"""
Image to Conv Embedding.
"""
def __init__(self, patch_size, num_channels, embed_dim, stride, padding):
super().__init__()
patch_size = patch_size if isinstance(patch_size, collections.abc.Iterable) else (patch_size, patch_size)
self.patch_size = patch_size
self.projection = nn.Conv2d(num_channels, embed_dim, kernel_size=patch_size, stride=stride, padding=padding)
self.normalization = nn.LayerNorm(embed_dim)
def forward(self, pixel_values):
pixel_values = self.projection(pixel_values)
batch_size, num_channels, height, width = pixel_values.shape
hidden_size = height * width
# rearrange "b c h w -> b (h w) c"
pixel_values = pixel_values.view(batch_size, num_channels, hidden_size).permute(0, 2, 1)
if self.normalization:
pixel_values = self.normalization(pixel_values)
# rearrange "b (h w) c" -> b c h w"
pixel_values = pixel_values.permute(0, 2, 1).view(batch_size, num_channels, height, width)
return pixel_values
class CvtSelfAttentionConvProjection(nn.Module):
def __init__(self, embed_dim, kernel_size, padding, stride):
super().__init__()
self.convolution = nn.Conv2d(
embed_dim,
embed_dim,
kernel_size=kernel_size,
padding=padding,
stride=stride,
bias=False,
groups=embed_dim,
)
self.normalization = nn.BatchNorm2d(embed_dim)
def forward(self, hidden_state):
hidden_state = self.convolution(hidden_state)
hidden_state = self.normalization(hidden_state)
return hidden_state
class CvtSelfAttentionLinearProjection(nn.Module):
def forward(self, hidden_state):
batch_size, num_channels, height, width = hidden_state.shape
hidden_size = height * width
# rearrange " b c h w -> b (h w) c"
hidden_state = hidden_state.view(batch_size, num_channels, hidden_size).permute(0, 2, 1)
return hidden_state
class CvtSelfAttentionProjection(nn.Module):
def __init__(self, embed_dim, kernel_size, padding, stride, projection_method="dw_bn"):
super().__init__()
if projection_method == "dw_bn":
self.convolution_projection = CvtSelfAttentionConvProjection(embed_dim, kernel_size, padding, stride)
self.linear_projection = CvtSelfAttentionLinearProjection()
def forward(self, hidden_state):
hidden_state = self.convolution_projection(hidden_state)
hidden_state = self.linear_projection(hidden_state)
return hidden_state
class CvtSelfAttention(nn.Module):
def __init__(
self,
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
with_cls_token=True,
**kwargs,
):
super().__init__()
self.scale = embed_dim**-0.5
self.with_cls_token = with_cls_token
self.embed_dim = embed_dim
self.num_heads = num_heads
self.convolution_projection_query = CvtSelfAttentionProjection(
embed_dim,
kernel_size,
padding_q,
stride_q,
projection_method="linear" if qkv_projection_method == "avg" else qkv_projection_method,
)
self.convolution_projection_key = CvtSelfAttentionProjection(
embed_dim, kernel_size, padding_kv, stride_kv, projection_method=qkv_projection_method
)
self.convolution_projection_value = CvtSelfAttentionProjection(
embed_dim, kernel_size, padding_kv, stride_kv, projection_method=qkv_projection_method
)
self.projection_query = nn.Linear(embed_dim, embed_dim, bias=qkv_bias)
self.projection_key = nn.Linear(embed_dim, embed_dim, bias=qkv_bias)
self.projection_value = nn.Linear(embed_dim, embed_dim, bias=qkv_bias)
self.dropout = nn.Dropout(attention_drop_rate)
def rearrange_for_multi_head_attention(self, hidden_state):
batch_size, hidden_size, _ = hidden_state.shape
head_dim = self.embed_dim // self.num_heads
# rearrange 'b t (h d) -> b h t d'
return hidden_state.view(batch_size, hidden_size, self.num_heads, head_dim).permute(0, 2, 1, 3)
def forward(self, hidden_state, height, width):
if self.with_cls_token:
cls_token, hidden_state = torch.split(hidden_state, [1, height * width], 1)
batch_size, hidden_size, num_channels = hidden_state.shape
# rearrange "b (h w) c -> b c h w"
hidden_state = hidden_state.permute(0, 2, 1).view(batch_size, num_channels, height, width)
key = self.convolution_projection_key(hidden_state)
query = self.convolution_projection_query(hidden_state)
value = self.convolution_projection_value(hidden_state)
if self.with_cls_token:
query = torch.cat((cls_token, query), dim=1)
key = torch.cat((cls_token, key), dim=1)
value = torch.cat((cls_token, value), dim=1)
head_dim = self.embed_dim // self.num_heads
query = self.rearrange_for_multi_head_attention(self.projection_query(query))
key = self.rearrange_for_multi_head_attention(self.projection_key(key))
value = self.rearrange_for_multi_head_attention(self.projection_value(value))
attention_score = torch.einsum("bhlk,bhtk->bhlt", [query, key]) * self.scale
attention_probs = torch.nn.functional.softmax(attention_score, dim=-1)
attention_probs = self.dropout(attention_probs)
context = torch.einsum("bhlt,bhtv->bhlv", [attention_probs, value])
# rearrange"b h t d -> b t (h d)"
_, _, hidden_size, _ = context.shape
context = context.permute(0, 2, 1, 3).contiguous().view(batch_size, hidden_size, self.num_heads * head_dim)
return context
class CvtSelfOutput(nn.Module):
"""
The residual connection is defined in CvtLayer instead of here (as is the case with other models), due to the
layernorm applied before each block.
"""
def __init__(self, embed_dim, drop_rate):
super().__init__()
self.dense = nn.Linear(embed_dim, embed_dim)
self.dropout = nn.Dropout(drop_rate)
def forward(self, hidden_state, input_tensor):
hidden_state = self.dense(hidden_state)
hidden_state = self.dropout(hidden_state)
return hidden_state
class CvtAttention(nn.Module):
def __init__(
self,
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
with_cls_token=True,
):
super().__init__()
self.attention = CvtSelfAttention(
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
with_cls_token,
)
self.output = CvtSelfOutput(embed_dim, drop_rate)
self.pruned_heads = set()
def prune_heads(self, heads):
if len(heads) == 0:
return
heads, index = find_pruneable_heads_and_indices(
heads, self.attention.num_attention_heads, self.attention.attention_head_size, self.pruned_heads
)
# Prune linear layers
self.attention.query = prune_linear_layer(self.attention.query, index)
self.attention.key = prune_linear_layer(self.attention.key, index)
self.attention.value = prune_linear_layer(self.attention.value, index)
self.output.dense = prune_linear_layer(self.output.dense, index, dim=1)
# Update hyper params and store pruned heads
self.attention.num_attention_heads = self.attention.num_attention_heads - len(heads)
self.attention.all_head_size = self.attention.attention_head_size * self.attention.num_attention_heads
self.pruned_heads = self.pruned_heads.union(heads)
def forward(self, hidden_state, height, width):
self_output = self.attention(hidden_state, height, width)
attention_output = self.output(self_output, hidden_state)
return attention_output
class CvtIntermediate(nn.Module):
def __init__(self, embed_dim, mlp_ratio):
super().__init__()
self.dense = nn.Linear(embed_dim, int(embed_dim * mlp_ratio))
self.activation = nn.GELU()
def forward(self, hidden_state):
hidden_state = self.dense(hidden_state)
hidden_state = self.activation(hidden_state)
return hidden_state
class CvtOutput(nn.Module):
def __init__(self, embed_dim, mlp_ratio, drop_rate):
super().__init__()
self.dense = nn.Linear(int(embed_dim * mlp_ratio), embed_dim)
self.dropout = nn.Dropout(drop_rate)
def forward(self, hidden_state, input_tensor):
hidden_state = self.dense(hidden_state)
hidden_state = self.dropout(hidden_state)
hidden_state = hidden_state + input_tensor
return hidden_state
class CvtLayer(nn.Module):
"""
CvtLayer composed by attention layers, normalization and multi-layer perceptrons (mlps).
"""
def __init__(
self,
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
mlp_ratio,
drop_path_rate,
with_cls_token=True,
):
super().__init__()
self.attention = CvtAttention(
num_heads,
embed_dim,
kernel_size,
padding_q,
padding_kv,
stride_q,
stride_kv,
qkv_projection_method,
qkv_bias,
attention_drop_rate,
drop_rate,
with_cls_token,
)
self.intermediate = CvtIntermediate(embed_dim, mlp_ratio)
self.output = CvtOutput(embed_dim, mlp_ratio, drop_rate)
self.drop_path = CvtDropPath(drop_prob=drop_path_rate) if drop_path_rate > 0.0 else nn.Identity()
self.layernorm_before = nn.LayerNorm(embed_dim)
self.layernorm_after = nn.LayerNorm(embed_dim)
def forward(self, hidden_state, height, width):
self_attention_output = self.attention(
self.layernorm_before(hidden_state), # in Cvt, layernorm is applied before self-attention
height,
width,
)
attention_output = self_attention_output
attention_output = self.drop_path(attention_output)
# first residual connection
hidden_state = attention_output + hidden_state
# in Cvt, layernorm is also applied after self-attention
layer_output = self.layernorm_after(hidden_state)
layer_output = self.intermediate(layer_output)
# second residual connection is done here
layer_output = self.output(layer_output, hidden_state)
layer_output = self.drop_path(layer_output)
return layer_output
class CvtStage(nn.Module):
def __init__(self, config, stage):
super().__init__()
self.config = config
self.stage = stage
if self.config.cls_token[self.stage]:
self.cls_token = nn.Parameter(torch.randn(1, 1, self.config.embed_dim[-1]))
self.embedding = CvtEmbeddings(
patch_size=config.patch_sizes[self.stage],
stride=config.patch_stride[self.stage],
num_channels=config.num_channels if self.stage == 0 else config.embed_dim[self.stage - 1],
embed_dim=config.embed_dim[self.stage],
padding=config.patch_padding[self.stage],
dropout_rate=config.drop_rate[self.stage],
)
drop_path_rates = [x.item() for x in torch.linspace(0, config.drop_path_rate[self.stage], config.depth[stage])]
self.layers = nn.Sequential(
*[
CvtLayer(
num_heads=config.num_heads[self.stage],
embed_dim=config.embed_dim[self.stage],
kernel_size=config.kernel_qkv[self.stage],
padding_q=config.padding_q[self.stage],
padding_kv=config.padding_kv[self.stage],
stride_kv=config.stride_kv[self.stage],
stride_q=config.stride_q[self.stage],
qkv_projection_method=config.qkv_projection_method[self.stage],
qkv_bias=config.qkv_bias[self.stage],
attention_drop_rate=config.attention_drop_rate[self.stage],
drop_rate=config.drop_rate[self.stage],
drop_path_rate=drop_path_rates[self.stage],
mlp_ratio=config.mlp_ratio[self.stage],
with_cls_token=config.cls_token[self.stage],
)
for _ in range(config.depth[self.stage])
]
)
def forward(self, hidden_state):
cls_token = None
hidden_state = self.embedding(hidden_state)
batch_size, num_channels, height, width = hidden_state.shape
# rearrange b c h w -> b (h w) c"
hidden_state = hidden_state.view(batch_size, num_channels, height * width).permute(0, 2, 1)
if self.config.cls_token[self.stage]:
cls_token = self.cls_token.expand(batch_size, -1, -1)
hidden_state = torch.cat((cls_token, hidden_state), dim=1)
for layer in self.layers:
layer_outputs = layer(hidden_state, height, width)
hidden_state = layer_outputs
if self.config.cls_token[self.stage]:
cls_token, hidden_state = torch.split(hidden_state, [1, height * width], 1)
hidden_state = hidden_state.permute(0, 2, 1).view(batch_size, num_channels, height, width)
return hidden_state, cls_token
class CvtEncoder(nn.Module):
def __init__(self, config):
super().__init__()
self.config = config
self.stages = nn.ModuleList([])
for stage_idx in range(len(config.depth)):
self.stages.append(CvtStage(config, stage_idx))
def forward(self, pixel_values, output_hidden_states=False, return_dict=True):
all_hidden_states = () if output_hidden_states else None
hidden_state = pixel_values
cls_token = None
for _, (stage_module) in enumerate(self.stages):
hidden_state, cls_token = stage_module(hidden_state)
if output_hidden_states:
all_hidden_states = all_hidden_states + (hidden_state,)
if not return_dict:
return tuple(v for v in [hidden_state, cls_token, all_hidden_states] if v is not None)
return BaseModelOutputWithCLSToken(
last_hidden_state=hidden_state,
cls_token_value=cls_token,
hidden_states=all_hidden_states,
)
class CvtPreTrainedModel(PreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = CvtConfig
base_model_prefix = "cvt"
main_input_name = "pixel_values"
def _init_weights(self, module):
"""Initialize the weights"""
if isinstance(module, (nn.Linear, nn.Conv2d)):
module.weight.data = nn.init.trunc_normal_(module.weight.data, mean=0.0, std=self.config.initializer_range)
if module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.LayerNorm):
module.bias.data.zero_()
module.weight.data.fill_(1.0)
elif isinstance(module, CvtStage):
if self.config.cls_token[module.stage]:
module.cls_token.data = nn.init.trunc_normal_(
torch.zeros(1, 1, self.config.embed_dim[-1]), mean=0.0, std=self.config.initializer_range
)
CVT_START_DOCSTRING = r"""
This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass. Use it
as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and
behavior.
Parameters:
config ([`CvtConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""
CVT_INPUTS_DOCSTRING = r"""
Args:
pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See [`CvtImageProcessor.__call__`]
for details.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.
"""
@add_start_docstrings(
"The bare Cvt Model transformer outputting raw hidden-states without any specific head on top.",
CVT_START_DOCSTRING,
)
class CvtModel(CvtPreTrainedModel):
def __init__(self, config, add_pooling_layer=True):
super().__init__(config)
self.config = config
self.encoder = CvtEncoder(config)
self.post_init()
def _prune_heads(self, heads_to_prune):
"""
Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base
class PreTrainedModel
"""
for layer, heads in heads_to_prune.items():
self.encoder.layer[layer].attention.prune_heads(heads)
@add_start_docstrings_to_model_forward(CVT_INPUTS_DOCSTRING)
@add_code_sample_docstrings(
checkpoint=_CHECKPOINT_FOR_DOC,
output_type=BaseModelOutputWithCLSToken,
config_class=_CONFIG_FOR_DOC,
modality="vision",
expected_output=_EXPECTED_OUTPUT_SHAPE,
)
def forward(
self,
pixel_values: Optional[torch.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple, BaseModelOutputWithCLSToken]:
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if pixel_values is None:
raise ValueError("You have to specify pixel_values")
encoder_outputs = self.encoder(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = encoder_outputs[0]
if not return_dict:
return (sequence_output,) + encoder_outputs[1:]
return BaseModelOutputWithCLSToken(
last_hidden_state=sequence_output,
cls_token_value=encoder_outputs.cls_token_value,
hidden_states=encoder_outputs.hidden_states,
)
@add_start_docstrings(
"""
Cvt Model transformer with an image classification head on top (a linear layer on top of the final hidden state of
the [CLS] token) e.g. for ImageNet.
""",
CVT_START_DOCSTRING,
)
class CvtForImageClassification(CvtPreTrainedModel):
def __init__(self, config):
super().__init__(config)
self.num_labels = config.num_labels
self.cvt = CvtModel(config, add_pooling_layer=False)
self.layernorm = nn.LayerNorm(config.embed_dim[-1])
# Classifier head
self.classifier = (
nn.Linear(config.embed_dim[-1], config.num_labels) if config.num_labels > 0 else nn.Identity()
)
# Initialize weights and apply final processing
self.post_init()
@add_start_docstrings_to_model_forward(CVT_INPUTS_DOCSTRING)
@add_code_sample_docstrings(
checkpoint=_IMAGE_CLASS_CHECKPOINT,
output_type=ImageClassifierOutputWithNoAttention,
config_class=_CONFIG_FOR_DOC,
expected_output=_IMAGE_CLASS_EXPECTED_OUTPUT,
)
def forward(
self,
pixel_values: Optional[torch.Tensor] = None,
labels: Optional[torch.Tensor] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple, ImageClassifierOutputWithNoAttention]:
r"""
labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
Labels for computing the image classification/regression loss. Indices should be in `[0, ...,
config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
outputs = self.cvt(
pixel_values,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
cls_token = outputs[1]
if self.config.cls_token[-1]:
sequence_output = self.layernorm(cls_token)
else:
batch_size, num_channels, height, width = sequence_output.shape
# rearrange "b c h w -> b (h w) c"
sequence_output = sequence_output.view(batch_size, num_channels, height * width).permute(0, 2, 1)
sequence_output = self.layernorm(sequence_output)
sequence_output_mean = sequence_output.mean(dim=1)
logits = self.classifier(sequence_output_mean)
loss = None
if labels is not None:
if self.config.problem_type is None:
if self.config.num_labels == 1:
self.config.problem_type = "regression"
elif self.config.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
self.config.problem_type = "single_label_classification"
else:
self.config.problem_type = "multi_label_classification"
if self.config.problem_type == "regression":
loss_fct = MSELoss()
if self.config.num_labels == 1:
loss = loss_fct(logits.squeeze(), labels.squeeze())
else:
loss = loss_fct(logits, labels)
elif self.config.problem_type == "single_label_classification":
loss_fct = CrossEntropyLoss()
loss = loss_fct(logits.view(-1, self.config.num_labels), labels.view(-1))
elif self.config.problem_type == "multi_label_classification":
loss_fct = BCEWithLogitsLoss()
loss = loss_fct(logits, labels)
if not return_dict:
output = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return ImageClassifierOutputWithNoAttention(loss=loss, logits=logits, hidden_states=outputs.hidden_states)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./ISSUES.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
# How To Request Support
This is an Open Source Project so please be mindful that like in any other project of this kind there is no obligation to answer all requests for help.
However, we want to encourage you to ask for help whenever you think it's needed! We are happy about every question we get because it allows us to better understand your needs, possible misunderstandings, and most importantly a way for you to help us make this library better. That being said, this document's main purpose is to provide guidelines at how you can formulate your requests to increase your chances to be understood and to get support.
There are two main venues to receive support: [the forums](https://discuss.huggingface.co/) and [the GitHub issues](https://github.com/huggingface/transformers/issues).
## The Forums
[The user forums](https://discuss.huggingface.co/) are supported by the wide community of the library users and backed up by developers when needed.
If you have a difficulty with deploying this library or some questions, or you'd like to discuss a new feature, please first consider discussing those things at the forums. Only when you feel your subject matter has been crystalized and you still need support from the library developers do proceed to file an [issue](https://github.com/huggingface/transformers/issues).
In particular all "Please explain" questions or objectively very user-specific feature requests belong to the forums. Here are some example of such questions:
* "I would like to use a BertModel within a RL-Agent for a customer support service. How can I use a BertForMaskedLM in my ChatBotModel?"
* "Could you please explain why T5 has no positional embedding matrix under T5Model?"
* "How should I set my generation parameters for translation?"
* "How to train T5 on De->En translation?"
## The GitHub Issues
Everything which hints at a bug should be opened as an [issue](https://github.com/huggingface/transformers/issues).
You are not required to read the following guidelines before opening an issue. However, if you notice that your issue doesn't get any replies, chances are that the developers have one or several difficulties with its quality. In this case, reading the following points and adjusting your issue accordingly could help.
1. Before posting an issue, first search for already posted issues, since chances are someone has already asked a similar question before you.
If you use Google your search query should be:
```
"huggingface" "transformers" your query
```
The first two quoted words tell Google to limit the search to the context of the Huggingface Transformers. The remainder is your query - most commonly this would be the error message the software fails with. We will go deeper into details shortly.
The results of such a query will typically match GitHub issues, Hugging Face forums, StackExchange, and blogs.
If you find relevant hints, you may choose to continue the discussion there if you have follow up questions.
If what you found is similar but doesn't quite answer your problem, please, post a new issue and do include links to similar issues or forum discussions you may have found.
Let's look at some examples:
The error message, often referred to as an assertion, tells us what went wrong. Here is an example of an assertion:
```python
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/transformers/src/transformers/__init__.py", line 34, in <module>
from . import dependency_versions_check
File "/transformers/src/transformers/dependency_versions_check.py", line 34, in <module>
from .utils import is_tokenizers_available
File "/transformers/src/transformers/utils/import_utils.py", line 40, in <module>
from tqdm.auto import tqdm
ModuleNotFoundError: No module named 'tqdm.auto'
```
and it typically includes a traceback, so that we can see the full stack of calls the program made before it fails. This gives us the context to know why the program failed.
Going back to the above example. If you received this error search, look at the very last line of the error which is:
```python
ModuleNotFoundError: No module named 'tqdm.auto'
```
And now we can use it to do the searching on your favorite search engine:
1. first for `"huggingface" "transformers" "ModuleNotFoundError: No module named 'tqdm.auto'"`
2. if you don't find relevant results, then search for just `"ModuleNotFoundError: No module named 'tqdm.auto'"`
3. and finally if nothing still comes up, then remove the outside quotes: `ModuleNotFoundError: No module named 'tqdm.auto'`
If the error includes any messages that include bits unique to your filesystem, always remove those in the search query since other users will not have the same filesystem as yours. For example:
```bash
python -c 'open("/tmp/wrong_path.txt", "r")'
Traceback (most recent call last):
File "<string>", line 1, in <module>
FileNotFoundError: [Errno 2] No such file or directory: '/tmp/wrong_path.txt'
```
Here you'd search for just: `"FileNotFoundError: [Errno 2] No such file or directory"`
If the local information that you removed were inside the error message and you removed them you may need to remove double quotes since your query is no longer exact. So if the error message was something like:
```bash
ValueError: '/tmp/wrong_path.txt' cannot be found
```
then you'd search for `"ValueError" "cannot be found"`
As you search you will notice that when you don't use quotes often the search engines will return a variety of unrelated hits, which may or may not be what you want.
Experiment with different ways and find which approach gives the most satisfactory results.
2. Keep the issue short, providing the information that you think will aid the developers to understand your situation. Put yourself in the shoes of the person who has never seen your code or knows anything about your custom setup. This mental exercise will help to develop an intuition to what/what not to share"
3. If there is a software failure, always provide the full traceback, for example:
```python
$ python -c 'import transformers'
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/transformers/src/transformers/__init__.py", line 34, in <module>
from . import dependency_versions_check
File "/transformers/src/transformers/dependency_versions_check.py", line 34, in <module>
from .utils import is_tokenizers_available
File "/transformers/src/transformers/utils/import_utils.py", line 40, in <module>
from tqdm.auto import tqdm
ModuleNotFoundError: No module named 'tqdm.auto'
```
As compared to providing just the last line of the error message, e.g.:
```python
ModuleNotFoundError: No module named 'tqdm.auto'
```
which is not sufficient.
If your application is running on more than one GPU (e.g. under `DistributedDataParallel`) and typically getting every log and traceback printed multiple times, please make sure that you paste only one copy of it. At times the traceback from parallel processes may get interleaved - so either disentangle these or change the loggers to log only for `local_rank==0` so that only one process logs things.
4. When quoting a traceback, command line instructions and any type of code always enclose it in triple backticks inside the editor window, that is:
````
```
git clone https://github.com/huggingface/transformers
cd transformers
pip install .
```
````
If it's a command line with a long argument list, please consider breaking it down using backslashes and new lines. Here is an example of a good command line quote:
```bash
cd examples/seq2seq
python -m torch.distributed.launch --nproc_per_node=2 ./finetune_trainer.py \
--model_name_or_path sshleifer/distill-mbart-en-ro-12-4 --data_dir wmt_en_ro \
--output_dir output_dir --overwrite_output_dir \
--do_train --n_train 500 --num_train_epochs 1 \
--per_device_train_batch_size 1 --freeze_embeds \
--src_lang en_XX --tgt_lang ro_RO --task translation \
--fp16 --sharded_ddp
```
If you don't break it up, one has to scroll horizontally which often makes it quite difficult to quickly see what's happening.
The backslashes allow us to copy the command directly into the console to run it, without needing to edit it.
5. Include only the important information that you think will help the developer to quickly identify the problem.
For example applications often create huge amounts of logs. Ask yourself whether providing all or parts of the log is useful.
Pasting a 100-1000 lines of log into the issue is an immediate turn off, since it will take a lot of time to figure out where the pertinent parts of the log are.
Attaching a full log can be helpful if it's done as an attachment, if it's enclosed in the following html code in the comment editor window:
```
<details>
<summary>Full log</summary>
<pre>
many
lines
go
here
</pre>
</details>
```
which would result in the following entry, which can be opened if desired, but otherwise takes little space.
<details>
<summary>Full log</summary>
<pre>
many
lines
go
here
</pre>
</details>
You could also provide a link to a pastebin service, but this is less beneficial since those links tend to expire quickly and future readers of your issue might not be able to access that log file anymore and may lack some context.
6. If this is an issue in your code, do try to reduce that code to a minimal example that still demonstrates the problem. Please ask at the forums if you have a hard time figuring how to do that. Please realize that we don't have the luxury of having time to try and understand all of your custom code.
If you really tried to make a short reproducible code but couldn't figure it out, it might be that having a traceback will give the developer enough information to know what's going on. But if it is not enough and we can't reproduce the problem, we can't really solve it.
Do not despair if you can't figure it out from the beginning, just share what you can and perhaps someone else will be able to help you at the forums.
If your setup involves any custom datasets, the best way to help us reproduce the problem is to create a [Google Colab notebook](https://colab.research.google.com/) that demonstrates the issue and once you verify that the issue still exists, include a link to that notebook in the Issue. Just make sure that you don't copy and paste the location bar url of the open notebook - as this is private and we won't be able to open it. Instead, you need to click on `Share` in the right upper corner of the notebook, select `Get Link` and then copy and paste the public link it will give to you.
7. If you forked off some of this project's code or example applications, please, do not ask us to go into your code repository and figure out what you may have done. The code is already very complex and unless there is an easy way to do a diff and it's a small diff, it won't be possible to find someone with time on their hands to make a lengthy investigation. Albeit, you might find someone at the forums who will be generous to do this for you.
8. Before reporting an issue, first, always try to update your environment to the latest official version of this library. We have no resources to go and debug older revisions, which could easily have bugs that have been fixed in the latest released version.
We understand that this is not always possible, especially when APIs change, in which case file an issue against the highest library version your environment can support.
Of course, if you upgrade the library, always retest that the problem is still there.
9. Please do not ask us to reproduce an issue with your custom data, since we don't have it. So, either you should use some existing dataset supported by HF datasets or you need to supply a code that generates a small sample on the fly, or some another quick and simple way to get it.
Please do not send us any non-public domain data that may require a license or a permission to be used.
10. Do not tag multiple developers on the issue unless you know this is expected, either because you asked them and they gave you an explicit permission to tag them or the issue template instructs you to do so.
The "who to tag for what domain" part of the issue template is there to help users direct their questions to the right developers who are designated maintainers of project's specific domains. They can then decide at their own discretion to tag other developers if they feel it'd help move the issue forward.
We currently don't have a triage service and we trust your capacity to identify the right domain and thus the persons to tag in your issue. If you are not sure, please use the forums to ask for guidance.
When in doubt, err on the side of not tagging a given person. If you tag multiple people out of context or permission don't be surprised if you get no response at all. Please remember that every time you tag someone, they get a notification and you're taking their time without their permission. Please be sensitive to that.
If you got helped by one of the developers in the past please don't tag them in future issues, unless they are listed in the issue template for the domain you are asking about or that developer gave you an explicit permission to tag them in future issues.
If you see a certain developer doing multiple and/or recent commits into a specific area of the project that you feel is relevant to your issue, it is not a good reason to tag them. Various developers may be fixing things that prevent them from moving forward, but often their work is focused on a totally different domain. And while they may or may not know how to help you with the problem at hand, it would benefit the whole community much more if they focus on the domain of their unique expertise.
11. Use the Edit button. Take your time, and re-read and improve the wording and formatting to make your posts and comments as easy to understand as possible.
Avoid posting multiple comments in a row, as each comment generates a notification for the developers tagged in that issue. If you happened to post multiple comments in a row, and nobody followed up yet - consider merging those into one or a few comments while editing the combined content to be coherent.
If you choose to edit your older comments after others posted follow up comments you need to be aware that your modifications might not be noticed, so if it's not a typo fixing, try to write a new comment flagging that something has been changed in the previous comments.
For example, the very first comment is the most important one. If while the thread unfolds you realize that things aren't as they seemed to you originally you may want to edit the first post to reflect the up-to-date understanding of the issue at hand so that it helps those who read your issue in the future quickly understand what's going on and not need to sift through dozens of comments. It also helps to indicate that the post was edited. So, those reading the thread later can understand why there might be certain discontinuity in the information flow.
Use bullets and items if you have lists of items and the outcome improves overall readability.
Use backticks to refer to class and function names, e.g. `BartModel` and `generate` as these stand out and improve the speed of a reader's comprehension.
Try not use italics and bold text too much as these often make the text more difficult to read.
12. If you are cross-referencing a specific comment in a given thread or another issue, always link to that specific comment, rather than using the issue link. If you do the latter it could be quite impossible to find which specific comment you're referring to.
To get the link to the specific comment do not copy the url from the location bar of your browser, but instead, click the `...` icon in the upper right corner of the comment and then select "Copy Link".
For example the first link is a link to an issue, and the second to a specific comment in the same issue:
1. https://github.com/huggingface/transformers/issues/9257
2. https://github.com/huggingface/transformers/issues/9257#issuecomment-749945162
13. If you are replying to a last comment, it's totally fine to make your reply with just your comment in it. The readers can follow the information flow here.
But if you're replying to a comment that happened some comments back it's always a good practice to quote just the relevant lines you're replying it. The `>` is used for quoting, or you can always use the menu to do so. For example your editor box will look like:
```
> How big is your gpu cluster?
Our cluster is made of 256 gpus.
```
If you are addressing multiple comments, quote the relevant parts of each before your answer. Some people use the same comment to do multiple replies, others separate them into separate comments. Either way works. The latter approach helps for linking to a specific comment.
In general the best way to figure out what works the best is learn from issues posted by other people - see which issues get great responses and which get little to no response - observe what the posters who received great responses did differently from those who did not.
Thank you for reading this somewhat lengthy document. We would like to conclude that these are not absolute rules, but a friendly advice that will help maximize the chances for us to understand what you are trying to communicate, reproduce the problem then resolve it to your satisfaction and the benefit of the whole community.
If after reading this document there are remaining questions on how and why or there is a need for further elucidation, please, don't hesitate to ask your question in [this thread](https://discuss.huggingface.co/t/how-to-request-support/3128).
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
# How To Request Support
This is an Open Source Project so please be mindful that like in any other project of this kind there is no obligation to answer all requests for help.
However, we want to encourage you to ask for help whenever you think it's needed! We are happy about every question we get because it allows us to better understand your needs, possible misunderstandings, and most importantly a way for you to help us make this library better. That being said, this document's main purpose is to provide guidelines at how you can formulate your requests to increase your chances to be understood and to get support.
There are two main venues to receive support: [the forums](https://discuss.huggingface.co/) and [the GitHub issues](https://github.com/huggingface/transformers/issues).
## The Forums
[The user forums](https://discuss.huggingface.co/) are supported by the wide community of the library users and backed up by developers when needed.
If you have a difficulty with deploying this library or some questions, or you'd like to discuss a new feature, please first consider discussing those things at the forums. Only when you feel your subject matter has been crystalized and you still need support from the library developers do proceed to file an [issue](https://github.com/huggingface/transformers/issues).
In particular all "Please explain" questions or objectively very user-specific feature requests belong to the forums. Here are some example of such questions:
* "I would like to use a BertModel within a RL-Agent for a customer support service. How can I use a BertForMaskedLM in my ChatBotModel?"
* "Could you please explain why T5 has no positional embedding matrix under T5Model?"
* "How should I set my generation parameters for translation?"
* "How to train T5 on De->En translation?"
## The GitHub Issues
Everything which hints at a bug should be opened as an [issue](https://github.com/huggingface/transformers/issues).
You are not required to read the following guidelines before opening an issue. However, if you notice that your issue doesn't get any replies, chances are that the developers have one or several difficulties with its quality. In this case, reading the following points and adjusting your issue accordingly could help.
1. Before posting an issue, first search for already posted issues, since chances are someone has already asked a similar question before you.
If you use Google your search query should be:
```
"huggingface" "transformers" your query
```
The first two quoted words tell Google to limit the search to the context of the Huggingface Transformers. The remainder is your query - most commonly this would be the error message the software fails with. We will go deeper into details shortly.
The results of such a query will typically match GitHub issues, Hugging Face forums, StackExchange, and blogs.
If you find relevant hints, you may choose to continue the discussion there if you have follow up questions.
If what you found is similar but doesn't quite answer your problem, please, post a new issue and do include links to similar issues or forum discussions you may have found.
Let's look at some examples:
The error message, often referred to as an assertion, tells us what went wrong. Here is an example of an assertion:
```python
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/transformers/src/transformers/__init__.py", line 34, in <module>
from . import dependency_versions_check
File "/transformers/src/transformers/dependency_versions_check.py", line 34, in <module>
from .utils import is_tokenizers_available
File "/transformers/src/transformers/utils/import_utils.py", line 40, in <module>
from tqdm.auto import tqdm
ModuleNotFoundError: No module named 'tqdm.auto'
```
and it typically includes a traceback, so that we can see the full stack of calls the program made before it fails. This gives us the context to know why the program failed.
Going back to the above example. If you received this error search, look at the very last line of the error which is:
```python
ModuleNotFoundError: No module named 'tqdm.auto'
```
And now we can use it to do the searching on your favorite search engine:
1. first for `"huggingface" "transformers" "ModuleNotFoundError: No module named 'tqdm.auto'"`
2. if you don't find relevant results, then search for just `"ModuleNotFoundError: No module named 'tqdm.auto'"`
3. and finally if nothing still comes up, then remove the outside quotes: `ModuleNotFoundError: No module named 'tqdm.auto'`
If the error includes any messages that include bits unique to your filesystem, always remove those in the search query since other users will not have the same filesystem as yours. For example:
```bash
python -c 'open("/tmp/wrong_path.txt", "r")'
Traceback (most recent call last):
File "<string>", line 1, in <module>
FileNotFoundError: [Errno 2] No such file or directory: '/tmp/wrong_path.txt'
```
Here you'd search for just: `"FileNotFoundError: [Errno 2] No such file or directory"`
If the local information that you removed were inside the error message and you removed them you may need to remove double quotes since your query is no longer exact. So if the error message was something like:
```bash
ValueError: '/tmp/wrong_path.txt' cannot be found
```
then you'd search for `"ValueError" "cannot be found"`
As you search you will notice that when you don't use quotes often the search engines will return a variety of unrelated hits, which may or may not be what you want.
Experiment with different ways and find which approach gives the most satisfactory results.
2. Keep the issue short, providing the information that you think will aid the developers to understand your situation. Put yourself in the shoes of the person who has never seen your code or knows anything about your custom setup. This mental exercise will help to develop an intuition to what/what not to share"
3. If there is a software failure, always provide the full traceback, for example:
```python
$ python -c 'import transformers'
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/transformers/src/transformers/__init__.py", line 34, in <module>
from . import dependency_versions_check
File "/transformers/src/transformers/dependency_versions_check.py", line 34, in <module>
from .utils import is_tokenizers_available
File "/transformers/src/transformers/utils/import_utils.py", line 40, in <module>
from tqdm.auto import tqdm
ModuleNotFoundError: No module named 'tqdm.auto'
```
As compared to providing just the last line of the error message, e.g.:
```python
ModuleNotFoundError: No module named 'tqdm.auto'
```
which is not sufficient.
If your application is running on more than one GPU (e.g. under `DistributedDataParallel`) and typically getting every log and traceback printed multiple times, please make sure that you paste only one copy of it. At times the traceback from parallel processes may get interleaved - so either disentangle these or change the loggers to log only for `local_rank==0` so that only one process logs things.
4. When quoting a traceback, command line instructions and any type of code always enclose it in triple backticks inside the editor window, that is:
````
```
git clone https://github.com/huggingface/transformers
cd transformers
pip install .
```
````
If it's a command line with a long argument list, please consider breaking it down using backslashes and new lines. Here is an example of a good command line quote:
```bash
cd examples/seq2seq
python -m torch.distributed.launch --nproc_per_node=2 ./finetune_trainer.py \
--model_name_or_path sshleifer/distill-mbart-en-ro-12-4 --data_dir wmt_en_ro \
--output_dir output_dir --overwrite_output_dir \
--do_train --n_train 500 --num_train_epochs 1 \
--per_device_train_batch_size 1 --freeze_embeds \
--src_lang en_XX --tgt_lang ro_RO --task translation \
--fp16 --sharded_ddp
```
If you don't break it up, one has to scroll horizontally which often makes it quite difficult to quickly see what's happening.
The backslashes allow us to copy the command directly into the console to run it, without needing to edit it.
5. Include only the important information that you think will help the developer to quickly identify the problem.
For example applications often create huge amounts of logs. Ask yourself whether providing all or parts of the log is useful.
Pasting a 100-1000 lines of log into the issue is an immediate turn off, since it will take a lot of time to figure out where the pertinent parts of the log are.
Attaching a full log can be helpful if it's done as an attachment, if it's enclosed in the following html code in the comment editor window:
```
<details>
<summary>Full log</summary>
<pre>
many
lines
go
here
</pre>
</details>
```
which would result in the following entry, which can be opened if desired, but otherwise takes little space.
<details>
<summary>Full log</summary>
<pre>
many
lines
go
here
</pre>
</details>
You could also provide a link to a pastebin service, but this is less beneficial since those links tend to expire quickly and future readers of your issue might not be able to access that log file anymore and may lack some context.
6. If this is an issue in your code, do try to reduce that code to a minimal example that still demonstrates the problem. Please ask at the forums if you have a hard time figuring how to do that. Please realize that we don't have the luxury of having time to try and understand all of your custom code.
If you really tried to make a short reproducible code but couldn't figure it out, it might be that having a traceback will give the developer enough information to know what's going on. But if it is not enough and we can't reproduce the problem, we can't really solve it.
Do not despair if you can't figure it out from the beginning, just share what you can and perhaps someone else will be able to help you at the forums.
If your setup involves any custom datasets, the best way to help us reproduce the problem is to create a [Google Colab notebook](https://colab.research.google.com/) that demonstrates the issue and once you verify that the issue still exists, include a link to that notebook in the Issue. Just make sure that you don't copy and paste the location bar url of the open notebook - as this is private and we won't be able to open it. Instead, you need to click on `Share` in the right upper corner of the notebook, select `Get Link` and then copy and paste the public link it will give to you.
7. If you forked off some of this project's code or example applications, please, do not ask us to go into your code repository and figure out what you may have done. The code is already very complex and unless there is an easy way to do a diff and it's a small diff, it won't be possible to find someone with time on their hands to make a lengthy investigation. Albeit, you might find someone at the forums who will be generous to do this for you.
8. Before reporting an issue, first, always try to update your environment to the latest official version of this library. We have no resources to go and debug older revisions, which could easily have bugs that have been fixed in the latest released version.
We understand that this is not always possible, especially when APIs change, in which case file an issue against the highest library version your environment can support.
Of course, if you upgrade the library, always retest that the problem is still there.
9. Please do not ask us to reproduce an issue with your custom data, since we don't have it. So, either you should use some existing dataset supported by HF datasets or you need to supply a code that generates a small sample on the fly, or some another quick and simple way to get it.
Please do not send us any non-public domain data that may require a license or a permission to be used.
10. Do not tag multiple developers on the issue unless you know this is expected, either because you asked them and they gave you an explicit permission to tag them or the issue template instructs you to do so.
The "who to tag for what domain" part of the issue template is there to help users direct their questions to the right developers who are designated maintainers of project's specific domains. They can then decide at their own discretion to tag other developers if they feel it'd help move the issue forward.
We currently don't have a triage service and we trust your capacity to identify the right domain and thus the persons to tag in your issue. If you are not sure, please use the forums to ask for guidance.
When in doubt, err on the side of not tagging a given person. If you tag multiple people out of context or permission don't be surprised if you get no response at all. Please remember that every time you tag someone, they get a notification and you're taking their time without their permission. Please be sensitive to that.
If you got helped by one of the developers in the past please don't tag them in future issues, unless they are listed in the issue template for the domain you are asking about or that developer gave you an explicit permission to tag them in future issues.
If you see a certain developer doing multiple and/or recent commits into a specific area of the project that you feel is relevant to your issue, it is not a good reason to tag them. Various developers may be fixing things that prevent them from moving forward, but often their work is focused on a totally different domain. And while they may or may not know how to help you with the problem at hand, it would benefit the whole community much more if they focus on the domain of their unique expertise.
11. Use the Edit button. Take your time, and re-read and improve the wording and formatting to make your posts and comments as easy to understand as possible.
Avoid posting multiple comments in a row, as each comment generates a notification for the developers tagged in that issue. If you happened to post multiple comments in a row, and nobody followed up yet - consider merging those into one or a few comments while editing the combined content to be coherent.
If you choose to edit your older comments after others posted follow up comments you need to be aware that your modifications might not be noticed, so if it's not a typo fixing, try to write a new comment flagging that something has been changed in the previous comments.
For example, the very first comment is the most important one. If while the thread unfolds you realize that things aren't as they seemed to you originally you may want to edit the first post to reflect the up-to-date understanding of the issue at hand so that it helps those who read your issue in the future quickly understand what's going on and not need to sift through dozens of comments. It also helps to indicate that the post was edited. So, those reading the thread later can understand why there might be certain discontinuity in the information flow.
Use bullets and items if you have lists of items and the outcome improves overall readability.
Use backticks to refer to class and function names, e.g. `BartModel` and `generate` as these stand out and improve the speed of a reader's comprehension.
Try not use italics and bold text too much as these often make the text more difficult to read.
12. If you are cross-referencing a specific comment in a given thread or another issue, always link to that specific comment, rather than using the issue link. If you do the latter it could be quite impossible to find which specific comment you're referring to.
To get the link to the specific comment do not copy the url from the location bar of your browser, but instead, click the `...` icon in the upper right corner of the comment and then select "Copy Link".
For example the first link is a link to an issue, and the second to a specific comment in the same issue:
1. https://github.com/huggingface/transformers/issues/9257
2. https://github.com/huggingface/transformers/issues/9257#issuecomment-749945162
13. If you are replying to a last comment, it's totally fine to make your reply with just your comment in it. The readers can follow the information flow here.
But if you're replying to a comment that happened some comments back it's always a good practice to quote just the relevant lines you're replying it. The `>` is used for quoting, or you can always use the menu to do so. For example your editor box will look like:
```
> How big is your gpu cluster?
Our cluster is made of 256 gpus.
```
If you are addressing multiple comments, quote the relevant parts of each before your answer. Some people use the same comment to do multiple replies, others separate them into separate comments. Either way works. The latter approach helps for linking to a specific comment.
In general the best way to figure out what works the best is learn from issues posted by other people - see which issues get great responses and which get little to no response - observe what the posters who received great responses did differently from those who did not.
Thank you for reading this somewhat lengthy document. We would like to conclude that these are not absolute rules, but a friendly advice that will help maximize the chances for us to understand what you are trying to communicate, reproduce the problem then resolve it to your satisfaction and the benefit of the whole community.
If after reading this document there are remaining questions on how and why or there is a need for further elucidation, please, don't hesitate to ask your question in [this thread](https://discuss.huggingface.co/t/how-to-request-support/3128).
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/oneformer/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/bigbird_pegasus/test_modeling_bigbird_pegasus.py | # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch BigBirdPegasus model. """
import copy
import tempfile
import unittest
from transformers import BigBirdPegasusConfig, is_torch_available
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
BigBirdPegasusForCausalLM,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForQuestionAnswering,
BigBirdPegasusForSequenceClassification,
BigBirdPegasusModel,
PegasusTokenizer,
)
from transformers.models.bigbird_pegasus.modeling_bigbird_pegasus import (
BigBirdPegasusDecoder,
BigBirdPegasusEncoder,
)
MODEL_ID = "google/bigbird-pegasus-large-pubmed"
def prepare_bigbird_pegasus_inputs_dict(
config,
input_ids,
decoder_input_ids,
attention_mask=None,
decoder_attention_mask=None,
):
if attention_mask is None:
attention_mask = input_ids.ne(config.pad_token_id)
if decoder_attention_mask is None:
decoder_attention_mask = decoder_input_ids.ne(config.pad_token_id)
input_dict = {
"input_ids": input_ids,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"decoder_attention_mask": attention_mask,
}
input_dict = {k: input_dict[k].to(torch_device) for k in input_dict}
return input_dict
class BigBirdPegasusModelTester:
def __init__(
self,
parent,
batch_size=7,
seq_length=256,
is_training=True,
use_labels=False,
vocab_size=99,
hidden_size=32,
num_hidden_layers=2,
num_attention_heads=4,
intermediate_size=31,
hidden_act="gelu_fast",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=260,
eos_token_id=1,
pad_token_id=0,
bos_token_id=2,
attention_type="block_sparse",
use_bias=False,
block_size=16,
num_random_blocks=3,
scale_embedding=True,
):
self.parent = parent
self.batch_size = batch_size
self.seq_length = seq_length
self.is_training = is_training
self.use_labels = use_labels
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_act = hidden_act
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.eos_token_id = eos_token_id
self.pad_token_id = pad_token_id
self.bos_token_id = bos_token_id
self.attention_type = attention_type
self.use_bias = use_bias
self.block_size = block_size
self.num_random_blocks = num_random_blocks
self.scale_embedding = scale_embedding
def prepare_config_and_inputs(self):
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size).clamp(
3,
)
input_ids[:, -1] = self.eos_token_id # Eos Token
decoder_input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
config = self.get_config()
inputs_dict = prepare_bigbird_pegasus_inputs_dict(config, input_ids, decoder_input_ids)
return config, inputs_dict
def get_config(self):
return BigBirdPegasusConfig(
vocab_size=self.vocab_size,
d_model=self.hidden_size,
encoder_layers=self.num_hidden_layers,
decoder_layers=self.num_hidden_layers,
encoder_attention_heads=self.num_attention_heads,
decoder_attention_heads=self.num_attention_heads,
encoder_ffn_dim=self.intermediate_size,
decoder_ffn_dim=self.intermediate_size,
dropout=self.hidden_dropout_prob,
attention_dropout=self.attention_probs_dropout_prob,
max_position_embeddings=self.max_position_embeddings,
eos_token_id=self.eos_token_id,
bos_token_id=self.bos_token_id,
pad_token_id=self.pad_token_id,
attention_type=self.attention_type,
use_bias=self.use_bias,
block_size=self.block_size,
num_random_blocks=self.num_random_blocks,
scale_embedding=self.scale_embedding,
)
def prepare_config_and_inputs_for_common(self):
config, inputs_dict = self.prepare_config_and_inputs()
return config, inputs_dict
def create_and_check_decoder_model_past_large_inputs(self, config, inputs_dict):
model = BigBirdPegasusModel(config=config).get_decoder().to(torch_device).eval()
input_ids = inputs_dict["input_ids"]
attention_mask = inputs_dict["attention_mask"]
# first forward pass
outputs = model(input_ids, attention_mask=attention_mask, use_cache=True)
output, past_key_values = outputs.to_tuple()
# create hypothetical multiple next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
next_attn_mask = ids_tensor((self.batch_size, 3), 2)
# append to next input_ids and
next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
next_attention_mask = torch.cat([attention_mask, next_attn_mask], dim=-1)
output_from_no_past = model(next_input_ids, attention_mask=next_attention_mask)["last_hidden_state"]
output_from_past = model(next_tokens, attention_mask=next_attention_mask, past_key_values=past_key_values)[
"last_hidden_state"
]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, :, random_slice_idx].detach()
self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1])
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-2))
def check_encoder_decoder_model_standalone(self, config, inputs_dict):
model = BigBirdPegasusModel(config=config).to(torch_device).eval()
outputs = model(**inputs_dict)
encoder_last_hidden_state = outputs.encoder_last_hidden_state
last_hidden_state = outputs.last_hidden_state
with tempfile.TemporaryDirectory() as tmpdirname:
encoder = model.get_encoder()
encoder.save_pretrained(tmpdirname)
encoder = BigBirdPegasusEncoder.from_pretrained(tmpdirname).to(torch_device)
encoder_last_hidden_state_2 = encoder(inputs_dict["input_ids"], attention_mask=inputs_dict["attention_mask"])[
0
]
self.parent.assertTrue((encoder_last_hidden_state_2 - encoder_last_hidden_state).abs().max().item() < 1e-3)
with tempfile.TemporaryDirectory() as tmpdirname:
decoder = model.get_decoder()
decoder.save_pretrained(tmpdirname)
decoder = BigBirdPegasusDecoder.from_pretrained(tmpdirname).to(torch_device)
last_hidden_state_2 = decoder(
input_ids=inputs_dict["decoder_input_ids"],
attention_mask=inputs_dict["decoder_attention_mask"],
encoder_hidden_states=encoder_last_hidden_state,
encoder_attention_mask=inputs_dict["attention_mask"],
)[0]
self.parent.assertTrue((last_hidden_state_2 - last_hidden_state).abs().max().item() < 1e-3)
def create_and_check_model(self, config, inputs_dict):
model = BigBirdPegasusModel(config=config).to(torch_device).eval()
input_ids = inputs_dict["input_ids"]
decoder_input_ids = inputs_dict["decoder_input_ids"]
result = model(input_ids, decoder_input_ids=decoder_input_ids, use_cache=True)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
@require_torch
class BigBirdPegasusModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(
BigBirdPegasusModel,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForSequenceClassification,
BigBirdPegasusForQuestionAnswering,
)
if is_torch_available()
else ()
)
all_generative_model_classes = (BigBirdPegasusForConditionalGeneration,) if is_torch_available() else ()
pipeline_model_mapping = (
{
"conversational": BigBirdPegasusForConditionalGeneration,
"feature-extraction": BigBirdPegasusModel,
"question-answering": BigBirdPegasusForQuestionAnswering,
"summarization": BigBirdPegasusForConditionalGeneration,
"text2text-generation": BigBirdPegasusForConditionalGeneration,
"text-classification": BigBirdPegasusForSequenceClassification,
"text-generation": BigBirdPegasusForCausalLM,
"zero-shot": BigBirdPegasusForSequenceClassification,
}
if is_torch_available()
else {}
)
is_encoder_decoder = True
test_missing_keys = False
test_pruning = False
test_head_masking = False
# torchscript tests are not passing for now.
# Also torchscript is not an important feature to have in the beginning.
test_torchscript = False
# overwrite from GenerationTesterMixin to solve problem
# with conflicting random seeds
def _get_input_ids_and_config(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
config.attention_type = "original_full"
input_ids = inputs_dict[self.input_name]
attention_mask = torch.ones_like(input_ids, dtype=torch.long)
# cut to half length & take max batch_size 3
max_batch_size = 2
sequence_length = input_ids.shape[-1] // 2
input_ids = input_ids[:max_batch_size, :sequence_length]
attention_mask = attention_mask[:max_batch_size, :sequence_length]
# generate max 3 tokens
max_length = input_ids.shape[-1] + 3
if config.eos_token_id is not None and config.pad_token_id is None:
# hack to allow generate for models such as GPT2 as is done in `generate()`
config.pad_token_id = config.eos_token_id
return config, input_ids, attention_mask, max_length
def setUp(self):
self.model_tester = BigBirdPegasusModelTester(self)
self.config_tester = ConfigTester(self, config_class=BigBirdPegasusConfig)
def test_config(self):
self.config_tester.run_common_tests()
def test_save_load_strict(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs()
for model_class in self.all_model_classes:
model = model_class(config)
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(tmpdirname)
model2, info = model_class.from_pretrained(tmpdirname, output_loading_info=True)
self.assertEqual(info["missing_keys"], [])
def test_decoder_model_past_with_large_inputs(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_decoder_model_past_large_inputs(*config_and_inputs)
def test_encoder_decoder_model_standalone(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs_for_common()
self.model_tester.check_encoder_decoder_model_standalone(*config_and_inputs)
def test_model_various_attn_type(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
for type in ["original_full", "block_sparse"]:
config_and_inputs[0].attention_type = type
self.model_tester.create_and_check_model(*config_and_inputs)
def test_generate_without_input_ids(self):
if self.model_tester.attention_type == "block_sparse":
# this test can never pass for BigBird-block-sparse attention since input_ids must be multiple of block_size
return
super().test_generate_without_input_ids()
def test_retain_grad_hidden_states_attentions(self):
if self.model_tester.attention_type == "block_sparse":
# this test can't pass since attention matrix (which is getting returned) can't have gradients (& just 0 at many locations)
return
super().test_retain_grad_hidden_states_attentions()
# BigBirdPegasusForSequenceClassification does not support inputs_embeds
def test_inputs_embeds(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in (
BigBirdPegasusModel,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForQuestionAnswering,
):
model = model_class(config)
model.to(torch_device)
model.eval()
inputs = copy.deepcopy(self._prepare_for_class(inputs_dict, model_class))
if not self.is_encoder_decoder:
input_ids = inputs["input_ids"]
del inputs["input_ids"]
else:
encoder_input_ids = inputs["input_ids"]
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
del inputs["input_ids"]
inputs.pop("decoder_input_ids", None)
wte = model.get_input_embeddings()
if not self.is_encoder_decoder:
inputs["inputs_embeds"] = wte(input_ids)
else:
inputs["inputs_embeds"] = wte(encoder_input_ids)
inputs["decoder_inputs_embeds"] = wte(decoder_input_ids)
with torch.no_grad():
model(**inputs)[0]
def test_generate_fp16(self):
config, input_dict = self.model_tester.prepare_config_and_inputs()
input_dict.pop("decoder_attention_mask")
input_dict.pop("decoder_input_ids")
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
if torch_device == "cuda":
model.half()
model.generate(**input_dict)
model.generate(**input_dict, do_sample=True, early_stopping=False, num_return_sequences=3)
@slow
def test_batched_forward_original_full(self):
self._check_batched_forward(attn_type="original_full")
@slow
def test_batched_forward_block_sparse(self):
self._check_batched_forward(attn_type="block_sparse", tolerance=1e-1)
def _check_batched_forward(self, attn_type, tolerance=1e-3):
config, _ = self.model_tester.prepare_config_and_inputs()
config.max_position_embeddings = 128
config.block_size = 16
config.attention_type = attn_type
model = BigBirdPegasusForConditionalGeneration(config).to(torch_device)
model.eval()
chunk_length = 32
sample_with_padding = [3, 8, 11] * chunk_length + [0] * chunk_length
sample_without_padding = [4, 7, 9, 13] * chunk_length
target_ids_without_padding = [2, 3] * 8
target_ids_with_padding = [7, 8] * 6 + 4 * [-100]
attention_mask = torch.tensor(
[[1] * 3 * chunk_length + [0] * chunk_length, [1] * 4 * chunk_length],
device=torch_device,
dtype=torch.long,
)
input_ids = torch.tensor([sample_with_padding, sample_without_padding], device=torch_device, dtype=torch.long)
labels = torch.tensor(
[target_ids_without_padding, target_ids_with_padding], device=torch_device, dtype=torch.long
)
with torch.no_grad():
logits_batched = model(input_ids=input_ids, attention_mask=attention_mask, labels=labels).logits
with torch.no_grad():
logits_single_first = model(input_ids=input_ids[:1, :-chunk_length], labels=labels[:1]).logits
self.assertTrue(torch.allclose(logits_batched[0, -3:], logits_single_first[0, -3:], atol=tolerance))
with torch.no_grad():
logits_single_second = model(input_ids=input_ids[1:], labels=labels[1:, :-4]).logits
self.assertTrue(torch.allclose(logits_batched[1, :3], logits_single_second[0, :3], atol=tolerance))
def test_auto_padding(self):
ids = [[7, 6, 9] * 65]
config, _ = self.model_tester.prepare_config_and_inputs()
input_ids = torch.tensor(ids, device=torch_device, dtype=torch.long)
attention_mask = input_ids.new_ones(input_ids.shape)
decoder_input_ids = torch.tensor([[33, 5, 8] * 3], device=torch_device, dtype=torch.long)
config.block_size = 8
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
output1 = model(input_ids=input_ids, attention_mask=attention_mask, decoder_input_ids=decoder_input_ids)[
"logits"
]
ids = [[7, 6, 9] * 65 + [0] * 5]
input_ids = torch.tensor(ids, device=torch_device, dtype=torch.long)
attention_mask = torch.tensor([[1] * 3 * 65 + [0] * 5], device=torch_device, dtype=torch.long)
output2 = model(input_ids=input_ids, attention_mask=attention_mask, decoder_input_ids=decoder_input_ids)[
"logits"
]
self.assertTrue(torch.allclose(output1, output2, atol=1e-5))
def test_for_change_to_full_attn(self):
self.model_tester.seq_length = 9
config, input_dict = self.model_tester.prepare_config_and_inputs()
# automatic switch will happen
config.attention_type = "block_sparse"
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
state_dict = model.state_dict()
outputs1 = model(**input_dict)["logits"]
config.attention_type = "original_full"
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
model.load_state_dict(state_dict)
outputs2 = model(**input_dict)["logits"]
self.assertTrue(torch.allclose(outputs1, outputs2, atol=1e-5))
@require_torch
@require_sentencepiece
@require_tokenizers
@slow
class BigBirdPegasusModelIntegrationTests(unittest.TestCase):
def _get_dummy_input_ids(self):
# fmt: off
ids = torch.tensor(
[[685, 560, 630, 193, 836, 764, 708, 360, 10, 724, 278, 755, 805, 600, 71, 473, 601, 397, 315, 706, 487, 552, 88, 175, 601, 850, 678, 538, 846, 73, 778, 917, 116, 977, 756, 710, 1023, 848, 432, 449, 851, 100, 985, 178, 756, 798, 660, 148, 911, 424, 289, 962, 266, 698, 640, 545, 544, 715, 245, 152, 676, 511, 460, 883, 184, 29, 803, 129, 129, 933, 54, 902, 551, 489, 757, 274, 336, 389, 618, 43, 443, 544, 889, 258, 322, 1000, 938, 58, 292, 871, 120, 780, 431, 83, 92, 897, 399, 612, 566, 909, 634, 939, 85, 204, 325, 775, 965, 48, 640, 1013, 132, 973, 869, 181, 1001, 847, 144, 661, 228, 955, 792, 720, 910, 374, 854, 561, 306, 582, 170, 676, 449, 96, 198, 607, 257, 882, 691, 293, 931, 817, 862, 388, 611, 555, 974, 369, 1000, 918, 202, 384, 513, 907, 371, 556, 955, 384, 24, 700, 131, 378, 99, 575, 932, 735, 124, 964, 595, 943, 740, 149, 210, 563, 412, 783, 42, 59, 706, 37, 779, 87, 44, 873, 12, 771, 308, 81, 33, 183, 129, 807, 276, 175, 555, 372, 185, 445, 489, 590, 287, 281, 638, 771, 516, 95, 227, 876, 270, 881, 297, 329, 20, 608, 841, 411, 451, 249, 181, 324, 1005, 830, 783, 865, 261, 964, 750, 140, 1021, 599, 462, 890, 622, 844, 697, 529, 153, 926, 150, 111, 26, 465, 957, 890, 887, 118, 446, 596, 674, 873, 929, 229, 508, 764, 122, 327, 470, 288, 526, 840, 697, 153, 592, 42, 275, 553, 439, 208, 780, 167, 112, 350, 1018, 130, 736, 887, 813, 217, 382, 25, 68, 979, 1008, 772, 235, 717, 999, 292, 727, 1023, 702, 710, 728, 556, 33, 12, 617, 213, 139, 695, 1004, 422, 638, 669, 624, 489, 771, 540, 980, 218, 664, 822, 308, 175, 149, 950, 542, 580, 548, 808, 394, 74, 298, 920, 900, 815, 731, 947, 877, 772, 800, 778, 395, 540, 430, 200, 424, 62, 342, 866, 45, 803, 931, 89, 34, 646, 233, 768, 37, 769, 460, 291, 198, 895, 950, 255, 81, 447, 137, 190, 130, 210, 369, 292, 377, 348, 169, 885, 805, 177, 538, 324, 872, 509, 804, 115, 799, 30, 754, 290, 147, 274, 222, 341, 510, 515, 70, 358, 909, 557, 886, 766, 323, 624, 92, 342, 424, 552, 972, 663, 415, 658, 711, 968, 275, 861, 44, 84, 434, 810, 94, 175, 406, 202, 858, 499, 481, 988, 330, 541, 1004, 210, 618, 955, 897, 983, 576, 17, 107, 165, 607, 537, 629, 192, 196, 308, 137, 953, 860, 94, 892, 751, 88, 161, 148, 585, 456, 88, 14, 315, 594, 121, 885, 952, 833, 716, 733, 933, 282, 801, 427, 783, 471, 285, 277, 979, 325, 535, 228, 891, 596, 648, 969, 574, 654, 518, 257, 137, 208, 464, 950, 140, 5, 424, 349, 942, 283, 587, 821, 1007, 434, 220, 820, 740, 874, 787, 374, 291, 564, 671, 438, 827, 940, 824, 509, 1021, 787, 942, 856, 450, 327, 491, 54, 817, 95, 60, 337, 667, 637, 164, 571, 946, 107, 202, 301, 782, 890, 839, 551, 680, 649, 14, 1017, 904, 721, 1017, 535, 505, 848, 986, 777, 740, 775, 210, 456, 469, 474, 963, 573, 401, 57, 883, 750, 664, 281, 5, 613, 1005, 306, 344, 543, 567, 154, 789, 354, 358, 698, 408, 412, 30, 930, 372, 822, 632, 948, 855, 503, 8, 618, 1010, 138, 695, 897, 852, 377, 933, 722, 149, 886, 1009, 260, 127, 811, 578, 533, 805, 325, 977, 113, 944, 651, 238, 361, 991, 860, 556, 64, 928, 917, 455, 266, 445, 604, 624, 420, 340, 845, 275, 370, 843, 227, 226, 940, 644, 909, 229, 827, 898, 370, 129, 808, 25, 699, 293, 356, 838, 135, 4, 227, 890, 681, 445, 418, 285, 837, 27, 737, 249, 366, 948, 202, 438, 198, 930, 648, 638, 607, 73, 247, 853, 136, 708, 214, 476, 621, 324, 103, 853, 328, 596, 224, 257, 646, 348, 108, 927, 970, 980, 520, 150, 998, 477, 393, 684, 559, 1, 361, 692, 551, 90, 75, 500, 739, 636, 344, 97, 852, 283, 719, 33, 116, 455, 866, 429, 828, 826, 691, 174, 746, 133, 442, 94, 348, 402, 420, 707, 405, 942, 186, 976, 376, 677, 874, 703, 517, 498, 499, 206, 415, 366, 856, 739, 420, 586, 219, 952, 539, 375, 23, 461, 720, 355, 603, 52, 999, 815, 721, 574, 445, 816, 1019, 105, 641, 395, 972, 910, 328, 607, 519, 686, 246, 415, 528, 170, 167, 310, 940, 595, 392, 221, 834, 682, 835, 115, 861, 335, 742, 220, 247, 101, 416, 222, 179, 509, 175, 606, 627, 674, 781, 737, 746, 849, 67, 457, 1012, 126, 139, 625, 731, 156, 697, 121, 322, 449, 710, 857, 291, 976, 4, 701, 239, 678, 172, 724, 857, 583, 661, 903, 797, 628, 903, 835, 605, 989, 615, 870, 380, 710, 110, 330, 101, 695, 846, 918, 508, 672, 594, 36, 238, 244, 251, 393, 767, 282, 22, 430, 230, 983, 401, 154, 1007, 120, 678, 896, 386, 390, 711, 397, 347, 587, 1020, 951, 79, 831, 585, 200, 814, 134, 560, 700, 171, 452, 139, 755, 314, 476, 346, 388, 126, 719, 851, 198, 699, 901, 18, 710, 448, 351, 665, 644, 326, 425, 165, 571, 178, 440, 665, 674, 915, 866, 463, 754, 136, 950, 748, 47, 497, 1013, 640, 930, 338, 158, 525, 631, 815, 887, 289, 803, 116, 600, 637, 410, 175, 499, 876, 565, 1002, 623, 577, 333, 887, 586, 147, 773, 776, 644, 49, 77, 294, 117, 494, 561, 110, 979, 180, 562, 72, 859, 434, 1007, 286, 516, 75, 597, 491, 322, 888, 533, 209, 43, 499, 29, 411, 856, 181, 305, 963, 615, 778, 259, 373, 877, 746, 858, 381, 886, 613, 91, 69, 618, 523, 13, 617, 226, 422, 168, 929, 379, 290, 923, 100, 218, 307, 345, 211, 789, 735, 669, 585, 275, 410, 921, 552, 235, 636, 285, 665, 659, 708, 173, 724, 302, 823, 1, 139, 708, 903, 732, 868, 442, 967, 916, 163, 51, 243, 871]], # noqa: E231
dtype=torch.long,
device=torch_device,
)
# fmt: on
return ids
def _get_dummy_target_ids(self):
# fmt: off
ids = torch.tensor(
[[13, 6, 1, 4, 12, 4, 8, 10, 4, 6, 3, 5, 8, 7, 9, 9]], # noqa: E231
dtype=torch.long,
device=torch_device,
)
# fmt: on
return ids
def test_inference_block_sparse(self):
model = BigBirdPegasusForConditionalGeneration.from_pretrained(
MODEL_ID, attention_type="block_sparse", block_size=16, num_random_blocks=3
)
model.to(torch_device)
input_ids = self._get_dummy_input_ids()
target_ids = self._get_dummy_target_ids()
outputs = model(input_ids, labels=target_ids)
prediction_logits = outputs.logits
self.assertEqual(prediction_logits.shape, torch.Size((1, 16, 96103)))
# fmt: off
expected_prediction_logits_slice = torch.tensor(
[[1.7769, 5.8479, 6.2375, 2.2745, 8.6157, 4.7483, 5.0647, 6.5358, 2.3393, 7.8333, 3.8403, 0.0255, 7.219, 5.2759, 3.097, 6.387, 4.9341, 7.1409, 5.1179, 0.1144, 6.8268, 0.7598, 0.6258, 2.373, 0.4627, -1.9919, 1.8422, 3.4578], [1.8026, 5.9604, 5.954, 2.8642, 9.0608, 4.394, 5.3779, 7.0216, 1.543, 7.8744, 4.4231, -0.0398, 7.6091, 5.6611, 3.3536, 6.8624, 4.7699, 6.5241, 4.8893, 0.5791, 6.8368, 0.1034, 0.0338, 2.9393, 0.5034, -2.5509, 2.0172, 3.2858], [1.8426, 5.9151, 5.5374, 3.0426, 9.1762, 3.6287, 5.3916, 7.4621, 1.2582, 7.9244, 4.694, -0.1308, 7.4725, 5.5385, 3.4598, 7.0422, 4.2455, 5.797, 4.5927, 0.7478, 6.7467, -0.2695, -0.3207, 3.0269, 0.4714, -2.8134, 2.0406, 3.1089], [1.6527, 5.8416, 5.4558, 3.0044, 9.3478, 3.2607, 5.3887, 7.52, 0.9362, 7.8877, 4.8465, -0.1705, 7.3932, 5.6352, 3.5744, 7.2623, 4.0485, 5.2788, 4.5859, 0.8325, 6.6088, -0.3676, -0.6287, 3.1731, 0.4483, -3.1573, 2.0522, 2.8868]], # noqa: E231
device=torch_device,
)
# fmt: on
self.assertTrue(
torch.allclose(prediction_logits[0, 4:8, 128:156], expected_prediction_logits_slice, atol=1e-4)
)
def test_inference_full_attn(self):
model = BigBirdPegasusForConditionalGeneration.from_pretrained(MODEL_ID, attention_type="original_full")
model.to(torch_device)
input_ids = self._get_dummy_input_ids()
target_ids = self._get_dummy_target_ids()
outputs = model(input_ids, labels=target_ids)
prediction_logits = outputs.logits
self.assertEqual(prediction_logits.shape, torch.Size((1, 16, 96103)))
# fmt: off
expected_prediction_logits_slice = torch.tensor(
[[1.3418, 5.8304, 6.5662, 2.0448, 8.7702, 4.6579, 4.9947, 6.429, 2.4296, 7.9431, 4.217, 0.0672, 7.334, 5.1966, 2.9603, 6.0814, 4.6756, 7.5522, 5.076, 0.213, 6.6638, 0.6577, 0.244, 2.1221, 0.7531, -2.4076, 1.8731, 3.5594], [1.5525, 6.0524, 6.309, 2.6245, 9.229, 4.5213, 5.0913, 7.0622, 1.7992, 8.0962, 4.7994, -0.0248, 7.7168, 5.5878, 3.0883, 6.5248, 4.7895, 6.9974, 4.8787, 0.5445, 6.6686, 0.0102, -0.1659, 2.6195, 0.7389, -2.8956, 1.9928, 3.3777], [1.6407, 6.2104, 6.0331, 2.8076, 9.4074, 3.9772, 5.0574, 7.5316, 1.4201, 8.3035, 5.0212, -0.1031, 7.553, 5.5023, 3.1427, 6.7674, 4.4409, 6.457, 4.525, 0.728, 6.5422, -0.6234, -0.4726, 2.7486, 0.6985, -3.0804, 1.9669, 3.2365], [1.5065, 6.1271, 5.8296, 2.8405, 9.5649, 3.6834, 5.1214, 7.546, 0.9758, 8.3335, 5.1952, -0.1395, 7.4348, 5.6893, 3.2942, 7.0356, 4.1665, 5.9695, 4.3898, 0.8931, 6.3988, -0.8957, -0.7522, 2.8924, 0.6498, -3.4358, 1.8654, 2.9735]], # noqa: E231
device=torch_device,
)
# fmt: on
self.assertTrue(
torch.allclose(prediction_logits[0, 4:8, 128:156], expected_prediction_logits_slice, atol=1e-4)
)
def test_seq_to_seq_generation(self):
MODEL_ID = "google/bigbird-pegasus-large-arxiv"
model = BigBirdPegasusForConditionalGeneration.from_pretrained(MODEL_ID).to(torch_device)
tokenizer = PegasusTokenizer.from_pretrained(MODEL_ID)
ARTICLE_LEP = r"""the lep experiments at the resonance of @xmath1-boson have tested the standard model ( sm ) at quantum level , measuring the @xmath1-decay into fermion pairs with an accuracy of one part in ten thousands . the good agreement of the lep data with the sm predictions have severely constrained the behavior of new physics at the @xmath1-pole . taking these achievements into account one can imagine that the physics of @xmath1-boson will again play the central role in the frontier of particle physics if the next generation @xmath1 factory comes true with the generated @xmath1 events several orders of magnitude higher than that of the lep . this factory can be realized in the gigaz option of the international linear collider ( ilc)@xcite . the ilc is a proposed electron - positron collider with tunable energy ranging from @xmath12 to @xmath13 and polarized beams in its first phase , and the gigaz option corresponds to its operation on top of the resonance of @xmath1 boson by adding a bypass to its main beam line . given the high luminosity , @xmath14 , and the cross section at the resonance of @xmath1 boson , @xmath15 , about @xmath16 @xmath1 events can be generated in an operational year of @xmath17 of gigaz , which implies that the expected sensitivity to the branching ratio of @xmath1-decay can be improved from @xmath18 at the lep to @xmath19 at the gigaz@xcite . in light of this , the @xmath1-boson properties , especially its exotic or rare decays which are widely believed to be sensitive to new physics , should be investigated comprehensively to evaluate their potential in probing new physics . among the rare @xmath1-decays , the flavor changing ( fc ) processes were most extensively studied to explore the flavor texture in new physics @xcite , and it was found that , although these processes are severely suppressed in the sm , their branching ratios in new physics models can be greatly enhanced to @xmath19 for lepton flavor violation decays @xcite and @xmath20 for quark flavor violation decays @xcite . besides the fc processes , the @xmath1-decay into light higgs boson(s ) is another type of rare process that was widely studied , e.g. the decay @xmath21 ( @xmath22 ) with the particle @xmath0 denoting a light higgs boson was studied in @xcite , the decay @xmath23 was studied in the two higgs doublet model ( 2hdm)@xcite and the minimal supersymmetric standard model ( mssm)@xcite , and the decay @xmath4 was studied in a model independent way @xcite , in 2hdm@xcite and also in mssm@xcite . these studies indicate that , in contrast with the kinematic forbidden of these decays in the sm , the rates of these decays can be as large as @xmath18 in new physics models , which lie within the expected sensitivity of the gigaz . in this work , we extend the previous studies of these decays to some new models and investigate these decays altogether . we are motivated by some recent studies on the singlet extension of the mssm , such as the next - to - minimal supersymmetric standard model ( nmssm ) @xcite and the nearly minimal supersymmetric standard model ( nmssm ) @xcite , where a light cp - odd higgs boson @xmath0 with singlet - dominant component may naturally arise from the spontaneous breaking of some approximate global symmetry like @xmath24 or peccei - quuin symmetry @xcite . these non - minimal supersymmetric models can not only avoid the @xmath25-problem , but also alleviate the little hierarchy by having such a light higgs boson @xmath0 @xcite . we are also motivated by that , with the latest experiments , the properties of the light higgs boson are more stringently constrained than before . so it is worth updating the previous studies . so far there is no model - independent lower bound on the lightest higgs boson mass . in the sm , it must be heavier than @xmath26 gev , obtained from the null observation of the higgs boson at lep experiments . however , due to the more complex structure of the higgs sector in the extensions of the sm , this lower bound can be significantly relaxed according to recent studies , e.g. , for the cp - odd higgs boson @xmath0 we have @xmath27 gev in the nmssm @xcite , @xmath28 gev in the nmssm @xcite , and @xmath29 gev in the lepton - specific 2hdm ( l2hdm ) @xcite . with such a light cp - odd higgs boson , the z - decay into one or more @xmath0 is open up . noting that the decay @xmath30 is forbidden due to bose symmetry , we in this work study the rare @xmath1-decays @xmath6 ( @xmath22 ) , @xmath31 and @xmath4 in a comparative way for four models , namely the type - ii 2hdm@xcite , the l2hdm @xcite , the nmssm and the nmssm . in our study , we examine carefully the constraints on the light @xmath0 from many latest experimental results . this work is organized as follows . in sec . ii we briefly describe the four new physics models . in sec . iii we present the calculations of the rare @xmath1-decays . in sec . iv we list the constraints on the four new physics models . in sec . v we show the numerical results for the branching ratios of the rare @xmath1-decays in various models . finally , the conclusion is given in sec . as the most economical way , the sm utilizes one higgs doublet to break the electroweak symmetry . as a result , the sm predicts only one physical higgs boson with its properties totally determined by two free parameters . in new physics models , the higgs sector is usually extended by adding higgs doublets and/or singlets , and consequently , more physical higgs bosons are predicted along with more free parameters involved in . the general 2hdm contains two @xmath32 doublet higgs fields @xmath33 and @xmath34 , and with the assumption of cp - conserving , its scalar potential can be parameterized as@xcite : @xmath35,\end{aligned}\ ] ] where @xmath36 ( @xmath37 ) are free dimensionless parameters , and @xmath38 ( @xmath39 ) are the parameters with mass dimension . after the electroweak symmetry breaking , the spectrum of this higgs sector includes three massless goldstone modes , which become the longitudinal modes of @xmath40 and @xmath1 bosons , and five massive physical states : two cp - even higgs bosons @xmath41 and @xmath42 , one neutral cp - odd higgs particle @xmath0 and a pair of charged higgs bosons @xmath43 . noting the constraint @xmath44 with @xmath45 and @xmath46 denoting the vacuum expectation values ( vev ) of @xmath33 and @xmath34 respectively , we choose @xmath47 as the input parameters with @xmath48 , and @xmath49 being the mixing angle that diagonalizes the mass matrix of the cp - even higgs fields . the difference between the type - ii 2hdm and the l2hdm comes from the yukawa coupling of the higgs bosons to quark / lepton . in the type - ii 2hdm , one higgs doublet @xmath34 generates the masses of up - type quarks and the other doublet @xmath33 generates the masses of down - type quarks and charged leptons ; while in the l2hdm one higgs doublet @xmath33 couples only to leptons and the other doublet @xmath34 couples only to quarks . so the yukawa interactions of @xmath0 to fermions in these two models are given by @xcite @xmath50 with @xmath51 denoting generation index . obviously , in the type - ii 2hdm the @xmath52 coupling and the @xmath53 coupling can be simultaneously enhanced by @xmath54 , while in the l2hdm only the @xmath53 coupling is enhanced by @xmath55 . the structures of the nmssm and the nmssm are described by their superpotentials and corresponding soft - breaking terms , which are given by @xcite @xmath56 where @xmath57 is the superpotential of the mssm without the @xmath25 term , @xmath58 and @xmath59 are higgs doublet and singlet superfields with @xmath60 and @xmath61 being their scalar component respectively , @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 and @xmath67 are soft breaking parameters , and @xmath68 and @xmath69 are coefficients of the higgs self interactions . with the superpotentials and the soft - breaking terms , one can get the higgs potentials of the nmssm and the nmssm respectively . like the 2hdm , the higgs bosons with same cp property will mix and the mass eigenstates are obtained by diagonalizing the corresponding mass matrices : @xmath70 where the fields on the right hands of the equations are component fields of @xmath71 , @xmath72 and @xmath61 defined by @xmath73 @xmath74 and @xmath75 are respectively the cp - even and cp - odd neutral higgs bosons , @xmath76 and @xmath77 are goldstone bosons eaten by @xmath1 and @xmath78 , and @xmath79 is the charged higgs boson . so both the nmssm and nmssm predict three cp - even higgs bosons , two cp - odd higgs bosons and one pair of charged higgs bosons . in general , the lighter cp - odd higgs @xmath0 in these model is the mixture of the singlet field @xmath80 and the doublet field combination , @xmath81 , i.e. @xmath82 and its couplings to down - type quarks are then proportional to @xmath83 . so for singlet dominated @xmath0 , @xmath84 is small and the couplings are suppressed . as a comparison , the interactions of @xmath0 with the squarks are given by@xcite @xmath85 i.e. the interaction does not vanish when @xmath86 approaches zero . just like the 2hdm where we use the vevs of the higgs fields as fundamental parameters , we choose @xmath68 , @xmath69 , @xmath87 , @xmath88 , @xmath66 and @xmath89 as input parameters for the nmssm@xcite and @xmath68 , @xmath54 , @xmath88 , @xmath65 , @xmath90 and @xmath91 as input parameters for the nmssm@xcite . about the nmssm and the nmssm , three points should be noted . the first is for the two models , there is no explicit @xmath92term , and the effective @xmath25 parameter ( @xmath93 ) is generated when the scalar component of @xmath59 develops a vev . the second is , the nmssm is actually same as the nmssm with @xmath94@xcite , because the tadpole terms @xmath95 and its soft breaking term @xmath96 in the nmssm do not induce any interactions , except for the tree - level higgs boson masses and the minimization conditions . and the last is despite of the similarities , the nmssm has its own peculiarity , which comes from its neutralino sector . in the basis @xmath97 , its neutralino mass matrix is given by @xcite @xmath98 where @xmath99 and @xmath100 are @xmath101 and @xmath102 gaugino masses respectively , @xmath103 , @xmath104 , @xmath105 and @xmath106 . after diagonalizing this matrix one can get the mass eigenstate of the lightest neutralino @xmath107 with mass taking the following form @xcite @xmath108 this expression implies that @xmath107 must be lighter than about @xmath109 gev for @xmath110 ( from lower bound on chargnio mass ) and @xmath111 ( perturbativity bound ) . like the other supersymmetric models , @xmath107 as the lightest sparticle acts as the dark matter in the universe , but due to its singlino - dominated nature , it is difficult to annihilate sufficiently to get the correct density in the current universe . so the relic density of @xmath107 plays a crucial way in selecting the model parameters . for example , as shown in @xcite , for @xmath112 , there is no way to get the correct relic density , and for the other cases , @xmath107 mainly annihilates by exchanging @xmath1 boson for @xmath113 , or by exchanging a light cp - odd higgs boson @xmath0 with mass satisfying the relation @xmath114 for @xmath115 . for the annihilation , @xmath54 and @xmath25 are required to be less than 10 and @xmath116 respectively because through eq.([mass - exp ] ) a large @xmath87 or @xmath25 will suppress @xmath117 to make the annihilation more difficult . the properties of the lightest cp - odd higgs boson @xmath0 , such as its mass and couplings , are also limited tightly since @xmath0 plays an important role in @xmath107 annihilation . the phenomenology of the nmssm is also rather special , and this was discussed in detail in @xcite . in the type - ii 2hdm , l2hdm , nmssm and nmssm , the rare @xmath1-decays @xmath118 ( @xmath22 ) , @xmath3 and @xmath4 may proceed by the feynman diagrams shown in fig.[fig1 ] , fig.[fig2 ] and fig.[fig3 ] respectively . for these diagrams , the intermediate state @xmath119 represents all possible cp - even higgs bosons in the corresponding model , i.e. @xmath41 and @xmath42 in type - ii 2hdm and l2hdm and @xmath41 , @xmath42 and @xmath120 in nmssm and nmssm . in order to take into account the possible resonance effects of @xmath119 in fig.[fig1](c ) for @xmath2 and fig.[fig3 ] ( a ) for @xmath11 , we have calculated all the decay modes of @xmath119 and properly included the width effect in its propagator . as to the decay @xmath121 , two points should be noted . one is , unlike the decays @xmath6 and @xmath11 , this process proceeds only through loops mediated by quarks / leptons in the type - ii 2hdm and l2hdm , and additionally by sparticles in the nmssm and nmssm . so in most cases its rate should be much smaller than the other two . the other is due to cp - invariance , loops mediated by squarks / sleptons give no contribution to the decay@xcite . in actual calculation , this is reflected by the fact that the coupling coefficient of @xmath122 differs from that of @xmath123 by a minus sign ( see eq.([asqsq ] ) ) , and as a result , the squark - mediated contributions to @xmath121 are completely canceled out . with regard to the rare decay @xmath11 , we have more explanations . in the lowest order , this decay proceeds by the diagram shown in fig.[fig3 ] ( a ) , and hence one may think that , as a rough estimate , it is enough to only consider the contributions from fig.[fig3](a ) . however , we note that in some cases of the type - ii 2hdm and l2hdm , due to the cancelation of the contributions from different @xmath119 in fig.[fig3 ] ( a ) and also due to the potentially largeness of @xmath124 couplings ( i.e. larger than the electroweak scale @xmath125 ) , the radiative correction from the higgs - mediated loops may dominate over the tree level contribution even when the tree level prediction of the rate , @xmath126 , exceeds @xmath20 . on the other hand , we find the contribution from quark / lepton - mediated loops can be safely neglected if @xmath127 in the type - ii 2hdm and the l2hdm . in the nmssm and the nmssm , besides the corrections from the higgs- and quark / lepton - mediated loops , loops involving sparticles such as squarks , charginos and neutralinos can also contribute to the decay . we numerically checked that the contributions from squarks and charginos can be safely neglected if @xmath127 . we also calculated part of potentially large neutralino correction ( note that there are totally about @xmath128 diagrams for such correction ! ) and found they can be neglected too . since considering all the radiative corrections will make our numerical calculation rather slow , we only include the most important correction , namely that from higgs - mediated loops , in presenting our results for the four models . one can intuitively understand the relative smallness of the sparticle contribution to @xmath11 as follows . first consider the squark contribution which is induced by the @xmath129 interaction ( @xmath130 denotes the squark in chirality state ) and the @xmath131 interaction through box diagrams . because the @xmath132 interaction conserves the chirality of the squarks while the @xmath133 interaction violates the chirality , to get non - zero contribution to @xmath11 from the squark loops , at least four chiral flippings are needed , with three of them provided by @xmath131 interaction and the rest provided by the left - right squark mixing . this means that , if one calculates the amplitude in the chirality basis with the mass insertion method , the amplitude is suppressed by the mixing factor @xmath134 with @xmath135 being the off diagonal element in squark mass matrix . next consider the chargino / neutralino contributions . since for a light @xmath0 , its doublet component , parameterized by @xmath84 in eq.([mixing ] ) , is usually small , the couplings of @xmath0 with the sparticles will never be tremendously large@xcite . so the chargino / neutralino contributions are not important too . in our calculation of the decays , we work in the mass eigenstates of sparticles instead of in the chirality basis . for the type - ii 2hdm and the l2hdm , we consider the following constraints @xcite : * theoretical constraints on @xmath136 from perturbativity , unitarity and requirements that the scalar potential is finit at large field values and contains no flat directions @xcite , which imply that @xmath137 * the constraints from the lep search for neutral higgs bosons . we compute the signals from the higgs - strahlung production @xmath138 ( @xmath139 ) with @xmath140 @xcite and from the associated production @xmath141 with @xmath142 @xcite , and compare them with the corresponding lep data which have been inputted into our code . we also consider the constraints from @xmath138 by looking for a peak of @xmath143 recoil mass distribution of @xmath1-boson @xcite and the constraint of @xmath144 mev when @xmath145 @xcite . + these constraints limit the quantities such as @xmath146 \times br ( h_i \to \bar{b } b ) $ ] on the @xmath147 plane with the the subscript @xmath148 denoting the coupling coefficient of the @xmath149 interaction . they also impose a model - dependent lower bound on @xmath150 , e.g. , @xmath151 for the type - ii 2hdm ( from our scan results ) , @xmath152 for the l2hdm@xcite , and @xmath153 for the nmssm @xcite . these bounds are significantly lower than that of the sm , i.e. @xmath154 , partially because in new physics models , unconventional decay modes of @xmath155 such as @xmath156 are open up . as to the nmssm , another specific reason for allowing a significantly lighter cp - even higgs boson is that the boson may be singlet - dominated in this model . + with regard to the lightest cp - odd higgs boson @xmath0 , we checked that there is no lower bound on its mass so long as the @xmath157 interaction is weak or @xmath155 is sufficiently heavy . * the constraints from the lep search for a light higgs boson via the yukawa process @xmath158 with @xmath22 and @xmath61 denoting a scalar @xcite . these constraints can limit the @xmath159 coupling versus @xmath160 in new physics models . * the constraints from the cleo - iii limit on @xmath161 and the latest babar limits on @xmath162 . these constraints will put very tight constraints on the @xmath163 coupling for @xmath164 . in our analysis , we use the results of fig.8 in the second paper of @xcite to excluded the unfavored points . * the constraints from @xmath165 couplings . since the higgs sector can give sizable higher order corrections to @xmath165 couplings , we calculate them to one loop level and require the corrected @xmath165 couplings to lie within the @xmath166 range of their fitted value . the sm predictions for the couplings at @xmath1-pole are given by @xmath167 and @xmath168 @xcite , and the fitted values are given by @xmath169 and @xmath170 , respectively@xcite . we adopt the formula in @xcite to the 2hdm in our calculation . * the constraints from @xmath171 leptonic decay . we require the new physics correction to the branching ratio @xmath172 to be in the range of @xmath173 @xcite . we use the formula in @xcite in our calculation . + about the constraints ( 5 ) and ( 6 ) , two points should be noted . one is all higgs bosons are involved in the constraints by entering the self energy of @xmath171 lepton , the @xmath174 vertex correction or the @xmath175 vertex correction , and also the box diagrams for @xmath176@xcite . since the yukawa couplings of the higgs bosons to @xmath171 lepton get enhanced by @xmath54 and so do the corrections , @xmath54 must be upper bounded for given spectrum of the higgs sector . generally speaking , the lighter @xmath0 is , the more tightly @xmath54 is limited@xcite . the other point is in the type - ii 2hdm , @xmath177 , b - physics observables as well as @xmath178 decays discussed above can constraint the model in a tighter way than the constraints ( 5 ) and ( 6 ) since the yukawa couplings of @xmath171 lepton and @xmath179 quark are simultaneously enhanced by @xmath54 . but for the l2hdm , because only the yukawa couplings of @xmath171 lepton get enhanced ( see eq.[yukawa ] ) , the constraints ( 5 ) and ( 6 ) are more important in limiting @xmath54 . * indirect constraints from the precision electroweak observables such as @xmath180 , @xmath181 and @xmath182 , or their combinations @xmath183 @xcite . we require @xmath184 to be compatible with the lep / sld data at @xmath185 confidence level@xcite . we also require new physics prediction of @xmath186 is within the @xmath187 range of its experimental value . the latest results for @xmath188 are @xmath189 ( measured value ) and @xmath190 ( sm prediction ) for @xmath191 gev @xcite . in our code , we adopt the formula for these observables presented in @xcite to the type - ii 2hdm and the l2hdm respectively . + in calculating @xmath180 , @xmath181 and @xmath182 , we note that these observables get dominant contributions from the self energies of the gauge bosons @xmath1 , @xmath192 and @xmath193 . since there is no @xmath194 coupling or @xmath195 coupling , @xmath0 must be associated with the other higgs bosons to contribute to the self energies . so by the uv convergence of these quantities , one can infer that , for the case of a light @xmath0 and @xmath196 , these quantities depend on the spectrum of the higgs sector in a way like @xmath197 at leading order , which implies that a light @xmath0 can still survive the constraints from the precision electroweak observables given the splitting between @xmath150 and @xmath198 is moderate@xcite . * the constraints from b physics observables such as the branching ratios for @xmath199 , @xmath200 and @xmath201 , and the mass differences @xmath202 and @xmath203 . we require their theoretical predications to agree with the corresponding experimental values at @xmath187 level . + in the type - ii 2hdm and the l2hdm , only the charged higgs boson contributes to these observables by loops , so one can expect that @xmath198 versus @xmath54 is to be limited . combined analysis of the limits in the type - ii 2hdm has been done by the ckmfitter group , and the lower bound of @xmath204 as a function of @xmath87 was given in fig.11 of @xcite . this analysis indicates that @xmath198 must be heavier than @xmath205 at @xmath185 c.l . regardless the value of @xmath54 . in this work , we use the results of fig.11 in @xcite to exclude the unfavored points . as for the l2hdm , b physics actually can not put any constraints@xcite because in this model the couplings of the charged higgs boson to quarks are proportional to @xmath206 and in the case of large @xmath54 which we are interested in , they are suppressed . in our analysis of the l2hdm , we impose the lep bound on @xmath198 , i.e. @xmath207@xcite . * the constraints from the muon anomalous magnetic moment @xmath208 . now both the theoretical prediction and the experimental measured value of @xmath208 have reached a remarkable precision , but a significant deviation still exists : @xmath209 @xcite . in the 2hdm , @xmath208 gets additional contributions from the one - loop diagrams induced by the higgs bosons and also from the two - loop barr - zee diagrams mediated by @xmath0 and @xmath155@xcite . if the higgs bosons are much heavier than @xmath25 lepton mass , the contributions from the barr - zee diagrams are more important , and to efficiently alleviate the discrepancy of @xmath208 , one needs a light @xmath0 along with its enhanced couplings to @xmath25 lepton and also to heavy fermions such as bottom quark and @xmath171 lepton to push up the effects of the barr - zee diagram@xcite . the cp - even higgs bosons are usually preferred to be heavy since their contributions to @xmath208 are negative . + in the type - ii 2hdm , because @xmath54 is tightly constrained by the process @xmath210 at the lep@xcite and the @xmath178 decay@xcite , the barr - zee diagram contribution is insufficient to enhance @xmath208 to @xmath187 range around its measured value@xcite . so in our analysis , we require the type - ii 2hdm to explain @xmath208 at @xmath211 level . while for the l2hdm , @xmath54 is less constrained compared with the type - ii 2hdm , and the barr - zee diagram involving the @xmath171-loop is capable to push up greatly the theoretical prediction of @xmath208@xcite . therefore , we require the l2hdm to explain the discrepancy at @xmath187 level . + unlike the other constraints discussed above , the @xmath208 constraint will put a two - sided bound on @xmath54 since on the one hand , it needs a large @xmath54 to enhance the barr - zee contribution , but on the other hand , too large @xmath54 will result in an unacceptable large @xmath208 . * since this paper concentrates on a light @xmath0 , the decay @xmath212 is open up with a possible large decay width . we require the width of any higgs boson to be smaller than its mass to avoid a too fat higgs boson@xcite . we checked that for the scenario characterized by @xmath213 , the coefficient of @xmath214 interaction is usually larger than the electroweak scale @xmath125 , and consequently a large decay width is resulted . for the nmssm and nmssm , the above constraints become more complicated because in these models , not only more higgs bosons are involved in , but also sparticles enter the constraints . so it is not easy to understand some of the constraints intuitively . take the process @xmath199 as an example . in the supersymmetric models , besides the charged higgs contribution , chargino loops , gluino loops as well as neutralino loops also contribute to the process@xcite , and depending on the susy parameters , any of these contributions may become dominated over or be canceled by other contributions . as a result , although the charged higgs affects the process in the same way as that in the type - ii 2hdm , charged higgs as light as @xmath215 is still allowed even for @xmath216@xcite . since among the constraints , @xmath208 is rather peculiar in that it needs new physics to explain the discrepancy between @xmath217 and @xmath218 , we discuss more about its dependence on susy parameters . in the nmssm and the nmssm , @xmath208 receives contributions from higgs loops and neutralino / chargino loops . for the higgs contribution , it is quite similar to that of the type - ii 2hdm except that more higgs bosons are involved in@xcite . for the neutralino / chargino contribution , in the light bino limit ( i.e. @xmath219 ) , it can be approximated by@xcite @xmath220 for @xmath221 with @xmath222 being smuon mass . so combining the two contributions together , one can learn that a light @xmath0 along with large @xmath54 and/or light smuon with moderate @xmath87 are favored to dilute the discrepancy . because more parameters are involved in the constraints on the supersymmetric models , we consider following additional constraints to further limit their parameters : * direct bounds on sparticle masses from the lep1 , the lep2 and the tevatron experiments @xcite . * the lep1 bound on invisible z decay @xmath223 ; the lep2 bound on neutralino production @xmath224 and @xmath225@xcite . * dark matter constraints from the wmap relic density 0.0975 @xmath226 0.1213 @xcite . note that among the above constraints , the constraint ( 2 ) on higgs sector and the constraint ( c ) on neutralino sector are very important . this is because in the supersymmetric models , the sm - like higgs is upper bounded by about @xmath227 at tree level and by about @xmath228 at loop level , and that the relic density restricts the lsp annihilation cross section in a certain narrow range . in our analysis of the nmssm , we calculate the constraints ( 3 ) and ( 5 - 7 ) by ourselves and utilize the code nmssmtools @xcite to implement the rest constraints . we also extend nmssmtools to the nmssm to implement the constraints . for the extension , the most difficult thing we faced is how to adapt the code micromegas@xcite to the nmssm case . we solve this problem by noting the following facts : * as we mentioned before , the nmssm is actually same as the nmssm with the trilinear singlet term setting to zero . so we can utilize the model file of the nmssm as the input of the micromegas and set @xmath229 . * since in the nmssm , the lsp is too light to annihilate into higgs pairs , there is no need to reconstruct the effective higgs potential to calculate precisely the annihilation channel @xmath230 with @xmath61 denoting any of higgs bosons@xcite . we thank the authors of the nmssmtools for helpful discussion on this issue when we finish such extension@xcite . with the above constraints , we perform four independent random scans over the parameter space of the type - ii 2hdm , the l2hdm , the nmssm and the nmssm respectively . we vary the parameters in following ranges : @xmath231 for the type - ii 2hdm , @xmath232 for the l2hdm , @xmath233 for the nmssm , and @xmath234 for the nmssm . in performing the scans , we note that for the nmssm and the nmssm , some constraints also rely on the gaugino masses and the soft breaking parameters in the squark sector and the slepton sector . since these parameters affect little on the properties of @xmath0 , we fix them to reduce the number of free parameters in our scan . for the squark sector , we adopt the @xmath235 scenario which assumes that the soft mass parameters for the third generation squarks are degenerate : @xmath236 800 gev , and that the trilinear couplings of the third generation squarks are also degenerate , @xmath237 with @xmath238 . for the slepton sector , we assume all the soft - breaking masses and trilinear parameters to be 100 gev . this setting is necessary for the nmssm since this model is difficult to explain the muon anomalous moment at @xmath239 level for heavy sleptons@xcite . finally , we assume the grand unification relation @xmath240 for the gaugino masses with @xmath241 being fine structure constants of the different gauge group . with large number of random points in the scans , we finally get about @xmath242 , @xmath243 , @xmath244 and @xmath242 samples for the type - ii 2hdm , the l2hdm , the nmssm and the nmssm respectively which survive the constraints and satisfy @xmath245 . analyzing the properties of the @xmath0 indicates that for most of the surviving points in the nmssm and the nmssm , its dominant component is the singlet field ( numerically speaking , @xmath246 ) so that its couplings to the sm fermions are suppressed@xcite . our analysis also indicates that the main decay products of @xmath0 are @xmath247 for the l2hdm@xcite , @xmath248 ( dominant ) and @xmath247 ( subdominant ) for the type - ii 2hdm , the nmssm and the nmssm , and in some rare cases , neutralino pairs in the nmssm@xcite . in fig.[fig4 ] , we project the surviving samples on the @xmath249 plane . this figure shows that the allowed range of @xmath54 is from @xmath250 to @xmath251 in the type - ii 2hdm , and from @xmath252 to @xmath253 in the l2hdm . just as we introduced before , the lower bounds of @xmath254 come from the fact that we require the models to explain the muon anomalous moment , while the upper bound is due to we have imposed the constraint from the lep process @xmath255 , which have limited the upper reach of the @xmath256 coupling for light @xmath61 @xcite(for the dependence of @xmath256 coupling on @xmath54 , see sec . this figure also indicates that for the nmssm and the nmssm , @xmath54 is upper bounded by @xmath257 . for the nmssm , this is because large @xmath87 can suppress the dark matter mass to make its annihilation difficult ( see @xcite and also sec . ii ) , but for the nmssm , this is because we choose a light slepton mass so that large @xmath54 can enhance @xmath208 too significantly to be experimentally unacceptable . we checked that for the slepton mass as heavy as @xmath258 , @xmath259 is still allowed for the nmssm . in fig.[fig5 ] and fig.[fig6 ] , we show the branching ratios of @xmath260 and @xmath261 respectively . fig.[fig5 ] indicates , among the four models , the type - ii 2hdm predicts the largest ratio for @xmath260 with its value varying from @xmath262 to @xmath263 . the underlying reason is in the type - ii 2hdm , the @xmath264 coupling is enhanced by @xmath54 ( see fig.[fig4 ] ) , while in the other three model , the coupling is suppressed either by @xmath265 or by the singlet component of the @xmath0 . fig.[fig6 ] shows that the l2hdm predicts the largest rate for @xmath266 with its value reaching @xmath5 in optimum case , and for the other three models , the ratio of @xmath261 is at least about one order smaller than that of @xmath267 . this feature can be easily understood from the @xmath268 coupling introduced in sect . we emphasize that , if the nature prefers a light @xmath0 , @xmath260 and/or @xmath269 in the type - ii 2hdm and the l2hdm will be observable at the gigaz . then by the rates of the two decays , one can determine whether the type - ii 2hdm or the l2hdm is the right theory . on the other hand , if both decays are observed with small rates or fail to be observed , the singlet extensions of the mssm are favored . in fig.[fig7 ] , we show the rate of @xmath3 as the function of @xmath270 . this figure indicates that the branching ratio of @xmath121 can reach @xmath271 , @xmath272 , @xmath273 and @xmath274 for the optimal cases of the type - ii 2hdm , the l2hdm , the nmssm and the nmssm respectively , which implies that the decay @xmath121 will never be observable at the gigaz if the studied model is chosen by nature . the reason for the smallness is , as we pointed out before , that the decay @xmath121 proceeds only at loop level . comparing the optimum cases of the type - ii 2hdm , the nmssm and the nmssm shown in fig.5 - 7 , one may find that the relation @xmath275 holds for any of the decays . this is because the decays are all induced by the yukawa couplings with similar structure for the models . in the supersymmetric models , the large singlet component of the light @xmath0 is to suppress the yukawa couplings , and the @xmath0 in the nmssm has more singlet component than that in the nmssm . next we consider the decay @xmath11 , which , unlike the above decays , depends on the higgs self interactions . in fig.[fig8 ] we plot its rate as a function of @xmath270 and this figure indicates that the @xmath276 may be the largest among the ratios of the exotic @xmath1 decays , reaching @xmath277 in the optimum cases of the type - ii 2hdm , the l2hdm and the nmssm . the underlying reason is , in some cases , the intermediate state @xmath119 in fig.[fig3 ] ( a ) may be on - shell . in fact , we find this is one of the main differences between the nmssm and the nmssm , that is , in the nmssm , @xmath119 in fig.[fig3 ] ( a ) may be on - shell ( corresponds to the points with large @xmath278 ) while in the nmssm , this seems impossible . so we conclude that the decay @xmath11 may serve as an alternative channel to test new physics models , especially it may be used to distinguish the nmssm from the nmssm if the supersymmetry is found at the lhc and the @xmath11 is observed at the gigaz with large rate . before we end our discussion , we note that in the nmssm , the higgs boson @xmath0 may be lighter than @xmath279 without conflicting with low energy data from @xmath178 decays and the other observables ( see fig.[fig4]-[fig8 ] ) . in this case , @xmath0 is axion - like as pointed out in @xcite . we checked that , among the rare @xmath1 decays discussed in this paper , the largest branching ratio comes from @xmath280 which can reach @xmath281 . since in this case , the decay product of @xmath0 is highly collinear muon pair , detecting the decay @xmath280 may need some knowledge about detectors , which is beyond our discussion . in this paper , we studied the rare @xmath1-decays @xmath2 ( @xmath7 ) , @xmath282 and @xmath4 in the type - ii 2hdm , lepton - specific 2hdm , nmssm and nmssm , which predict a light cp - odd higgs boson @xmath0 . in the parameter space allowed by current experiments , the branching ratio can be as large as @xmath5 for @xmath118 , @xmath8 for @xmath3 and @xmath9 for @xmath4 , which implies that the decays @xmath2 and @xmath283 may be accessible at the gigaz option . since different models predict different size of branching ratios , these decays can be used to distinguish different model through the measurement of these rare decays . this work was supported in part by hastit under grant no . 2009hastit004 , by the national natural science foundation of china ( nnsfc ) under grant nos . 10821504 , 10725526 , 10635030 , 10775039 , 11075045 and by the project of knowledge innovation program ( pkip ) of chinese academy of sciences under grant no . . for some reviews , see , e.g. , m. a. perez , g. tavares - velasco and j. j. toscano , int . j. mod . a * 19 * , 159 ( 2004 ) ; j. m. yang , arxiv:1006.2594 . j. i. illana , m. masip , 67 , 035004 ( 2003 ) ; j. cao , z. xiong , j. m. yang , 32 , 245 ( 2004 ) . d. atwood _ et al_. , 66 , 093005 ( 2002 ) . j. kalinowski , and s. pokorski , 219 , 116 ( 1989 ) ; a. djouadi , p. m. zerwas and j. zunft , 259 , 175 ( 1991 ) ; a. djouadi , j. kalinowski , and p. m. zerwas , z. phys . c * 54 * , 255 ( 1992 ) . m. krawczyk , _ et al . _ , 19 , 463 ( 2001 ) ; 8 , 495 ( 1999 ) . j. f. gunion , g. gamberini and s. f. novaes , 38 , 3481 ( 1988 ) ; thomas j. weiler and tzu - chiang yuan , 318 , 337 ( 1989 ) ; a. djouadi , _ et al . _ , 1 , 163 ( 1998)[hep - ph/9701342 ] . d. chang and w. y. keung , phys . lett . * 77 * , 3732 ( 1996 ) . e. keith and e. ma , 57 , 2017 ( 1998 ) ; m. a. perez , g. tavares - velasco and j. j. toscano , int . j. mod.phys . a * 19 * , 159 ( 2004 ) . f. larios , g. tavares - velasco and c. p. yuan , 64 , 055004 ( 2001 ) ; 66 , 075006 ( 2002 ) . a. djouadi , _ et al . _ , 10 , 27 ( 1999 ) [ hep - ph/9903229 ] . for a detailed introduction of the nmssm , see f. franke and h. fraas , int . j. mod . a * 12 * ( 1997 ) 479 ; for a recent review of the nmssm , see for example , u. ellwanger , c. hugonie , and a. m. teixeira , arxiv : 0910.1785 . see , e.g. , j. r. ellis , j. f. gunion , h. e. haber , l. roszkowski and f. zwirner , phys . rev . d * 39 * ( 1989 ) 844 ; m. drees , int . j. mod . phys . a * 4 * ( 1989 ) 3635 ; u. ellwanger , m. rausch de traubenberg and c. a. savoy , phys . b * 315 * ( 1993 ) 331 ; nucl . b * 492 * ( 1997 ) 21 ; d.j . miller , r. nevzorov , p.m. zerwas , 681 , 3 ( 2004 ) . c. panagiotakopoulos , k. tamvakis , 446 , 224 ( 1999 ) ; 469 , 145 ( 1999 ) ; c. panagiotakopoulos , a. pilaftsis , 63 , 055003 ( 2001 ) ; a. dedes , _ et al . _ , 63 , 055009 ( 2001 ) ; a. menon , _ et al . _ , 70 , 035005 ( 2004 ) ; v. barger , _ et al . _ , 630 , 85 ( 2005 ) . c. balazs , _ et al . _ , 0706 , 066 ( 2007 ) . b. a. dobrescu , k. t. matchev , 0009 , 031 ( 2000 ) ; a. arhrib , k. cheung , t. j. hou , k. w. song , hep - ph/0611211 ; 0703 , 073 ( 2007 ) ; x. g. he , j. tandean , and g. valencia , 98 , 081802 ( 2007 ) ; 0806 , 002 ( 2008 ) ; f. domingo _ et al_. , 0901 , 061 ( 2009 ) ; gudrun hiller , 70 , 034018 ( 2004 ) ; r. dermisek , and john f. gunion , 75 , 075019 ( 2007 ) ; 79 , 055014 ( 2009 ) ; 81 , 055001 ( 2010 ) ; r. dermisek , john f. gunion , and b. mcelrath , 76 , 051105 ( 2007 ) ; z. heng , _ et al_. , 77 , 095012 ( 2008 ) ; a. belyaev _ et al_. , 81 , 075021 ( 2010 ) ; d. das and u. ellwanger , arxiv:1007.1151 [ hep - ph ] . s. andreas , o. lebedev , s. ramos - sanchez and a. ringwald , arxiv:1005.3978 [ hep - ph ] . j. f. gunion , jhep * 0908 * , 032 ( 2009 ) ; r. dermisek and j. f. gunion , phys . rev . d * 81 * , 075003 ( 2010 ) . r. dermisek and j. f. gunion , phys . lett . * 95 * , 041801 ( 2005 ) ; phys . d * 73 * , 111701 ( 2006 ) . j. cao , h. e. logan , j. m. yang , 79 , 091701 ( 2009 ) . j. cao , p. wan , l. wu , j. m. yang , 80 , 071701 ( 2009 ) . j. f. gunion and h. e. haber , 67 , 075019 ( 2003 ) . r. m. barnett , _ et al . _ , phys . b * 136 * , 191 ( 1984 ) ; r. m. barnett , g. senjanovic and d. wyler , phys . d * 30 * , 1529 ( 1984 ) ; y. grossman , nucl . b * 426 * , 355 ( 1994 ) . h. s. goh , l. j. hall and p. kumar , jhep * 0905 * , 097 ( 2009 ) ; a. g. akeroyd and w. j. stirling , nucl . b * 447 * , 3 ( 1995 ) ; a. g. akeroyd , phys . b * 377 * , 95 ( 1996 ) ; h. e. logan and d. maclennan , phys . rev . d * 79 * , 115022 ( 2009 ) ; m. aoki , _ et al . _ , arxiv:0902.4665 [ hep - ph ] . v. barger , p. langacker , h. s. lee and g. shaughnessy , phys . d * 73 * , 115010 ( 2006 ) . s. hesselbach , _ et . _ , arxiv:0810.0511v2 [ hep - ph ] . de vivie and p. janot [ aleph collaboration ] , pa13 - 027 contribution to the international conference on high energy physics , warsaw , poland , 2531 july 1996 ; j. kurowska , o. grajek and p. zalewski [ delphi collaboration ] , cern - open-99 - 385 . [ aleph collaboration and delphi collaboration and l3 collaboration ] , phys . rept . * 427 * , 257 ( 2006 ) . j. cao and j. m. yang , jhep * 0812 * , 006 ( 2008 ) . m. krawczyk and d. temes , eur . j. c * 44 * , 435 ( 2005 ) . g. altarelli and r. barbieri , 253 , 161 ( 1991 ) ; m. e. peskin , t. takeuchi , 46 , 381 ( 1992 ) . c. amsler , _ et al . _ , ( particle data group ) , 667 , 1 ( 2008 ) . o. deschamps , s. descotes - genon , s. monteil , v. niess , s. tjampens and v. tisserand , arxiv:0907.5135 [ hep - ph ] . s. su and b. thomas , phys . d * 79 * , 095014 ( 2009 ) . g. abbiendi , _ et al . _ , eur . phys . j. c * 32 * , 453 ( 2004 ) . m. davier , _ et al . _ , 66 , 1 ( 2010 ) . k. cheung , _ et al . _ , phys . d * 64 * , 111301 ( 2001 ) . k. cheung and o. c. w. kong , phys . d * 68 * , 053003 ( 2003 ) . t. besmer , c. greub , t.hurth , 609 , 359 ( 2001 ) ; f. borzumati , _ et al . _ , 62 , 075005(2000 ) . j. cao , k. i. hikasa , w. wang , j. m. yang and l. x. yu , phys . d * 82 * , 051701 ( 2010 ) [ arxiv:1006.4811 [ hep - ph ] ] . j. f. gunion , _ et . d * 73 * , 015011 ( 2006 ) . martin and j. d. wells , phys . d * 64 * , 035003 ( 2001 ) . j. abdallah _ et al . _ , eur . j. c * 31 * , 421 ( 2004 ) ; g. abbiendi _ et al . _ , eur . j. c * 35 * , 1 ( 2004 ) . j. dunkley _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 180 * , 306 ( 2009 ) [ arxiv:0803.0586 [ astro - ph ] ] . u. ellwanger _ et al . _ , 02 , 066 ( 2005 ) . g. belanger , f. boudjema , a. pukhov and a. semenov , comput . commun . * 174 * , 577 ( 2006 ) ; comput . phys . commun . * 176 * , 367 ( 2007 ) . g. belanger , f. boudjema , c. hugonie , a. pukhov and a. semenov , jcap * 0509 * , 001 ( 2005 ) ."""
ARTICLE_MAGNET = r"""it is well known that the classical magnetoresistance ( mr ) in metals or semiconductors with a closed free electron fermi surface increases quadratically with increasing magnetic field @xmath2 for @xmath3 and saturates when @xmath4 . here @xmath5 is the zero - magnetic - field mobility . hence , the extraordinarily high and linear mr ( lmr ) , which breaks this familiar rule , has been gaining much attention as soon as its discovery . in the past decade , this unexpected lmr has been reported in silver chalcogenide,@xcite indium antimonide,@xcite silicon,@xcite mnas - gaas composite material,@xcite and graphene.@xcite kapitza s linear law@xcite indicates that the metal shows a magnetoresistance linear in perpendicular magnetic field when it has an open fermi surface and a mean free path longer than the electronic larmor radius . recently , another two models , irrespective of the open fermi surface , have been constructed to provide possible mechanisms for the lmr phenomenon . abrikosov suggested a quantum - limit origin of lmr for the homogenous system with a gapless linear energy spectrum.@xcite his model requires that landau levels are well formed and the carrier concentration is small that all electrons occupy only the lowest landau band . alternatively , parish and littlewood developed a classical model without involving linear spectrum.@xcite ignoring the concrete microscopic mechanism , they attributed this unusual mr to the mobility fluctuations in a strongly inhomogenous system . topological insulators@xcite ( tis ) are novel materials with a full energy gap in bulk , while there are gapless surface states . due to its unique band structure with only one helical dirac cone and linear energy dispersion,@xcite the surface states of the ti bi@xmath0se@xmath1 become an excellent platform for the study of quantum - limit lmr . the recent experiment in this flat surface system , however , reported that a large positive mr , which becomes very linear above a characteristic field of @xmath6@xmath7@xmath8 t , was observed even in an opposite situation where the carrier sheet density is high that electrons occupy more than one landau levels.@xcite moreover , they found that raising temperature to room temperature almost has no influence on the observed lmr . it is striking that this observation is in conflict with abrikosov s model and also with the classical parish - littlewood model . so far a reliable theoretical scheme capable of explaining this novel experiment has still been lacking . in this paper , we generalize the balance - equation approach@xcite to a system modeling the surface states of a three - dimensional ti to investigate the two - dimensional magnetotransport in it . we find that a positive , nonsaturating and dominantly linear magnetoresistance can appear within quite wide magnetic - field range in the ti surface state having a positive and finite effective g - factor . this linear magnetoresistance shows up in the system of high carrier concentration and low mobility when electrons are in extended states and spread over many smeared landau levels , and persists up to room temperature , providing a possible mechanism for the recently observed linear magnetoresistance in topological insulator bi@xmath0se@xmath1 nanoribbons.@xcite we consider the surface state of a bi@xmath0se@xmath1-type large bulk gap ti in the @xmath9-@xmath10 plane under the influence of a uniform magnetic field @xmath11 applied along the @xmath12 direction.@xcite following the experimental observation,@xcite we assume that the fermi energy locates in the gap of the bulk band and above the dirac point , i.e. the surface carriers are electrons . further , the separations of the fermi energy from the bottom of bulk band and dirac point are much larger than the highest temperature ( @xmath13 ) considered in this work . hence , the contribution from the bulk band to the magnetotransport is negligible . these electrons , scattered by randomly distributed impurities and by phonons , are driven by a uniform in - plane electric field @xmath14 in the topological surface . the hamiltonian of this many - electron and phonon system consists of an electron part @xmath15 , a phonon part @xmath16 , and electron - impurity and electron - phonon interactions @xmath17 and @xmath18 : @xmath19 here , the electron hamiltonian is taken in the form @xmath20 , \ ] ] in which @xmath21 , @xmath22 , @xmath23 and @xmath24 , stand , respectively , for the canonical momentum , coordinate , momentum and spin operators of the @xmath25th electron having charge @xmath26 , @xmath27 is the vector potential of the perpendicular magnetic field @xmath28 in the landau gauge , @xmath29 is the fermi velocity , @xmath30 is the effective g - factor of the surface electron , and @xmath31 is the bohr magneton with @xmath32 the free electron mass . the sum index @xmath25 in eq.([helectron ] ) goes over all electrons of total number @xmath33 in the surface state of unit area . in the frame work of balance equation approach,@xcite the two - dimensional center - of - mass ( c.m . ) momentum and coordinate @xmath34 and @xmath35 , and the relative - electron momenta and coordinates @xmath36 and @xmath37 are introduced to write the hamiltonian @xmath15 into the sum of a single - particle c.m . part @xmath38 and a many - particle relative - electron part @xmath39 : @xmath40 , with @xmath41.\end{aligned}\ ] ] in this , @xmath42 is the canonical momentum of the center - of - mass and @xmath43 is the canonical momentum for the @xmath25th relative electron . here we have also introduced c.m . spin operators @xmath44 and @xmath45 . the commutation relations between the c.m . spin operators @xmath46 and @xmath47 and the spin operators @xmath48 , @xmath49 and @xmath50 of the @xmath25th electron are of order of @xmath51 : @xmath52= n^{-1}2\,{\rm i}\,\varepsi lon_{\beta_1\beta_2\beta_3}\sigma_j^{\beta_3}$ ] with @xmath53 . therefore , for a macroscopic large @xmath33 system , the c.m . part @xmath38 actually commutes with the relative - electron part @xmath54 in the hamiltonian , i.e. the c.m . motion and the relative motion of electrons are truly separated from each other . the couplings between the two emerge only through the electron impurity and electron phonon interactions . furthermore , the electric field @xmath55 shows up only in @xmath38 . and , in view of @xmath56={\rm i}\delta_{\alpha \beta}(\delta_{ij}-1/n)\simeq { \rm i}\delta_{\alpha\beta}\delta_{ij}$ ] , i.e. the relative - electron momenta and coordinates can be treated as canonical conjugate variables , the relative - motion part @xmath54 is just the hamiltonian of @xmath33 electrons in the surface state of ti in the magnetic field without the presence of the electric field . in terms of the c.m . coordinate @xmath57 and the relative electron density operator @xmath58 , the electron impurity and electron phonon interactions can be written as@xcite @xmath59 here @xmath60 and @xmath61 are respectively the impurity potential ( an impurity at randomly distributed position @xmath62 ) and electron phonon coupling matrix element in the plane - wave representation , and @xmath63 with @xmath64 and @xmath65 being the creation and annihilation operators for a phonon of wavevector @xmath66 in branch @xmath67 having frequency @xmath68 . velocity ( operator ) @xmath69 is the time variation of its coordinate : @xmath70= v_{\rm f}(\sigma_{\rm c}^y\ , \hat{i}-\sigma_{\rm c}^x\ , \hat{j})$ ] . to derive a force - balance equation for steady state transport we consider the heisenberg equation for the rate of change of the c.m . canonical momentum @xmath71 : @xmath72= - n e({\bm v}\times { \bm b})- n e{\bm e}+{\bm { f}}_{\rm i}+{\bm { f}}_{\rm p},\ ] ] in which the frictional forces @xmath73 and @xmath74 share the same expressions as given in ref .. the statistical average of the operator equation can be determined to linear order in the electron impurity and electron phonon interactions @xmath17 and @xmath18 with the initial density matrix @xmath75 at temperature @xmath76 when the in - plane electric field @xmath77 is not strong . for steady - transport states we have @xmath78 , leading to a force - balance equation of the form @xmath79 here @xmath80 , the statistically averaged velocity of the moving center - of - mass , is identified as the average rate of change of its position , i.e. the drift velocity of the electron system driven by the electric field @xmath77 , and @xmath81 and @xmath82 are frictional forces experienced by the center - of - mass due to impurity and phonon scatterings : @xmath83,\label{fp}\end{aligned}\ ] ] in which @xmath84 is the bose distribution function , @xmath85 , and @xmath86 stands for the imaginary part of the fourier spectrum of the relative - electron density correlation function defined by @xmath87\big\rangle_{0},\ ] ] where @xmath88 and @xmath89 denotes the statistical averaging over the initial density matrix @xmath90.@xcite the force - balance equation describes the steady - state two - dimensional magnetotransport in the surface state of a ti . note that the frictional forces @xmath81 and @xmath82 are in the opposite direction of the drift velocity @xmath91 and their magnitudes are functions of @xmath92 only . with the drift velocity @xmath93 in the @xmath9 direction , the force - balance equation eq . yields a transverse resistivity @xmath94 , and a longitudinal resistivity @xmath95 . the linear one is in the form @xmath96 for calculating the electron density correlation function @xmath97 we proceed in the landau representation.@xcite the landau levels of the single - particle hamiltonian @xmath98 of the relative - electron system in the absence of electric field are composed of a positive `` @xmath99 '' and a negative `` @xmath100 '' branch@xcite @xmath101 with @xmath102 and @xmath103 , and a zero ( @xmath104 ) level @xmath105 the corresponding landau wave functions are @xmath106 and @xmath107 for @xmath108 ; and @xmath109 for @xmath104 . here @xmath110 is the wavevector of the system along @xmath9 direction ; @xmath111 with @xmath112 ; and @xmath113 is the harmonic oscillator eigenfunction with @xmath114 being the hermite polynomial , @xmath115 , and @xmath116 . each landau level contains @xmath117 electron states for system of unit surface area . the positive branch @xmath118 and the @xmath104 level @xmath119 of the above energy spectra are indeed quite close to those of the surface states in the bulk gap of bi@xmath0se@xmath1-family materials derived from microscopic band calculation.@xcite the landau levels are broadened due to impurity , phonon and electron - electron scatterings . we model the imaginary part of the retarded green s function , or the density - of - states , of the broadened landau level @xmath120 ( written for `` + ' ' -branch and @xmath104 levels ) , using a gaussian - type form:@xcite @xmath121,\ ] ] with a half - width @xmath122 of the form:@xcite @xmath123^{1/2}$ ] . here @xmath124 is the single - particle lifetime and @xmath125 is the cyclotron frequency of linear - energy - dispersion system with @xmath126 being the zero - temperature fermi level . using a semi - empirical parameter @xmath127 to relate @xmath124 with the transport scattering time @xmath128 , and expressing @xmath129 with the zero - field mobility @xmath5 at finite temperature,@xcite we can write the landau - level broadening as @xmath130^{1/2}.\ ] ] in the present study we consider the case of @xmath120-doping , i.e. the fermi level is high enough above the energy zero of the dirac cone in the range of `` + ' ' -branch levels and the states of `` @xmath100''-branch levels are completely filled , that they are irrelevant to electron transport . special attention has to be paid to the @xmath104 level , since , depending on the direction of exchange potential the effective g - factor of a ti surface state , @xmath30 , can be positive , zero or negative.@xcite the sign and magnitude of the effective g - factor determines how many states of the zero level should be included in or excluded from the available states for electron occupation in the case of @xmath120-doping at a magnetic field . ( i ) if @xmath131 , the @xmath104 level center is exactly at @xmath132 and the system is electron - hole symmetric . the total number of negative energy states ( including the states of the lower half of the @xmath104 level and states of the @xmath100"-branch levels ) and that of positive energy states ( including the states of the upper half of the @xmath104 level and states of the @xmath99"-branch levels ) do not change when changing magnetic field . therefore , the lower - half negative energy states of this level are always filled and the upper - half positive - energy states of it are available for the occupation of particles which are counted as electrons participating in transport in the case of @xmath120-doping . ( ii ) for a finite positive @xmath133 , the @xmath104 level @xmath134 moves downward to negative energy and its distance to the nearest @xmath100"-branch level is @xmath135 closer than to the nearest + " -branch level at finite magnetic field strength @xmath2 . this is equivalent to the opening of an increasingly enlarged ( with increasing @xmath2 ) energy gap between the + " -branch states and the states of the zero - level and the @xmath100"-branch levels . the opening of a sufficient energy gap implies that with increasing magnetic field the states in the + " -branch levels would no longer shrink into the zero - level , and thus the @xmath104 level should be completely excluded from the conduction band , i.e. only particles occupying the + " -branch states are counted as electrons participating in transport in the case of @xmath120-doping , when the magnetic field @xmath2 gets larger than a certain value ( depending on the magnitude of @xmath30 ) . ( iii ) for a finite negative @xmath136 , the @xmath104 level @xmath134 moves upward to positive energy and an increasingly enlarged energy gap will be opened between the states of the zero - level and the + " -branch and the states of @xmath100"-branch levels , and particles occupying the @xmath104 level and + " -branch states are electrons participating in transport when the magnetic field @xmath2 gets larger than a certain value . as a result , the experimentally accessible sheet density @xmath33 of electrons participating in transport is related to the fermi energy @xmath137 by the following equation valid at finite @xmath30 for the magnetic field @xmath2 larger than a certain value : @xmath138 in which @xmath139 + 1\}^{-1}$ ] is the fermi distribution function at temperature @xmath76 and the summation index @xmath120 goes over @xmath140 for @xmath133 , or @xmath141 for @xmath136 . in the case of @xmath131 , @xmath142\ ] ] valid for arbitrary magnetic field , in which @xmath143 . the imaginary part of relative - electron density correlation function in the presence of a magnetic field , @xmath86 , can be expressed in the landau representation as@xcite @xmath144 in which the transform factor @xmath145 ^ 2,\end{aligned}\ ] ] with @xmath146 , @xmath147 , @xmath148 , and @xmath149 being associated laguerre polynomials . the landau - representation correlation function @xmath150 in eq.([piqw ] ) can be constructed with the imaginary part of the retarded green s function @xmath151 , or the density - of - states , of the @xmath120th landau level as@xcite @xmath152\nonumber\\ & \hspace{1.2cm}\times{\rm im}g_n(\epsilon+\omega){\rm im}g_{n'}(\epsilon).\end{aligned}\ ] ] the summation indices @xmath120 and @xmath153 in eq.([piqw ] ) are taken over @xmath140 for @xmath133 , or @xmath154 for @xmath136 . in the case of @xmath131 , eq.([piqw ] ) still works and the summation indices @xmath120 and @xmath153 go over @xmath154 but with @xmath155 replaced by @xmath156 in eq.([p2nn ] ) . numerical calculations are performed for the magnetoresistivity @xmath157 of surface state in a uniform ti bi@xmath0se@xmath1 . at zero temperature the elastic scattering contributing to the resistivity is modeled by a coulomb potential due to charged impurities:@xcite @xmath158 with @xmath159 being the impurity density , which is determined by the zero - magnetic - field mobility @xmath5 . at temperatures higher than @xmath160,@xcite phonon scatterings play increasingly important role and the dominant inelastic contribution comes from optical phonons . for this polar material , the scattering by optical phonons via the deformation potential can be neglected . hence , we take account of inelastic scattering from optical phonons via frhlich coupling : @xmath161 . in the numerical calculation we use the following parameters:@xcite fermi velocity @xmath162 , static dielectric constant @xmath163 , optical dielectric constant @xmath164 , and phonon energy @xmath165 . the broadening parameter is taken to be @xmath166 . as a function of the magnetic field @xmath2 having different effective g - factors : @xmath167 and @xmath168 for a ti surface system with electron sheet density @xmath169 in the cases of zero - magnetic - field mobility @xmath170 ( a ) and @xmath171 ( b ) . several integer - number positions of filling factor @xmath172 are marked in ( b).,scaledwidth=40.0% ] fig.[diffg ] shows the calculated magnetoresistivity @xmath157 versus the magnetic field strength @xmath2 for a ti surface system with electron sheet density @xmath169 but having different effective g - factors : @xmath167 and @xmath168 for two values of zero - magnetic - field mobility @xmath170 and @xmath171 , representing different degree of landau - level broadening . in the case without zeeman splitting ( @xmath131 ) the resistivity @xmath157 exhibits almost no change with changing magnetic field up to 10 t , except the shubnikov - de haas ( sdh ) oscillation showing up in the case of @xmath171 . this kind of magnetoresistance behavior was indeed seen experimentally in the electron - hole symmetrical massless system of single - layer graphene.@xcite in the case of a positive g - factor , @xmath173 , the magnetoresistivity increases linearly with increasing magnetic field ; while for a negative g - factor , @xmath174 , the magnetoresistivity decreases linearly with increasing magnetic field . is shown as a function of the magnetic field @xmath2 for different values of zero - magnetic - field mobility : ( a ) @xmath175 , ( b ) @xmath176 , ( c ) @xmath177 , ( d ) @xmath178 , ( e ) @xmath179 , and ( f ) @xmath180 . the inset of ( a ) illustrates the same for a larger magnetic - field range @xmath181 . the filling factor @xmath182 is plotted versus the magnetic field in ( f ) ; and several integer - number positions of @xmath182 are also marked in ( d ) and ( e ) . here the surface electron density @xmath169 and the lattice temperature @xmath183.,scaledwidth=47.0% ] in the following we will give more detailed examination on the linearly increasing magnetoresistance in the positive @xmath30 case . fig.[rhob ] shows the calculated resistivity @xmath157 versus the magnetic field strength @xmath2 at lattice temperature @xmath183 for system of carrier sheet density @xmath169 and @xmath173 , having different zero - field mobility @xmath184 and @xmath180 . all resistivity curves for mobility @xmath185 exhibit clear linearity in the magnetic - field range and appear no tendency of saturation at the highest field shown in the figure . especially , for the case @xmath170 , the linear behavior extends even up to the magnetic field of @xmath186 , as illustrated in the inset of fig.[rhob](a ) . this feature contradicts the classical mr which saturates at sufficiently large magnetic field @xmath187 . note that here we only present the calculated @xmath157 for magnetic field @xmath2 larger than @xmath188 t , for which a sufficient energy gap @xmath135 is assumed to open that with further increase of the magnetic field the states in the `` + ' ' -branch levels no longer shrink into the zero level and thus it should be excluded from the conduction band . this is of course not true for very weak magnetic field . when @xmath189 the energy gap @xmath190 , the situation becomes similar to the case of @xmath131 : the whole upper half of the zero - level states are available to electron occupation and we should have a flat resistivity @xmath157 when changing magnetic field . with increasing @xmath2 the portion of the zero - level states available to conduction electrons decreases until the magnetic field reaches @xmath191 . as a result the resistivity @xmath157 should exhibit a crossover from a flat changing at small @xmath2 to positively linear increasing at @xmath192 . this is just the behavior observed in the ti bi@xmath0se@xmath1.@xcite note that in the case of @xmath170 , the broadened landau - level widths are always larger than the neighboring level interval : @xmath193 , which requires @xmath194 ^ 2 $ ] , even for the lowest landau level @xmath195 , i.e. the whole landau - level spectrum is smeared . with increasing the zero - field mobility the magnitude of resistivity @xmath157 decreases , and when the broadened landau - level width becomes smaller than the neighboring level interval , @xmath196 , a weak sdh oscillation begin to occur around the linearly - dependent average value of @xmath157 at higher portion of the magnetic field range , as seen in fig.[rhob](c ) , ( d ) and ( e ) for @xmath197 and @xmath198 . on the other hand , in the case of large mobility , e.g. @xmath199 , where the broadened landau - level widths @xmath200 are much smaller than the neighboring level interval even for level index @xmath120 as large as @xmath201 , the magnetoresistivity shows pronounced sdh oscillation and the linear - dependent behavior disappears , before the appearance of quantum hall effect,@xcite as shown in fig.[rhob](f ) . abrikosov s model for the lmr requires the applied magnetic field large enough to reach the quantum limit at which all the carriers are within the lowest landau level,@xcite while it is obvious that more than one landau levels are occupied in the experimental samples in the field range in which the linear and non - saturating magnetoresistivity was observed.@xcite for the given electron surface density @xmath202 , the number of occupied landau levels , or the filling factor @xmath172 , at different magnetic fields is shown in fig.[rhob](f ) , as well as in the fig.[rhob](d ) and ( e ) , where the integer - number positions of @xmath203 , i.e. filling up to entire @xmath182 landau levels , coincide with the minima of the density - of - states or the dips of sdh oscillation . this is in contrast with @xmath131 case , where the integer number of @xmath203 , which implies a filling up to the center position of the @xmath182th landau levels , locates at a peak of sdh oscillation , as shown in fig.[diffg]b . the observed sdh oscillations in the bi@xmath0se@xmath1 nanoribbon exhibiting nonsaturating surface lmr in the experiment@xcite favor the former case : a finite positive effective @xmath133 . is plotted as a function of the surface electron density @xmath33 at magnetic field @xmath204 : ( a ) at different values of zero - field mobility @xmath5 , and ( b ) at different values of zero - field conductivity @xmath205.,scaledwidth=40.0% ] at various lattice temperatures . here the zero - magnetic - field mobility at zero temperature is @xmath206.,scaledwidth=35.0% ] next , we examine the density - dependence of the linear magnetoresistivity . to compare with abrikosov s quantum magnetoresistance which suggests a @xmath207 behavior,@xcite we show the calculated @xmath208 for above lmr versus the carrier sheet density @xmath33 in fig.[rhon ] at fixed magnetic field @xmath209 t . the mobility is taken respectively to be @xmath210 and @xmath211m@xmath212/vs to make the resistivity in the lmr regime . a clearly linear dependence of @xmath213 on the surface density @xmath33 is seen in all cases , indicating that this non - saturating linear resistivity is almost inversely proportional to the carrier density . in the figure we also show @xmath208 versus @xmath33 under the condition of different given conductivity @xmath214 and @xmath215 . in this case the half - width @xmath216 is independent of surface density . the linear dependence still holds , indicating that this linear behavior is not sensitive to the modest @xmath33-dependence of landau level broadening @xmath216 as long as the system is in the overlapped landau level regime . from the above discussion , it is obvious that lmr shows up in the system having overlapped landau levels and the separation of landau levels makes the mr departure from the linear increase . at high temperature , the thermal energy would smear the level separation and phonon scatterings further broaden landau levels . hence , it is believed that this lmr will be robust against raising temperature . this is indeed the case as seen in fig.[rhot ] , where we plot the calculated magnetoresistivity @xmath157 for the above system with zero - temperature linear mobility @xmath217m@xmath212/vs versus the magnetic field at different lattice temperatures . we can see that raising temperature to room temperature has little effect on the linearity of mr . due to the decreased mobility at higher temperature from phonon scattering , the weak sdh oscillation on the linear background tends to vanish . these features are in good agreement with the experimental report.@xcite in summary , we have studied the two - dimensional magnetotransport in the flat surface of a three - dimensional ti , which arises from the surface states with a wavevector - linear energy dispersion and a finite , positive zeeman splitting within the bulk energy gap . when the level broadening is comparable to or larger than the landau - level separation and the conduction electrons spread over many landau levels , a positive , dominantly linear and non - saturating magnetoresistance appears within a quite wide range of magnetic field and persists up to room temperature . this remarkable lmr provides a possible mechanism for the recently observed linear magnetoresistance in topological insulator bi@xmath0se@xmath1 nanoribbons.@xcite in contrast to quantum hall effect which appears in the case of well formed landau levels and to abrikosov s quantum magnetotransport,@xcite which is limited to the extreme quantum limit that all electrons coalesce into the lowest landau level , the discussed lmr is a phenomena of pure classical two - dimensional magnetotransport in a system having linear - energy - dispersion , appearing in the regime of overlapped landau levels , irrespective of its showing up in relatively high magnetic field range . furthermore , the present scheme deals with spatially uniform case without invoking the mobility fluctuation in a strongly inhomogeneous system , which is required in the classical parish and littlewood model to produce a lmr.@xcite the appearance of this significant positive - increasing linear magnetoresistance depends on the existence of a positive and sizable effective g - factor . if the zeeman energy splitting is quite small the resistivity @xmath157 would exhibit little change with changing magnetic field . in the case of a negative and sizable effective g - factor the magnetoresistivity would decrease linearly with increasing magnetic field . therefore , the behavior of the longitudinal resistivity versus magnetic field may provide a useful way for judging the direction and the size of the effective zeeman energy splitting in ti surface states . this work was supported by the national science foundation of china ( grant no . 11104002 ) , the national basic research program of china ( grant no . 2012cb927403 ) and by the program for science&technology innovation talents in universities of henan province ( grant no . 2012hastit029 ) ."""
inputs = tokenizer(
[ARTICLE_LEP, ARTICLE_MAGNET],
max_length=1024,
padding="max_length",
truncation=True,
return_tensors="pt",
)
inputs = {k: inputs[k].to(torch_device) for k in inputs}
hypotheses_batch = model.generate(**inputs)
EXPECTED_LEP = (
"motivated by some recent studies on the light cp - odd higgs boson @xmath0 in non - minimal"
" supersymmetric models, we investigate the rare @xmath1-decays @xmath2 ( @xmath3 ) in the two higgs"
" doublet model ( 2hdm ), the nearly minimal supersymmetric standard model ( nmssm ), the next - to -"
" minimal supersymmetric standard model ( nmssm ) and the minimal supersymmetric standard model ( mssm"
" ).<n> we find that the branching ratios of @xmath4 can reach @xmath5 in 2hdm, @xmath6 in nmssm and"
" @xmath7 in mssm, which are at the level of @xmath8 in 2hdm, @xmath9 in nmssm and @xmath10 in mssm,"
" respectively.<n> these rates can be significantly enhanced in new physics models which lie within the"
" expected sensitivity of the gigaz option of the international linear collider ( ilc ). <n> = # 1,nucl."
" <n> phys. <n> b * # 1"
)
EXPECTED_MAGNET = (
"a positive, nonsaturating and dominantly linear magnetoresistance can appear within quite wide magnetic -"
" field range in the surface state of a topological insulator having a positive and finite effective g -"
" factor. this linear magnetoresistance shows up in the system of high carrier concentration and low"
" mobility when electrons are in extended states and spread over many smeared landau levels, and persists"
" up to room temperature, providing a possible mechanism for the recently observed linear"
" magnetoresistance in topological insulator bi@xmath0se@xmath1 nanoribbons."
)
generated = tokenizer.batch_decode(
hypotheses_batch.tolist(), clean_up_tokenization_spaces=True, skip_special_tokens=True
)
self.assertTrue(generated == [EXPECTED_LEP, EXPECTED_MAGNET])
class BigBirdPegasusStandaloneDecoderModelTester:
def __init__(
self,
parent,
vocab_size=99,
batch_size=7,
d_model=32,
decoder_seq_length=7,
is_training=True,
is_decoder=True,
use_attention_mask=True,
use_cache=False,
use_labels=True,
decoder_start_token_id=2,
decoder_ffn_dim=32,
decoder_layers=4,
encoder_attention_heads=4,
decoder_attention_heads=4,
max_position_embeddings=30,
is_encoder_decoder=False,
pad_token_id=0,
bos_token_id=1,
eos_token_id=2,
scope=None,
attention_type="original_full",
use_bias=True,
block_size=16,
num_random_blocks=3,
):
self.parent = parent
self.batch_size = batch_size
self.decoder_seq_length = decoder_seq_length
# For common tests
self.seq_length = self.decoder_seq_length
self.is_training = is_training
self.use_attention_mask = use_attention_mask
self.use_labels = use_labels
self.vocab_size = vocab_size
self.d_model = d_model
self.hidden_size = d_model
self.num_hidden_layers = decoder_layers
self.decoder_layers = decoder_layers
self.decoder_ffn_dim = decoder_ffn_dim
self.encoder_attention_heads = encoder_attention_heads
self.decoder_attention_heads = decoder_attention_heads
self.num_attention_heads = decoder_attention_heads
self.eos_token_id = eos_token_id
self.bos_token_id = bos_token_id
self.pad_token_id = pad_token_id
self.decoder_start_token_id = decoder_start_token_id
self.use_cache = use_cache
self.max_position_embeddings = max_position_embeddings
self.is_encoder_decoder = is_encoder_decoder
self.scope = None
self.decoder_key_length = decoder_seq_length
self.base_model_out_len = 2
self.decoder_attention_idx = 1
self.attention_type = attention_type
self.use_bias = use_bias
self.block_size = block_size
self.num_random_blocks = num_random_blocks
def prepare_config_and_inputs(self):
input_ids = ids_tensor([self.batch_size, self.decoder_seq_length], self.vocab_size)
attention_mask = None
if self.use_attention_mask:
attention_mask = ids_tensor([self.batch_size, self.decoder_seq_length], vocab_size=2)
lm_labels = None
if self.use_labels:
lm_labels = ids_tensor([self.batch_size, self.decoder_seq_length], self.vocab_size)
config = BigBirdPegasusConfig(
vocab_size=self.vocab_size,
d_model=self.d_model,
decoder_layers=self.decoder_layers,
decoder_ffn_dim=self.decoder_ffn_dim,
encoder_attention_heads=self.encoder_attention_heads,
decoder_attention_heads=self.decoder_attention_heads,
eos_token_id=self.eos_token_id,
bos_token_id=self.bos_token_id,
use_cache=self.use_cache,
pad_token_id=self.pad_token_id,
decoder_start_token_id=self.decoder_start_token_id,
max_position_embeddings=self.max_position_embeddings,
is_encoder_decoder=self.is_encoder_decoder,
attention_type=self.attention_type,
use_bias=self.use_bias,
block_size=self.block_size,
num_random_blocks=self.num_random_blocks,
)
return (
config,
input_ids,
attention_mask,
lm_labels,
)
def create_and_check_decoder_model_past(
self,
config,
input_ids,
attention_mask,
lm_labels,
):
config.use_cache = True
model = BigBirdPegasusDecoder(config=config).to(torch_device).eval()
# first forward pass
outputs = model(input_ids, use_cache=True)
outputs_use_cache_conf = model(input_ids)
outputs_no_past = model(input_ids, use_cache=False)
self.parent.assertTrue(len(outputs) == len(outputs_use_cache_conf))
self.parent.assertTrue(len(outputs) == len(outputs_no_past) + 1)
past_key_values = outputs["past_key_values"]
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)
# append to next input_ids and
next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
output_from_no_past = model(next_input_ids)["last_hidden_state"]
output_from_past = model(next_tokens, past_key_values=past_key_values)["last_hidden_state"]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, next_input_ids.shape[-1] - 1, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
assert torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3)
def create_and_check_decoder_model_attention_mask_past(
self,
config,
input_ids,
attention_mask,
lm_labels,
):
model = BigBirdPegasusDecoder(config=config).to(torch_device).eval()
# create attention mask
attn_mask = torch.ones(input_ids.shape, dtype=torch.long, device=torch_device)
half_seq_length = input_ids.shape[-1] // 2
attn_mask[:, half_seq_length:] = 0
# first forward pass
past_key_values = model(input_ids, attention_mask=attn_mask, use_cache=True)["past_key_values"]
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)
# change a random masked slice from input_ids
random_seq_idx_to_change = ids_tensor((1,), half_seq_length).item() + 1
random_other_next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size).squeeze(-1)
input_ids[:, -random_seq_idx_to_change] = random_other_next_tokens
# append to next input_ids and attn_mask
next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
attn_mask = torch.cat(
[attn_mask, torch.ones((attn_mask.shape[0], 1), dtype=torch.long, device=torch_device)],
dim=1,
)
# get two different outputs
output_from_no_past = model(next_input_ids)["last_hidden_state"]
output_from_past = model(next_tokens, past_key_values=past_key_values)["last_hidden_state"]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, next_input_ids.shape[-1] - 1, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
# big bird has extremely high logits which requires
# such a high error tolerance here
assert torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=5e-1)
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
config, input_ids, attention_mask, lm_labels = config_and_inputs
inputs_dict = {"input_ids": input_ids, "attention_mask": attention_mask}
return config, inputs_dict
@require_torch
class BigBirdPegasusStandaloneDecoderModelTest(ModelTesterMixin, GenerationTesterMixin, unittest.TestCase):
all_model_classes = (BigBirdPegasusDecoder, BigBirdPegasusForCausalLM) if is_torch_available() else ()
all_generative_model_classes = (BigBirdPegasusForCausalLM,) if is_torch_available() else ()
test_pruning = False
is_encoder_decoder = False
def setUp(
self,
):
self.model_tester = BigBirdPegasusStandaloneDecoderModelTester(self, is_training=False)
self.config_tester = ConfigTester(self, config_class=BigBirdPegasusConfig)
def test_config(self):
self.config_tester.run_common_tests()
def test_decoder_model_past(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_decoder_model_past(*config_and_inputs)
def test_decoder_model_attn_mask_past(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_decoder_model_attention_mask_past(*config_and_inputs)
def test_retain_grad_hidden_states_attentions(self):
# decoder cannot keep gradients
return
| # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch BigBirdPegasus model. """
import copy
import tempfile
import unittest
from transformers import BigBirdPegasusConfig, is_torch_available
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
BigBirdPegasusForCausalLM,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForQuestionAnswering,
BigBirdPegasusForSequenceClassification,
BigBirdPegasusModel,
PegasusTokenizer,
)
from transformers.models.bigbird_pegasus.modeling_bigbird_pegasus import (
BigBirdPegasusDecoder,
BigBirdPegasusEncoder,
)
MODEL_ID = "google/bigbird-pegasus-large-pubmed"
def prepare_bigbird_pegasus_inputs_dict(
config,
input_ids,
decoder_input_ids,
attention_mask=None,
decoder_attention_mask=None,
):
if attention_mask is None:
attention_mask = input_ids.ne(config.pad_token_id)
if decoder_attention_mask is None:
decoder_attention_mask = decoder_input_ids.ne(config.pad_token_id)
input_dict = {
"input_ids": input_ids,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"decoder_attention_mask": attention_mask,
}
input_dict = {k: input_dict[k].to(torch_device) for k in input_dict}
return input_dict
class BigBirdPegasusModelTester:
def __init__(
self,
parent,
batch_size=7,
seq_length=256,
is_training=True,
use_labels=False,
vocab_size=99,
hidden_size=32,
num_hidden_layers=2,
num_attention_heads=4,
intermediate_size=31,
hidden_act="gelu_fast",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
max_position_embeddings=260,
eos_token_id=1,
pad_token_id=0,
bos_token_id=2,
attention_type="block_sparse",
use_bias=False,
block_size=16,
num_random_blocks=3,
scale_embedding=True,
):
self.parent = parent
self.batch_size = batch_size
self.seq_length = seq_length
self.is_training = is_training
self.use_labels = use_labels
self.vocab_size = vocab_size
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_act = hidden_act
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.max_position_embeddings = max_position_embeddings
self.eos_token_id = eos_token_id
self.pad_token_id = pad_token_id
self.bos_token_id = bos_token_id
self.attention_type = attention_type
self.use_bias = use_bias
self.block_size = block_size
self.num_random_blocks = num_random_blocks
self.scale_embedding = scale_embedding
def prepare_config_and_inputs(self):
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size).clamp(
3,
)
input_ids[:, -1] = self.eos_token_id # Eos Token
decoder_input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
config = self.get_config()
inputs_dict = prepare_bigbird_pegasus_inputs_dict(config, input_ids, decoder_input_ids)
return config, inputs_dict
def get_config(self):
return BigBirdPegasusConfig(
vocab_size=self.vocab_size,
d_model=self.hidden_size,
encoder_layers=self.num_hidden_layers,
decoder_layers=self.num_hidden_layers,
encoder_attention_heads=self.num_attention_heads,
decoder_attention_heads=self.num_attention_heads,
encoder_ffn_dim=self.intermediate_size,
decoder_ffn_dim=self.intermediate_size,
dropout=self.hidden_dropout_prob,
attention_dropout=self.attention_probs_dropout_prob,
max_position_embeddings=self.max_position_embeddings,
eos_token_id=self.eos_token_id,
bos_token_id=self.bos_token_id,
pad_token_id=self.pad_token_id,
attention_type=self.attention_type,
use_bias=self.use_bias,
block_size=self.block_size,
num_random_blocks=self.num_random_blocks,
scale_embedding=self.scale_embedding,
)
def prepare_config_and_inputs_for_common(self):
config, inputs_dict = self.prepare_config_and_inputs()
return config, inputs_dict
def create_and_check_decoder_model_past_large_inputs(self, config, inputs_dict):
model = BigBirdPegasusModel(config=config).get_decoder().to(torch_device).eval()
input_ids = inputs_dict["input_ids"]
attention_mask = inputs_dict["attention_mask"]
# first forward pass
outputs = model(input_ids, attention_mask=attention_mask, use_cache=True)
output, past_key_values = outputs.to_tuple()
# create hypothetical multiple next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
next_attn_mask = ids_tensor((self.batch_size, 3), 2)
# append to next input_ids and
next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
next_attention_mask = torch.cat([attention_mask, next_attn_mask], dim=-1)
output_from_no_past = model(next_input_ids, attention_mask=next_attention_mask)["last_hidden_state"]
output_from_past = model(next_tokens, attention_mask=next_attention_mask, past_key_values=past_key_values)[
"last_hidden_state"
]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, :, random_slice_idx].detach()
self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1])
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-2))
def check_encoder_decoder_model_standalone(self, config, inputs_dict):
model = BigBirdPegasusModel(config=config).to(torch_device).eval()
outputs = model(**inputs_dict)
encoder_last_hidden_state = outputs.encoder_last_hidden_state
last_hidden_state = outputs.last_hidden_state
with tempfile.TemporaryDirectory() as tmpdirname:
encoder = model.get_encoder()
encoder.save_pretrained(tmpdirname)
encoder = BigBirdPegasusEncoder.from_pretrained(tmpdirname).to(torch_device)
encoder_last_hidden_state_2 = encoder(inputs_dict["input_ids"], attention_mask=inputs_dict["attention_mask"])[
0
]
self.parent.assertTrue((encoder_last_hidden_state_2 - encoder_last_hidden_state).abs().max().item() < 1e-3)
with tempfile.TemporaryDirectory() as tmpdirname:
decoder = model.get_decoder()
decoder.save_pretrained(tmpdirname)
decoder = BigBirdPegasusDecoder.from_pretrained(tmpdirname).to(torch_device)
last_hidden_state_2 = decoder(
input_ids=inputs_dict["decoder_input_ids"],
attention_mask=inputs_dict["decoder_attention_mask"],
encoder_hidden_states=encoder_last_hidden_state,
encoder_attention_mask=inputs_dict["attention_mask"],
)[0]
self.parent.assertTrue((last_hidden_state_2 - last_hidden_state).abs().max().item() < 1e-3)
def create_and_check_model(self, config, inputs_dict):
model = BigBirdPegasusModel(config=config).to(torch_device).eval()
input_ids = inputs_dict["input_ids"]
decoder_input_ids = inputs_dict["decoder_input_ids"]
result = model(input_ids, decoder_input_ids=decoder_input_ids, use_cache=True)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
@require_torch
class BigBirdPegasusModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (
(
BigBirdPegasusModel,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForSequenceClassification,
BigBirdPegasusForQuestionAnswering,
)
if is_torch_available()
else ()
)
all_generative_model_classes = (BigBirdPegasusForConditionalGeneration,) if is_torch_available() else ()
pipeline_model_mapping = (
{
"conversational": BigBirdPegasusForConditionalGeneration,
"feature-extraction": BigBirdPegasusModel,
"question-answering": BigBirdPegasusForQuestionAnswering,
"summarization": BigBirdPegasusForConditionalGeneration,
"text2text-generation": BigBirdPegasusForConditionalGeneration,
"text-classification": BigBirdPegasusForSequenceClassification,
"text-generation": BigBirdPegasusForCausalLM,
"zero-shot": BigBirdPegasusForSequenceClassification,
}
if is_torch_available()
else {}
)
is_encoder_decoder = True
test_missing_keys = False
test_pruning = False
test_head_masking = False
# torchscript tests are not passing for now.
# Also torchscript is not an important feature to have in the beginning.
test_torchscript = False
# overwrite from GenerationTesterMixin to solve problem
# with conflicting random seeds
def _get_input_ids_and_config(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
config.attention_type = "original_full"
input_ids = inputs_dict[self.input_name]
attention_mask = torch.ones_like(input_ids, dtype=torch.long)
# cut to half length & take max batch_size 3
max_batch_size = 2
sequence_length = input_ids.shape[-1] // 2
input_ids = input_ids[:max_batch_size, :sequence_length]
attention_mask = attention_mask[:max_batch_size, :sequence_length]
# generate max 3 tokens
max_length = input_ids.shape[-1] + 3
if config.eos_token_id is not None and config.pad_token_id is None:
# hack to allow generate for models such as GPT2 as is done in `generate()`
config.pad_token_id = config.eos_token_id
return config, input_ids, attention_mask, max_length
def setUp(self):
self.model_tester = BigBirdPegasusModelTester(self)
self.config_tester = ConfigTester(self, config_class=BigBirdPegasusConfig)
def test_config(self):
self.config_tester.run_common_tests()
def test_save_load_strict(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs()
for model_class in self.all_model_classes:
model = model_class(config)
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(tmpdirname)
model2, info = model_class.from_pretrained(tmpdirname, output_loading_info=True)
self.assertEqual(info["missing_keys"], [])
def test_decoder_model_past_with_large_inputs(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_decoder_model_past_large_inputs(*config_and_inputs)
def test_encoder_decoder_model_standalone(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs_for_common()
self.model_tester.check_encoder_decoder_model_standalone(*config_and_inputs)
def test_model_various_attn_type(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
for type in ["original_full", "block_sparse"]:
config_and_inputs[0].attention_type = type
self.model_tester.create_and_check_model(*config_and_inputs)
def test_generate_without_input_ids(self):
if self.model_tester.attention_type == "block_sparse":
# this test can never pass for BigBird-block-sparse attention since input_ids must be multiple of block_size
return
super().test_generate_without_input_ids()
def test_retain_grad_hidden_states_attentions(self):
if self.model_tester.attention_type == "block_sparse":
# this test can't pass since attention matrix (which is getting returned) can't have gradients (& just 0 at many locations)
return
super().test_retain_grad_hidden_states_attentions()
# BigBirdPegasusForSequenceClassification does not support inputs_embeds
def test_inputs_embeds(self):
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in (
BigBirdPegasusModel,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForQuestionAnswering,
):
model = model_class(config)
model.to(torch_device)
model.eval()
inputs = copy.deepcopy(self._prepare_for_class(inputs_dict, model_class))
if not self.is_encoder_decoder:
input_ids = inputs["input_ids"]
del inputs["input_ids"]
else:
encoder_input_ids = inputs["input_ids"]
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
del inputs["input_ids"]
inputs.pop("decoder_input_ids", None)
wte = model.get_input_embeddings()
if not self.is_encoder_decoder:
inputs["inputs_embeds"] = wte(input_ids)
else:
inputs["inputs_embeds"] = wte(encoder_input_ids)
inputs["decoder_inputs_embeds"] = wte(decoder_input_ids)
with torch.no_grad():
model(**inputs)[0]
def test_generate_fp16(self):
config, input_dict = self.model_tester.prepare_config_and_inputs()
input_dict.pop("decoder_attention_mask")
input_dict.pop("decoder_input_ids")
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
if torch_device == "cuda":
model.half()
model.generate(**input_dict)
model.generate(**input_dict, do_sample=True, early_stopping=False, num_return_sequences=3)
@slow
def test_batched_forward_original_full(self):
self._check_batched_forward(attn_type="original_full")
@slow
def test_batched_forward_block_sparse(self):
self._check_batched_forward(attn_type="block_sparse", tolerance=1e-1)
def _check_batched_forward(self, attn_type, tolerance=1e-3):
config, _ = self.model_tester.prepare_config_and_inputs()
config.max_position_embeddings = 128
config.block_size = 16
config.attention_type = attn_type
model = BigBirdPegasusForConditionalGeneration(config).to(torch_device)
model.eval()
chunk_length = 32
sample_with_padding = [3, 8, 11] * chunk_length + [0] * chunk_length
sample_without_padding = [4, 7, 9, 13] * chunk_length
target_ids_without_padding = [2, 3] * 8
target_ids_with_padding = [7, 8] * 6 + 4 * [-100]
attention_mask = torch.tensor(
[[1] * 3 * chunk_length + [0] * chunk_length, [1] * 4 * chunk_length],
device=torch_device,
dtype=torch.long,
)
input_ids = torch.tensor([sample_with_padding, sample_without_padding], device=torch_device, dtype=torch.long)
labels = torch.tensor(
[target_ids_without_padding, target_ids_with_padding], device=torch_device, dtype=torch.long
)
with torch.no_grad():
logits_batched = model(input_ids=input_ids, attention_mask=attention_mask, labels=labels).logits
with torch.no_grad():
logits_single_first = model(input_ids=input_ids[:1, :-chunk_length], labels=labels[:1]).logits
self.assertTrue(torch.allclose(logits_batched[0, -3:], logits_single_first[0, -3:], atol=tolerance))
with torch.no_grad():
logits_single_second = model(input_ids=input_ids[1:], labels=labels[1:, :-4]).logits
self.assertTrue(torch.allclose(logits_batched[1, :3], logits_single_second[0, :3], atol=tolerance))
def test_auto_padding(self):
ids = [[7, 6, 9] * 65]
config, _ = self.model_tester.prepare_config_and_inputs()
input_ids = torch.tensor(ids, device=torch_device, dtype=torch.long)
attention_mask = input_ids.new_ones(input_ids.shape)
decoder_input_ids = torch.tensor([[33, 5, 8] * 3], device=torch_device, dtype=torch.long)
config.block_size = 8
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
output1 = model(input_ids=input_ids, attention_mask=attention_mask, decoder_input_ids=decoder_input_ids)[
"logits"
]
ids = [[7, 6, 9] * 65 + [0] * 5]
input_ids = torch.tensor(ids, device=torch_device, dtype=torch.long)
attention_mask = torch.tensor([[1] * 3 * 65 + [0] * 5], device=torch_device, dtype=torch.long)
output2 = model(input_ids=input_ids, attention_mask=attention_mask, decoder_input_ids=decoder_input_ids)[
"logits"
]
self.assertTrue(torch.allclose(output1, output2, atol=1e-5))
def test_for_change_to_full_attn(self):
self.model_tester.seq_length = 9
config, input_dict = self.model_tester.prepare_config_and_inputs()
# automatic switch will happen
config.attention_type = "block_sparse"
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
state_dict = model.state_dict()
outputs1 = model(**input_dict)["logits"]
config.attention_type = "original_full"
model = BigBirdPegasusForConditionalGeneration(config).eval().to(torch_device)
model.load_state_dict(state_dict)
outputs2 = model(**input_dict)["logits"]
self.assertTrue(torch.allclose(outputs1, outputs2, atol=1e-5))
@require_torch
@require_sentencepiece
@require_tokenizers
@slow
class BigBirdPegasusModelIntegrationTests(unittest.TestCase):
def _get_dummy_input_ids(self):
# fmt: off
ids = torch.tensor(
[[685, 560, 630, 193, 836, 764, 708, 360, 10, 724, 278, 755, 805, 600, 71, 473, 601, 397, 315, 706, 487, 552, 88, 175, 601, 850, 678, 538, 846, 73, 778, 917, 116, 977, 756, 710, 1023, 848, 432, 449, 851, 100, 985, 178, 756, 798, 660, 148, 911, 424, 289, 962, 266, 698, 640, 545, 544, 715, 245, 152, 676, 511, 460, 883, 184, 29, 803, 129, 129, 933, 54, 902, 551, 489, 757, 274, 336, 389, 618, 43, 443, 544, 889, 258, 322, 1000, 938, 58, 292, 871, 120, 780, 431, 83, 92, 897, 399, 612, 566, 909, 634, 939, 85, 204, 325, 775, 965, 48, 640, 1013, 132, 973, 869, 181, 1001, 847, 144, 661, 228, 955, 792, 720, 910, 374, 854, 561, 306, 582, 170, 676, 449, 96, 198, 607, 257, 882, 691, 293, 931, 817, 862, 388, 611, 555, 974, 369, 1000, 918, 202, 384, 513, 907, 371, 556, 955, 384, 24, 700, 131, 378, 99, 575, 932, 735, 124, 964, 595, 943, 740, 149, 210, 563, 412, 783, 42, 59, 706, 37, 779, 87, 44, 873, 12, 771, 308, 81, 33, 183, 129, 807, 276, 175, 555, 372, 185, 445, 489, 590, 287, 281, 638, 771, 516, 95, 227, 876, 270, 881, 297, 329, 20, 608, 841, 411, 451, 249, 181, 324, 1005, 830, 783, 865, 261, 964, 750, 140, 1021, 599, 462, 890, 622, 844, 697, 529, 153, 926, 150, 111, 26, 465, 957, 890, 887, 118, 446, 596, 674, 873, 929, 229, 508, 764, 122, 327, 470, 288, 526, 840, 697, 153, 592, 42, 275, 553, 439, 208, 780, 167, 112, 350, 1018, 130, 736, 887, 813, 217, 382, 25, 68, 979, 1008, 772, 235, 717, 999, 292, 727, 1023, 702, 710, 728, 556, 33, 12, 617, 213, 139, 695, 1004, 422, 638, 669, 624, 489, 771, 540, 980, 218, 664, 822, 308, 175, 149, 950, 542, 580, 548, 808, 394, 74, 298, 920, 900, 815, 731, 947, 877, 772, 800, 778, 395, 540, 430, 200, 424, 62, 342, 866, 45, 803, 931, 89, 34, 646, 233, 768, 37, 769, 460, 291, 198, 895, 950, 255, 81, 447, 137, 190, 130, 210, 369, 292, 377, 348, 169, 885, 805, 177, 538, 324, 872, 509, 804, 115, 799, 30, 754, 290, 147, 274, 222, 341, 510, 515, 70, 358, 909, 557, 886, 766, 323, 624, 92, 342, 424, 552, 972, 663, 415, 658, 711, 968, 275, 861, 44, 84, 434, 810, 94, 175, 406, 202, 858, 499, 481, 988, 330, 541, 1004, 210, 618, 955, 897, 983, 576, 17, 107, 165, 607, 537, 629, 192, 196, 308, 137, 953, 860, 94, 892, 751, 88, 161, 148, 585, 456, 88, 14, 315, 594, 121, 885, 952, 833, 716, 733, 933, 282, 801, 427, 783, 471, 285, 277, 979, 325, 535, 228, 891, 596, 648, 969, 574, 654, 518, 257, 137, 208, 464, 950, 140, 5, 424, 349, 942, 283, 587, 821, 1007, 434, 220, 820, 740, 874, 787, 374, 291, 564, 671, 438, 827, 940, 824, 509, 1021, 787, 942, 856, 450, 327, 491, 54, 817, 95, 60, 337, 667, 637, 164, 571, 946, 107, 202, 301, 782, 890, 839, 551, 680, 649, 14, 1017, 904, 721, 1017, 535, 505, 848, 986, 777, 740, 775, 210, 456, 469, 474, 963, 573, 401, 57, 883, 750, 664, 281, 5, 613, 1005, 306, 344, 543, 567, 154, 789, 354, 358, 698, 408, 412, 30, 930, 372, 822, 632, 948, 855, 503, 8, 618, 1010, 138, 695, 897, 852, 377, 933, 722, 149, 886, 1009, 260, 127, 811, 578, 533, 805, 325, 977, 113, 944, 651, 238, 361, 991, 860, 556, 64, 928, 917, 455, 266, 445, 604, 624, 420, 340, 845, 275, 370, 843, 227, 226, 940, 644, 909, 229, 827, 898, 370, 129, 808, 25, 699, 293, 356, 838, 135, 4, 227, 890, 681, 445, 418, 285, 837, 27, 737, 249, 366, 948, 202, 438, 198, 930, 648, 638, 607, 73, 247, 853, 136, 708, 214, 476, 621, 324, 103, 853, 328, 596, 224, 257, 646, 348, 108, 927, 970, 980, 520, 150, 998, 477, 393, 684, 559, 1, 361, 692, 551, 90, 75, 500, 739, 636, 344, 97, 852, 283, 719, 33, 116, 455, 866, 429, 828, 826, 691, 174, 746, 133, 442, 94, 348, 402, 420, 707, 405, 942, 186, 976, 376, 677, 874, 703, 517, 498, 499, 206, 415, 366, 856, 739, 420, 586, 219, 952, 539, 375, 23, 461, 720, 355, 603, 52, 999, 815, 721, 574, 445, 816, 1019, 105, 641, 395, 972, 910, 328, 607, 519, 686, 246, 415, 528, 170, 167, 310, 940, 595, 392, 221, 834, 682, 835, 115, 861, 335, 742, 220, 247, 101, 416, 222, 179, 509, 175, 606, 627, 674, 781, 737, 746, 849, 67, 457, 1012, 126, 139, 625, 731, 156, 697, 121, 322, 449, 710, 857, 291, 976, 4, 701, 239, 678, 172, 724, 857, 583, 661, 903, 797, 628, 903, 835, 605, 989, 615, 870, 380, 710, 110, 330, 101, 695, 846, 918, 508, 672, 594, 36, 238, 244, 251, 393, 767, 282, 22, 430, 230, 983, 401, 154, 1007, 120, 678, 896, 386, 390, 711, 397, 347, 587, 1020, 951, 79, 831, 585, 200, 814, 134, 560, 700, 171, 452, 139, 755, 314, 476, 346, 388, 126, 719, 851, 198, 699, 901, 18, 710, 448, 351, 665, 644, 326, 425, 165, 571, 178, 440, 665, 674, 915, 866, 463, 754, 136, 950, 748, 47, 497, 1013, 640, 930, 338, 158, 525, 631, 815, 887, 289, 803, 116, 600, 637, 410, 175, 499, 876, 565, 1002, 623, 577, 333, 887, 586, 147, 773, 776, 644, 49, 77, 294, 117, 494, 561, 110, 979, 180, 562, 72, 859, 434, 1007, 286, 516, 75, 597, 491, 322, 888, 533, 209, 43, 499, 29, 411, 856, 181, 305, 963, 615, 778, 259, 373, 877, 746, 858, 381, 886, 613, 91, 69, 618, 523, 13, 617, 226, 422, 168, 929, 379, 290, 923, 100, 218, 307, 345, 211, 789, 735, 669, 585, 275, 410, 921, 552, 235, 636, 285, 665, 659, 708, 173, 724, 302, 823, 1, 139, 708, 903, 732, 868, 442, 967, 916, 163, 51, 243, 871]], # noqa: E231
dtype=torch.long,
device=torch_device,
)
# fmt: on
return ids
def _get_dummy_target_ids(self):
# fmt: off
ids = torch.tensor(
[[13, 6, 1, 4, 12, 4, 8, 10, 4, 6, 3, 5, 8, 7, 9, 9]], # noqa: E231
dtype=torch.long,
device=torch_device,
)
# fmt: on
return ids
def test_inference_block_sparse(self):
model = BigBirdPegasusForConditionalGeneration.from_pretrained(
MODEL_ID, attention_type="block_sparse", block_size=16, num_random_blocks=3
)
model.to(torch_device)
input_ids = self._get_dummy_input_ids()
target_ids = self._get_dummy_target_ids()
outputs = model(input_ids, labels=target_ids)
prediction_logits = outputs.logits
self.assertEqual(prediction_logits.shape, torch.Size((1, 16, 96103)))
# fmt: off
expected_prediction_logits_slice = torch.tensor(
[[1.7769, 5.8479, 6.2375, 2.2745, 8.6157, 4.7483, 5.0647, 6.5358, 2.3393, 7.8333, 3.8403, 0.0255, 7.219, 5.2759, 3.097, 6.387, 4.9341, 7.1409, 5.1179, 0.1144, 6.8268, 0.7598, 0.6258, 2.373, 0.4627, -1.9919, 1.8422, 3.4578], [1.8026, 5.9604, 5.954, 2.8642, 9.0608, 4.394, 5.3779, 7.0216, 1.543, 7.8744, 4.4231, -0.0398, 7.6091, 5.6611, 3.3536, 6.8624, 4.7699, 6.5241, 4.8893, 0.5791, 6.8368, 0.1034, 0.0338, 2.9393, 0.5034, -2.5509, 2.0172, 3.2858], [1.8426, 5.9151, 5.5374, 3.0426, 9.1762, 3.6287, 5.3916, 7.4621, 1.2582, 7.9244, 4.694, -0.1308, 7.4725, 5.5385, 3.4598, 7.0422, 4.2455, 5.797, 4.5927, 0.7478, 6.7467, -0.2695, -0.3207, 3.0269, 0.4714, -2.8134, 2.0406, 3.1089], [1.6527, 5.8416, 5.4558, 3.0044, 9.3478, 3.2607, 5.3887, 7.52, 0.9362, 7.8877, 4.8465, -0.1705, 7.3932, 5.6352, 3.5744, 7.2623, 4.0485, 5.2788, 4.5859, 0.8325, 6.6088, -0.3676, -0.6287, 3.1731, 0.4483, -3.1573, 2.0522, 2.8868]], # noqa: E231
device=torch_device,
)
# fmt: on
self.assertTrue(
torch.allclose(prediction_logits[0, 4:8, 128:156], expected_prediction_logits_slice, atol=1e-4)
)
def test_inference_full_attn(self):
model = BigBirdPegasusForConditionalGeneration.from_pretrained(MODEL_ID, attention_type="original_full")
model.to(torch_device)
input_ids = self._get_dummy_input_ids()
target_ids = self._get_dummy_target_ids()
outputs = model(input_ids, labels=target_ids)
prediction_logits = outputs.logits
self.assertEqual(prediction_logits.shape, torch.Size((1, 16, 96103)))
# fmt: off
expected_prediction_logits_slice = torch.tensor(
[[1.3418, 5.8304, 6.5662, 2.0448, 8.7702, 4.6579, 4.9947, 6.429, 2.4296, 7.9431, 4.217, 0.0672, 7.334, 5.1966, 2.9603, 6.0814, 4.6756, 7.5522, 5.076, 0.213, 6.6638, 0.6577, 0.244, 2.1221, 0.7531, -2.4076, 1.8731, 3.5594], [1.5525, 6.0524, 6.309, 2.6245, 9.229, 4.5213, 5.0913, 7.0622, 1.7992, 8.0962, 4.7994, -0.0248, 7.7168, 5.5878, 3.0883, 6.5248, 4.7895, 6.9974, 4.8787, 0.5445, 6.6686, 0.0102, -0.1659, 2.6195, 0.7389, -2.8956, 1.9928, 3.3777], [1.6407, 6.2104, 6.0331, 2.8076, 9.4074, 3.9772, 5.0574, 7.5316, 1.4201, 8.3035, 5.0212, -0.1031, 7.553, 5.5023, 3.1427, 6.7674, 4.4409, 6.457, 4.525, 0.728, 6.5422, -0.6234, -0.4726, 2.7486, 0.6985, -3.0804, 1.9669, 3.2365], [1.5065, 6.1271, 5.8296, 2.8405, 9.5649, 3.6834, 5.1214, 7.546, 0.9758, 8.3335, 5.1952, -0.1395, 7.4348, 5.6893, 3.2942, 7.0356, 4.1665, 5.9695, 4.3898, 0.8931, 6.3988, -0.8957, -0.7522, 2.8924, 0.6498, -3.4358, 1.8654, 2.9735]], # noqa: E231
device=torch_device,
)
# fmt: on
self.assertTrue(
torch.allclose(prediction_logits[0, 4:8, 128:156], expected_prediction_logits_slice, atol=1e-4)
)
def test_seq_to_seq_generation(self):
MODEL_ID = "google/bigbird-pegasus-large-arxiv"
model = BigBirdPegasusForConditionalGeneration.from_pretrained(MODEL_ID).to(torch_device)
tokenizer = PegasusTokenizer.from_pretrained(MODEL_ID)
ARTICLE_LEP = r"""the lep experiments at the resonance of @xmath1-boson have tested the standard model ( sm ) at quantum level , measuring the @xmath1-decay into fermion pairs with an accuracy of one part in ten thousands . the good agreement of the lep data with the sm predictions have severely constrained the behavior of new physics at the @xmath1-pole . taking these achievements into account one can imagine that the physics of @xmath1-boson will again play the central role in the frontier of particle physics if the next generation @xmath1 factory comes true with the generated @xmath1 events several orders of magnitude higher than that of the lep . this factory can be realized in the gigaz option of the international linear collider ( ilc)@xcite . the ilc is a proposed electron - positron collider with tunable energy ranging from @xmath12 to @xmath13 and polarized beams in its first phase , and the gigaz option corresponds to its operation on top of the resonance of @xmath1 boson by adding a bypass to its main beam line . given the high luminosity , @xmath14 , and the cross section at the resonance of @xmath1 boson , @xmath15 , about @xmath16 @xmath1 events can be generated in an operational year of @xmath17 of gigaz , which implies that the expected sensitivity to the branching ratio of @xmath1-decay can be improved from @xmath18 at the lep to @xmath19 at the gigaz@xcite . in light of this , the @xmath1-boson properties , especially its exotic or rare decays which are widely believed to be sensitive to new physics , should be investigated comprehensively to evaluate their potential in probing new physics . among the rare @xmath1-decays , the flavor changing ( fc ) processes were most extensively studied to explore the flavor texture in new physics @xcite , and it was found that , although these processes are severely suppressed in the sm , their branching ratios in new physics models can be greatly enhanced to @xmath19 for lepton flavor violation decays @xcite and @xmath20 for quark flavor violation decays @xcite . besides the fc processes , the @xmath1-decay into light higgs boson(s ) is another type of rare process that was widely studied , e.g. the decay @xmath21 ( @xmath22 ) with the particle @xmath0 denoting a light higgs boson was studied in @xcite , the decay @xmath23 was studied in the two higgs doublet model ( 2hdm)@xcite and the minimal supersymmetric standard model ( mssm)@xcite , and the decay @xmath4 was studied in a model independent way @xcite , in 2hdm@xcite and also in mssm@xcite . these studies indicate that , in contrast with the kinematic forbidden of these decays in the sm , the rates of these decays can be as large as @xmath18 in new physics models , which lie within the expected sensitivity of the gigaz . in this work , we extend the previous studies of these decays to some new models and investigate these decays altogether . we are motivated by some recent studies on the singlet extension of the mssm , such as the next - to - minimal supersymmetric standard model ( nmssm ) @xcite and the nearly minimal supersymmetric standard model ( nmssm ) @xcite , where a light cp - odd higgs boson @xmath0 with singlet - dominant component may naturally arise from the spontaneous breaking of some approximate global symmetry like @xmath24 or peccei - quuin symmetry @xcite . these non - minimal supersymmetric models can not only avoid the @xmath25-problem , but also alleviate the little hierarchy by having such a light higgs boson @xmath0 @xcite . we are also motivated by that , with the latest experiments , the properties of the light higgs boson are more stringently constrained than before . so it is worth updating the previous studies . so far there is no model - independent lower bound on the lightest higgs boson mass . in the sm , it must be heavier than @xmath26 gev , obtained from the null observation of the higgs boson at lep experiments . however , due to the more complex structure of the higgs sector in the extensions of the sm , this lower bound can be significantly relaxed according to recent studies , e.g. , for the cp - odd higgs boson @xmath0 we have @xmath27 gev in the nmssm @xcite , @xmath28 gev in the nmssm @xcite , and @xmath29 gev in the lepton - specific 2hdm ( l2hdm ) @xcite . with such a light cp - odd higgs boson , the z - decay into one or more @xmath0 is open up . noting that the decay @xmath30 is forbidden due to bose symmetry , we in this work study the rare @xmath1-decays @xmath6 ( @xmath22 ) , @xmath31 and @xmath4 in a comparative way for four models , namely the type - ii 2hdm@xcite , the l2hdm @xcite , the nmssm and the nmssm . in our study , we examine carefully the constraints on the light @xmath0 from many latest experimental results . this work is organized as follows . in sec . ii we briefly describe the four new physics models . in sec . iii we present the calculations of the rare @xmath1-decays . in sec . iv we list the constraints on the four new physics models . in sec . v we show the numerical results for the branching ratios of the rare @xmath1-decays in various models . finally , the conclusion is given in sec . as the most economical way , the sm utilizes one higgs doublet to break the electroweak symmetry . as a result , the sm predicts only one physical higgs boson with its properties totally determined by two free parameters . in new physics models , the higgs sector is usually extended by adding higgs doublets and/or singlets , and consequently , more physical higgs bosons are predicted along with more free parameters involved in . the general 2hdm contains two @xmath32 doublet higgs fields @xmath33 and @xmath34 , and with the assumption of cp - conserving , its scalar potential can be parameterized as@xcite : @xmath35,\end{aligned}\ ] ] where @xmath36 ( @xmath37 ) are free dimensionless parameters , and @xmath38 ( @xmath39 ) are the parameters with mass dimension . after the electroweak symmetry breaking , the spectrum of this higgs sector includes three massless goldstone modes , which become the longitudinal modes of @xmath40 and @xmath1 bosons , and five massive physical states : two cp - even higgs bosons @xmath41 and @xmath42 , one neutral cp - odd higgs particle @xmath0 and a pair of charged higgs bosons @xmath43 . noting the constraint @xmath44 with @xmath45 and @xmath46 denoting the vacuum expectation values ( vev ) of @xmath33 and @xmath34 respectively , we choose @xmath47 as the input parameters with @xmath48 , and @xmath49 being the mixing angle that diagonalizes the mass matrix of the cp - even higgs fields . the difference between the type - ii 2hdm and the l2hdm comes from the yukawa coupling of the higgs bosons to quark / lepton . in the type - ii 2hdm , one higgs doublet @xmath34 generates the masses of up - type quarks and the other doublet @xmath33 generates the masses of down - type quarks and charged leptons ; while in the l2hdm one higgs doublet @xmath33 couples only to leptons and the other doublet @xmath34 couples only to quarks . so the yukawa interactions of @xmath0 to fermions in these two models are given by @xcite @xmath50 with @xmath51 denoting generation index . obviously , in the type - ii 2hdm the @xmath52 coupling and the @xmath53 coupling can be simultaneously enhanced by @xmath54 , while in the l2hdm only the @xmath53 coupling is enhanced by @xmath55 . the structures of the nmssm and the nmssm are described by their superpotentials and corresponding soft - breaking terms , which are given by @xcite @xmath56 where @xmath57 is the superpotential of the mssm without the @xmath25 term , @xmath58 and @xmath59 are higgs doublet and singlet superfields with @xmath60 and @xmath61 being their scalar component respectively , @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 and @xmath67 are soft breaking parameters , and @xmath68 and @xmath69 are coefficients of the higgs self interactions . with the superpotentials and the soft - breaking terms , one can get the higgs potentials of the nmssm and the nmssm respectively . like the 2hdm , the higgs bosons with same cp property will mix and the mass eigenstates are obtained by diagonalizing the corresponding mass matrices : @xmath70 where the fields on the right hands of the equations are component fields of @xmath71 , @xmath72 and @xmath61 defined by @xmath73 @xmath74 and @xmath75 are respectively the cp - even and cp - odd neutral higgs bosons , @xmath76 and @xmath77 are goldstone bosons eaten by @xmath1 and @xmath78 , and @xmath79 is the charged higgs boson . so both the nmssm and nmssm predict three cp - even higgs bosons , two cp - odd higgs bosons and one pair of charged higgs bosons . in general , the lighter cp - odd higgs @xmath0 in these model is the mixture of the singlet field @xmath80 and the doublet field combination , @xmath81 , i.e. @xmath82 and its couplings to down - type quarks are then proportional to @xmath83 . so for singlet dominated @xmath0 , @xmath84 is small and the couplings are suppressed . as a comparison , the interactions of @xmath0 with the squarks are given by@xcite @xmath85 i.e. the interaction does not vanish when @xmath86 approaches zero . just like the 2hdm where we use the vevs of the higgs fields as fundamental parameters , we choose @xmath68 , @xmath69 , @xmath87 , @xmath88 , @xmath66 and @xmath89 as input parameters for the nmssm@xcite and @xmath68 , @xmath54 , @xmath88 , @xmath65 , @xmath90 and @xmath91 as input parameters for the nmssm@xcite . about the nmssm and the nmssm , three points should be noted . the first is for the two models , there is no explicit @xmath92term , and the effective @xmath25 parameter ( @xmath93 ) is generated when the scalar component of @xmath59 develops a vev . the second is , the nmssm is actually same as the nmssm with @xmath94@xcite , because the tadpole terms @xmath95 and its soft breaking term @xmath96 in the nmssm do not induce any interactions , except for the tree - level higgs boson masses and the minimization conditions . and the last is despite of the similarities , the nmssm has its own peculiarity , which comes from its neutralino sector . in the basis @xmath97 , its neutralino mass matrix is given by @xcite @xmath98 where @xmath99 and @xmath100 are @xmath101 and @xmath102 gaugino masses respectively , @xmath103 , @xmath104 , @xmath105 and @xmath106 . after diagonalizing this matrix one can get the mass eigenstate of the lightest neutralino @xmath107 with mass taking the following form @xcite @xmath108 this expression implies that @xmath107 must be lighter than about @xmath109 gev for @xmath110 ( from lower bound on chargnio mass ) and @xmath111 ( perturbativity bound ) . like the other supersymmetric models , @xmath107 as the lightest sparticle acts as the dark matter in the universe , but due to its singlino - dominated nature , it is difficult to annihilate sufficiently to get the correct density in the current universe . so the relic density of @xmath107 plays a crucial way in selecting the model parameters . for example , as shown in @xcite , for @xmath112 , there is no way to get the correct relic density , and for the other cases , @xmath107 mainly annihilates by exchanging @xmath1 boson for @xmath113 , or by exchanging a light cp - odd higgs boson @xmath0 with mass satisfying the relation @xmath114 for @xmath115 . for the annihilation , @xmath54 and @xmath25 are required to be less than 10 and @xmath116 respectively because through eq.([mass - exp ] ) a large @xmath87 or @xmath25 will suppress @xmath117 to make the annihilation more difficult . the properties of the lightest cp - odd higgs boson @xmath0 , such as its mass and couplings , are also limited tightly since @xmath0 plays an important role in @xmath107 annihilation . the phenomenology of the nmssm is also rather special , and this was discussed in detail in @xcite . in the type - ii 2hdm , l2hdm , nmssm and nmssm , the rare @xmath1-decays @xmath118 ( @xmath22 ) , @xmath3 and @xmath4 may proceed by the feynman diagrams shown in fig.[fig1 ] , fig.[fig2 ] and fig.[fig3 ] respectively . for these diagrams , the intermediate state @xmath119 represents all possible cp - even higgs bosons in the corresponding model , i.e. @xmath41 and @xmath42 in type - ii 2hdm and l2hdm and @xmath41 , @xmath42 and @xmath120 in nmssm and nmssm . in order to take into account the possible resonance effects of @xmath119 in fig.[fig1](c ) for @xmath2 and fig.[fig3 ] ( a ) for @xmath11 , we have calculated all the decay modes of @xmath119 and properly included the width effect in its propagator . as to the decay @xmath121 , two points should be noted . one is , unlike the decays @xmath6 and @xmath11 , this process proceeds only through loops mediated by quarks / leptons in the type - ii 2hdm and l2hdm , and additionally by sparticles in the nmssm and nmssm . so in most cases its rate should be much smaller than the other two . the other is due to cp - invariance , loops mediated by squarks / sleptons give no contribution to the decay@xcite . in actual calculation , this is reflected by the fact that the coupling coefficient of @xmath122 differs from that of @xmath123 by a minus sign ( see eq.([asqsq ] ) ) , and as a result , the squark - mediated contributions to @xmath121 are completely canceled out . with regard to the rare decay @xmath11 , we have more explanations . in the lowest order , this decay proceeds by the diagram shown in fig.[fig3 ] ( a ) , and hence one may think that , as a rough estimate , it is enough to only consider the contributions from fig.[fig3](a ) . however , we note that in some cases of the type - ii 2hdm and l2hdm , due to the cancelation of the contributions from different @xmath119 in fig.[fig3 ] ( a ) and also due to the potentially largeness of @xmath124 couplings ( i.e. larger than the electroweak scale @xmath125 ) , the radiative correction from the higgs - mediated loops may dominate over the tree level contribution even when the tree level prediction of the rate , @xmath126 , exceeds @xmath20 . on the other hand , we find the contribution from quark / lepton - mediated loops can be safely neglected if @xmath127 in the type - ii 2hdm and the l2hdm . in the nmssm and the nmssm , besides the corrections from the higgs- and quark / lepton - mediated loops , loops involving sparticles such as squarks , charginos and neutralinos can also contribute to the decay . we numerically checked that the contributions from squarks and charginos can be safely neglected if @xmath127 . we also calculated part of potentially large neutralino correction ( note that there are totally about @xmath128 diagrams for such correction ! ) and found they can be neglected too . since considering all the radiative corrections will make our numerical calculation rather slow , we only include the most important correction , namely that from higgs - mediated loops , in presenting our results for the four models . one can intuitively understand the relative smallness of the sparticle contribution to @xmath11 as follows . first consider the squark contribution which is induced by the @xmath129 interaction ( @xmath130 denotes the squark in chirality state ) and the @xmath131 interaction through box diagrams . because the @xmath132 interaction conserves the chirality of the squarks while the @xmath133 interaction violates the chirality , to get non - zero contribution to @xmath11 from the squark loops , at least four chiral flippings are needed , with three of them provided by @xmath131 interaction and the rest provided by the left - right squark mixing . this means that , if one calculates the amplitude in the chirality basis with the mass insertion method , the amplitude is suppressed by the mixing factor @xmath134 with @xmath135 being the off diagonal element in squark mass matrix . next consider the chargino / neutralino contributions . since for a light @xmath0 , its doublet component , parameterized by @xmath84 in eq.([mixing ] ) , is usually small , the couplings of @xmath0 with the sparticles will never be tremendously large@xcite . so the chargino / neutralino contributions are not important too . in our calculation of the decays , we work in the mass eigenstates of sparticles instead of in the chirality basis . for the type - ii 2hdm and the l2hdm , we consider the following constraints @xcite : * theoretical constraints on @xmath136 from perturbativity , unitarity and requirements that the scalar potential is finit at large field values and contains no flat directions @xcite , which imply that @xmath137 * the constraints from the lep search for neutral higgs bosons . we compute the signals from the higgs - strahlung production @xmath138 ( @xmath139 ) with @xmath140 @xcite and from the associated production @xmath141 with @xmath142 @xcite , and compare them with the corresponding lep data which have been inputted into our code . we also consider the constraints from @xmath138 by looking for a peak of @xmath143 recoil mass distribution of @xmath1-boson @xcite and the constraint of @xmath144 mev when @xmath145 @xcite . + these constraints limit the quantities such as @xmath146 \times br ( h_i \to \bar{b } b ) $ ] on the @xmath147 plane with the the subscript @xmath148 denoting the coupling coefficient of the @xmath149 interaction . they also impose a model - dependent lower bound on @xmath150 , e.g. , @xmath151 for the type - ii 2hdm ( from our scan results ) , @xmath152 for the l2hdm@xcite , and @xmath153 for the nmssm @xcite . these bounds are significantly lower than that of the sm , i.e. @xmath154 , partially because in new physics models , unconventional decay modes of @xmath155 such as @xmath156 are open up . as to the nmssm , another specific reason for allowing a significantly lighter cp - even higgs boson is that the boson may be singlet - dominated in this model . + with regard to the lightest cp - odd higgs boson @xmath0 , we checked that there is no lower bound on its mass so long as the @xmath157 interaction is weak or @xmath155 is sufficiently heavy . * the constraints from the lep search for a light higgs boson via the yukawa process @xmath158 with @xmath22 and @xmath61 denoting a scalar @xcite . these constraints can limit the @xmath159 coupling versus @xmath160 in new physics models . * the constraints from the cleo - iii limit on @xmath161 and the latest babar limits on @xmath162 . these constraints will put very tight constraints on the @xmath163 coupling for @xmath164 . in our analysis , we use the results of fig.8 in the second paper of @xcite to excluded the unfavored points . * the constraints from @xmath165 couplings . since the higgs sector can give sizable higher order corrections to @xmath165 couplings , we calculate them to one loop level and require the corrected @xmath165 couplings to lie within the @xmath166 range of their fitted value . the sm predictions for the couplings at @xmath1-pole are given by @xmath167 and @xmath168 @xcite , and the fitted values are given by @xmath169 and @xmath170 , respectively@xcite . we adopt the formula in @xcite to the 2hdm in our calculation . * the constraints from @xmath171 leptonic decay . we require the new physics correction to the branching ratio @xmath172 to be in the range of @xmath173 @xcite . we use the formula in @xcite in our calculation . + about the constraints ( 5 ) and ( 6 ) , two points should be noted . one is all higgs bosons are involved in the constraints by entering the self energy of @xmath171 lepton , the @xmath174 vertex correction or the @xmath175 vertex correction , and also the box diagrams for @xmath176@xcite . since the yukawa couplings of the higgs bosons to @xmath171 lepton get enhanced by @xmath54 and so do the corrections , @xmath54 must be upper bounded for given spectrum of the higgs sector . generally speaking , the lighter @xmath0 is , the more tightly @xmath54 is limited@xcite . the other point is in the type - ii 2hdm , @xmath177 , b - physics observables as well as @xmath178 decays discussed above can constraint the model in a tighter way than the constraints ( 5 ) and ( 6 ) since the yukawa couplings of @xmath171 lepton and @xmath179 quark are simultaneously enhanced by @xmath54 . but for the l2hdm , because only the yukawa couplings of @xmath171 lepton get enhanced ( see eq.[yukawa ] ) , the constraints ( 5 ) and ( 6 ) are more important in limiting @xmath54 . * indirect constraints from the precision electroweak observables such as @xmath180 , @xmath181 and @xmath182 , or their combinations @xmath183 @xcite . we require @xmath184 to be compatible with the lep / sld data at @xmath185 confidence level@xcite . we also require new physics prediction of @xmath186 is within the @xmath187 range of its experimental value . the latest results for @xmath188 are @xmath189 ( measured value ) and @xmath190 ( sm prediction ) for @xmath191 gev @xcite . in our code , we adopt the formula for these observables presented in @xcite to the type - ii 2hdm and the l2hdm respectively . + in calculating @xmath180 , @xmath181 and @xmath182 , we note that these observables get dominant contributions from the self energies of the gauge bosons @xmath1 , @xmath192 and @xmath193 . since there is no @xmath194 coupling or @xmath195 coupling , @xmath0 must be associated with the other higgs bosons to contribute to the self energies . so by the uv convergence of these quantities , one can infer that , for the case of a light @xmath0 and @xmath196 , these quantities depend on the spectrum of the higgs sector in a way like @xmath197 at leading order , which implies that a light @xmath0 can still survive the constraints from the precision electroweak observables given the splitting between @xmath150 and @xmath198 is moderate@xcite . * the constraints from b physics observables such as the branching ratios for @xmath199 , @xmath200 and @xmath201 , and the mass differences @xmath202 and @xmath203 . we require their theoretical predications to agree with the corresponding experimental values at @xmath187 level . + in the type - ii 2hdm and the l2hdm , only the charged higgs boson contributes to these observables by loops , so one can expect that @xmath198 versus @xmath54 is to be limited . combined analysis of the limits in the type - ii 2hdm has been done by the ckmfitter group , and the lower bound of @xmath204 as a function of @xmath87 was given in fig.11 of @xcite . this analysis indicates that @xmath198 must be heavier than @xmath205 at @xmath185 c.l . regardless the value of @xmath54 . in this work , we use the results of fig.11 in @xcite to exclude the unfavored points . as for the l2hdm , b physics actually can not put any constraints@xcite because in this model the couplings of the charged higgs boson to quarks are proportional to @xmath206 and in the case of large @xmath54 which we are interested in , they are suppressed . in our analysis of the l2hdm , we impose the lep bound on @xmath198 , i.e. @xmath207@xcite . * the constraints from the muon anomalous magnetic moment @xmath208 . now both the theoretical prediction and the experimental measured value of @xmath208 have reached a remarkable precision , but a significant deviation still exists : @xmath209 @xcite . in the 2hdm , @xmath208 gets additional contributions from the one - loop diagrams induced by the higgs bosons and also from the two - loop barr - zee diagrams mediated by @xmath0 and @xmath155@xcite . if the higgs bosons are much heavier than @xmath25 lepton mass , the contributions from the barr - zee diagrams are more important , and to efficiently alleviate the discrepancy of @xmath208 , one needs a light @xmath0 along with its enhanced couplings to @xmath25 lepton and also to heavy fermions such as bottom quark and @xmath171 lepton to push up the effects of the barr - zee diagram@xcite . the cp - even higgs bosons are usually preferred to be heavy since their contributions to @xmath208 are negative . + in the type - ii 2hdm , because @xmath54 is tightly constrained by the process @xmath210 at the lep@xcite and the @xmath178 decay@xcite , the barr - zee diagram contribution is insufficient to enhance @xmath208 to @xmath187 range around its measured value@xcite . so in our analysis , we require the type - ii 2hdm to explain @xmath208 at @xmath211 level . while for the l2hdm , @xmath54 is less constrained compared with the type - ii 2hdm , and the barr - zee diagram involving the @xmath171-loop is capable to push up greatly the theoretical prediction of @xmath208@xcite . therefore , we require the l2hdm to explain the discrepancy at @xmath187 level . + unlike the other constraints discussed above , the @xmath208 constraint will put a two - sided bound on @xmath54 since on the one hand , it needs a large @xmath54 to enhance the barr - zee contribution , but on the other hand , too large @xmath54 will result in an unacceptable large @xmath208 . * since this paper concentrates on a light @xmath0 , the decay @xmath212 is open up with a possible large decay width . we require the width of any higgs boson to be smaller than its mass to avoid a too fat higgs boson@xcite . we checked that for the scenario characterized by @xmath213 , the coefficient of @xmath214 interaction is usually larger than the electroweak scale @xmath125 , and consequently a large decay width is resulted . for the nmssm and nmssm , the above constraints become more complicated because in these models , not only more higgs bosons are involved in , but also sparticles enter the constraints . so it is not easy to understand some of the constraints intuitively . take the process @xmath199 as an example . in the supersymmetric models , besides the charged higgs contribution , chargino loops , gluino loops as well as neutralino loops also contribute to the process@xcite , and depending on the susy parameters , any of these contributions may become dominated over or be canceled by other contributions . as a result , although the charged higgs affects the process in the same way as that in the type - ii 2hdm , charged higgs as light as @xmath215 is still allowed even for @xmath216@xcite . since among the constraints , @xmath208 is rather peculiar in that it needs new physics to explain the discrepancy between @xmath217 and @xmath218 , we discuss more about its dependence on susy parameters . in the nmssm and the nmssm , @xmath208 receives contributions from higgs loops and neutralino / chargino loops . for the higgs contribution , it is quite similar to that of the type - ii 2hdm except that more higgs bosons are involved in@xcite . for the neutralino / chargino contribution , in the light bino limit ( i.e. @xmath219 ) , it can be approximated by@xcite @xmath220 for @xmath221 with @xmath222 being smuon mass . so combining the two contributions together , one can learn that a light @xmath0 along with large @xmath54 and/or light smuon with moderate @xmath87 are favored to dilute the discrepancy . because more parameters are involved in the constraints on the supersymmetric models , we consider following additional constraints to further limit their parameters : * direct bounds on sparticle masses from the lep1 , the lep2 and the tevatron experiments @xcite . * the lep1 bound on invisible z decay @xmath223 ; the lep2 bound on neutralino production @xmath224 and @xmath225@xcite . * dark matter constraints from the wmap relic density 0.0975 @xmath226 0.1213 @xcite . note that among the above constraints , the constraint ( 2 ) on higgs sector and the constraint ( c ) on neutralino sector are very important . this is because in the supersymmetric models , the sm - like higgs is upper bounded by about @xmath227 at tree level and by about @xmath228 at loop level , and that the relic density restricts the lsp annihilation cross section in a certain narrow range . in our analysis of the nmssm , we calculate the constraints ( 3 ) and ( 5 - 7 ) by ourselves and utilize the code nmssmtools @xcite to implement the rest constraints . we also extend nmssmtools to the nmssm to implement the constraints . for the extension , the most difficult thing we faced is how to adapt the code micromegas@xcite to the nmssm case . we solve this problem by noting the following facts : * as we mentioned before , the nmssm is actually same as the nmssm with the trilinear singlet term setting to zero . so we can utilize the model file of the nmssm as the input of the micromegas and set @xmath229 . * since in the nmssm , the lsp is too light to annihilate into higgs pairs , there is no need to reconstruct the effective higgs potential to calculate precisely the annihilation channel @xmath230 with @xmath61 denoting any of higgs bosons@xcite . we thank the authors of the nmssmtools for helpful discussion on this issue when we finish such extension@xcite . with the above constraints , we perform four independent random scans over the parameter space of the type - ii 2hdm , the l2hdm , the nmssm and the nmssm respectively . we vary the parameters in following ranges : @xmath231 for the type - ii 2hdm , @xmath232 for the l2hdm , @xmath233 for the nmssm , and @xmath234 for the nmssm . in performing the scans , we note that for the nmssm and the nmssm , some constraints also rely on the gaugino masses and the soft breaking parameters in the squark sector and the slepton sector . since these parameters affect little on the properties of @xmath0 , we fix them to reduce the number of free parameters in our scan . for the squark sector , we adopt the @xmath235 scenario which assumes that the soft mass parameters for the third generation squarks are degenerate : @xmath236 800 gev , and that the trilinear couplings of the third generation squarks are also degenerate , @xmath237 with @xmath238 . for the slepton sector , we assume all the soft - breaking masses and trilinear parameters to be 100 gev . this setting is necessary for the nmssm since this model is difficult to explain the muon anomalous moment at @xmath239 level for heavy sleptons@xcite . finally , we assume the grand unification relation @xmath240 for the gaugino masses with @xmath241 being fine structure constants of the different gauge group . with large number of random points in the scans , we finally get about @xmath242 , @xmath243 , @xmath244 and @xmath242 samples for the type - ii 2hdm , the l2hdm , the nmssm and the nmssm respectively which survive the constraints and satisfy @xmath245 . analyzing the properties of the @xmath0 indicates that for most of the surviving points in the nmssm and the nmssm , its dominant component is the singlet field ( numerically speaking , @xmath246 ) so that its couplings to the sm fermions are suppressed@xcite . our analysis also indicates that the main decay products of @xmath0 are @xmath247 for the l2hdm@xcite , @xmath248 ( dominant ) and @xmath247 ( subdominant ) for the type - ii 2hdm , the nmssm and the nmssm , and in some rare cases , neutralino pairs in the nmssm@xcite . in fig.[fig4 ] , we project the surviving samples on the @xmath249 plane . this figure shows that the allowed range of @xmath54 is from @xmath250 to @xmath251 in the type - ii 2hdm , and from @xmath252 to @xmath253 in the l2hdm . just as we introduced before , the lower bounds of @xmath254 come from the fact that we require the models to explain the muon anomalous moment , while the upper bound is due to we have imposed the constraint from the lep process @xmath255 , which have limited the upper reach of the @xmath256 coupling for light @xmath61 @xcite(for the dependence of @xmath256 coupling on @xmath54 , see sec . this figure also indicates that for the nmssm and the nmssm , @xmath54 is upper bounded by @xmath257 . for the nmssm , this is because large @xmath87 can suppress the dark matter mass to make its annihilation difficult ( see @xcite and also sec . ii ) , but for the nmssm , this is because we choose a light slepton mass so that large @xmath54 can enhance @xmath208 too significantly to be experimentally unacceptable . we checked that for the slepton mass as heavy as @xmath258 , @xmath259 is still allowed for the nmssm . in fig.[fig5 ] and fig.[fig6 ] , we show the branching ratios of @xmath260 and @xmath261 respectively . fig.[fig5 ] indicates , among the four models , the type - ii 2hdm predicts the largest ratio for @xmath260 with its value varying from @xmath262 to @xmath263 . the underlying reason is in the type - ii 2hdm , the @xmath264 coupling is enhanced by @xmath54 ( see fig.[fig4 ] ) , while in the other three model , the coupling is suppressed either by @xmath265 or by the singlet component of the @xmath0 . fig.[fig6 ] shows that the l2hdm predicts the largest rate for @xmath266 with its value reaching @xmath5 in optimum case , and for the other three models , the ratio of @xmath261 is at least about one order smaller than that of @xmath267 . this feature can be easily understood from the @xmath268 coupling introduced in sect . we emphasize that , if the nature prefers a light @xmath0 , @xmath260 and/or @xmath269 in the type - ii 2hdm and the l2hdm will be observable at the gigaz . then by the rates of the two decays , one can determine whether the type - ii 2hdm or the l2hdm is the right theory . on the other hand , if both decays are observed with small rates or fail to be observed , the singlet extensions of the mssm are favored . in fig.[fig7 ] , we show the rate of @xmath3 as the function of @xmath270 . this figure indicates that the branching ratio of @xmath121 can reach @xmath271 , @xmath272 , @xmath273 and @xmath274 for the optimal cases of the type - ii 2hdm , the l2hdm , the nmssm and the nmssm respectively , which implies that the decay @xmath121 will never be observable at the gigaz if the studied model is chosen by nature . the reason for the smallness is , as we pointed out before , that the decay @xmath121 proceeds only at loop level . comparing the optimum cases of the type - ii 2hdm , the nmssm and the nmssm shown in fig.5 - 7 , one may find that the relation @xmath275 holds for any of the decays . this is because the decays are all induced by the yukawa couplings with similar structure for the models . in the supersymmetric models , the large singlet component of the light @xmath0 is to suppress the yukawa couplings , and the @xmath0 in the nmssm has more singlet component than that in the nmssm . next we consider the decay @xmath11 , which , unlike the above decays , depends on the higgs self interactions . in fig.[fig8 ] we plot its rate as a function of @xmath270 and this figure indicates that the @xmath276 may be the largest among the ratios of the exotic @xmath1 decays , reaching @xmath277 in the optimum cases of the type - ii 2hdm , the l2hdm and the nmssm . the underlying reason is , in some cases , the intermediate state @xmath119 in fig.[fig3 ] ( a ) may be on - shell . in fact , we find this is one of the main differences between the nmssm and the nmssm , that is , in the nmssm , @xmath119 in fig.[fig3 ] ( a ) may be on - shell ( corresponds to the points with large @xmath278 ) while in the nmssm , this seems impossible . so we conclude that the decay @xmath11 may serve as an alternative channel to test new physics models , especially it may be used to distinguish the nmssm from the nmssm if the supersymmetry is found at the lhc and the @xmath11 is observed at the gigaz with large rate . before we end our discussion , we note that in the nmssm , the higgs boson @xmath0 may be lighter than @xmath279 without conflicting with low energy data from @xmath178 decays and the other observables ( see fig.[fig4]-[fig8 ] ) . in this case , @xmath0 is axion - like as pointed out in @xcite . we checked that , among the rare @xmath1 decays discussed in this paper , the largest branching ratio comes from @xmath280 which can reach @xmath281 . since in this case , the decay product of @xmath0 is highly collinear muon pair , detecting the decay @xmath280 may need some knowledge about detectors , which is beyond our discussion . in this paper , we studied the rare @xmath1-decays @xmath2 ( @xmath7 ) , @xmath282 and @xmath4 in the type - ii 2hdm , lepton - specific 2hdm , nmssm and nmssm , which predict a light cp - odd higgs boson @xmath0 . in the parameter space allowed by current experiments , the branching ratio can be as large as @xmath5 for @xmath118 , @xmath8 for @xmath3 and @xmath9 for @xmath4 , which implies that the decays @xmath2 and @xmath283 may be accessible at the gigaz option . since different models predict different size of branching ratios , these decays can be used to distinguish different model through the measurement of these rare decays . this work was supported in part by hastit under grant no . 2009hastit004 , by the national natural science foundation of china ( nnsfc ) under grant nos . 10821504 , 10725526 , 10635030 , 10775039 , 11075045 and by the project of knowledge innovation program ( pkip ) of chinese academy of sciences under grant no . . for some reviews , see , e.g. , m. a. perez , g. tavares - velasco and j. j. toscano , int . j. mod . a * 19 * , 159 ( 2004 ) ; j. m. yang , arxiv:1006.2594 . j. i. illana , m. masip , 67 , 035004 ( 2003 ) ; j. cao , z. xiong , j. m. yang , 32 , 245 ( 2004 ) . d. atwood _ et al_. , 66 , 093005 ( 2002 ) . j. kalinowski , and s. pokorski , 219 , 116 ( 1989 ) ; a. djouadi , p. m. zerwas and j. zunft , 259 , 175 ( 1991 ) ; a. djouadi , j. kalinowski , and p. m. zerwas , z. phys . c * 54 * , 255 ( 1992 ) . m. krawczyk , _ et al . _ , 19 , 463 ( 2001 ) ; 8 , 495 ( 1999 ) . j. f. gunion , g. gamberini and s. f. novaes , 38 , 3481 ( 1988 ) ; thomas j. weiler and tzu - chiang yuan , 318 , 337 ( 1989 ) ; a. djouadi , _ et al . _ , 1 , 163 ( 1998)[hep - ph/9701342 ] . d. chang and w. y. keung , phys . lett . * 77 * , 3732 ( 1996 ) . e. keith and e. ma , 57 , 2017 ( 1998 ) ; m. a. perez , g. tavares - velasco and j. j. toscano , int . j. mod.phys . a * 19 * , 159 ( 2004 ) . f. larios , g. tavares - velasco and c. p. yuan , 64 , 055004 ( 2001 ) ; 66 , 075006 ( 2002 ) . a. djouadi , _ et al . _ , 10 , 27 ( 1999 ) [ hep - ph/9903229 ] . for a detailed introduction of the nmssm , see f. franke and h. fraas , int . j. mod . a * 12 * ( 1997 ) 479 ; for a recent review of the nmssm , see for example , u. ellwanger , c. hugonie , and a. m. teixeira , arxiv : 0910.1785 . see , e.g. , j. r. ellis , j. f. gunion , h. e. haber , l. roszkowski and f. zwirner , phys . rev . d * 39 * ( 1989 ) 844 ; m. drees , int . j. mod . phys . a * 4 * ( 1989 ) 3635 ; u. ellwanger , m. rausch de traubenberg and c. a. savoy , phys . b * 315 * ( 1993 ) 331 ; nucl . b * 492 * ( 1997 ) 21 ; d.j . miller , r. nevzorov , p.m. zerwas , 681 , 3 ( 2004 ) . c. panagiotakopoulos , k. tamvakis , 446 , 224 ( 1999 ) ; 469 , 145 ( 1999 ) ; c. panagiotakopoulos , a. pilaftsis , 63 , 055003 ( 2001 ) ; a. dedes , _ et al . _ , 63 , 055009 ( 2001 ) ; a. menon , _ et al . _ , 70 , 035005 ( 2004 ) ; v. barger , _ et al . _ , 630 , 85 ( 2005 ) . c. balazs , _ et al . _ , 0706 , 066 ( 2007 ) . b. a. dobrescu , k. t. matchev , 0009 , 031 ( 2000 ) ; a. arhrib , k. cheung , t. j. hou , k. w. song , hep - ph/0611211 ; 0703 , 073 ( 2007 ) ; x. g. he , j. tandean , and g. valencia , 98 , 081802 ( 2007 ) ; 0806 , 002 ( 2008 ) ; f. domingo _ et al_. , 0901 , 061 ( 2009 ) ; gudrun hiller , 70 , 034018 ( 2004 ) ; r. dermisek , and john f. gunion , 75 , 075019 ( 2007 ) ; 79 , 055014 ( 2009 ) ; 81 , 055001 ( 2010 ) ; r. dermisek , john f. gunion , and b. mcelrath , 76 , 051105 ( 2007 ) ; z. heng , _ et al_. , 77 , 095012 ( 2008 ) ; a. belyaev _ et al_. , 81 , 075021 ( 2010 ) ; d. das and u. ellwanger , arxiv:1007.1151 [ hep - ph ] . s. andreas , o. lebedev , s. ramos - sanchez and a. ringwald , arxiv:1005.3978 [ hep - ph ] . j. f. gunion , jhep * 0908 * , 032 ( 2009 ) ; r. dermisek and j. f. gunion , phys . rev . d * 81 * , 075003 ( 2010 ) . r. dermisek and j. f. gunion , phys . lett . * 95 * , 041801 ( 2005 ) ; phys . d * 73 * , 111701 ( 2006 ) . j. cao , h. e. logan , j. m. yang , 79 , 091701 ( 2009 ) . j. cao , p. wan , l. wu , j. m. yang , 80 , 071701 ( 2009 ) . j. f. gunion and h. e. haber , 67 , 075019 ( 2003 ) . r. m. barnett , _ et al . _ , phys . b * 136 * , 191 ( 1984 ) ; r. m. barnett , g. senjanovic and d. wyler , phys . d * 30 * , 1529 ( 1984 ) ; y. grossman , nucl . b * 426 * , 355 ( 1994 ) . h. s. goh , l. j. hall and p. kumar , jhep * 0905 * , 097 ( 2009 ) ; a. g. akeroyd and w. j. stirling , nucl . b * 447 * , 3 ( 1995 ) ; a. g. akeroyd , phys . b * 377 * , 95 ( 1996 ) ; h. e. logan and d. maclennan , phys . rev . d * 79 * , 115022 ( 2009 ) ; m. aoki , _ et al . _ , arxiv:0902.4665 [ hep - ph ] . v. barger , p. langacker , h. s. lee and g. shaughnessy , phys . d * 73 * , 115010 ( 2006 ) . s. hesselbach , _ et . _ , arxiv:0810.0511v2 [ hep - ph ] . de vivie and p. janot [ aleph collaboration ] , pa13 - 027 contribution to the international conference on high energy physics , warsaw , poland , 2531 july 1996 ; j. kurowska , o. grajek and p. zalewski [ delphi collaboration ] , cern - open-99 - 385 . [ aleph collaboration and delphi collaboration and l3 collaboration ] , phys . rept . * 427 * , 257 ( 2006 ) . j. cao and j. m. yang , jhep * 0812 * , 006 ( 2008 ) . m. krawczyk and d. temes , eur . j. c * 44 * , 435 ( 2005 ) . g. altarelli and r. barbieri , 253 , 161 ( 1991 ) ; m. e. peskin , t. takeuchi , 46 , 381 ( 1992 ) . c. amsler , _ et al . _ , ( particle data group ) , 667 , 1 ( 2008 ) . o. deschamps , s. descotes - genon , s. monteil , v. niess , s. tjampens and v. tisserand , arxiv:0907.5135 [ hep - ph ] . s. su and b. thomas , phys . d * 79 * , 095014 ( 2009 ) . g. abbiendi , _ et al . _ , eur . phys . j. c * 32 * , 453 ( 2004 ) . m. davier , _ et al . _ , 66 , 1 ( 2010 ) . k. cheung , _ et al . _ , phys . d * 64 * , 111301 ( 2001 ) . k. cheung and o. c. w. kong , phys . d * 68 * , 053003 ( 2003 ) . t. besmer , c. greub , t.hurth , 609 , 359 ( 2001 ) ; f. borzumati , _ et al . _ , 62 , 075005(2000 ) . j. cao , k. i. hikasa , w. wang , j. m. yang and l. x. yu , phys . d * 82 * , 051701 ( 2010 ) [ arxiv:1006.4811 [ hep - ph ] ] . j. f. gunion , _ et . d * 73 * , 015011 ( 2006 ) . martin and j. d. wells , phys . d * 64 * , 035003 ( 2001 ) . j. abdallah _ et al . _ , eur . j. c * 31 * , 421 ( 2004 ) ; g. abbiendi _ et al . _ , eur . j. c * 35 * , 1 ( 2004 ) . j. dunkley _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 180 * , 306 ( 2009 ) [ arxiv:0803.0586 [ astro - ph ] ] . u. ellwanger _ et al . _ , 02 , 066 ( 2005 ) . g. belanger , f. boudjema , a. pukhov and a. semenov , comput . commun . * 174 * , 577 ( 2006 ) ; comput . phys . commun . * 176 * , 367 ( 2007 ) . g. belanger , f. boudjema , c. hugonie , a. pukhov and a. semenov , jcap * 0509 * , 001 ( 2005 ) ."""
ARTICLE_MAGNET = r"""it is well known that the classical magnetoresistance ( mr ) in metals or semiconductors with a closed free electron fermi surface increases quadratically with increasing magnetic field @xmath2 for @xmath3 and saturates when @xmath4 . here @xmath5 is the zero - magnetic - field mobility . hence , the extraordinarily high and linear mr ( lmr ) , which breaks this familiar rule , has been gaining much attention as soon as its discovery . in the past decade , this unexpected lmr has been reported in silver chalcogenide,@xcite indium antimonide,@xcite silicon,@xcite mnas - gaas composite material,@xcite and graphene.@xcite kapitza s linear law@xcite indicates that the metal shows a magnetoresistance linear in perpendicular magnetic field when it has an open fermi surface and a mean free path longer than the electronic larmor radius . recently , another two models , irrespective of the open fermi surface , have been constructed to provide possible mechanisms for the lmr phenomenon . abrikosov suggested a quantum - limit origin of lmr for the homogenous system with a gapless linear energy spectrum.@xcite his model requires that landau levels are well formed and the carrier concentration is small that all electrons occupy only the lowest landau band . alternatively , parish and littlewood developed a classical model without involving linear spectrum.@xcite ignoring the concrete microscopic mechanism , they attributed this unusual mr to the mobility fluctuations in a strongly inhomogenous system . topological insulators@xcite ( tis ) are novel materials with a full energy gap in bulk , while there are gapless surface states . due to its unique band structure with only one helical dirac cone and linear energy dispersion,@xcite the surface states of the ti bi@xmath0se@xmath1 become an excellent platform for the study of quantum - limit lmr . the recent experiment in this flat surface system , however , reported that a large positive mr , which becomes very linear above a characteristic field of @xmath6@xmath7@xmath8 t , was observed even in an opposite situation where the carrier sheet density is high that electrons occupy more than one landau levels.@xcite moreover , they found that raising temperature to room temperature almost has no influence on the observed lmr . it is striking that this observation is in conflict with abrikosov s model and also with the classical parish - littlewood model . so far a reliable theoretical scheme capable of explaining this novel experiment has still been lacking . in this paper , we generalize the balance - equation approach@xcite to a system modeling the surface states of a three - dimensional ti to investigate the two - dimensional magnetotransport in it . we find that a positive , nonsaturating and dominantly linear magnetoresistance can appear within quite wide magnetic - field range in the ti surface state having a positive and finite effective g - factor . this linear magnetoresistance shows up in the system of high carrier concentration and low mobility when electrons are in extended states and spread over many smeared landau levels , and persists up to room temperature , providing a possible mechanism for the recently observed linear magnetoresistance in topological insulator bi@xmath0se@xmath1 nanoribbons.@xcite we consider the surface state of a bi@xmath0se@xmath1-type large bulk gap ti in the @xmath9-@xmath10 plane under the influence of a uniform magnetic field @xmath11 applied along the @xmath12 direction.@xcite following the experimental observation,@xcite we assume that the fermi energy locates in the gap of the bulk band and above the dirac point , i.e. the surface carriers are electrons . further , the separations of the fermi energy from the bottom of bulk band and dirac point are much larger than the highest temperature ( @xmath13 ) considered in this work . hence , the contribution from the bulk band to the magnetotransport is negligible . these electrons , scattered by randomly distributed impurities and by phonons , are driven by a uniform in - plane electric field @xmath14 in the topological surface . the hamiltonian of this many - electron and phonon system consists of an electron part @xmath15 , a phonon part @xmath16 , and electron - impurity and electron - phonon interactions @xmath17 and @xmath18 : @xmath19 here , the electron hamiltonian is taken in the form @xmath20 , \ ] ] in which @xmath21 , @xmath22 , @xmath23 and @xmath24 , stand , respectively , for the canonical momentum , coordinate , momentum and spin operators of the @xmath25th electron having charge @xmath26 , @xmath27 is the vector potential of the perpendicular magnetic field @xmath28 in the landau gauge , @xmath29 is the fermi velocity , @xmath30 is the effective g - factor of the surface electron , and @xmath31 is the bohr magneton with @xmath32 the free electron mass . the sum index @xmath25 in eq.([helectron ] ) goes over all electrons of total number @xmath33 in the surface state of unit area . in the frame work of balance equation approach,@xcite the two - dimensional center - of - mass ( c.m . ) momentum and coordinate @xmath34 and @xmath35 , and the relative - electron momenta and coordinates @xmath36 and @xmath37 are introduced to write the hamiltonian @xmath15 into the sum of a single - particle c.m . part @xmath38 and a many - particle relative - electron part @xmath39 : @xmath40 , with @xmath41.\end{aligned}\ ] ] in this , @xmath42 is the canonical momentum of the center - of - mass and @xmath43 is the canonical momentum for the @xmath25th relative electron . here we have also introduced c.m . spin operators @xmath44 and @xmath45 . the commutation relations between the c.m . spin operators @xmath46 and @xmath47 and the spin operators @xmath48 , @xmath49 and @xmath50 of the @xmath25th electron are of order of @xmath51 : @xmath52= n^{-1}2\,{\rm i}\,\varepsi lon_{\beta_1\beta_2\beta_3}\sigma_j^{\beta_3}$ ] with @xmath53 . therefore , for a macroscopic large @xmath33 system , the c.m . part @xmath38 actually commutes with the relative - electron part @xmath54 in the hamiltonian , i.e. the c.m . motion and the relative motion of electrons are truly separated from each other . the couplings between the two emerge only through the electron impurity and electron phonon interactions . furthermore , the electric field @xmath55 shows up only in @xmath38 . and , in view of @xmath56={\rm i}\delta_{\alpha \beta}(\delta_{ij}-1/n)\simeq { \rm i}\delta_{\alpha\beta}\delta_{ij}$ ] , i.e. the relative - electron momenta and coordinates can be treated as canonical conjugate variables , the relative - motion part @xmath54 is just the hamiltonian of @xmath33 electrons in the surface state of ti in the magnetic field without the presence of the electric field . in terms of the c.m . coordinate @xmath57 and the relative electron density operator @xmath58 , the electron impurity and electron phonon interactions can be written as@xcite @xmath59 here @xmath60 and @xmath61 are respectively the impurity potential ( an impurity at randomly distributed position @xmath62 ) and electron phonon coupling matrix element in the plane - wave representation , and @xmath63 with @xmath64 and @xmath65 being the creation and annihilation operators for a phonon of wavevector @xmath66 in branch @xmath67 having frequency @xmath68 . velocity ( operator ) @xmath69 is the time variation of its coordinate : @xmath70= v_{\rm f}(\sigma_{\rm c}^y\ , \hat{i}-\sigma_{\rm c}^x\ , \hat{j})$ ] . to derive a force - balance equation for steady state transport we consider the heisenberg equation for the rate of change of the c.m . canonical momentum @xmath71 : @xmath72= - n e({\bm v}\times { \bm b})- n e{\bm e}+{\bm { f}}_{\rm i}+{\bm { f}}_{\rm p},\ ] ] in which the frictional forces @xmath73 and @xmath74 share the same expressions as given in ref .. the statistical average of the operator equation can be determined to linear order in the electron impurity and electron phonon interactions @xmath17 and @xmath18 with the initial density matrix @xmath75 at temperature @xmath76 when the in - plane electric field @xmath77 is not strong . for steady - transport states we have @xmath78 , leading to a force - balance equation of the form @xmath79 here @xmath80 , the statistically averaged velocity of the moving center - of - mass , is identified as the average rate of change of its position , i.e. the drift velocity of the electron system driven by the electric field @xmath77 , and @xmath81 and @xmath82 are frictional forces experienced by the center - of - mass due to impurity and phonon scatterings : @xmath83,\label{fp}\end{aligned}\ ] ] in which @xmath84 is the bose distribution function , @xmath85 , and @xmath86 stands for the imaginary part of the fourier spectrum of the relative - electron density correlation function defined by @xmath87\big\rangle_{0},\ ] ] where @xmath88 and @xmath89 denotes the statistical averaging over the initial density matrix @xmath90.@xcite the force - balance equation describes the steady - state two - dimensional magnetotransport in the surface state of a ti . note that the frictional forces @xmath81 and @xmath82 are in the opposite direction of the drift velocity @xmath91 and their magnitudes are functions of @xmath92 only . with the drift velocity @xmath93 in the @xmath9 direction , the force - balance equation eq . yields a transverse resistivity @xmath94 , and a longitudinal resistivity @xmath95 . the linear one is in the form @xmath96 for calculating the electron density correlation function @xmath97 we proceed in the landau representation.@xcite the landau levels of the single - particle hamiltonian @xmath98 of the relative - electron system in the absence of electric field are composed of a positive `` @xmath99 '' and a negative `` @xmath100 '' branch@xcite @xmath101 with @xmath102 and @xmath103 , and a zero ( @xmath104 ) level @xmath105 the corresponding landau wave functions are @xmath106 and @xmath107 for @xmath108 ; and @xmath109 for @xmath104 . here @xmath110 is the wavevector of the system along @xmath9 direction ; @xmath111 with @xmath112 ; and @xmath113 is the harmonic oscillator eigenfunction with @xmath114 being the hermite polynomial , @xmath115 , and @xmath116 . each landau level contains @xmath117 electron states for system of unit surface area . the positive branch @xmath118 and the @xmath104 level @xmath119 of the above energy spectra are indeed quite close to those of the surface states in the bulk gap of bi@xmath0se@xmath1-family materials derived from microscopic band calculation.@xcite the landau levels are broadened due to impurity , phonon and electron - electron scatterings . we model the imaginary part of the retarded green s function , or the density - of - states , of the broadened landau level @xmath120 ( written for `` + ' ' -branch and @xmath104 levels ) , using a gaussian - type form:@xcite @xmath121,\ ] ] with a half - width @xmath122 of the form:@xcite @xmath123^{1/2}$ ] . here @xmath124 is the single - particle lifetime and @xmath125 is the cyclotron frequency of linear - energy - dispersion system with @xmath126 being the zero - temperature fermi level . using a semi - empirical parameter @xmath127 to relate @xmath124 with the transport scattering time @xmath128 , and expressing @xmath129 with the zero - field mobility @xmath5 at finite temperature,@xcite we can write the landau - level broadening as @xmath130^{1/2}.\ ] ] in the present study we consider the case of @xmath120-doping , i.e. the fermi level is high enough above the energy zero of the dirac cone in the range of `` + ' ' -branch levels and the states of `` @xmath100''-branch levels are completely filled , that they are irrelevant to electron transport . special attention has to be paid to the @xmath104 level , since , depending on the direction of exchange potential the effective g - factor of a ti surface state , @xmath30 , can be positive , zero or negative.@xcite the sign and magnitude of the effective g - factor determines how many states of the zero level should be included in or excluded from the available states for electron occupation in the case of @xmath120-doping at a magnetic field . ( i ) if @xmath131 , the @xmath104 level center is exactly at @xmath132 and the system is electron - hole symmetric . the total number of negative energy states ( including the states of the lower half of the @xmath104 level and states of the @xmath100"-branch levels ) and that of positive energy states ( including the states of the upper half of the @xmath104 level and states of the @xmath99"-branch levels ) do not change when changing magnetic field . therefore , the lower - half negative energy states of this level are always filled and the upper - half positive - energy states of it are available for the occupation of particles which are counted as electrons participating in transport in the case of @xmath120-doping . ( ii ) for a finite positive @xmath133 , the @xmath104 level @xmath134 moves downward to negative energy and its distance to the nearest @xmath100"-branch level is @xmath135 closer than to the nearest + " -branch level at finite magnetic field strength @xmath2 . this is equivalent to the opening of an increasingly enlarged ( with increasing @xmath2 ) energy gap between the + " -branch states and the states of the zero - level and the @xmath100"-branch levels . the opening of a sufficient energy gap implies that with increasing magnetic field the states in the + " -branch levels would no longer shrink into the zero - level , and thus the @xmath104 level should be completely excluded from the conduction band , i.e. only particles occupying the + " -branch states are counted as electrons participating in transport in the case of @xmath120-doping , when the magnetic field @xmath2 gets larger than a certain value ( depending on the magnitude of @xmath30 ) . ( iii ) for a finite negative @xmath136 , the @xmath104 level @xmath134 moves upward to positive energy and an increasingly enlarged energy gap will be opened between the states of the zero - level and the + " -branch and the states of @xmath100"-branch levels , and particles occupying the @xmath104 level and + " -branch states are electrons participating in transport when the magnetic field @xmath2 gets larger than a certain value . as a result , the experimentally accessible sheet density @xmath33 of electrons participating in transport is related to the fermi energy @xmath137 by the following equation valid at finite @xmath30 for the magnetic field @xmath2 larger than a certain value : @xmath138 in which @xmath139 + 1\}^{-1}$ ] is the fermi distribution function at temperature @xmath76 and the summation index @xmath120 goes over @xmath140 for @xmath133 , or @xmath141 for @xmath136 . in the case of @xmath131 , @xmath142\ ] ] valid for arbitrary magnetic field , in which @xmath143 . the imaginary part of relative - electron density correlation function in the presence of a magnetic field , @xmath86 , can be expressed in the landau representation as@xcite @xmath144 in which the transform factor @xmath145 ^ 2,\end{aligned}\ ] ] with @xmath146 , @xmath147 , @xmath148 , and @xmath149 being associated laguerre polynomials . the landau - representation correlation function @xmath150 in eq.([piqw ] ) can be constructed with the imaginary part of the retarded green s function @xmath151 , or the density - of - states , of the @xmath120th landau level as@xcite @xmath152\nonumber\\ & \hspace{1.2cm}\times{\rm im}g_n(\epsilon+\omega){\rm im}g_{n'}(\epsilon).\end{aligned}\ ] ] the summation indices @xmath120 and @xmath153 in eq.([piqw ] ) are taken over @xmath140 for @xmath133 , or @xmath154 for @xmath136 . in the case of @xmath131 , eq.([piqw ] ) still works and the summation indices @xmath120 and @xmath153 go over @xmath154 but with @xmath155 replaced by @xmath156 in eq.([p2nn ] ) . numerical calculations are performed for the magnetoresistivity @xmath157 of surface state in a uniform ti bi@xmath0se@xmath1 . at zero temperature the elastic scattering contributing to the resistivity is modeled by a coulomb potential due to charged impurities:@xcite @xmath158 with @xmath159 being the impurity density , which is determined by the zero - magnetic - field mobility @xmath5 . at temperatures higher than @xmath160,@xcite phonon scatterings play increasingly important role and the dominant inelastic contribution comes from optical phonons . for this polar material , the scattering by optical phonons via the deformation potential can be neglected . hence , we take account of inelastic scattering from optical phonons via frhlich coupling : @xmath161 . in the numerical calculation we use the following parameters:@xcite fermi velocity @xmath162 , static dielectric constant @xmath163 , optical dielectric constant @xmath164 , and phonon energy @xmath165 . the broadening parameter is taken to be @xmath166 . as a function of the magnetic field @xmath2 having different effective g - factors : @xmath167 and @xmath168 for a ti surface system with electron sheet density @xmath169 in the cases of zero - magnetic - field mobility @xmath170 ( a ) and @xmath171 ( b ) . several integer - number positions of filling factor @xmath172 are marked in ( b).,scaledwidth=40.0% ] fig.[diffg ] shows the calculated magnetoresistivity @xmath157 versus the magnetic field strength @xmath2 for a ti surface system with electron sheet density @xmath169 but having different effective g - factors : @xmath167 and @xmath168 for two values of zero - magnetic - field mobility @xmath170 and @xmath171 , representing different degree of landau - level broadening . in the case without zeeman splitting ( @xmath131 ) the resistivity @xmath157 exhibits almost no change with changing magnetic field up to 10 t , except the shubnikov - de haas ( sdh ) oscillation showing up in the case of @xmath171 . this kind of magnetoresistance behavior was indeed seen experimentally in the electron - hole symmetrical massless system of single - layer graphene.@xcite in the case of a positive g - factor , @xmath173 , the magnetoresistivity increases linearly with increasing magnetic field ; while for a negative g - factor , @xmath174 , the magnetoresistivity decreases linearly with increasing magnetic field . is shown as a function of the magnetic field @xmath2 for different values of zero - magnetic - field mobility : ( a ) @xmath175 , ( b ) @xmath176 , ( c ) @xmath177 , ( d ) @xmath178 , ( e ) @xmath179 , and ( f ) @xmath180 . the inset of ( a ) illustrates the same for a larger magnetic - field range @xmath181 . the filling factor @xmath182 is plotted versus the magnetic field in ( f ) ; and several integer - number positions of @xmath182 are also marked in ( d ) and ( e ) . here the surface electron density @xmath169 and the lattice temperature @xmath183.,scaledwidth=47.0% ] in the following we will give more detailed examination on the linearly increasing magnetoresistance in the positive @xmath30 case . fig.[rhob ] shows the calculated resistivity @xmath157 versus the magnetic field strength @xmath2 at lattice temperature @xmath183 for system of carrier sheet density @xmath169 and @xmath173 , having different zero - field mobility @xmath184 and @xmath180 . all resistivity curves for mobility @xmath185 exhibit clear linearity in the magnetic - field range and appear no tendency of saturation at the highest field shown in the figure . especially , for the case @xmath170 , the linear behavior extends even up to the magnetic field of @xmath186 , as illustrated in the inset of fig.[rhob](a ) . this feature contradicts the classical mr which saturates at sufficiently large magnetic field @xmath187 . note that here we only present the calculated @xmath157 for magnetic field @xmath2 larger than @xmath188 t , for which a sufficient energy gap @xmath135 is assumed to open that with further increase of the magnetic field the states in the `` + ' ' -branch levels no longer shrink into the zero level and thus it should be excluded from the conduction band . this is of course not true for very weak magnetic field . when @xmath189 the energy gap @xmath190 , the situation becomes similar to the case of @xmath131 : the whole upper half of the zero - level states are available to electron occupation and we should have a flat resistivity @xmath157 when changing magnetic field . with increasing @xmath2 the portion of the zero - level states available to conduction electrons decreases until the magnetic field reaches @xmath191 . as a result the resistivity @xmath157 should exhibit a crossover from a flat changing at small @xmath2 to positively linear increasing at @xmath192 . this is just the behavior observed in the ti bi@xmath0se@xmath1.@xcite note that in the case of @xmath170 , the broadened landau - level widths are always larger than the neighboring level interval : @xmath193 , which requires @xmath194 ^ 2 $ ] , even for the lowest landau level @xmath195 , i.e. the whole landau - level spectrum is smeared . with increasing the zero - field mobility the magnitude of resistivity @xmath157 decreases , and when the broadened landau - level width becomes smaller than the neighboring level interval , @xmath196 , a weak sdh oscillation begin to occur around the linearly - dependent average value of @xmath157 at higher portion of the magnetic field range , as seen in fig.[rhob](c ) , ( d ) and ( e ) for @xmath197 and @xmath198 . on the other hand , in the case of large mobility , e.g. @xmath199 , where the broadened landau - level widths @xmath200 are much smaller than the neighboring level interval even for level index @xmath120 as large as @xmath201 , the magnetoresistivity shows pronounced sdh oscillation and the linear - dependent behavior disappears , before the appearance of quantum hall effect,@xcite as shown in fig.[rhob](f ) . abrikosov s model for the lmr requires the applied magnetic field large enough to reach the quantum limit at which all the carriers are within the lowest landau level,@xcite while it is obvious that more than one landau levels are occupied in the experimental samples in the field range in which the linear and non - saturating magnetoresistivity was observed.@xcite for the given electron surface density @xmath202 , the number of occupied landau levels , or the filling factor @xmath172 , at different magnetic fields is shown in fig.[rhob](f ) , as well as in the fig.[rhob](d ) and ( e ) , where the integer - number positions of @xmath203 , i.e. filling up to entire @xmath182 landau levels , coincide with the minima of the density - of - states or the dips of sdh oscillation . this is in contrast with @xmath131 case , where the integer number of @xmath203 , which implies a filling up to the center position of the @xmath182th landau levels , locates at a peak of sdh oscillation , as shown in fig.[diffg]b . the observed sdh oscillations in the bi@xmath0se@xmath1 nanoribbon exhibiting nonsaturating surface lmr in the experiment@xcite favor the former case : a finite positive effective @xmath133 . is plotted as a function of the surface electron density @xmath33 at magnetic field @xmath204 : ( a ) at different values of zero - field mobility @xmath5 , and ( b ) at different values of zero - field conductivity @xmath205.,scaledwidth=40.0% ] at various lattice temperatures . here the zero - magnetic - field mobility at zero temperature is @xmath206.,scaledwidth=35.0% ] next , we examine the density - dependence of the linear magnetoresistivity . to compare with abrikosov s quantum magnetoresistance which suggests a @xmath207 behavior,@xcite we show the calculated @xmath208 for above lmr versus the carrier sheet density @xmath33 in fig.[rhon ] at fixed magnetic field @xmath209 t . the mobility is taken respectively to be @xmath210 and @xmath211m@xmath212/vs to make the resistivity in the lmr regime . a clearly linear dependence of @xmath213 on the surface density @xmath33 is seen in all cases , indicating that this non - saturating linear resistivity is almost inversely proportional to the carrier density . in the figure we also show @xmath208 versus @xmath33 under the condition of different given conductivity @xmath214 and @xmath215 . in this case the half - width @xmath216 is independent of surface density . the linear dependence still holds , indicating that this linear behavior is not sensitive to the modest @xmath33-dependence of landau level broadening @xmath216 as long as the system is in the overlapped landau level regime . from the above discussion , it is obvious that lmr shows up in the system having overlapped landau levels and the separation of landau levels makes the mr departure from the linear increase . at high temperature , the thermal energy would smear the level separation and phonon scatterings further broaden landau levels . hence , it is believed that this lmr will be robust against raising temperature . this is indeed the case as seen in fig.[rhot ] , where we plot the calculated magnetoresistivity @xmath157 for the above system with zero - temperature linear mobility @xmath217m@xmath212/vs versus the magnetic field at different lattice temperatures . we can see that raising temperature to room temperature has little effect on the linearity of mr . due to the decreased mobility at higher temperature from phonon scattering , the weak sdh oscillation on the linear background tends to vanish . these features are in good agreement with the experimental report.@xcite in summary , we have studied the two - dimensional magnetotransport in the flat surface of a three - dimensional ti , which arises from the surface states with a wavevector - linear energy dispersion and a finite , positive zeeman splitting within the bulk energy gap . when the level broadening is comparable to or larger than the landau - level separation and the conduction electrons spread over many landau levels , a positive , dominantly linear and non - saturating magnetoresistance appears within a quite wide range of magnetic field and persists up to room temperature . this remarkable lmr provides a possible mechanism for the recently observed linear magnetoresistance in topological insulator bi@xmath0se@xmath1 nanoribbons.@xcite in contrast to quantum hall effect which appears in the case of well formed landau levels and to abrikosov s quantum magnetotransport,@xcite which is limited to the extreme quantum limit that all electrons coalesce into the lowest landau level , the discussed lmr is a phenomena of pure classical two - dimensional magnetotransport in a system having linear - energy - dispersion , appearing in the regime of overlapped landau levels , irrespective of its showing up in relatively high magnetic field range . furthermore , the present scheme deals with spatially uniform case without invoking the mobility fluctuation in a strongly inhomogeneous system , which is required in the classical parish and littlewood model to produce a lmr.@xcite the appearance of this significant positive - increasing linear magnetoresistance depends on the existence of a positive and sizable effective g - factor . if the zeeman energy splitting is quite small the resistivity @xmath157 would exhibit little change with changing magnetic field . in the case of a negative and sizable effective g - factor the magnetoresistivity would decrease linearly with increasing magnetic field . therefore , the behavior of the longitudinal resistivity versus magnetic field may provide a useful way for judging the direction and the size of the effective zeeman energy splitting in ti surface states . this work was supported by the national science foundation of china ( grant no . 11104002 ) , the national basic research program of china ( grant no . 2012cb927403 ) and by the program for science&technology innovation talents in universities of henan province ( grant no . 2012hastit029 ) ."""
inputs = tokenizer(
[ARTICLE_LEP, ARTICLE_MAGNET],
max_length=1024,
padding="max_length",
truncation=True,
return_tensors="pt",
)
inputs = {k: inputs[k].to(torch_device) for k in inputs}
hypotheses_batch = model.generate(**inputs)
EXPECTED_LEP = (
"motivated by some recent studies on the light cp - odd higgs boson @xmath0 in non - minimal"
" supersymmetric models, we investigate the rare @xmath1-decays @xmath2 ( @xmath3 ) in the two higgs"
" doublet model ( 2hdm ), the nearly minimal supersymmetric standard model ( nmssm ), the next - to -"
" minimal supersymmetric standard model ( nmssm ) and the minimal supersymmetric standard model ( mssm"
" ).<n> we find that the branching ratios of @xmath4 can reach @xmath5 in 2hdm, @xmath6 in nmssm and"
" @xmath7 in mssm, which are at the level of @xmath8 in 2hdm, @xmath9 in nmssm and @xmath10 in mssm,"
" respectively.<n> these rates can be significantly enhanced in new physics models which lie within the"
" expected sensitivity of the gigaz option of the international linear collider ( ilc ). <n> = # 1,nucl."
" <n> phys. <n> b * # 1"
)
EXPECTED_MAGNET = (
"a positive, nonsaturating and dominantly linear magnetoresistance can appear within quite wide magnetic -"
" field range in the surface state of a topological insulator having a positive and finite effective g -"
" factor. this linear magnetoresistance shows up in the system of high carrier concentration and low"
" mobility when electrons are in extended states and spread over many smeared landau levels, and persists"
" up to room temperature, providing a possible mechanism for the recently observed linear"
" magnetoresistance in topological insulator bi@xmath0se@xmath1 nanoribbons."
)
generated = tokenizer.batch_decode(
hypotheses_batch.tolist(), clean_up_tokenization_spaces=True, skip_special_tokens=True
)
self.assertTrue(generated == [EXPECTED_LEP, EXPECTED_MAGNET])
class BigBirdPegasusStandaloneDecoderModelTester:
def __init__(
self,
parent,
vocab_size=99,
batch_size=7,
d_model=32,
decoder_seq_length=7,
is_training=True,
is_decoder=True,
use_attention_mask=True,
use_cache=False,
use_labels=True,
decoder_start_token_id=2,
decoder_ffn_dim=32,
decoder_layers=4,
encoder_attention_heads=4,
decoder_attention_heads=4,
max_position_embeddings=30,
is_encoder_decoder=False,
pad_token_id=0,
bos_token_id=1,
eos_token_id=2,
scope=None,
attention_type="original_full",
use_bias=True,
block_size=16,
num_random_blocks=3,
):
self.parent = parent
self.batch_size = batch_size
self.decoder_seq_length = decoder_seq_length
# For common tests
self.seq_length = self.decoder_seq_length
self.is_training = is_training
self.use_attention_mask = use_attention_mask
self.use_labels = use_labels
self.vocab_size = vocab_size
self.d_model = d_model
self.hidden_size = d_model
self.num_hidden_layers = decoder_layers
self.decoder_layers = decoder_layers
self.decoder_ffn_dim = decoder_ffn_dim
self.encoder_attention_heads = encoder_attention_heads
self.decoder_attention_heads = decoder_attention_heads
self.num_attention_heads = decoder_attention_heads
self.eos_token_id = eos_token_id
self.bos_token_id = bos_token_id
self.pad_token_id = pad_token_id
self.decoder_start_token_id = decoder_start_token_id
self.use_cache = use_cache
self.max_position_embeddings = max_position_embeddings
self.is_encoder_decoder = is_encoder_decoder
self.scope = None
self.decoder_key_length = decoder_seq_length
self.base_model_out_len = 2
self.decoder_attention_idx = 1
self.attention_type = attention_type
self.use_bias = use_bias
self.block_size = block_size
self.num_random_blocks = num_random_blocks
def prepare_config_and_inputs(self):
input_ids = ids_tensor([self.batch_size, self.decoder_seq_length], self.vocab_size)
attention_mask = None
if self.use_attention_mask:
attention_mask = ids_tensor([self.batch_size, self.decoder_seq_length], vocab_size=2)
lm_labels = None
if self.use_labels:
lm_labels = ids_tensor([self.batch_size, self.decoder_seq_length], self.vocab_size)
config = BigBirdPegasusConfig(
vocab_size=self.vocab_size,
d_model=self.d_model,
decoder_layers=self.decoder_layers,
decoder_ffn_dim=self.decoder_ffn_dim,
encoder_attention_heads=self.encoder_attention_heads,
decoder_attention_heads=self.decoder_attention_heads,
eos_token_id=self.eos_token_id,
bos_token_id=self.bos_token_id,
use_cache=self.use_cache,
pad_token_id=self.pad_token_id,
decoder_start_token_id=self.decoder_start_token_id,
max_position_embeddings=self.max_position_embeddings,
is_encoder_decoder=self.is_encoder_decoder,
attention_type=self.attention_type,
use_bias=self.use_bias,
block_size=self.block_size,
num_random_blocks=self.num_random_blocks,
)
return (
config,
input_ids,
attention_mask,
lm_labels,
)
def create_and_check_decoder_model_past(
self,
config,
input_ids,
attention_mask,
lm_labels,
):
config.use_cache = True
model = BigBirdPegasusDecoder(config=config).to(torch_device).eval()
# first forward pass
outputs = model(input_ids, use_cache=True)
outputs_use_cache_conf = model(input_ids)
outputs_no_past = model(input_ids, use_cache=False)
self.parent.assertTrue(len(outputs) == len(outputs_use_cache_conf))
self.parent.assertTrue(len(outputs) == len(outputs_no_past) + 1)
past_key_values = outputs["past_key_values"]
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)
# append to next input_ids and
next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
output_from_no_past = model(next_input_ids)["last_hidden_state"]
output_from_past = model(next_tokens, past_key_values=past_key_values)["last_hidden_state"]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, next_input_ids.shape[-1] - 1, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
assert torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3)
def create_and_check_decoder_model_attention_mask_past(
self,
config,
input_ids,
attention_mask,
lm_labels,
):
model = BigBirdPegasusDecoder(config=config).to(torch_device).eval()
# create attention mask
attn_mask = torch.ones(input_ids.shape, dtype=torch.long, device=torch_device)
half_seq_length = input_ids.shape[-1] // 2
attn_mask[:, half_seq_length:] = 0
# first forward pass
past_key_values = model(input_ids, attention_mask=attn_mask, use_cache=True)["past_key_values"]
# create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size)
# change a random masked slice from input_ids
random_seq_idx_to_change = ids_tensor((1,), half_seq_length).item() + 1
random_other_next_tokens = ids_tensor((self.batch_size, 1), config.vocab_size).squeeze(-1)
input_ids[:, -random_seq_idx_to_change] = random_other_next_tokens
# append to next input_ids and attn_mask
next_input_ids = torch.cat([input_ids, next_tokens], dim=-1)
attn_mask = torch.cat(
[attn_mask, torch.ones((attn_mask.shape[0], 1), dtype=torch.long, device=torch_device)],
dim=1,
)
# get two different outputs
output_from_no_past = model(next_input_ids)["last_hidden_state"]
output_from_past = model(next_tokens, past_key_values=past_key_values)["last_hidden_state"]
# select random slice
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
output_from_no_past_slice = output_from_no_past[:, next_input_ids.shape[-1] - 1, random_slice_idx].detach()
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
# big bird has extremely high logits which requires
# such a high error tolerance here
assert torch.allclose(output_from_past_slice, output_from_no_past_slice, atol=5e-1)
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
config, input_ids, attention_mask, lm_labels = config_and_inputs
inputs_dict = {"input_ids": input_ids, "attention_mask": attention_mask}
return config, inputs_dict
@require_torch
class BigBirdPegasusStandaloneDecoderModelTest(ModelTesterMixin, GenerationTesterMixin, unittest.TestCase):
all_model_classes = (BigBirdPegasusDecoder, BigBirdPegasusForCausalLM) if is_torch_available() else ()
all_generative_model_classes = (BigBirdPegasusForCausalLM,) if is_torch_available() else ()
test_pruning = False
is_encoder_decoder = False
def setUp(
self,
):
self.model_tester = BigBirdPegasusStandaloneDecoderModelTester(self, is_training=False)
self.config_tester = ConfigTester(self, config_class=BigBirdPegasusConfig)
def test_config(self):
self.config_tester.run_common_tests()
def test_decoder_model_past(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_decoder_model_past(*config_and_inputs)
def test_decoder_model_attn_mask_past(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_decoder_model_attention_mask_past(*config_and_inputs)
def test_retain_grad_hidden_states_attentions(self):
# decoder cannot keep gradients
return
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/tapas/__init__.py | # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tf_available, is_torch_available
_import_structure = {
"configuration_tapas": ["TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP", "TapasConfig"],
"tokenization_tapas": ["TapasTokenizer"],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tapas"] = [
"TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TapasForMaskedLM",
"TapasForQuestionAnswering",
"TapasForSequenceClassification",
"TapasModel",
"TapasPreTrainedModel",
"load_tf_weights_in_tapas",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tf_tapas"] = [
"TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFTapasForMaskedLM",
"TFTapasForQuestionAnswering",
"TFTapasForSequenceClassification",
"TFTapasModel",
"TFTapasPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_tapas import TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP, TapasConfig
from .tokenization_tapas import TapasTokenizer
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tapas import (
TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TapasForMaskedLM,
TapasForQuestionAnswering,
TapasForSequenceClassification,
TapasModel,
TapasPreTrainedModel,
load_tf_weights_in_tapas,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_tapas import (
TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TFTapasForMaskedLM,
TFTapasForQuestionAnswering,
TFTapasForSequenceClassification,
TFTapasModel,
TFTapasPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tf_available, is_torch_available
_import_structure = {
"configuration_tapas": ["TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP", "TapasConfig"],
"tokenization_tapas": ["TapasTokenizer"],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tapas"] = [
"TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TapasForMaskedLM",
"TapasForQuestionAnswering",
"TapasForSequenceClassification",
"TapasModel",
"TapasPreTrainedModel",
"load_tf_weights_in_tapas",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tf_tapas"] = [
"TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFTapasForMaskedLM",
"TFTapasForQuestionAnswering",
"TFTapasForSequenceClassification",
"TFTapasModel",
"TFTapasPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_tapas import TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP, TapasConfig
from .tokenization_tapas import TapasTokenizer
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tapas import (
TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TapasForMaskedLM,
TapasForQuestionAnswering,
TapasForSequenceClassification,
TapasModel,
TapasPreTrainedModel,
load_tf_weights_in_tapas,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_tapas import (
TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TFTapasForMaskedLM,
TFTapasForQuestionAnswering,
TFTapasForSequenceClassification,
TFTapasModel,
TFTapasPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/legacy/token-classification/utils_ner.py | # coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Named entity recognition fine-tuning: utilities to work with CoNLL-2003 task. """
import logging
import os
from dataclasses import dataclass
from enum import Enum
from typing import List, Optional, Union
from filelock import FileLock
from transformers import PreTrainedTokenizer, is_tf_available, is_torch_available
logger = logging.getLogger(__name__)
@dataclass
class InputExample:
"""
A single training/test example for token classification.
Args:
guid: Unique id for the example.
words: list. The words of the sequence.
labels: (Optional) list. The labels for each word of the sequence. This should be
specified for train and dev examples, but not for test examples.
"""
guid: str
words: List[str]
labels: Optional[List[str]]
@dataclass
class InputFeatures:
"""
A single set of features of data.
Property names are the same names as the corresponding inputs to a model.
"""
input_ids: List[int]
attention_mask: List[int]
token_type_ids: Optional[List[int]] = None
label_ids: Optional[List[int]] = None
class Split(Enum):
train = "train"
dev = "dev"
test = "test"
class TokenClassificationTask:
@staticmethod
def read_examples_from_file(data_dir, mode: Union[Split, str]) -> List[InputExample]:
raise NotImplementedError
@staticmethod
def get_labels(path: str) -> List[str]:
raise NotImplementedError
@staticmethod
def convert_examples_to_features(
examples: List[InputExample],
label_list: List[str],
max_seq_length: int,
tokenizer: PreTrainedTokenizer,
cls_token_at_end=False,
cls_token="[CLS]",
cls_token_segment_id=1,
sep_token="[SEP]",
sep_token_extra=False,
pad_on_left=False,
pad_token=0,
pad_token_segment_id=0,
pad_token_label_id=-100,
sequence_a_segment_id=0,
mask_padding_with_zero=True,
) -> List[InputFeatures]:
"""Loads a data file into a list of `InputFeatures`
`cls_token_at_end` define the location of the CLS token:
- False (Default, BERT/XLM pattern): [CLS] + A + [SEP] + B + [SEP]
- True (XLNet/GPT pattern): A + [SEP] + B + [SEP] + [CLS]
`cls_token_segment_id` define the segment id associated to the CLS token (0 for BERT, 2 for XLNet)
"""
# TODO clean up all this to leverage built-in features of tokenizers
label_map = {label: i for i, label in enumerate(label_list)}
features = []
for ex_index, example in enumerate(examples):
if ex_index % 10_000 == 0:
logger.info("Writing example %d of %d", ex_index, len(examples))
tokens = []
label_ids = []
for word, label in zip(example.words, example.labels):
word_tokens = tokenizer.tokenize(word)
# bert-base-multilingual-cased sometimes output "nothing ([]) when calling tokenize with just a space.
if len(word_tokens) > 0:
tokens.extend(word_tokens)
# Use the real label id for the first token of the word, and padding ids for the remaining tokens
label_ids.extend([label_map[label]] + [pad_token_label_id] * (len(word_tokens) - 1))
# Account for [CLS] and [SEP] with "- 2" and with "- 3" for RoBERTa.
special_tokens_count = tokenizer.num_special_tokens_to_add()
if len(tokens) > max_seq_length - special_tokens_count:
tokens = tokens[: (max_seq_length - special_tokens_count)]
label_ids = label_ids[: (max_seq_length - special_tokens_count)]
# The convention in BERT is:
# (a) For sequence pairs:
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
# (b) For single sequences:
# tokens: [CLS] the dog is hairy . [SEP]
# type_ids: 0 0 0 0 0 0 0
#
# Where "type_ids" are used to indicate whether this is the first
# sequence or the second sequence. The embedding vectors for `type=0` and
# `type=1` were learned during pre-training and are added to the wordpiece
# embedding vector (and position vector). This is not *strictly* necessary
# since the [SEP] token unambiguously separates the sequences, but it makes
# it easier for the model to learn the concept of sequences.
#
# For classification tasks, the first vector (corresponding to [CLS]) is
# used as the "sentence vector". Note that this only makes sense because
# the entire model is fine-tuned.
tokens += [sep_token]
label_ids += [pad_token_label_id]
if sep_token_extra:
# roberta uses an extra separator b/w pairs of sentences
tokens += [sep_token]
label_ids += [pad_token_label_id]
segment_ids = [sequence_a_segment_id] * len(tokens)
if cls_token_at_end:
tokens += [cls_token]
label_ids += [pad_token_label_id]
segment_ids += [cls_token_segment_id]
else:
tokens = [cls_token] + tokens
label_ids = [pad_token_label_id] + label_ids
segment_ids = [cls_token_segment_id] + segment_ids
input_ids = tokenizer.convert_tokens_to_ids(tokens)
# The mask has 1 for real tokens and 0 for padding tokens. Only real
# tokens are attended to.
input_mask = [1 if mask_padding_with_zero else 0] * len(input_ids)
# Zero-pad up to the sequence length.
padding_length = max_seq_length - len(input_ids)
if pad_on_left:
input_ids = ([pad_token] * padding_length) + input_ids
input_mask = ([0 if mask_padding_with_zero else 1] * padding_length) + input_mask
segment_ids = ([pad_token_segment_id] * padding_length) + segment_ids
label_ids = ([pad_token_label_id] * padding_length) + label_ids
else:
input_ids += [pad_token] * padding_length
input_mask += [0 if mask_padding_with_zero else 1] * padding_length
segment_ids += [pad_token_segment_id] * padding_length
label_ids += [pad_token_label_id] * padding_length
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
assert len(label_ids) == max_seq_length
if ex_index < 5:
logger.info("*** Example ***")
logger.info("guid: %s", example.guid)
logger.info("tokens: %s", " ".join([str(x) for x in tokens]))
logger.info("input_ids: %s", " ".join([str(x) for x in input_ids]))
logger.info("input_mask: %s", " ".join([str(x) for x in input_mask]))
logger.info("segment_ids: %s", " ".join([str(x) for x in segment_ids]))
logger.info("label_ids: %s", " ".join([str(x) for x in label_ids]))
if "token_type_ids" not in tokenizer.model_input_names:
segment_ids = None
features.append(
InputFeatures(
input_ids=input_ids, attention_mask=input_mask, token_type_ids=segment_ids, label_ids=label_ids
)
)
return features
if is_torch_available():
import torch
from torch import nn
from torch.utils.data import Dataset
class TokenClassificationDataset(Dataset):
"""
This will be superseded by a framework-agnostic approach
soon.
"""
features: List[InputFeatures]
pad_token_label_id: int = nn.CrossEntropyLoss().ignore_index
# Use cross entropy ignore_index as padding label id so that only
# real label ids contribute to the loss later.
def __init__(
self,
token_classification_task: TokenClassificationTask,
data_dir: str,
tokenizer: PreTrainedTokenizer,
labels: List[str],
model_type: str,
max_seq_length: Optional[int] = None,
overwrite_cache=False,
mode: Split = Split.train,
):
# Load data features from cache or dataset file
cached_features_file = os.path.join(
data_dir,
"cached_{}_{}_{}".format(mode.value, tokenizer.__class__.__name__, str(max_seq_length)),
)
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
lock_path = cached_features_file + ".lock"
with FileLock(lock_path):
if os.path.exists(cached_features_file) and not overwrite_cache:
logger.info(f"Loading features from cached file {cached_features_file}")
self.features = torch.load(cached_features_file)
else:
logger.info(f"Creating features from dataset file at {data_dir}")
examples = token_classification_task.read_examples_from_file(data_dir, mode)
# TODO clean up all this to leverage built-in features of tokenizers
self.features = token_classification_task.convert_examples_to_features(
examples,
labels,
max_seq_length,
tokenizer,
cls_token_at_end=bool(model_type in ["xlnet"]),
# xlnet has a cls token at the end
cls_token=tokenizer.cls_token,
cls_token_segment_id=2 if model_type in ["xlnet"] else 0,
sep_token=tokenizer.sep_token,
sep_token_extra=False,
# roberta uses an extra separator b/w pairs of sentences, cf. github.com/pytorch/fairseq/commit/1684e166e3da03f5b600dbb7855cb98ddfcd0805
pad_on_left=bool(tokenizer.padding_side == "left"),
pad_token=tokenizer.pad_token_id,
pad_token_segment_id=tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
logger.info(f"Saving features into cached file {cached_features_file}")
torch.save(self.features, cached_features_file)
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
if is_tf_available():
import tensorflow as tf
class TFTokenClassificationDataset:
"""
This will be superseded by a framework-agnostic approach
soon.
"""
features: List[InputFeatures]
pad_token_label_id: int = -100
# Use cross entropy ignore_index as padding label id so that only
# real label ids contribute to the loss later.
def __init__(
self,
token_classification_task: TokenClassificationTask,
data_dir: str,
tokenizer: PreTrainedTokenizer,
labels: List[str],
model_type: str,
max_seq_length: Optional[int] = None,
overwrite_cache=False,
mode: Split = Split.train,
):
examples = token_classification_task.read_examples_from_file(data_dir, mode)
# TODO clean up all this to leverage built-in features of tokenizers
self.features = token_classification_task.convert_examples_to_features(
examples,
labels,
max_seq_length,
tokenizer,
cls_token_at_end=bool(model_type in ["xlnet"]),
# xlnet has a cls token at the end
cls_token=tokenizer.cls_token,
cls_token_segment_id=2 if model_type in ["xlnet"] else 0,
sep_token=tokenizer.sep_token,
sep_token_extra=False,
# roberta uses an extra separator b/w pairs of sentences, cf. github.com/pytorch/fairseq/commit/1684e166e3da03f5b600dbb7855cb98ddfcd0805
pad_on_left=bool(tokenizer.padding_side == "left"),
pad_token=tokenizer.pad_token_id,
pad_token_segment_id=tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
def gen():
for ex in self.features:
if ex.token_type_ids is None:
yield (
{"input_ids": ex.input_ids, "attention_mask": ex.attention_mask},
ex.label_ids,
)
else:
yield (
{
"input_ids": ex.input_ids,
"attention_mask": ex.attention_mask,
"token_type_ids": ex.token_type_ids,
},
ex.label_ids,
)
if "token_type_ids" not in tokenizer.model_input_names:
self.dataset = tf.data.Dataset.from_generator(
gen,
({"input_ids": tf.int32, "attention_mask": tf.int32}, tf.int64),
(
{"input_ids": tf.TensorShape([None]), "attention_mask": tf.TensorShape([None])},
tf.TensorShape([None]),
),
)
else:
self.dataset = tf.data.Dataset.from_generator(
gen,
({"input_ids": tf.int32, "attention_mask": tf.int32, "token_type_ids": tf.int32}, tf.int64),
(
{
"input_ids": tf.TensorShape([None]),
"attention_mask": tf.TensorShape([None]),
"token_type_ids": tf.TensorShape([None]),
},
tf.TensorShape([None]),
),
)
def get_dataset(self):
self.dataset = self.dataset.apply(tf.data.experimental.assert_cardinality(len(self.features)))
return self.dataset
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
| # coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Named entity recognition fine-tuning: utilities to work with CoNLL-2003 task. """
import logging
import os
from dataclasses import dataclass
from enum import Enum
from typing import List, Optional, Union
from filelock import FileLock
from transformers import PreTrainedTokenizer, is_tf_available, is_torch_available
logger = logging.getLogger(__name__)
@dataclass
class InputExample:
"""
A single training/test example for token classification.
Args:
guid: Unique id for the example.
words: list. The words of the sequence.
labels: (Optional) list. The labels for each word of the sequence. This should be
specified for train and dev examples, but not for test examples.
"""
guid: str
words: List[str]
labels: Optional[List[str]]
@dataclass
class InputFeatures:
"""
A single set of features of data.
Property names are the same names as the corresponding inputs to a model.
"""
input_ids: List[int]
attention_mask: List[int]
token_type_ids: Optional[List[int]] = None
label_ids: Optional[List[int]] = None
class Split(Enum):
train = "train"
dev = "dev"
test = "test"
class TokenClassificationTask:
@staticmethod
def read_examples_from_file(data_dir, mode: Union[Split, str]) -> List[InputExample]:
raise NotImplementedError
@staticmethod
def get_labels(path: str) -> List[str]:
raise NotImplementedError
@staticmethod
def convert_examples_to_features(
examples: List[InputExample],
label_list: List[str],
max_seq_length: int,
tokenizer: PreTrainedTokenizer,
cls_token_at_end=False,
cls_token="[CLS]",
cls_token_segment_id=1,
sep_token="[SEP]",
sep_token_extra=False,
pad_on_left=False,
pad_token=0,
pad_token_segment_id=0,
pad_token_label_id=-100,
sequence_a_segment_id=0,
mask_padding_with_zero=True,
) -> List[InputFeatures]:
"""Loads a data file into a list of `InputFeatures`
`cls_token_at_end` define the location of the CLS token:
- False (Default, BERT/XLM pattern): [CLS] + A + [SEP] + B + [SEP]
- True (XLNet/GPT pattern): A + [SEP] + B + [SEP] + [CLS]
`cls_token_segment_id` define the segment id associated to the CLS token (0 for BERT, 2 for XLNet)
"""
# TODO clean up all this to leverage built-in features of tokenizers
label_map = {label: i for i, label in enumerate(label_list)}
features = []
for ex_index, example in enumerate(examples):
if ex_index % 10_000 == 0:
logger.info("Writing example %d of %d", ex_index, len(examples))
tokens = []
label_ids = []
for word, label in zip(example.words, example.labels):
word_tokens = tokenizer.tokenize(word)
# bert-base-multilingual-cased sometimes output "nothing ([]) when calling tokenize with just a space.
if len(word_tokens) > 0:
tokens.extend(word_tokens)
# Use the real label id for the first token of the word, and padding ids for the remaining tokens
label_ids.extend([label_map[label]] + [pad_token_label_id] * (len(word_tokens) - 1))
# Account for [CLS] and [SEP] with "- 2" and with "- 3" for RoBERTa.
special_tokens_count = tokenizer.num_special_tokens_to_add()
if len(tokens) > max_seq_length - special_tokens_count:
tokens = tokens[: (max_seq_length - special_tokens_count)]
label_ids = label_ids[: (max_seq_length - special_tokens_count)]
# The convention in BERT is:
# (a) For sequence pairs:
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
# (b) For single sequences:
# tokens: [CLS] the dog is hairy . [SEP]
# type_ids: 0 0 0 0 0 0 0
#
# Where "type_ids" are used to indicate whether this is the first
# sequence or the second sequence. The embedding vectors for `type=0` and
# `type=1` were learned during pre-training and are added to the wordpiece
# embedding vector (and position vector). This is not *strictly* necessary
# since the [SEP] token unambiguously separates the sequences, but it makes
# it easier for the model to learn the concept of sequences.
#
# For classification tasks, the first vector (corresponding to [CLS]) is
# used as the "sentence vector". Note that this only makes sense because
# the entire model is fine-tuned.
tokens += [sep_token]
label_ids += [pad_token_label_id]
if sep_token_extra:
# roberta uses an extra separator b/w pairs of sentences
tokens += [sep_token]
label_ids += [pad_token_label_id]
segment_ids = [sequence_a_segment_id] * len(tokens)
if cls_token_at_end:
tokens += [cls_token]
label_ids += [pad_token_label_id]
segment_ids += [cls_token_segment_id]
else:
tokens = [cls_token] + tokens
label_ids = [pad_token_label_id] + label_ids
segment_ids = [cls_token_segment_id] + segment_ids
input_ids = tokenizer.convert_tokens_to_ids(tokens)
# The mask has 1 for real tokens and 0 for padding tokens. Only real
# tokens are attended to.
input_mask = [1 if mask_padding_with_zero else 0] * len(input_ids)
# Zero-pad up to the sequence length.
padding_length = max_seq_length - len(input_ids)
if pad_on_left:
input_ids = ([pad_token] * padding_length) + input_ids
input_mask = ([0 if mask_padding_with_zero else 1] * padding_length) + input_mask
segment_ids = ([pad_token_segment_id] * padding_length) + segment_ids
label_ids = ([pad_token_label_id] * padding_length) + label_ids
else:
input_ids += [pad_token] * padding_length
input_mask += [0 if mask_padding_with_zero else 1] * padding_length
segment_ids += [pad_token_segment_id] * padding_length
label_ids += [pad_token_label_id] * padding_length
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
assert len(label_ids) == max_seq_length
if ex_index < 5:
logger.info("*** Example ***")
logger.info("guid: %s", example.guid)
logger.info("tokens: %s", " ".join([str(x) for x in tokens]))
logger.info("input_ids: %s", " ".join([str(x) for x in input_ids]))
logger.info("input_mask: %s", " ".join([str(x) for x in input_mask]))
logger.info("segment_ids: %s", " ".join([str(x) for x in segment_ids]))
logger.info("label_ids: %s", " ".join([str(x) for x in label_ids]))
if "token_type_ids" not in tokenizer.model_input_names:
segment_ids = None
features.append(
InputFeatures(
input_ids=input_ids, attention_mask=input_mask, token_type_ids=segment_ids, label_ids=label_ids
)
)
return features
if is_torch_available():
import torch
from torch import nn
from torch.utils.data import Dataset
class TokenClassificationDataset(Dataset):
"""
This will be superseded by a framework-agnostic approach
soon.
"""
features: List[InputFeatures]
pad_token_label_id: int = nn.CrossEntropyLoss().ignore_index
# Use cross entropy ignore_index as padding label id so that only
# real label ids contribute to the loss later.
def __init__(
self,
token_classification_task: TokenClassificationTask,
data_dir: str,
tokenizer: PreTrainedTokenizer,
labels: List[str],
model_type: str,
max_seq_length: Optional[int] = None,
overwrite_cache=False,
mode: Split = Split.train,
):
# Load data features from cache or dataset file
cached_features_file = os.path.join(
data_dir,
"cached_{}_{}_{}".format(mode.value, tokenizer.__class__.__name__, str(max_seq_length)),
)
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
lock_path = cached_features_file + ".lock"
with FileLock(lock_path):
if os.path.exists(cached_features_file) and not overwrite_cache:
logger.info(f"Loading features from cached file {cached_features_file}")
self.features = torch.load(cached_features_file)
else:
logger.info(f"Creating features from dataset file at {data_dir}")
examples = token_classification_task.read_examples_from_file(data_dir, mode)
# TODO clean up all this to leverage built-in features of tokenizers
self.features = token_classification_task.convert_examples_to_features(
examples,
labels,
max_seq_length,
tokenizer,
cls_token_at_end=bool(model_type in ["xlnet"]),
# xlnet has a cls token at the end
cls_token=tokenizer.cls_token,
cls_token_segment_id=2 if model_type in ["xlnet"] else 0,
sep_token=tokenizer.sep_token,
sep_token_extra=False,
# roberta uses an extra separator b/w pairs of sentences, cf. github.com/pytorch/fairseq/commit/1684e166e3da03f5b600dbb7855cb98ddfcd0805
pad_on_left=bool(tokenizer.padding_side == "left"),
pad_token=tokenizer.pad_token_id,
pad_token_segment_id=tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
logger.info(f"Saving features into cached file {cached_features_file}")
torch.save(self.features, cached_features_file)
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
if is_tf_available():
import tensorflow as tf
class TFTokenClassificationDataset:
"""
This will be superseded by a framework-agnostic approach
soon.
"""
features: List[InputFeatures]
pad_token_label_id: int = -100
# Use cross entropy ignore_index as padding label id so that only
# real label ids contribute to the loss later.
def __init__(
self,
token_classification_task: TokenClassificationTask,
data_dir: str,
tokenizer: PreTrainedTokenizer,
labels: List[str],
model_type: str,
max_seq_length: Optional[int] = None,
overwrite_cache=False,
mode: Split = Split.train,
):
examples = token_classification_task.read_examples_from_file(data_dir, mode)
# TODO clean up all this to leverage built-in features of tokenizers
self.features = token_classification_task.convert_examples_to_features(
examples,
labels,
max_seq_length,
tokenizer,
cls_token_at_end=bool(model_type in ["xlnet"]),
# xlnet has a cls token at the end
cls_token=tokenizer.cls_token,
cls_token_segment_id=2 if model_type in ["xlnet"] else 0,
sep_token=tokenizer.sep_token,
sep_token_extra=False,
# roberta uses an extra separator b/w pairs of sentences, cf. github.com/pytorch/fairseq/commit/1684e166e3da03f5b600dbb7855cb98ddfcd0805
pad_on_left=bool(tokenizer.padding_side == "left"),
pad_token=tokenizer.pad_token_id,
pad_token_segment_id=tokenizer.pad_token_type_id,
pad_token_label_id=self.pad_token_label_id,
)
def gen():
for ex in self.features:
if ex.token_type_ids is None:
yield (
{"input_ids": ex.input_ids, "attention_mask": ex.attention_mask},
ex.label_ids,
)
else:
yield (
{
"input_ids": ex.input_ids,
"attention_mask": ex.attention_mask,
"token_type_ids": ex.token_type_ids,
},
ex.label_ids,
)
if "token_type_ids" not in tokenizer.model_input_names:
self.dataset = tf.data.Dataset.from_generator(
gen,
({"input_ids": tf.int32, "attention_mask": tf.int32}, tf.int64),
(
{"input_ids": tf.TensorShape([None]), "attention_mask": tf.TensorShape([None])},
tf.TensorShape([None]),
),
)
else:
self.dataset = tf.data.Dataset.from_generator(
gen,
({"input_ids": tf.int32, "attention_mask": tf.int32, "token_type_ids": tf.int32}, tf.int64),
(
{
"input_ids": tf.TensorShape([None]),
"attention_mask": tf.TensorShape([None]),
"token_type_ids": tf.TensorShape([None]),
},
tf.TensorShape([None]),
),
)
def get_dataset(self):
self.dataset = self.dataset.apply(tf.data.experimental.assert_cardinality(len(self.features)))
return self.dataset
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./.git/logs/refs/heads/main | 0000000000000000000000000000000000000000 976189a6df796a2ff442dd81b022626c840d8c27 jupyter <[email protected]> 1704826873 +0000 clone: from https://github.com/huggingface/transformers.git
| 0000000000000000000000000000000000000000 976189a6df796a2ff442dd81b022626c840d8c27 jupyter <[email protected]> 1704826873 +0000 clone: from https://github.com/huggingface/transformers.git
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/bertweet/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/pytorch/translation/README.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
## Translation
This directory contains examples for finetuning and evaluating transformers on translation tasks.
Please tag @patil-suraj with any issues/unexpected behaviors, or send a PR!
For deprecated `bertabs` instructions, see [`bertabs/README.md`](https://github.com/huggingface/transformers/blob/main/examples/research_projects/bertabs/README.md).
For the old `finetune_trainer.py` and related utils, see [`examples/legacy/seq2seq`](https://github.com/huggingface/transformers/blob/main/examples/legacy/seq2seq).
### Supported Architectures
- `BartForConditionalGeneration`
- `FSMTForConditionalGeneration` (translation only)
- `MBartForConditionalGeneration`
- `MarianMTModel`
- `PegasusForConditionalGeneration`
- `T5ForConditionalGeneration`
- `MT5ForConditionalGeneration`
`run_translation.py` is a lightweight examples of how to download and preprocess a dataset from the [🤗 Datasets](https://github.com/huggingface/datasets) library or use your own files (jsonlines or csv), then fine-tune one of the architectures above on it.
For custom datasets in `jsonlines` format please see: https://huggingface.co/docs/datasets/loading_datasets.html#json-files
and you also will find examples of these below.
## With Trainer
Here is an example of a translation fine-tuning with a MarianMT model:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path Helsinki-NLP/opus-mt-en-ro \
--do_train \
--do_eval \
--source_lang en \
--target_lang ro \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
MBart and some T5 models require special handling.
T5 models `t5-small`, `t5-base`, `t5-large`, `t5-3b` and `t5-11b` must use an additional argument: `--source_prefix "translate {source_lang} to {target_lang}"`. For example:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path t5-small \
--do_train \
--do_eval \
--source_lang en \
--target_lang ro \
--source_prefix "translate English to Romanian: " \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
If you get a terrible BLEU score, make sure that you didn't forget to use the `--source_prefix` argument.
For the aforementioned group of T5 models it's important to remember that if you switch to a different language pair, make sure to adjust the source and target values in all 3 language-specific command line argument: `--source_lang`, `--target_lang` and `--source_prefix`.
MBart models require a different format for `--source_lang` and `--target_lang` values, e.g. instead of `en` it expects `en_XX`, for `ro` it expects `ro_RO`. The full MBart specification for language codes can be found [here](https://huggingface.co/facebook/mbart-large-cc25). For example:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path facebook/mbart-large-en-ro \
--do_train \
--do_eval \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--source_lang en_XX \
--target_lang ro_RO \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
And here is how you would use the translation finetuning on your own files, after adjusting the
values for the arguments `--train_file`, `--validation_file` to match your setup:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path t5-small \
--do_train \
--do_eval \
--source_lang en \
--target_lang ro \
--source_prefix "translate English to Romanian: " \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--train_file path_to_jsonlines_file \
--validation_file path_to_jsonlines_file \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
The task of translation supports only custom JSONLINES files, with each line being a dictionary with a key `"translation"` and its value another dictionary whose keys is the language pair. For example:
```json
{ "translation": { "en": "Others have dismissed him as a joke.", "ro": "Alții l-au numit o glumă." } }
{ "translation": { "en": "And some are holding out for an implosion.", "ro": "Iar alții așteaptă implozia." } }
```
Here the languages are Romanian (`ro`) and English (`en`).
If you want to use a pre-processed dataset that leads to high BLEU scores, but for the `en-de` language pair, you can use `--dataset_name stas/wmt14-en-de-pre-processed`, as following:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path t5-small \
--do_train \
--do_eval \
--source_lang en \
--target_lang de \
--source_prefix "translate English to German: " \
--dataset_name stas/wmt14-en-de-pre-processed \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
## With Accelerate
Based on the script [`run_translation_no_trainer.py`](https://github.com/huggingface/transformers/blob/main/examples/pytorch/translation/run_translation_no_trainer.py).
Like `run_translation.py`, this script allows you to fine-tune any of the models supported on a
translation task, the main difference is that this
script exposes the bare training loop, to allow you to quickly experiment and add any customization you would like.
It offers less options than the script with `Trainer` (for instance you can easily change the options for the optimizer
or the dataloaders directly in the script) but still run in a distributed setup, on TPU and supports mixed precision by
the mean of the [🤗 `Accelerate`](https://github.com/huggingface/accelerate) library. You can use the script normally
after installing it:
```bash
pip install git+https://github.com/huggingface/accelerate
```
then
```bash
python run_translation_no_trainer.py \
--model_name_or_path Helsinki-NLP/opus-mt-en-ro \
--source_lang en \
--target_lang ro \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir ~/tmp/tst-translation
```
You can then use your usual launchers to run in it in a distributed environment, but the easiest way is to run
```bash
accelerate config
```
and reply to the questions asked. Then
```bash
accelerate test
```
that will check everything is ready for training. Finally, you can launch training with
```bash
accelerate launch run_translation_no_trainer.py \
--model_name_or_path Helsinki-NLP/opus-mt-en-ro \
--source_lang en \
--target_lang ro \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir ~/tmp/tst-translation
```
This command is the same and will work for:
- a CPU-only setup
- a setup with one GPU
- a distributed training with several GPUs (single or multi node)
- a training on TPUs
Note that this library is in alpha release so your feedback is more than welcome if you encounter any problem using it.
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
## Translation
This directory contains examples for finetuning and evaluating transformers on translation tasks.
Please tag @patil-suraj with any issues/unexpected behaviors, or send a PR!
For deprecated `bertabs` instructions, see [`bertabs/README.md`](https://github.com/huggingface/transformers/blob/main/examples/research_projects/bertabs/README.md).
For the old `finetune_trainer.py` and related utils, see [`examples/legacy/seq2seq`](https://github.com/huggingface/transformers/blob/main/examples/legacy/seq2seq).
### Supported Architectures
- `BartForConditionalGeneration`
- `FSMTForConditionalGeneration` (translation only)
- `MBartForConditionalGeneration`
- `MarianMTModel`
- `PegasusForConditionalGeneration`
- `T5ForConditionalGeneration`
- `MT5ForConditionalGeneration`
`run_translation.py` is a lightweight examples of how to download and preprocess a dataset from the [🤗 Datasets](https://github.com/huggingface/datasets) library or use your own files (jsonlines or csv), then fine-tune one of the architectures above on it.
For custom datasets in `jsonlines` format please see: https://huggingface.co/docs/datasets/loading_datasets.html#json-files
and you also will find examples of these below.
## With Trainer
Here is an example of a translation fine-tuning with a MarianMT model:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path Helsinki-NLP/opus-mt-en-ro \
--do_train \
--do_eval \
--source_lang en \
--target_lang ro \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
MBart and some T5 models require special handling.
T5 models `t5-small`, `t5-base`, `t5-large`, `t5-3b` and `t5-11b` must use an additional argument: `--source_prefix "translate {source_lang} to {target_lang}"`. For example:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path t5-small \
--do_train \
--do_eval \
--source_lang en \
--target_lang ro \
--source_prefix "translate English to Romanian: " \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
If you get a terrible BLEU score, make sure that you didn't forget to use the `--source_prefix` argument.
For the aforementioned group of T5 models it's important to remember that if you switch to a different language pair, make sure to adjust the source and target values in all 3 language-specific command line argument: `--source_lang`, `--target_lang` and `--source_prefix`.
MBart models require a different format for `--source_lang` and `--target_lang` values, e.g. instead of `en` it expects `en_XX`, for `ro` it expects `ro_RO`. The full MBart specification for language codes can be found [here](https://huggingface.co/facebook/mbart-large-cc25). For example:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path facebook/mbart-large-en-ro \
--do_train \
--do_eval \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--source_lang en_XX \
--target_lang ro_RO \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
And here is how you would use the translation finetuning on your own files, after adjusting the
values for the arguments `--train_file`, `--validation_file` to match your setup:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path t5-small \
--do_train \
--do_eval \
--source_lang en \
--target_lang ro \
--source_prefix "translate English to Romanian: " \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--train_file path_to_jsonlines_file \
--validation_file path_to_jsonlines_file \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
The task of translation supports only custom JSONLINES files, with each line being a dictionary with a key `"translation"` and its value another dictionary whose keys is the language pair. For example:
```json
{ "translation": { "en": "Others have dismissed him as a joke.", "ro": "Alții l-au numit o glumă." } }
{ "translation": { "en": "And some are holding out for an implosion.", "ro": "Iar alții așteaptă implozia." } }
```
Here the languages are Romanian (`ro`) and English (`en`).
If you want to use a pre-processed dataset that leads to high BLEU scores, but for the `en-de` language pair, you can use `--dataset_name stas/wmt14-en-de-pre-processed`, as following:
```bash
python examples/pytorch/translation/run_translation.py \
--model_name_or_path t5-small \
--do_train \
--do_eval \
--source_lang en \
--target_lang de \
--source_prefix "translate English to German: " \
--dataset_name stas/wmt14-en-de-pre-processed \
--output_dir /tmp/tst-translation \
--per_device_train_batch_size=4 \
--per_device_eval_batch_size=4 \
--overwrite_output_dir \
--predict_with_generate
```
## With Accelerate
Based on the script [`run_translation_no_trainer.py`](https://github.com/huggingface/transformers/blob/main/examples/pytorch/translation/run_translation_no_trainer.py).
Like `run_translation.py`, this script allows you to fine-tune any of the models supported on a
translation task, the main difference is that this
script exposes the bare training loop, to allow you to quickly experiment and add any customization you would like.
It offers less options than the script with `Trainer` (for instance you can easily change the options for the optimizer
or the dataloaders directly in the script) but still run in a distributed setup, on TPU and supports mixed precision by
the mean of the [🤗 `Accelerate`](https://github.com/huggingface/accelerate) library. You can use the script normally
after installing it:
```bash
pip install git+https://github.com/huggingface/accelerate
```
then
```bash
python run_translation_no_trainer.py \
--model_name_or_path Helsinki-NLP/opus-mt-en-ro \
--source_lang en \
--target_lang ro \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir ~/tmp/tst-translation
```
You can then use your usual launchers to run in it in a distributed environment, but the easiest way is to run
```bash
accelerate config
```
and reply to the questions asked. Then
```bash
accelerate test
```
that will check everything is ready for training. Finally, you can launch training with
```bash
accelerate launch run_translation_no_trainer.py \
--model_name_or_path Helsinki-NLP/opus-mt-en-ro \
--source_lang en \
--target_lang ro \
--dataset_name wmt16 \
--dataset_config_name ro-en \
--output_dir ~/tmp/tst-translation
```
This command is the same and will work for:
- a CPU-only setup
- a setup with one GPU
- a distributed training with several GPUs (single or multi node)
- a training on TPUs
Note that this library is in alpha release so your feedback is more than welcome if you encounter any problem using it.
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/en/model_doc/bort.mdx | <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# BORT
## Overview
The BORT model was proposed in [Optimal Subarchitecture Extraction for BERT](https://arxiv.org/abs/2010.10499) by
Adrian de Wynter and Daniel J. Perry. It is an optimal subset of architectural parameters for the BERT, which the
authors refer to as "Bort".
The abstract from the paper is the following:
*We extract an optimal subset of architectural parameters for the BERT architecture from Devlin et al. (2018) by
applying recent breakthroughs in algorithms for neural architecture search. This optimal subset, which we refer to as
"Bort", is demonstrably smaller, having an effective (that is, not counting the embedding layer) size of 5.5% the
original BERT-large architecture, and 16% of the net size. Bort is also able to be pretrained in 288 GPU hours, which
is 1.2% of the time required to pretrain the highest-performing BERT parametric architectural variant, RoBERTa-large
(Liu et al., 2019), and about 33% of that of the world-record, in GPU hours, required to train BERT-large on the same
hardware. It is also 7.9x faster on a CPU, as well as being better performing than other compressed variants of the
architecture, and some of the non-compressed variants: it obtains performance improvements of between 0.3% and 31%,
absolute, with respect to BERT-large, on multiple public natural language understanding (NLU) benchmarks.*
Tips:
- BORT's model architecture is based on BERT, so one can refer to [BERT's documentation page](bert) for the
model's API as well as usage examples.
- BORT uses the RoBERTa tokenizer instead of the BERT tokenizer, so one can refer to [RoBERTa's documentation page](roberta) for the tokenizer's API as well as usage examples.
- BORT requires a specific fine-tuning algorithm, called [Agora](https://adewynter.github.io/notes/bort_algorithms_and_applications.html#fine-tuning-with-algebraic-topology) ,
that is sadly not open-sourced yet. It would be very useful for the community, if someone tries to implement the
algorithm to make BORT fine-tuning work.
This model was contributed by [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/alexa/bort/).
| <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# BORT
## Overview
The BORT model was proposed in [Optimal Subarchitecture Extraction for BERT](https://arxiv.org/abs/2010.10499) by
Adrian de Wynter and Daniel J. Perry. It is an optimal subset of architectural parameters for the BERT, which the
authors refer to as "Bort".
The abstract from the paper is the following:
*We extract an optimal subset of architectural parameters for the BERT architecture from Devlin et al. (2018) by
applying recent breakthroughs in algorithms for neural architecture search. This optimal subset, which we refer to as
"Bort", is demonstrably smaller, having an effective (that is, not counting the embedding layer) size of 5.5% the
original BERT-large architecture, and 16% of the net size. Bort is also able to be pretrained in 288 GPU hours, which
is 1.2% of the time required to pretrain the highest-performing BERT parametric architectural variant, RoBERTa-large
(Liu et al., 2019), and about 33% of that of the world-record, in GPU hours, required to train BERT-large on the same
hardware. It is also 7.9x faster on a CPU, as well as being better performing than other compressed variants of the
architecture, and some of the non-compressed variants: it obtains performance improvements of between 0.3% and 31%,
absolute, with respect to BERT-large, on multiple public natural language understanding (NLU) benchmarks.*
Tips:
- BORT's model architecture is based on BERT, so one can refer to [BERT's documentation page](bert) for the
model's API as well as usage examples.
- BORT uses the RoBERTa tokenizer instead of the BERT tokenizer, so one can refer to [RoBERTa's documentation page](roberta) for the tokenizer's API as well as usage examples.
- BORT requires a specific fine-tuning algorithm, called [Agora](https://adewynter.github.io/notes/bort_algorithms_and_applications.html#fine-tuning-with-algebraic-topology) ,
that is sadly not open-sourced yet. It would be very useful for the community, if someone tries to implement the
algorithm to make BORT fine-tuning work.
This model was contributed by [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/alexa/bort/).
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/tapas/test_tokenization_tapas.py | # coding=utf-8
# Copyright 2020 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import inspect
import os
import shutil
import tempfile
import unittest
from typing import List
import numpy as np
import pandas as pd
from transformers import AddedToken
from transformers.models.tapas.tokenization_tapas import (
VOCAB_FILES_NAMES,
BasicTokenizer,
TapasTokenizer,
WordpieceTokenizer,
_is_control,
_is_punctuation,
_is_whitespace,
)
from transformers.testing_utils import (
is_pt_tf_cross_test,
require_pandas,
require_tensorflow_probability,
require_tokenizers,
require_torch,
slow,
)
from ...test_tokenization_common import TokenizerTesterMixin, filter_non_english, merge_model_tokenizer_mappings
@require_tokenizers
@require_pandas
class TapasTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
tokenizer_class = TapasTokenizer
test_rust_tokenizer = False
space_between_special_tokens = True
from_pretrained_filter = filter_non_english
test_seq2seq = False
def get_table(
self,
tokenizer: TapasTokenizer,
length=5,
):
toks = [tokenizer.decode([i], clean_up_tokenization_spaces=False) for i in range(len(tokenizer))]
if length == 0:
data = {}
else:
data = {toks[0]: [toks[tok] for tok in range(1, length)]}
table = pd.DataFrame.from_dict(data)
return table
def get_table_and_query(
self,
tokenizer: TapasTokenizer,
length=5,
):
toks = [tokenizer.decode([i], clean_up_tokenization_spaces=False) for i in range(len(tokenizer))]
table = self.get_table(tokenizer, length=length - 3)
query = " ".join(toks[:3])
return table, query
def get_clean_sequence(
self,
tokenizer: TapasTokenizer,
with_prefix_space=False,
max_length=20,
min_length=5,
empty_table: bool = False,
add_special_tokens: bool = True,
return_table_and_query: bool = False,
):
toks = [tokenizer.decode([i], clean_up_tokenization_spaces=False) for i in range(len(tokenizer))]
if empty_table:
table = pd.DataFrame.from_dict({})
query = " ".join(toks[:min_length])
else:
data = {toks[0]: [toks[tok] for tok in range(1, min_length - 3)]}
table = pd.DataFrame.from_dict(data)
query = " ".join(toks[:3])
output_ids = tokenizer.encode(table, query, add_special_tokens=add_special_tokens)
output_txt = tokenizer.decode(output_ids)
assert len(output_ids) >= min_length, "Update the code to generate the sequences so that they are larger"
assert len(output_ids) <= max_length, "Update the code to generate the sequences so that they are smaller"
if return_table_and_query:
return output_txt, output_ids, table, query
return output_txt, output_ids
def setUp(self):
super().setUp()
vocab_tokens = [
"[UNK]",
"[CLS]",
"[SEP]",
"[PAD]",
"[MASK]",
"want",
"##want",
"##ed",
"wa",
"un",
"runn",
"##ing",
",",
"low",
"lowest",
]
self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
def get_input_output_texts(self, tokenizer):
input_text = "UNwant\u00E9d,running"
output_text = "unwanted, running"
return input_text, output_text
@require_tensorflow_probability
@slow
def test_tf_encode_plus_sent_to_model(self):
from transformers import TF_MODEL_MAPPING, TOKENIZER_MAPPING
MODEL_TOKENIZER_MAPPING = merge_model_tokenizer_mappings(TF_MODEL_MAPPING, TOKENIZER_MAPPING)
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
if tokenizer.__class__ not in MODEL_TOKENIZER_MAPPING:
return
config_class, model_class = MODEL_TOKENIZER_MAPPING[tokenizer.__class__]
config = config_class()
if config.is_encoder_decoder or config.pad_token_id is None:
return
model = model_class(config)
# Make sure the model contains at least the full vocabulary size in its embedding matrix
self.assertGreaterEqual(model.config.vocab_size, len(tokenizer))
# Build sequence
first_ten_tokens = list(tokenizer.get_vocab().keys())[:10]
sequence = " ".join(first_ten_tokens)
table = self.get_table(tokenizer, length=0)
encoded_sequence = tokenizer.encode_plus(table, sequence, return_tensors="tf")
batch_encoded_sequence = tokenizer.batch_encode_plus(table, [sequence, sequence], return_tensors="tf")
# This should not fail
model(encoded_sequence)
model(batch_encoded_sequence)
def test_rust_and_python_full_tokenizers(self):
if not self.test_rust_tokenizer:
return
tokenizer = self.get_tokenizer()
rust_tokenizer = self.get_rust_tokenizer()
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
# With lower casing
tokenizer = self.get_tokenizer(do_lower_case=True)
rust_tokenizer = self.get_rust_tokenizer(do_lower_case=True)
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
def test_chinese(self):
tokenizer = BasicTokenizer()
self.assertListEqual(tokenizer.tokenize("ah\u535A\u63A8zz"), ["ah", "\u535A", "\u63A8", "zz"])
def test_basic_tokenizer_lower(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["hello", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hällo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["h\u00E9llo"])
def test_basic_tokenizer_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_default(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_no_lower(self):
tokenizer = BasicTokenizer(do_lower_case=False)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["HeLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HäLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HaLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_respects_never_split_tokens(self):
tokenizer = BasicTokenizer(do_lower_case=False, never_split=["[UNK]"])
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? [UNK]"), ["HeLLo", "!", "how", "Are", "yoU", "?", "[UNK]"]
)
def test_wordpiece_tokenizer(self):
vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing"]
vocab = {}
for i, token in enumerate(vocab_tokens):
vocab[token] = i
tokenizer = WordpieceTokenizer(vocab=vocab, unk_token="[UNK]")
self.assertListEqual(tokenizer.tokenize(""), [])
self.assertListEqual(tokenizer.tokenize("unwanted running"), ["un", "##want", "##ed", "runn", "##ing"])
self.assertListEqual(tokenizer.tokenize("unwantedX running"), ["[UNK]", "runn", "##ing"])
def test_is_whitespace(self):
self.assertTrue(_is_whitespace(" "))
self.assertTrue(_is_whitespace("\t"))
self.assertTrue(_is_whitespace("\r"))
self.assertTrue(_is_whitespace("\n"))
self.assertTrue(_is_whitespace("\u00A0"))
self.assertFalse(_is_whitespace("A"))
self.assertFalse(_is_whitespace("-"))
def test_is_control(self):
self.assertTrue(_is_control("\u0005"))
self.assertFalse(_is_control("A"))
self.assertFalse(_is_control(" "))
self.assertFalse(_is_control("\t"))
self.assertFalse(_is_control("\r"))
def test_is_punctuation(self):
self.assertTrue(_is_punctuation("-"))
self.assertTrue(_is_punctuation("$"))
self.assertTrue(_is_punctuation("`"))
self.assertTrue(_is_punctuation("."))
self.assertFalse(_is_punctuation("A"))
self.assertFalse(_is_punctuation(" "))
def test_clean_text(self):
tokenizer = self.get_tokenizer()
# Example taken from the issue https://github.com/huggingface/tokenizers/issues/340
self.assertListEqual(
[tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], ["[EMPTY]"], ["[UNK]"]]
)
@slow
def test_sequence_builders(self):
tokenizer = self.tokenizer_class.from_pretrained("google/tapas-base-finetuned-wtq")
empty_table = self.get_table(tokenizer, length=0)
table = self.get_table(tokenizer, length=10)
text = tokenizer.encode(table, add_special_tokens=False)
text_2 = tokenizer.encode(empty_table, "multi-sequence build", add_special_tokens=False)
encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)
assert encoded_pair == [101] + text + [102] + text_2
def test_offsets_with_special_characters(self):
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
sentence = f"A, naïve {tokenizer_r.mask_token} AllenNLP sentence."
tokens = tokenizer_r.encode_plus(
sentence,
return_attention_mask=False,
return_token_type_ids=False,
return_offsets_mapping=True,
add_special_tokens=True,
)
do_lower_case = tokenizer_r.do_lower_case if hasattr(tokenizer_r, "do_lower_case") else False
expected_results = (
[
((0, 0), tokenizer_r.cls_token),
((0, 1), "A"),
((1, 2), ","),
((3, 5), "na"),
((5, 6), "##ï"),
((6, 8), "##ve"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "Allen"),
((21, 23), "##NL"),
((23, 24), "##P"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
if not do_lower_case
else [
((0, 0), tokenizer_r.cls_token),
((0, 1), "a"),
((1, 2), ","),
((3, 8), "naive"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "allen"),
((21, 23), "##nl"),
((23, 24), "##p"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
)
self.assertEqual(
[e[1] for e in expected_results], tokenizer_r.convert_ids_to_tokens(tokens["input_ids"])
)
self.assertEqual([e[0] for e in expected_results], tokens["offset_mapping"])
def test_add_special_tokens(self):
tokenizers: List[TapasTokenizer] = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
input_table = self.get_table(tokenizer, length=0)
special_token = "[SPECIAL_TOKEN]"
tokenizer.add_special_tokens({"cls_token": special_token})
encoded_special_token = tokenizer.encode(input_table, special_token, add_special_tokens=False)
self.assertEqual(len(encoded_special_token), 1)
decoded = tokenizer.decode(encoded_special_token, skip_special_tokens=True)
self.assertTrue(special_token not in decoded)
def test_add_tokens_tokenizer(self):
tokenizers: List[TapasTokenizer] = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
vocab_size = tokenizer.vocab_size
all_size = len(tokenizer)
self.assertNotEqual(vocab_size, 0)
# We usually have added tokens from the start in tests because our vocab fixtures are
# smaller than the original vocabs - let's not assert this
# self.assertEqual(vocab_size, all_size)
new_toks = ["aaaaa bbbbbb", "cccccccccdddddddd"]
added_toks = tokenizer.add_tokens(new_toks)
vocab_size_2 = tokenizer.vocab_size
all_size_2 = len(tokenizer)
self.assertNotEqual(vocab_size_2, 0)
self.assertEqual(vocab_size, vocab_size_2)
self.assertEqual(added_toks, len(new_toks))
self.assertEqual(all_size_2, all_size + len(new_toks))
tokens = tokenizer.encode(table, "aaaaa bbbbbb low cccccccccdddddddd l", add_special_tokens=False)
self.assertGreaterEqual(len(tokens), 4)
self.assertGreater(tokens[0], tokenizer.vocab_size - 1)
self.assertGreater(tokens[-2], tokenizer.vocab_size - 1)
new_toks_2 = {"eos_token": ">>>>|||<||<<|<<", "pad_token": "<<<<<|||>|>>>>|>"}
added_toks_2 = tokenizer.add_special_tokens(new_toks_2)
vocab_size_3 = tokenizer.vocab_size
all_size_3 = len(tokenizer)
self.assertNotEqual(vocab_size_3, 0)
self.assertEqual(vocab_size, vocab_size_3)
self.assertEqual(added_toks_2, len(new_toks_2))
self.assertEqual(all_size_3, all_size_2 + len(new_toks_2))
tokens = tokenizer.encode(
table,
">>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l",
add_special_tokens=False,
)
self.assertGreaterEqual(len(tokens), 6)
self.assertGreater(tokens[0], tokenizer.vocab_size - 1)
self.assertGreater(tokens[0], tokens[1])
self.assertGreater(tokens[-2], tokenizer.vocab_size - 1)
self.assertGreater(tokens[-2], tokens[-3])
self.assertEqual(tokens[0], tokenizer.eos_token_id)
self.assertEqual(tokens[-2], tokenizer.pad_token_id)
@require_tokenizers
def test_encode_decode_with_spaces(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
new_toks = [AddedToken("[ABC]", normalized=False), AddedToken("[DEF]", normalized=False)]
tokenizer.add_tokens(new_toks)
input = "[ABC][DEF][ABC][DEF]"
if self.space_between_special_tokens:
output = "[ABC] [DEF] [ABC] [DEF]"
else:
output = input
encoded = tokenizer.encode(table, input, add_special_tokens=False)
decoded = tokenizer.decode(encoded, spaces_between_special_tokens=self.space_between_special_tokens)
self.assertIn(decoded, [output, output.lower()])
def test_encode_plus_with_padding(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequence = "Sequence"
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequence)
padding_size = 10
padding_idx = tokenizer.pad_token_id
token_type_padding_idx = tokenizer.pad_token_type_id
encoded_sequence = tokenizer.encode_plus(table, sequence, return_special_tokens_mask=True)
input_ids = encoded_sequence["input_ids"]
special_tokens_mask = encoded_sequence["special_tokens_mask"]
sequence_length = len(input_ids)
# Test 'longest' and 'no_padding' don't do anything
tokenizer.padding_side = "right"
not_padded_sequence = tokenizer.encode_plus(
table,
sequence,
padding=False,
return_special_tokens_mask=True,
)
not_padded_input_ids = not_padded_sequence["input_ids"]
not_padded_special_tokens_mask = not_padded_sequence["special_tokens_mask"]
not_padded_sequence_length = len(not_padded_input_ids)
assert sequence_length == not_padded_sequence_length
assert input_ids == not_padded_input_ids
assert special_tokens_mask == not_padded_special_tokens_mask
not_padded_sequence = tokenizer.encode_plus(
table,
sequence,
padding=False,
return_special_tokens_mask=True,
)
not_padded_input_ids = not_padded_sequence["input_ids"]
not_padded_special_tokens_mask = not_padded_sequence["special_tokens_mask"]
not_padded_sequence_length = len(not_padded_input_ids)
assert sequence_length == not_padded_sequence_length
assert input_ids == not_padded_input_ids
assert special_tokens_mask == not_padded_special_tokens_mask
# Test right padding
tokenizer.padding_side = "right"
right_padded_sequence = tokenizer.encode_plus(
table,
sequence,
max_length=sequence_length + padding_size,
padding="max_length",
return_special_tokens_mask=True,
)
right_padded_input_ids = right_padded_sequence["input_ids"]
right_padded_special_tokens_mask = right_padded_sequence["special_tokens_mask"]
right_padded_sequence_length = len(right_padded_input_ids)
assert sequence_length + padding_size == right_padded_sequence_length
assert input_ids + [padding_idx] * padding_size == right_padded_input_ids
assert special_tokens_mask + [1] * padding_size == right_padded_special_tokens_mask
# Test left padding
tokenizer.padding_side = "left"
left_padded_sequence = tokenizer.encode_plus(
table,
sequence,
max_length=sequence_length + padding_size,
padding="max_length",
return_special_tokens_mask=True,
)
left_padded_input_ids = left_padded_sequence["input_ids"]
left_padded_special_tokens_mask = left_padded_sequence["special_tokens_mask"]
left_padded_sequence_length = len(left_padded_input_ids)
assert sequence_length + padding_size == left_padded_sequence_length
assert [padding_idx] * padding_size + input_ids == left_padded_input_ids
assert [1] * padding_size + special_tokens_mask == left_padded_special_tokens_mask
if "token_type_ids" in tokenizer.model_input_names:
token_type_ids = encoded_sequence["token_type_ids"]
left_padded_token_type_ids = left_padded_sequence["token_type_ids"]
right_padded_token_type_ids = right_padded_sequence["token_type_ids"]
assert (
token_type_ids + [[token_type_padding_idx] * 7] * padding_size == right_padded_token_type_ids
)
assert [[token_type_padding_idx] * 7] * padding_size + token_type_ids == left_padded_token_type_ids
if "attention_mask" in tokenizer.model_input_names:
attention_mask = encoded_sequence["attention_mask"]
right_padded_attention_mask = right_padded_sequence["attention_mask"]
left_padded_attention_mask = left_padded_sequence["attention_mask"]
assert attention_mask + [0] * padding_size == right_padded_attention_mask
assert [0] * padding_size + attention_mask == left_padded_attention_mask
def test_internal_consistency(self):
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
input_text, output_text = self.get_input_output_texts(tokenizer)
tokens = tokenizer.tokenize(input_text)
ids = tokenizer.convert_tokens_to_ids(tokens)
ids_2 = tokenizer.encode(table, input_text, add_special_tokens=False)
self.assertListEqual(ids, ids_2)
tokens_2 = tokenizer.convert_ids_to_tokens(ids)
self.assertNotEqual(len(tokens_2), 0)
text_2 = tokenizer.decode(ids)
self.assertIsInstance(text_2, str)
self.assertEqual(text_2, output_text)
def test_mask_output(self):
tokenizers = self.get_tokenizers(fast=False, do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table, query = self.get_table_and_query(tokenizer)
if (
tokenizer.build_inputs_with_special_tokens.__qualname__.split(".")[0] != "PreTrainedTokenizer"
and "token_type_ids" in tokenizer.model_input_names
):
information = tokenizer.encode_plus(table, query, add_special_tokens=True)
sequences, mask = information["input_ids"], information["token_type_ids"]
self.assertEqual(len(sequences), len(mask))
@unittest.skip("TAPAS tokenizer only handles two sequences.")
def test_maximum_encoding_length_pair_input(self):
pass
@unittest.skip("TAPAS tokenizer only handles two sequences.")
def test_maximum_encoding_length_single_input(self):
pass
def test_number_of_added_tokens(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table, query = self.get_table_and_query(tokenizer)
sequences = tokenizer.encode(table, query, add_special_tokens=False)
attached_sequences = tokenizer.encode(table, query, add_special_tokens=True)
# Method is implemented (e.g. not GPT-2)
if len(attached_sequences) != 2:
self.assertEqual(
tokenizer.num_special_tokens_to_add(pair=True), len(attached_sequences) - len(sequences)
)
def test_padding_to_max_length(self):
"""We keep this test for backward compatibility but it should be removed when `pad_to_max_length` will be deprecated"""
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer)
sequence = "Sequence"
padding_size = 10
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequence)
padding_idx = tokenizer.pad_token_id
# Check that it correctly pads when a maximum length is specified along with the padding flag set to True
tokenizer.padding_side = "right"
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
# FIXME: the next line should be padding(max_length) to avoid warning
padded_sequence = tokenizer.encode(
table, sequence, max_length=sequence_length + padding_size, padding=True
)
padded_sequence_length = len(padded_sequence)
assert sequence_length + padding_size == padded_sequence_length
assert encoded_sequence + [padding_idx] * padding_size == padded_sequence
# Check that nothing is done when a maximum length is not specified
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
tokenizer.padding_side = "right"
padded_sequence_right = tokenizer.encode(table, sequence, pad_to_max_length=True)
padded_sequence_right_length = len(padded_sequence_right)
assert sequence_length == padded_sequence_right_length
assert encoded_sequence == padded_sequence_right
def test_call(self):
# Tests that all call wrap to encode_plus and batch_encode_plus
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
# Test not batched
table = self.get_table(tokenizer, length=0)
encoded_sequences_1 = tokenizer.encode_plus(table, sequences[0])
encoded_sequences_2 = tokenizer(table, sequences[0])
self.assertEqual(encoded_sequences_1, encoded_sequences_2)
# Test not batched pairs
table = self.get_table(tokenizer, length=10)
encoded_sequences_1 = tokenizer.encode_plus(table, sequences[1])
encoded_sequences_2 = tokenizer(table, sequences[1])
self.assertEqual(encoded_sequences_1, encoded_sequences_2)
# Test batched
table = self.get_table(tokenizer, length=0)
encoded_sequences_1 = tokenizer.batch_encode_plus(table, sequences)
encoded_sequences_2 = tokenizer(table, sequences)
self.assertEqual(encoded_sequences_1, encoded_sequences_2)
def test_batch_encode_plus_batch_sequence_length(self):
# Tests that all encoded values have the correct size
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
encoded_sequences = [tokenizer.encode_plus(table, sequence) for sequence in sequences]
encoded_sequences_batch = tokenizer.batch_encode_plus(table, sequences, padding=False)
self.assertListEqual(
encoded_sequences, self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch)
)
maximum_length = len(
max([encoded_sequence["input_ids"] for encoded_sequence in encoded_sequences], key=len)
)
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequences)
encoded_sequences_padded = [
tokenizer.encode_plus(table, sequence, max_length=maximum_length, padding="max_length")
for sequence in sequences
]
encoded_sequences_batch_padded = tokenizer.batch_encode_plus(table, sequences, padding=True)
self.assertListEqual(
encoded_sequences_padded,
self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch_padded),
)
# check 'longest' is unsensitive to a max length
encoded_sequences_batch_padded_1 = tokenizer.batch_encode_plus(table, sequences, padding=True)
encoded_sequences_batch_padded_2 = tokenizer.batch_encode_plus(
table, sequences, max_length=maximum_length + 10, padding="longest"
)
for key in encoded_sequences_batch_padded_1.keys():
self.assertListEqual(
encoded_sequences_batch_padded_1[key],
encoded_sequences_batch_padded_2[key],
)
# check 'no_padding' is unsensitive to a max length
encoded_sequences_batch_padded_1 = tokenizer.batch_encode_plus(table, sequences, padding=False)
encoded_sequences_batch_padded_2 = tokenizer.batch_encode_plus(
table, sequences, max_length=maximum_length + 10, padding=False
)
for key in encoded_sequences_batch_padded_1.keys():
self.assertListEqual(
encoded_sequences_batch_padded_1[key],
encoded_sequences_batch_padded_2[key],
)
@unittest.skip("batch_encode_plus does not handle overflowing tokens.")
def test_batch_encode_plus_overflowing_tokens(self):
pass
def test_batch_encode_plus_padding(self):
# Test that padded sequences are equivalent between batch_encode_plus and encode_plus
# Right padding tests
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
max_length = 100
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequences)
encoded_sequences = [
tokenizer.encode_plus(table, sequence, max_length=max_length, padding="max_length")
for sequence in sequences
]
encoded_sequences_batch = tokenizer.batch_encode_plus(
table, sequences, max_length=max_length, padding="max_length"
)
self.assertListEqual(
encoded_sequences, self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch)
)
# Left padding tests
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
tokenizer.padding_side = "left"
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
max_length = 100
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequences)
encoded_sequences = [
tokenizer.encode_plus(table, sequence, max_length=max_length, padding="max_length")
for sequence in sequences
]
encoded_sequences_batch = tokenizer.batch_encode_plus(
table, sequences, max_length=max_length, padding="max_length"
)
self.assertListEqual(
encoded_sequences, self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch)
)
def test_padding_to_multiple_of(self):
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
if tokenizer.pad_token is None:
self.skipTest("No padding token.")
else:
empty_tokens = tokenizer(table, padding=True, pad_to_multiple_of=8)
normal_tokens = tokenizer(table, "This is a sample input", padding=True, pad_to_multiple_of=8)
for key, value in empty_tokens.items():
self.assertEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
for key, value in normal_tokens.items():
self.assertEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
normal_tokens = tokenizer(table, "This", pad_to_multiple_of=8)
for key, value in normal_tokens.items():
self.assertNotEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
# Should also work with truncation
normal_tokens = tokenizer(table, "This", padding=True, truncation=True, pad_to_multiple_of=8)
for key, value in normal_tokens.items():
self.assertEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
@unittest.skip("TAPAS cannot handle `prepare_for_model` without passing by `encode_plus` or `batch_encode_plus`")
def test_prepare_for_model(self):
pass
def test_tokenizer_slow_store_full_signature(self):
signature = inspect.signature(self.tokenizer_class.__init__)
tokenizer = self.get_tokenizer()
for parameter_name, parameter in signature.parameters.items():
if parameter.default != inspect.Parameter.empty:
self.assertIn(parameter_name, tokenizer.init_kwargs)
def test_special_tokens_mask_input_pairs(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
sequence_0 = "Encode this."
empty_table = self.get_table(tokenizer, length=0)
table = self.get_table(tokenizer, length=10)
encoded_sequence = tokenizer.encode(empty_table, sequence_0, add_special_tokens=False)
encoded_sequence += tokenizer.encode(table, "", add_special_tokens=False)
encoded_sequence_dict = tokenizer.encode_plus(
table,
sequence_0,
add_special_tokens=True,
return_special_tokens_mask=True,
# add_prefix_space=False,
)
encoded_sequence_w_special = encoded_sequence_dict["input_ids"]
special_tokens_mask = encoded_sequence_dict["special_tokens_mask"]
self.assertEqual(len(special_tokens_mask), len(encoded_sequence_w_special))
filtered_sequence = [
(x if not special_tokens_mask[i] else None) for i, x in enumerate(encoded_sequence_w_special)
]
filtered_sequence = [x for x in filtered_sequence if x is not None]
self.assertEqual(encoded_sequence, filtered_sequence)
def test_special_tokens_mask(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequence_0 = "Encode this."
# Testing single inputs
encoded_sequence = tokenizer.encode(table, sequence_0, add_special_tokens=False)
encoded_sequence_dict = tokenizer.encode_plus(
table, sequence_0, add_special_tokens=True, return_special_tokens_mask=True
)
encoded_sequence_w_special = encoded_sequence_dict["input_ids"]
special_tokens_mask = encoded_sequence_dict["special_tokens_mask"]
self.assertEqual(len(special_tokens_mask), len(encoded_sequence_w_special))
filtered_sequence = [x for i, x in enumerate(encoded_sequence_w_special) if not special_tokens_mask[i]]
self.assertEqual(encoded_sequence, filtered_sequence)
def test_save_and_load_tokenizer(self):
# safety check on max_len default value so we are sure the test works
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
self.assertNotEqual(tokenizer.model_max_length, 42)
# Now let's start the test
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
# Isolate this from the other tests because we save additional tokens/etc
table = self.get_table(tokenizer, length=0)
tmpdirname = tempfile.mkdtemp()
sample_text = " He is very happy, UNwant\u00E9d,running"
before_tokens = tokenizer.encode(table, sample_text, add_special_tokens=False)
before_vocab = tokenizer.get_vocab()
tokenizer.save_pretrained(tmpdirname)
after_tokenizer = tokenizer.__class__.from_pretrained(tmpdirname)
after_tokens = after_tokenizer.encode(table, sample_text, add_special_tokens=False)
after_vocab = after_tokenizer.get_vocab()
self.assertListEqual(before_tokens, after_tokens)
self.assertDictEqual(before_vocab, after_vocab)
shutil.rmtree(tmpdirname)
@unittest.skip("Not implemented")
def test_right_and_left_truncation(self):
pass
def test_right_and_left_padding(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequence = "Sequence"
padding_size = 10
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequence)
padding_idx = tokenizer.pad_token_id
# RIGHT PADDING - Check that it correctly pads when a maximum length is specified along with the padding flag set to True
tokenizer.padding_side = "right"
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
padded_sequence = tokenizer.encode(
table, sequence, max_length=sequence_length + padding_size, padding="max_length"
)
padded_sequence_length = len(padded_sequence)
assert sequence_length + padding_size == padded_sequence_length
assert encoded_sequence + [padding_idx] * padding_size == padded_sequence
# LEFT PADDING - Check that it correctly pads when a maximum length is specified along with the padding flag set to True
tokenizer.padding_side = "left"
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
padded_sequence = tokenizer.encode(
table, sequence, max_length=sequence_length + padding_size, padding="max_length"
)
padded_sequence_length = len(padded_sequence)
assert sequence_length + padding_size == padded_sequence_length
assert [padding_idx] * padding_size + encoded_sequence == padded_sequence
# RIGHT & LEFT PADDING - Check that nothing is done for 'longest' and 'no_padding'
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
tokenizer.padding_side = "right"
padded_sequence_right = tokenizer.encode(table, sequence, padding=True)
padded_sequence_right_length = len(padded_sequence_right)
assert sequence_length == padded_sequence_right_length
assert encoded_sequence == padded_sequence_right
tokenizer.padding_side = "left"
padded_sequence_left = tokenizer.encode(table, sequence, padding="longest")
padded_sequence_left_length = len(padded_sequence_left)
assert sequence_length == padded_sequence_left_length
assert encoded_sequence == padded_sequence_left
tokenizer.padding_side = "right"
padded_sequence_right = tokenizer.encode(table, sequence)
padded_sequence_right_length = len(padded_sequence_right)
assert sequence_length == padded_sequence_right_length
assert encoded_sequence == padded_sequence_right
tokenizer.padding_side = "left"
padded_sequence_left = tokenizer.encode(table, sequence, padding=False)
padded_sequence_left_length = len(padded_sequence_left)
assert sequence_length == padded_sequence_left_length
assert encoded_sequence == padded_sequence_left
def test_token_type_ids(self):
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
empty_table = self.get_table(tokenizer, length=0)
seq_0 = "Test this method."
# We want to have sequence 0 and sequence 1 are tagged
# respectively with 0 and 1 token_ids
# (regardless of whether the model use token type ids)
# We use this assumption in the QA pipeline among other place
output = tokenizer(empty_table, seq_0, return_token_type_ids=True)
# Assert that the token type IDs have the same length as the input IDs
self.assertEqual(len(output["token_type_ids"]), len(output["input_ids"]))
# Assert that each token type ID has 7 values
self.assertTrue(all(len(token_type_ids) == 7 for token_type_ids in output["token_type_ids"]))
# Do the same test as modeling common.
self.assertIn(0, output["token_type_ids"][0])
@require_torch
@slow
def test_torch_encode_plus_sent_to_model(self):
import torch
from transformers import MODEL_MAPPING, TOKENIZER_MAPPING
MODEL_TOKENIZER_MAPPING = merge_model_tokenizer_mappings(MODEL_MAPPING, TOKENIZER_MAPPING)
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
if tokenizer.__class__ not in MODEL_TOKENIZER_MAPPING:
return
config_class, model_class = MODEL_TOKENIZER_MAPPING[tokenizer.__class__]
config = config_class()
if config.is_encoder_decoder or config.pad_token_id is None:
return
model = model_class(config)
# Make sure the model contains at least the full vocabulary size in its embedding matrix
is_using_common_embeddings = hasattr(model.get_input_embeddings(), "weight")
assert (
(model.get_input_embeddings().weight.shape[0] >= len(tokenizer))
if is_using_common_embeddings
else True
)
# Build sequence
first_ten_tokens = list(tokenizer.get_vocab().keys())[:10]
sequence = " ".join(first_ten_tokens)
table = self.get_table(tokenizer, length=0)
encoded_sequence = tokenizer.encode_plus(table, sequence, return_tensors="pt")
batch_encoded_sequence = tokenizer.batch_encode_plus(table, [sequence, sequence], return_tensors="pt")
# This should not fail
with torch.no_grad(): # saves some time
model(**encoded_sequence)
model(**batch_encoded_sequence)
@unittest.skip("TAPAS doesn't handle pre-tokenized inputs.")
def test_pretokenized_inputs(self):
pass
@slow
def test_tapas_truncation_integration_test(self):
data = {
"Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
"Age": ["56", "45", "59"],
"Number of movies": ["87", "53", "69"],
"Date of birth": ["18 december 1963", "11 november 1974", "6 may 1961"],
}
queries = [
"When was Brad Pitt born?",
"Which actor appeared in the least number of movies?",
"What is the average number of movies?",
]
table = pd.DataFrame.from_dict(data)
tokenizer = TapasTokenizer.from_pretrained("lysandre/tapas-temporary-repo", model_max_length=512)
for i in range(12):
# The table cannot even encode the headers, so raise an error
with self.assertRaises(ValueError):
tokenizer.encode(table=table, query=queries[0], max_length=i, truncation="drop_rows_to_fit")
for i in range(12, 512):
new_encoded_inputs = tokenizer.encode(
table=table, query=queries[0], max_length=i, truncation="drop_rows_to_fit"
)
# Ensure that the input IDs are less than the max length defined.
self.assertLessEqual(len(new_encoded_inputs), i)
tokenizer.model_max_length = 20
new_encoded_inputs = tokenizer.encode(table=table, query=queries[0], truncation=True)
dropped_encoded_inputs = tokenizer.encode(table=table, query=queries[0], truncation="drop_rows_to_fit")
# Ensure that the input IDs are still truncated when no max_length is specified
self.assertListEqual(new_encoded_inputs, dropped_encoded_inputs)
self.assertLessEqual(len(new_encoded_inputs), 20)
@slow
def test_min_max_question_length(self):
data = {
"Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
"Age": ["56", "45", "59"],
"Number of movies": ["87", "53", "69"],
"Date of birth": ["18 december 1963", "11 november 1974", "6 may 1961"],
}
queries = "When was Brad Pitt born?"
table = pd.DataFrame.from_dict(data)
# test max_question_length
tokenizer = TapasTokenizer.from_pretrained("lysandre/tapas-temporary-repo", max_question_length=2)
encoding = tokenizer(table=table, queries=queries)
# query should not be tokenized as it's longer than the specified max_question_length
expected_results = [101, 102]
self.assertListEqual(encoding.input_ids[:2], expected_results)
# test min_question_length
tokenizer = TapasTokenizer.from_pretrained("lysandre/tapas-temporary-repo", min_question_length=30)
encoding = tokenizer(table=table, queries=queries)
# query should not be tokenized as it's shorter than the specified min_question_length
expected_results = [101, 102]
self.assertListEqual(encoding.input_ids[:2], expected_results)
@is_pt_tf_cross_test
def test_batch_encode_plus_tensors(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
table = self.get_table(tokenizer, length=0)
# A Tensor cannot be build by sequences which are not the same size
self.assertRaises(ValueError, tokenizer.batch_encode_plus, table, sequences, return_tensors="pt")
self.assertRaises(ValueError, tokenizer.batch_encode_plus, table, sequences, return_tensors="tf")
if tokenizer.pad_token_id is None:
self.assertRaises(
ValueError,
tokenizer.batch_encode_plus,
table,
sequences,
padding=True,
return_tensors="pt",
)
self.assertRaises(
ValueError,
tokenizer.batch_encode_plus,
table,
sequences,
padding="longest",
return_tensors="tf",
)
else:
pytorch_tensor = tokenizer.batch_encode_plus(table, sequences, padding=True, return_tensors="pt")
tensorflow_tensor = tokenizer.batch_encode_plus(
table, sequences, padding="longest", return_tensors="tf"
)
encoded_sequences = tokenizer.batch_encode_plus(table, sequences, padding=True)
for key in encoded_sequences.keys():
pytorch_value = pytorch_tensor[key].tolist()
tensorflow_value = tensorflow_tensor[key].numpy().tolist()
encoded_value = encoded_sequences[key]
self.assertEqual(pytorch_value, tensorflow_value, encoded_value)
@slow
def test_tapas_integration_test(self):
data = {
"Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
"Age": ["56", "45", "59"],
"Number of movies": ["87", "53", "69"],
"Date of birth": ["18 december 1963", "11 november 1974", "6 may 1961"],
}
queries = [
"When was Brad Pitt born?",
"Which actor appeared in the least number of movies?",
"What is the average number of movies?",
]
table = pd.DataFrame.from_dict(data)
tokenizer = TapasTokenizer.from_pretrained("google/tapas-base-finetuned-wtq", model_max_length=512)
# fmt: off
expected_results = {'input_ids':[101,2043,2001,8226,15091,2141,1029,102,5889,2287,2193,1997,5691,3058,1997,4182,8226,15091,5179,6584,2324,2285,3699,14720,4487,6178,9488,3429,5187,2340,2281,3326,2577,18856,7828,3240,5354,6353,1020,2089,3777],'attention_mask':[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1],'token_type_ids':[[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[1,1,0,0,0,0,0],[1,2,0,0,0,0,0],[1,3,0,0,0,0,0],[1,3,0,0,0,0,0],[1,3,0,0,0,0,0],[1,4,0,0,0,0,0],[1,4,0,0,0,0,0],[1,4,0,0,0,0,0],[1,1,1,0,0,0,0],[1,1,1,0,0,0,0],[1,2,1,0,2,2,0],[1,3,1,0,3,1,0],[1,4,1,0,2,2,0],[1,4,1,0,2,2,0],[1,4,1,0,2,2,0],[1,1,2,0,0,0,0],[1,1,2,0,0,0,0],[1,1,2,0,0,0,0],[1,1,2,0,0,0,0],[1,2,2,0,1,3,0],[1,3,2,0,1,3,0],[1,4,2,0,3,1,0],[1,4,2,0,3,1,0],[1,4,2,0,3,1,0],[1,1,3,0,0,0,0],[1,1,3,0,0,0,0],[1,1,3,0,0,0,0],[1,1,3,0,0,0,0],[1,2,3,0,3,1,0],[1,3,3,0,2,2,0],[1,4,3,0,1,3,0],[1,4,3,0,1,3,0],[1,4,3,0,1,3,0]]} # noqa: E231
# fmt: on
new_encoded_inputs = tokenizer.encode_plus(table=table, query=queries[0])
self.assertDictEqual(dict(new_encoded_inputs), expected_results)
@slow
def test_full_tokenizer(self):
data = [
["Pos", "No", "Driver", "Team", "Laps", "Time/Retired", "Grid", "Points"],
["1", "32", "Patrick Carpentier", "Team Player's", "87", "1:48:11.023", "1", "22"],
["2", "1", "Bruno Junqueira", "Newman/Haas Racing", "87", "+0.8 secs", "2", "17"],
["3", "3", "Paul Tracy", "Team Player's", "87", "+28.6 secs", "3", "14"],
["4", "9", "Michel Jourdain, Jr.", "Team Rahal", "87", "+40.8 secs", "13", "12"],
["5", "34", "Mario Haberfeld", "Mi-Jack Conquest Racing", "87", "+42.1 secs", "6", "10"],
["6", "20", "Oriol Servia", "Patrick Racing", "87", "+1:00.2", "10", "8"],
["7", "51", "Adrian Fernandez", "Fernandez Racing", "87", "+1:01.4", "5", "6"],
["8", "12", "Jimmy Vasser", "American Spirit Team Johansson", "87", "+1:01.8", "8", "5"],
["9", "7", "Tiago Monteiro", "Fittipaldi-Dingman Racing", "86", "+ 1 Lap", "15", "4"],
["10", "55", "Mario Dominguez", "Herdez Competition", "86", "+ 1 Lap", "11", "3"],
["11", "27", "Bryan Herta", "PK Racing", "86", "+ 1 Lap", "12", "2"],
["12", "31", "Ryan Hunter-Reay", "American Spirit Team Johansson", "86", "+ 1 Lap", "17", "1"],
["13", "19", "Joel Camathias", "Dale Coyne Racing", "85", "+ 2 Laps", "18", "0"],
["14", "33", "Alex Tagliani", "Rocketsports Racing", "85", "+ 2 Laps", "14", "0"],
["15", "4", "Roberto Moreno", "Herdez Competition", "85", "+ 2 Laps", "9", "0"],
["16", "11", "Geoff Boss", "Dale Coyne Racing", "83", "Mechanical", "19", "0"],
["17", "2", "Sebastien Bourdais", "Newman/Haas Racing", "77", "Mechanical", "4", "0"],
["18", "15", "Darren Manning", "Walker Racing", "12", "Mechanical", "7", "0"],
["19", "5", "Rodolfo Lavin", "Walker Racing", "10", "Mechanical", "16", "0"],
]
query = "what were the drivers names?"
table = pd.DataFrame.from_records(data[1:], columns=data[0])
tokenizer = TapasTokenizer.from_pretrained("google/tapas-base-finetuned-wtq", model_max_length=512)
model_inputs = tokenizer(table, query, padding="max_length")
input_ids = model_inputs["input_ids"]
token_type_ids = np.array(model_inputs["token_type_ids"])
segment_ids = token_type_ids[:, 0]
column_ids = token_type_ids[:, 1]
row_ids = token_type_ids[:, 2]
# fmt: off
expected_results = {'input_ids':[101,2054,2020,1996,6853,3415,1029,102,13433,2015,2053,4062,2136,10876,2051,1013,3394,8370,2685,1015,3590,4754,29267,4765,3771,2136,2447,1005,1055,6584,1015,1024,4466,1024,2340,1012,6185,2509,1015,2570,1016,1015,10391,12022,4226,7895,10625,1013,22996,3868,6584,1009,1014,1012,1022,10819,2015,1016,2459,1017,1017,2703,10555,2136,2447,1005,1055,6584,1009,2654,1012,1020,10819,2015,1017,2403,1018,1023,8709,8183,3126,21351,2078,1010,3781,1012,2136,10958,8865,6584,1009,2871,1012,1022,10819,2015,2410,2260,1019,4090,7986,5292,5677,8151,2771,1011,2990,9187,3868,6584,1009,4413,1012,1015,10819,2015,1020,2184,1020,2322,2030,20282,14262,9035,4754,3868,6584,1009,1015,1024,4002,1012,1016,2184,1022,1021,4868,7918,12023,12023,3868,6584,1009,1015,1024,5890,1012,1018,1019,1020,1022,2260,5261,12436,18116,2137,4382,2136,26447,6584,1009,1015,1024,5890,1012,1022,1022,1019,1023,1021,27339,3995,10125,9711,4906,25101,24657,1011,22033,2386,3868,6564,1009,1015,5001,2321,1018,2184,4583,7986,14383,2075,29488,14906,9351,2971,6564,1009,1015,5001,2340,1017,2340,2676,8527,2014,2696,1052,2243,3868,6564,1009,1015,5001,2260,1016,2260,2861,4575,4477,1011,2128,4710,2137,4382,2136,26447,6564,1009,1015,5001,2459,1015,2410,2539,8963,11503,25457,3022,8512,2522,9654,3868,5594,1009,1016,10876,2324,1014,2403,3943,4074,6415,15204,2072,12496,25378,3868,5594,1009,1016,10876,2403,1014,2321,1018,10704,17921,14906,9351,2971,5594,1009,1016,10876,1023,1014,2385,2340,14915,5795,8512,2522,9654,3868,6640,6228,2539,1014,2459,1016,28328,8945,3126,21351,2015,10625,1013,22996,3868,6255,6228,1018,1014,2324,2321,12270,11956,5232,3868,2260,6228,1021,1014,2539,1019,8473,28027,2080,2474,6371,5232,3868,2184,6228,2385,1014,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],'column_ids':[0,0,0,0,0,0,0,0,1,1,2,3,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,5,6,6,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,3,3,3,3,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,3,4,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,4,4,4,4,5,6,7,8,1,2,3,3,3,3,3,4,4,4,4,5,6,7,8,1,2,3,3,4,4,5,6,7,8,1,2,3,3,3,3,3,4,4,5,6,7,8,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],'row_ids':[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,11,11,11,11,11,11,11,11,11,11,11,11,11,11,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,14,14,14,14,14,14,14,14,14,14,14,14,14,14,14,15,15,15,15,15,15,15,15,15,15,15,15,15,16,16,16,16,16,16,16,16,16,16,16,16,17,17,17,17,17,17,17,17,17,17,17,17,17,17,17,18,18,18,18,18,18,18,18,18,18,19,19,19,19,19,19,19,19,19,19,19,19,19,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],'segment_ids':[0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]} # noqa: E231
# fmt: on
self.assertListEqual(input_ids, expected_results["input_ids"])
self.assertListEqual(segment_ids.tolist(), expected_results["segment_ids"])
self.assertListEqual(column_ids.tolist(), expected_results["column_ids"])
self.assertListEqual(row_ids.tolist(), expected_results["row_ids"])
@unittest.skip("Skip this test while all models are still to be uploaded.")
def test_pretrained_model_lists(self):
pass
@unittest.skip("Doesn't support another framework than PyTorch")
def test_np_encode_plus_sent_to_model(self):
pass
| # coding=utf-8
# Copyright 2020 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import inspect
import os
import shutil
import tempfile
import unittest
from typing import List
import numpy as np
import pandas as pd
from transformers import AddedToken
from transformers.models.tapas.tokenization_tapas import (
VOCAB_FILES_NAMES,
BasicTokenizer,
TapasTokenizer,
WordpieceTokenizer,
_is_control,
_is_punctuation,
_is_whitespace,
)
from transformers.testing_utils import (
is_pt_tf_cross_test,
require_pandas,
require_tensorflow_probability,
require_tokenizers,
require_torch,
slow,
)
from ...test_tokenization_common import TokenizerTesterMixin, filter_non_english, merge_model_tokenizer_mappings
@require_tokenizers
@require_pandas
class TapasTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
tokenizer_class = TapasTokenizer
test_rust_tokenizer = False
space_between_special_tokens = True
from_pretrained_filter = filter_non_english
test_seq2seq = False
def get_table(
self,
tokenizer: TapasTokenizer,
length=5,
):
toks = [tokenizer.decode([i], clean_up_tokenization_spaces=False) for i in range(len(tokenizer))]
if length == 0:
data = {}
else:
data = {toks[0]: [toks[tok] for tok in range(1, length)]}
table = pd.DataFrame.from_dict(data)
return table
def get_table_and_query(
self,
tokenizer: TapasTokenizer,
length=5,
):
toks = [tokenizer.decode([i], clean_up_tokenization_spaces=False) for i in range(len(tokenizer))]
table = self.get_table(tokenizer, length=length - 3)
query = " ".join(toks[:3])
return table, query
def get_clean_sequence(
self,
tokenizer: TapasTokenizer,
with_prefix_space=False,
max_length=20,
min_length=5,
empty_table: bool = False,
add_special_tokens: bool = True,
return_table_and_query: bool = False,
):
toks = [tokenizer.decode([i], clean_up_tokenization_spaces=False) for i in range(len(tokenizer))]
if empty_table:
table = pd.DataFrame.from_dict({})
query = " ".join(toks[:min_length])
else:
data = {toks[0]: [toks[tok] for tok in range(1, min_length - 3)]}
table = pd.DataFrame.from_dict(data)
query = " ".join(toks[:3])
output_ids = tokenizer.encode(table, query, add_special_tokens=add_special_tokens)
output_txt = tokenizer.decode(output_ids)
assert len(output_ids) >= min_length, "Update the code to generate the sequences so that they are larger"
assert len(output_ids) <= max_length, "Update the code to generate the sequences so that they are smaller"
if return_table_and_query:
return output_txt, output_ids, table, query
return output_txt, output_ids
def setUp(self):
super().setUp()
vocab_tokens = [
"[UNK]",
"[CLS]",
"[SEP]",
"[PAD]",
"[MASK]",
"want",
"##want",
"##ed",
"wa",
"un",
"runn",
"##ing",
",",
"low",
"lowest",
]
self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
def get_input_output_texts(self, tokenizer):
input_text = "UNwant\u00E9d,running"
output_text = "unwanted, running"
return input_text, output_text
@require_tensorflow_probability
@slow
def test_tf_encode_plus_sent_to_model(self):
from transformers import TF_MODEL_MAPPING, TOKENIZER_MAPPING
MODEL_TOKENIZER_MAPPING = merge_model_tokenizer_mappings(TF_MODEL_MAPPING, TOKENIZER_MAPPING)
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
if tokenizer.__class__ not in MODEL_TOKENIZER_MAPPING:
return
config_class, model_class = MODEL_TOKENIZER_MAPPING[tokenizer.__class__]
config = config_class()
if config.is_encoder_decoder or config.pad_token_id is None:
return
model = model_class(config)
# Make sure the model contains at least the full vocabulary size in its embedding matrix
self.assertGreaterEqual(model.config.vocab_size, len(tokenizer))
# Build sequence
first_ten_tokens = list(tokenizer.get_vocab().keys())[:10]
sequence = " ".join(first_ten_tokens)
table = self.get_table(tokenizer, length=0)
encoded_sequence = tokenizer.encode_plus(table, sequence, return_tensors="tf")
batch_encoded_sequence = tokenizer.batch_encode_plus(table, [sequence, sequence], return_tensors="tf")
# This should not fail
model(encoded_sequence)
model(batch_encoded_sequence)
def test_rust_and_python_full_tokenizers(self):
if not self.test_rust_tokenizer:
return
tokenizer = self.get_tokenizer()
rust_tokenizer = self.get_rust_tokenizer()
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
# With lower casing
tokenizer = self.get_tokenizer(do_lower_case=True)
rust_tokenizer = self.get_rust_tokenizer(do_lower_case=True)
sequence = "UNwant\u00E9d,running"
tokens = tokenizer.tokenize(sequence)
rust_tokens = rust_tokenizer.tokenize(sequence)
self.assertListEqual(tokens, rust_tokens)
ids = tokenizer.encode(sequence, add_special_tokens=False)
rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
self.assertListEqual(ids, rust_ids)
rust_tokenizer = self.get_rust_tokenizer()
ids = tokenizer.encode(sequence)
rust_ids = rust_tokenizer.encode(sequence)
self.assertListEqual(ids, rust_ids)
def test_chinese(self):
tokenizer = BasicTokenizer()
self.assertListEqual(tokenizer.tokenize("ah\u535A\u63A8zz"), ["ah", "\u535A", "\u63A8", "zz"])
def test_basic_tokenizer_lower(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["hello", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hällo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["h\u00E9llo"])
def test_basic_tokenizer_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=True, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_lower_strip_accents_default(self):
tokenizer = BasicTokenizer(do_lower_case=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["hallo", "!", "how", "are", "you", "?"]
)
self.assertListEqual(tokenizer.tokenize("H\u00E9llo"), ["hello"])
def test_basic_tokenizer_no_lower(self):
tokenizer = BasicTokenizer(do_lower_case=False)
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? "), ["HeLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_false(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=False)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HäLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_no_lower_strip_accents_true(self):
tokenizer = BasicTokenizer(do_lower_case=False, strip_accents=True)
self.assertListEqual(
tokenizer.tokenize(" \tHäLLo!how \n Are yoU? "), ["HaLLo", "!", "how", "Are", "yoU", "?"]
)
def test_basic_tokenizer_respects_never_split_tokens(self):
tokenizer = BasicTokenizer(do_lower_case=False, never_split=["[UNK]"])
self.assertListEqual(
tokenizer.tokenize(" \tHeLLo!how \n Are yoU? [UNK]"), ["HeLLo", "!", "how", "Are", "yoU", "?", "[UNK]"]
)
def test_wordpiece_tokenizer(self):
vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing"]
vocab = {}
for i, token in enumerate(vocab_tokens):
vocab[token] = i
tokenizer = WordpieceTokenizer(vocab=vocab, unk_token="[UNK]")
self.assertListEqual(tokenizer.tokenize(""), [])
self.assertListEqual(tokenizer.tokenize("unwanted running"), ["un", "##want", "##ed", "runn", "##ing"])
self.assertListEqual(tokenizer.tokenize("unwantedX running"), ["[UNK]", "runn", "##ing"])
def test_is_whitespace(self):
self.assertTrue(_is_whitespace(" "))
self.assertTrue(_is_whitespace("\t"))
self.assertTrue(_is_whitespace("\r"))
self.assertTrue(_is_whitespace("\n"))
self.assertTrue(_is_whitespace("\u00A0"))
self.assertFalse(_is_whitespace("A"))
self.assertFalse(_is_whitespace("-"))
def test_is_control(self):
self.assertTrue(_is_control("\u0005"))
self.assertFalse(_is_control("A"))
self.assertFalse(_is_control(" "))
self.assertFalse(_is_control("\t"))
self.assertFalse(_is_control("\r"))
def test_is_punctuation(self):
self.assertTrue(_is_punctuation("-"))
self.assertTrue(_is_punctuation("$"))
self.assertTrue(_is_punctuation("`"))
self.assertTrue(_is_punctuation("."))
self.assertFalse(_is_punctuation("A"))
self.assertFalse(_is_punctuation(" "))
def test_clean_text(self):
tokenizer = self.get_tokenizer()
# Example taken from the issue https://github.com/huggingface/tokenizers/issues/340
self.assertListEqual(
[tokenizer.tokenize(t) for t in ["Test", "\xad", "test"]], [["[UNK]"], ["[EMPTY]"], ["[UNK]"]]
)
@slow
def test_sequence_builders(self):
tokenizer = self.tokenizer_class.from_pretrained("google/tapas-base-finetuned-wtq")
empty_table = self.get_table(tokenizer, length=0)
table = self.get_table(tokenizer, length=10)
text = tokenizer.encode(table, add_special_tokens=False)
text_2 = tokenizer.encode(empty_table, "multi-sequence build", add_special_tokens=False)
encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)
assert encoded_pair == [101] + text + [102] + text_2
def test_offsets_with_special_characters(self):
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"{tokenizer.__class__.__name__} ({pretrained_name})"):
tokenizer_r = self.rust_tokenizer_class.from_pretrained(pretrained_name, **kwargs)
sentence = f"A, naïve {tokenizer_r.mask_token} AllenNLP sentence."
tokens = tokenizer_r.encode_plus(
sentence,
return_attention_mask=False,
return_token_type_ids=False,
return_offsets_mapping=True,
add_special_tokens=True,
)
do_lower_case = tokenizer_r.do_lower_case if hasattr(tokenizer_r, "do_lower_case") else False
expected_results = (
[
((0, 0), tokenizer_r.cls_token),
((0, 1), "A"),
((1, 2), ","),
((3, 5), "na"),
((5, 6), "##ï"),
((6, 8), "##ve"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "Allen"),
((21, 23), "##NL"),
((23, 24), "##P"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
if not do_lower_case
else [
((0, 0), tokenizer_r.cls_token),
((0, 1), "a"),
((1, 2), ","),
((3, 8), "naive"),
((9, 15), tokenizer_r.mask_token),
((16, 21), "allen"),
((21, 23), "##nl"),
((23, 24), "##p"),
((25, 33), "sentence"),
((33, 34), "."),
((0, 0), tokenizer_r.sep_token),
]
)
self.assertEqual(
[e[1] for e in expected_results], tokenizer_r.convert_ids_to_tokens(tokens["input_ids"])
)
self.assertEqual([e[0] for e in expected_results], tokens["offset_mapping"])
def test_add_special_tokens(self):
tokenizers: List[TapasTokenizer] = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
input_table = self.get_table(tokenizer, length=0)
special_token = "[SPECIAL_TOKEN]"
tokenizer.add_special_tokens({"cls_token": special_token})
encoded_special_token = tokenizer.encode(input_table, special_token, add_special_tokens=False)
self.assertEqual(len(encoded_special_token), 1)
decoded = tokenizer.decode(encoded_special_token, skip_special_tokens=True)
self.assertTrue(special_token not in decoded)
def test_add_tokens_tokenizer(self):
tokenizers: List[TapasTokenizer] = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
vocab_size = tokenizer.vocab_size
all_size = len(tokenizer)
self.assertNotEqual(vocab_size, 0)
# We usually have added tokens from the start in tests because our vocab fixtures are
# smaller than the original vocabs - let's not assert this
# self.assertEqual(vocab_size, all_size)
new_toks = ["aaaaa bbbbbb", "cccccccccdddddddd"]
added_toks = tokenizer.add_tokens(new_toks)
vocab_size_2 = tokenizer.vocab_size
all_size_2 = len(tokenizer)
self.assertNotEqual(vocab_size_2, 0)
self.assertEqual(vocab_size, vocab_size_2)
self.assertEqual(added_toks, len(new_toks))
self.assertEqual(all_size_2, all_size + len(new_toks))
tokens = tokenizer.encode(table, "aaaaa bbbbbb low cccccccccdddddddd l", add_special_tokens=False)
self.assertGreaterEqual(len(tokens), 4)
self.assertGreater(tokens[0], tokenizer.vocab_size - 1)
self.assertGreater(tokens[-2], tokenizer.vocab_size - 1)
new_toks_2 = {"eos_token": ">>>>|||<||<<|<<", "pad_token": "<<<<<|||>|>>>>|>"}
added_toks_2 = tokenizer.add_special_tokens(new_toks_2)
vocab_size_3 = tokenizer.vocab_size
all_size_3 = len(tokenizer)
self.assertNotEqual(vocab_size_3, 0)
self.assertEqual(vocab_size, vocab_size_3)
self.assertEqual(added_toks_2, len(new_toks_2))
self.assertEqual(all_size_3, all_size_2 + len(new_toks_2))
tokens = tokenizer.encode(
table,
">>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l",
add_special_tokens=False,
)
self.assertGreaterEqual(len(tokens), 6)
self.assertGreater(tokens[0], tokenizer.vocab_size - 1)
self.assertGreater(tokens[0], tokens[1])
self.assertGreater(tokens[-2], tokenizer.vocab_size - 1)
self.assertGreater(tokens[-2], tokens[-3])
self.assertEqual(tokens[0], tokenizer.eos_token_id)
self.assertEqual(tokens[-2], tokenizer.pad_token_id)
@require_tokenizers
def test_encode_decode_with_spaces(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
new_toks = [AddedToken("[ABC]", normalized=False), AddedToken("[DEF]", normalized=False)]
tokenizer.add_tokens(new_toks)
input = "[ABC][DEF][ABC][DEF]"
if self.space_between_special_tokens:
output = "[ABC] [DEF] [ABC] [DEF]"
else:
output = input
encoded = tokenizer.encode(table, input, add_special_tokens=False)
decoded = tokenizer.decode(encoded, spaces_between_special_tokens=self.space_between_special_tokens)
self.assertIn(decoded, [output, output.lower()])
def test_encode_plus_with_padding(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequence = "Sequence"
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequence)
padding_size = 10
padding_idx = tokenizer.pad_token_id
token_type_padding_idx = tokenizer.pad_token_type_id
encoded_sequence = tokenizer.encode_plus(table, sequence, return_special_tokens_mask=True)
input_ids = encoded_sequence["input_ids"]
special_tokens_mask = encoded_sequence["special_tokens_mask"]
sequence_length = len(input_ids)
# Test 'longest' and 'no_padding' don't do anything
tokenizer.padding_side = "right"
not_padded_sequence = tokenizer.encode_plus(
table,
sequence,
padding=False,
return_special_tokens_mask=True,
)
not_padded_input_ids = not_padded_sequence["input_ids"]
not_padded_special_tokens_mask = not_padded_sequence["special_tokens_mask"]
not_padded_sequence_length = len(not_padded_input_ids)
assert sequence_length == not_padded_sequence_length
assert input_ids == not_padded_input_ids
assert special_tokens_mask == not_padded_special_tokens_mask
not_padded_sequence = tokenizer.encode_plus(
table,
sequence,
padding=False,
return_special_tokens_mask=True,
)
not_padded_input_ids = not_padded_sequence["input_ids"]
not_padded_special_tokens_mask = not_padded_sequence["special_tokens_mask"]
not_padded_sequence_length = len(not_padded_input_ids)
assert sequence_length == not_padded_sequence_length
assert input_ids == not_padded_input_ids
assert special_tokens_mask == not_padded_special_tokens_mask
# Test right padding
tokenizer.padding_side = "right"
right_padded_sequence = tokenizer.encode_plus(
table,
sequence,
max_length=sequence_length + padding_size,
padding="max_length",
return_special_tokens_mask=True,
)
right_padded_input_ids = right_padded_sequence["input_ids"]
right_padded_special_tokens_mask = right_padded_sequence["special_tokens_mask"]
right_padded_sequence_length = len(right_padded_input_ids)
assert sequence_length + padding_size == right_padded_sequence_length
assert input_ids + [padding_idx] * padding_size == right_padded_input_ids
assert special_tokens_mask + [1] * padding_size == right_padded_special_tokens_mask
# Test left padding
tokenizer.padding_side = "left"
left_padded_sequence = tokenizer.encode_plus(
table,
sequence,
max_length=sequence_length + padding_size,
padding="max_length",
return_special_tokens_mask=True,
)
left_padded_input_ids = left_padded_sequence["input_ids"]
left_padded_special_tokens_mask = left_padded_sequence["special_tokens_mask"]
left_padded_sequence_length = len(left_padded_input_ids)
assert sequence_length + padding_size == left_padded_sequence_length
assert [padding_idx] * padding_size + input_ids == left_padded_input_ids
assert [1] * padding_size + special_tokens_mask == left_padded_special_tokens_mask
if "token_type_ids" in tokenizer.model_input_names:
token_type_ids = encoded_sequence["token_type_ids"]
left_padded_token_type_ids = left_padded_sequence["token_type_ids"]
right_padded_token_type_ids = right_padded_sequence["token_type_ids"]
assert (
token_type_ids + [[token_type_padding_idx] * 7] * padding_size == right_padded_token_type_ids
)
assert [[token_type_padding_idx] * 7] * padding_size + token_type_ids == left_padded_token_type_ids
if "attention_mask" in tokenizer.model_input_names:
attention_mask = encoded_sequence["attention_mask"]
right_padded_attention_mask = right_padded_sequence["attention_mask"]
left_padded_attention_mask = left_padded_sequence["attention_mask"]
assert attention_mask + [0] * padding_size == right_padded_attention_mask
assert [0] * padding_size + attention_mask == left_padded_attention_mask
def test_internal_consistency(self):
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
input_text, output_text = self.get_input_output_texts(tokenizer)
tokens = tokenizer.tokenize(input_text)
ids = tokenizer.convert_tokens_to_ids(tokens)
ids_2 = tokenizer.encode(table, input_text, add_special_tokens=False)
self.assertListEqual(ids, ids_2)
tokens_2 = tokenizer.convert_ids_to_tokens(ids)
self.assertNotEqual(len(tokens_2), 0)
text_2 = tokenizer.decode(ids)
self.assertIsInstance(text_2, str)
self.assertEqual(text_2, output_text)
def test_mask_output(self):
tokenizers = self.get_tokenizers(fast=False, do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table, query = self.get_table_and_query(tokenizer)
if (
tokenizer.build_inputs_with_special_tokens.__qualname__.split(".")[0] != "PreTrainedTokenizer"
and "token_type_ids" in tokenizer.model_input_names
):
information = tokenizer.encode_plus(table, query, add_special_tokens=True)
sequences, mask = information["input_ids"], information["token_type_ids"]
self.assertEqual(len(sequences), len(mask))
@unittest.skip("TAPAS tokenizer only handles two sequences.")
def test_maximum_encoding_length_pair_input(self):
pass
@unittest.skip("TAPAS tokenizer only handles two sequences.")
def test_maximum_encoding_length_single_input(self):
pass
def test_number_of_added_tokens(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table, query = self.get_table_and_query(tokenizer)
sequences = tokenizer.encode(table, query, add_special_tokens=False)
attached_sequences = tokenizer.encode(table, query, add_special_tokens=True)
# Method is implemented (e.g. not GPT-2)
if len(attached_sequences) != 2:
self.assertEqual(
tokenizer.num_special_tokens_to_add(pair=True), len(attached_sequences) - len(sequences)
)
def test_padding_to_max_length(self):
"""We keep this test for backward compatibility but it should be removed when `pad_to_max_length` will be deprecated"""
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer)
sequence = "Sequence"
padding_size = 10
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequence)
padding_idx = tokenizer.pad_token_id
# Check that it correctly pads when a maximum length is specified along with the padding flag set to True
tokenizer.padding_side = "right"
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
# FIXME: the next line should be padding(max_length) to avoid warning
padded_sequence = tokenizer.encode(
table, sequence, max_length=sequence_length + padding_size, padding=True
)
padded_sequence_length = len(padded_sequence)
assert sequence_length + padding_size == padded_sequence_length
assert encoded_sequence + [padding_idx] * padding_size == padded_sequence
# Check that nothing is done when a maximum length is not specified
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
tokenizer.padding_side = "right"
padded_sequence_right = tokenizer.encode(table, sequence, pad_to_max_length=True)
padded_sequence_right_length = len(padded_sequence_right)
assert sequence_length == padded_sequence_right_length
assert encoded_sequence == padded_sequence_right
def test_call(self):
# Tests that all call wrap to encode_plus and batch_encode_plus
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
# Test not batched
table = self.get_table(tokenizer, length=0)
encoded_sequences_1 = tokenizer.encode_plus(table, sequences[0])
encoded_sequences_2 = tokenizer(table, sequences[0])
self.assertEqual(encoded_sequences_1, encoded_sequences_2)
# Test not batched pairs
table = self.get_table(tokenizer, length=10)
encoded_sequences_1 = tokenizer.encode_plus(table, sequences[1])
encoded_sequences_2 = tokenizer(table, sequences[1])
self.assertEqual(encoded_sequences_1, encoded_sequences_2)
# Test batched
table = self.get_table(tokenizer, length=0)
encoded_sequences_1 = tokenizer.batch_encode_plus(table, sequences)
encoded_sequences_2 = tokenizer(table, sequences)
self.assertEqual(encoded_sequences_1, encoded_sequences_2)
def test_batch_encode_plus_batch_sequence_length(self):
# Tests that all encoded values have the correct size
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
encoded_sequences = [tokenizer.encode_plus(table, sequence) for sequence in sequences]
encoded_sequences_batch = tokenizer.batch_encode_plus(table, sequences, padding=False)
self.assertListEqual(
encoded_sequences, self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch)
)
maximum_length = len(
max([encoded_sequence["input_ids"] for encoded_sequence in encoded_sequences], key=len)
)
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequences)
encoded_sequences_padded = [
tokenizer.encode_plus(table, sequence, max_length=maximum_length, padding="max_length")
for sequence in sequences
]
encoded_sequences_batch_padded = tokenizer.batch_encode_plus(table, sequences, padding=True)
self.assertListEqual(
encoded_sequences_padded,
self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch_padded),
)
# check 'longest' is unsensitive to a max length
encoded_sequences_batch_padded_1 = tokenizer.batch_encode_plus(table, sequences, padding=True)
encoded_sequences_batch_padded_2 = tokenizer.batch_encode_plus(
table, sequences, max_length=maximum_length + 10, padding="longest"
)
for key in encoded_sequences_batch_padded_1.keys():
self.assertListEqual(
encoded_sequences_batch_padded_1[key],
encoded_sequences_batch_padded_2[key],
)
# check 'no_padding' is unsensitive to a max length
encoded_sequences_batch_padded_1 = tokenizer.batch_encode_plus(table, sequences, padding=False)
encoded_sequences_batch_padded_2 = tokenizer.batch_encode_plus(
table, sequences, max_length=maximum_length + 10, padding=False
)
for key in encoded_sequences_batch_padded_1.keys():
self.assertListEqual(
encoded_sequences_batch_padded_1[key],
encoded_sequences_batch_padded_2[key],
)
@unittest.skip("batch_encode_plus does not handle overflowing tokens.")
def test_batch_encode_plus_overflowing_tokens(self):
pass
def test_batch_encode_plus_padding(self):
# Test that padded sequences are equivalent between batch_encode_plus and encode_plus
# Right padding tests
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
max_length = 100
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequences)
encoded_sequences = [
tokenizer.encode_plus(table, sequence, max_length=max_length, padding="max_length")
for sequence in sequences
]
encoded_sequences_batch = tokenizer.batch_encode_plus(
table, sequences, max_length=max_length, padding="max_length"
)
self.assertListEqual(
encoded_sequences, self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch)
)
# Left padding tests
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
tokenizer.padding_side = "left"
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
max_length = 100
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequences)
encoded_sequences = [
tokenizer.encode_plus(table, sequence, max_length=max_length, padding="max_length")
for sequence in sequences
]
encoded_sequences_batch = tokenizer.batch_encode_plus(
table, sequences, max_length=max_length, padding="max_length"
)
self.assertListEqual(
encoded_sequences, self.convert_batch_encode_plus_format_to_encode_plus(encoded_sequences_batch)
)
def test_padding_to_multiple_of(self):
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
if tokenizer.pad_token is None:
self.skipTest("No padding token.")
else:
empty_tokens = tokenizer(table, padding=True, pad_to_multiple_of=8)
normal_tokens = tokenizer(table, "This is a sample input", padding=True, pad_to_multiple_of=8)
for key, value in empty_tokens.items():
self.assertEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
for key, value in normal_tokens.items():
self.assertEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
normal_tokens = tokenizer(table, "This", pad_to_multiple_of=8)
for key, value in normal_tokens.items():
self.assertNotEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
# Should also work with truncation
normal_tokens = tokenizer(table, "This", padding=True, truncation=True, pad_to_multiple_of=8)
for key, value in normal_tokens.items():
self.assertEqual(len(value) % 8, 0, f"BatchEncoding.{key} is not multiple of 8")
@unittest.skip("TAPAS cannot handle `prepare_for_model` without passing by `encode_plus` or `batch_encode_plus`")
def test_prepare_for_model(self):
pass
def test_tokenizer_slow_store_full_signature(self):
signature = inspect.signature(self.tokenizer_class.__init__)
tokenizer = self.get_tokenizer()
for parameter_name, parameter in signature.parameters.items():
if parameter.default != inspect.Parameter.empty:
self.assertIn(parameter_name, tokenizer.init_kwargs)
def test_special_tokens_mask_input_pairs(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
sequence_0 = "Encode this."
empty_table = self.get_table(tokenizer, length=0)
table = self.get_table(tokenizer, length=10)
encoded_sequence = tokenizer.encode(empty_table, sequence_0, add_special_tokens=False)
encoded_sequence += tokenizer.encode(table, "", add_special_tokens=False)
encoded_sequence_dict = tokenizer.encode_plus(
table,
sequence_0,
add_special_tokens=True,
return_special_tokens_mask=True,
# add_prefix_space=False,
)
encoded_sequence_w_special = encoded_sequence_dict["input_ids"]
special_tokens_mask = encoded_sequence_dict["special_tokens_mask"]
self.assertEqual(len(special_tokens_mask), len(encoded_sequence_w_special))
filtered_sequence = [
(x if not special_tokens_mask[i] else None) for i, x in enumerate(encoded_sequence_w_special)
]
filtered_sequence = [x for x in filtered_sequence if x is not None]
self.assertEqual(encoded_sequence, filtered_sequence)
def test_special_tokens_mask(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequence_0 = "Encode this."
# Testing single inputs
encoded_sequence = tokenizer.encode(table, sequence_0, add_special_tokens=False)
encoded_sequence_dict = tokenizer.encode_plus(
table, sequence_0, add_special_tokens=True, return_special_tokens_mask=True
)
encoded_sequence_w_special = encoded_sequence_dict["input_ids"]
special_tokens_mask = encoded_sequence_dict["special_tokens_mask"]
self.assertEqual(len(special_tokens_mask), len(encoded_sequence_w_special))
filtered_sequence = [x for i, x in enumerate(encoded_sequence_w_special) if not special_tokens_mask[i]]
self.assertEqual(encoded_sequence, filtered_sequence)
def test_save_and_load_tokenizer(self):
# safety check on max_len default value so we are sure the test works
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
self.assertNotEqual(tokenizer.model_max_length, 42)
# Now let's start the test
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
# Isolate this from the other tests because we save additional tokens/etc
table = self.get_table(tokenizer, length=0)
tmpdirname = tempfile.mkdtemp()
sample_text = " He is very happy, UNwant\u00E9d,running"
before_tokens = tokenizer.encode(table, sample_text, add_special_tokens=False)
before_vocab = tokenizer.get_vocab()
tokenizer.save_pretrained(tmpdirname)
after_tokenizer = tokenizer.__class__.from_pretrained(tmpdirname)
after_tokens = after_tokenizer.encode(table, sample_text, add_special_tokens=False)
after_vocab = after_tokenizer.get_vocab()
self.assertListEqual(before_tokens, after_tokens)
self.assertDictEqual(before_vocab, after_vocab)
shutil.rmtree(tmpdirname)
@unittest.skip("Not implemented")
def test_right_and_left_truncation(self):
pass
def test_right_and_left_padding(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
table = self.get_table(tokenizer, length=0)
sequence = "Sequence"
padding_size = 10
# check correct behaviour if no pad_token_id exists and add it eventually
self._check_no_pad_token_padding(tokenizer, sequence)
padding_idx = tokenizer.pad_token_id
# RIGHT PADDING - Check that it correctly pads when a maximum length is specified along with the padding flag set to True
tokenizer.padding_side = "right"
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
padded_sequence = tokenizer.encode(
table, sequence, max_length=sequence_length + padding_size, padding="max_length"
)
padded_sequence_length = len(padded_sequence)
assert sequence_length + padding_size == padded_sequence_length
assert encoded_sequence + [padding_idx] * padding_size == padded_sequence
# LEFT PADDING - Check that it correctly pads when a maximum length is specified along with the padding flag set to True
tokenizer.padding_side = "left"
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
padded_sequence = tokenizer.encode(
table, sequence, max_length=sequence_length + padding_size, padding="max_length"
)
padded_sequence_length = len(padded_sequence)
assert sequence_length + padding_size == padded_sequence_length
assert [padding_idx] * padding_size + encoded_sequence == padded_sequence
# RIGHT & LEFT PADDING - Check that nothing is done for 'longest' and 'no_padding'
encoded_sequence = tokenizer.encode(table, sequence)
sequence_length = len(encoded_sequence)
tokenizer.padding_side = "right"
padded_sequence_right = tokenizer.encode(table, sequence, padding=True)
padded_sequence_right_length = len(padded_sequence_right)
assert sequence_length == padded_sequence_right_length
assert encoded_sequence == padded_sequence_right
tokenizer.padding_side = "left"
padded_sequence_left = tokenizer.encode(table, sequence, padding="longest")
padded_sequence_left_length = len(padded_sequence_left)
assert sequence_length == padded_sequence_left_length
assert encoded_sequence == padded_sequence_left
tokenizer.padding_side = "right"
padded_sequence_right = tokenizer.encode(table, sequence)
padded_sequence_right_length = len(padded_sequence_right)
assert sequence_length == padded_sequence_right_length
assert encoded_sequence == padded_sequence_right
tokenizer.padding_side = "left"
padded_sequence_left = tokenizer.encode(table, sequence, padding=False)
padded_sequence_left_length = len(padded_sequence_left)
assert sequence_length == padded_sequence_left_length
assert encoded_sequence == padded_sequence_left
def test_token_type_ids(self):
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
empty_table = self.get_table(tokenizer, length=0)
seq_0 = "Test this method."
# We want to have sequence 0 and sequence 1 are tagged
# respectively with 0 and 1 token_ids
# (regardless of whether the model use token type ids)
# We use this assumption in the QA pipeline among other place
output = tokenizer(empty_table, seq_0, return_token_type_ids=True)
# Assert that the token type IDs have the same length as the input IDs
self.assertEqual(len(output["token_type_ids"]), len(output["input_ids"]))
# Assert that each token type ID has 7 values
self.assertTrue(all(len(token_type_ids) == 7 for token_type_ids in output["token_type_ids"]))
# Do the same test as modeling common.
self.assertIn(0, output["token_type_ids"][0])
@require_torch
@slow
def test_torch_encode_plus_sent_to_model(self):
import torch
from transformers import MODEL_MAPPING, TOKENIZER_MAPPING
MODEL_TOKENIZER_MAPPING = merge_model_tokenizer_mappings(MODEL_MAPPING, TOKENIZER_MAPPING)
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
if tokenizer.__class__ not in MODEL_TOKENIZER_MAPPING:
return
config_class, model_class = MODEL_TOKENIZER_MAPPING[tokenizer.__class__]
config = config_class()
if config.is_encoder_decoder or config.pad_token_id is None:
return
model = model_class(config)
# Make sure the model contains at least the full vocabulary size in its embedding matrix
is_using_common_embeddings = hasattr(model.get_input_embeddings(), "weight")
assert (
(model.get_input_embeddings().weight.shape[0] >= len(tokenizer))
if is_using_common_embeddings
else True
)
# Build sequence
first_ten_tokens = list(tokenizer.get_vocab().keys())[:10]
sequence = " ".join(first_ten_tokens)
table = self.get_table(tokenizer, length=0)
encoded_sequence = tokenizer.encode_plus(table, sequence, return_tensors="pt")
batch_encoded_sequence = tokenizer.batch_encode_plus(table, [sequence, sequence], return_tensors="pt")
# This should not fail
with torch.no_grad(): # saves some time
model(**encoded_sequence)
model(**batch_encoded_sequence)
@unittest.skip("TAPAS doesn't handle pre-tokenized inputs.")
def test_pretokenized_inputs(self):
pass
@slow
def test_tapas_truncation_integration_test(self):
data = {
"Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
"Age": ["56", "45", "59"],
"Number of movies": ["87", "53", "69"],
"Date of birth": ["18 december 1963", "11 november 1974", "6 may 1961"],
}
queries = [
"When was Brad Pitt born?",
"Which actor appeared in the least number of movies?",
"What is the average number of movies?",
]
table = pd.DataFrame.from_dict(data)
tokenizer = TapasTokenizer.from_pretrained("lysandre/tapas-temporary-repo", model_max_length=512)
for i in range(12):
# The table cannot even encode the headers, so raise an error
with self.assertRaises(ValueError):
tokenizer.encode(table=table, query=queries[0], max_length=i, truncation="drop_rows_to_fit")
for i in range(12, 512):
new_encoded_inputs = tokenizer.encode(
table=table, query=queries[0], max_length=i, truncation="drop_rows_to_fit"
)
# Ensure that the input IDs are less than the max length defined.
self.assertLessEqual(len(new_encoded_inputs), i)
tokenizer.model_max_length = 20
new_encoded_inputs = tokenizer.encode(table=table, query=queries[0], truncation=True)
dropped_encoded_inputs = tokenizer.encode(table=table, query=queries[0], truncation="drop_rows_to_fit")
# Ensure that the input IDs are still truncated when no max_length is specified
self.assertListEqual(new_encoded_inputs, dropped_encoded_inputs)
self.assertLessEqual(len(new_encoded_inputs), 20)
@slow
def test_min_max_question_length(self):
data = {
"Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
"Age": ["56", "45", "59"],
"Number of movies": ["87", "53", "69"],
"Date of birth": ["18 december 1963", "11 november 1974", "6 may 1961"],
}
queries = "When was Brad Pitt born?"
table = pd.DataFrame.from_dict(data)
# test max_question_length
tokenizer = TapasTokenizer.from_pretrained("lysandre/tapas-temporary-repo", max_question_length=2)
encoding = tokenizer(table=table, queries=queries)
# query should not be tokenized as it's longer than the specified max_question_length
expected_results = [101, 102]
self.assertListEqual(encoding.input_ids[:2], expected_results)
# test min_question_length
tokenizer = TapasTokenizer.from_pretrained("lysandre/tapas-temporary-repo", min_question_length=30)
encoding = tokenizer(table=table, queries=queries)
# query should not be tokenized as it's shorter than the specified min_question_length
expected_results = [101, 102]
self.assertListEqual(encoding.input_ids[:2], expected_results)
@is_pt_tf_cross_test
def test_batch_encode_plus_tensors(self):
tokenizers = self.get_tokenizers(do_lower_case=False)
for tokenizer in tokenizers:
with self.subTest(f"{tokenizer.__class__.__name__}"):
sequences = [
"Testing batch encode plus",
"Testing batch encode plus with different sequence lengths",
"Testing batch encode plus with different sequence lengths correctly pads",
]
table = self.get_table(tokenizer, length=0)
# A Tensor cannot be build by sequences which are not the same size
self.assertRaises(ValueError, tokenizer.batch_encode_plus, table, sequences, return_tensors="pt")
self.assertRaises(ValueError, tokenizer.batch_encode_plus, table, sequences, return_tensors="tf")
if tokenizer.pad_token_id is None:
self.assertRaises(
ValueError,
tokenizer.batch_encode_plus,
table,
sequences,
padding=True,
return_tensors="pt",
)
self.assertRaises(
ValueError,
tokenizer.batch_encode_plus,
table,
sequences,
padding="longest",
return_tensors="tf",
)
else:
pytorch_tensor = tokenizer.batch_encode_plus(table, sequences, padding=True, return_tensors="pt")
tensorflow_tensor = tokenizer.batch_encode_plus(
table, sequences, padding="longest", return_tensors="tf"
)
encoded_sequences = tokenizer.batch_encode_plus(table, sequences, padding=True)
for key in encoded_sequences.keys():
pytorch_value = pytorch_tensor[key].tolist()
tensorflow_value = tensorflow_tensor[key].numpy().tolist()
encoded_value = encoded_sequences[key]
self.assertEqual(pytorch_value, tensorflow_value, encoded_value)
@slow
def test_tapas_integration_test(self):
data = {
"Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
"Age": ["56", "45", "59"],
"Number of movies": ["87", "53", "69"],
"Date of birth": ["18 december 1963", "11 november 1974", "6 may 1961"],
}
queries = [
"When was Brad Pitt born?",
"Which actor appeared in the least number of movies?",
"What is the average number of movies?",
]
table = pd.DataFrame.from_dict(data)
tokenizer = TapasTokenizer.from_pretrained("google/tapas-base-finetuned-wtq", model_max_length=512)
# fmt: off
expected_results = {'input_ids':[101,2043,2001,8226,15091,2141,1029,102,5889,2287,2193,1997,5691,3058,1997,4182,8226,15091,5179,6584,2324,2285,3699,14720,4487,6178,9488,3429,5187,2340,2281,3326,2577,18856,7828,3240,5354,6353,1020,2089,3777],'attention_mask':[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1],'token_type_ids':[[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[0,0,0,0,0,0,0],[1,1,0,0,0,0,0],[1,2,0,0,0,0,0],[1,3,0,0,0,0,0],[1,3,0,0,0,0,0],[1,3,0,0,0,0,0],[1,4,0,0,0,0,0],[1,4,0,0,0,0,0],[1,4,0,0,0,0,0],[1,1,1,0,0,0,0],[1,1,1,0,0,0,0],[1,2,1,0,2,2,0],[1,3,1,0,3,1,0],[1,4,1,0,2,2,0],[1,4,1,0,2,2,0],[1,4,1,0,2,2,0],[1,1,2,0,0,0,0],[1,1,2,0,0,0,0],[1,1,2,0,0,0,0],[1,1,2,0,0,0,0],[1,2,2,0,1,3,0],[1,3,2,0,1,3,0],[1,4,2,0,3,1,0],[1,4,2,0,3,1,0],[1,4,2,0,3,1,0],[1,1,3,0,0,0,0],[1,1,3,0,0,0,0],[1,1,3,0,0,0,0],[1,1,3,0,0,0,0],[1,2,3,0,3,1,0],[1,3,3,0,2,2,0],[1,4,3,0,1,3,0],[1,4,3,0,1,3,0],[1,4,3,0,1,3,0]]} # noqa: E231
# fmt: on
new_encoded_inputs = tokenizer.encode_plus(table=table, query=queries[0])
self.assertDictEqual(dict(new_encoded_inputs), expected_results)
@slow
def test_full_tokenizer(self):
data = [
["Pos", "No", "Driver", "Team", "Laps", "Time/Retired", "Grid", "Points"],
["1", "32", "Patrick Carpentier", "Team Player's", "87", "1:48:11.023", "1", "22"],
["2", "1", "Bruno Junqueira", "Newman/Haas Racing", "87", "+0.8 secs", "2", "17"],
["3", "3", "Paul Tracy", "Team Player's", "87", "+28.6 secs", "3", "14"],
["4", "9", "Michel Jourdain, Jr.", "Team Rahal", "87", "+40.8 secs", "13", "12"],
["5", "34", "Mario Haberfeld", "Mi-Jack Conquest Racing", "87", "+42.1 secs", "6", "10"],
["6", "20", "Oriol Servia", "Patrick Racing", "87", "+1:00.2", "10", "8"],
["7", "51", "Adrian Fernandez", "Fernandez Racing", "87", "+1:01.4", "5", "6"],
["8", "12", "Jimmy Vasser", "American Spirit Team Johansson", "87", "+1:01.8", "8", "5"],
["9", "7", "Tiago Monteiro", "Fittipaldi-Dingman Racing", "86", "+ 1 Lap", "15", "4"],
["10", "55", "Mario Dominguez", "Herdez Competition", "86", "+ 1 Lap", "11", "3"],
["11", "27", "Bryan Herta", "PK Racing", "86", "+ 1 Lap", "12", "2"],
["12", "31", "Ryan Hunter-Reay", "American Spirit Team Johansson", "86", "+ 1 Lap", "17", "1"],
["13", "19", "Joel Camathias", "Dale Coyne Racing", "85", "+ 2 Laps", "18", "0"],
["14", "33", "Alex Tagliani", "Rocketsports Racing", "85", "+ 2 Laps", "14", "0"],
["15", "4", "Roberto Moreno", "Herdez Competition", "85", "+ 2 Laps", "9", "0"],
["16", "11", "Geoff Boss", "Dale Coyne Racing", "83", "Mechanical", "19", "0"],
["17", "2", "Sebastien Bourdais", "Newman/Haas Racing", "77", "Mechanical", "4", "0"],
["18", "15", "Darren Manning", "Walker Racing", "12", "Mechanical", "7", "0"],
["19", "5", "Rodolfo Lavin", "Walker Racing", "10", "Mechanical", "16", "0"],
]
query = "what were the drivers names?"
table = pd.DataFrame.from_records(data[1:], columns=data[0])
tokenizer = TapasTokenizer.from_pretrained("google/tapas-base-finetuned-wtq", model_max_length=512)
model_inputs = tokenizer(table, query, padding="max_length")
input_ids = model_inputs["input_ids"]
token_type_ids = np.array(model_inputs["token_type_ids"])
segment_ids = token_type_ids[:, 0]
column_ids = token_type_ids[:, 1]
row_ids = token_type_ids[:, 2]
# fmt: off
expected_results = {'input_ids':[101,2054,2020,1996,6853,3415,1029,102,13433,2015,2053,4062,2136,10876,2051,1013,3394,8370,2685,1015,3590,4754,29267,4765,3771,2136,2447,1005,1055,6584,1015,1024,4466,1024,2340,1012,6185,2509,1015,2570,1016,1015,10391,12022,4226,7895,10625,1013,22996,3868,6584,1009,1014,1012,1022,10819,2015,1016,2459,1017,1017,2703,10555,2136,2447,1005,1055,6584,1009,2654,1012,1020,10819,2015,1017,2403,1018,1023,8709,8183,3126,21351,2078,1010,3781,1012,2136,10958,8865,6584,1009,2871,1012,1022,10819,2015,2410,2260,1019,4090,7986,5292,5677,8151,2771,1011,2990,9187,3868,6584,1009,4413,1012,1015,10819,2015,1020,2184,1020,2322,2030,20282,14262,9035,4754,3868,6584,1009,1015,1024,4002,1012,1016,2184,1022,1021,4868,7918,12023,12023,3868,6584,1009,1015,1024,5890,1012,1018,1019,1020,1022,2260,5261,12436,18116,2137,4382,2136,26447,6584,1009,1015,1024,5890,1012,1022,1022,1019,1023,1021,27339,3995,10125,9711,4906,25101,24657,1011,22033,2386,3868,6564,1009,1015,5001,2321,1018,2184,4583,7986,14383,2075,29488,14906,9351,2971,6564,1009,1015,5001,2340,1017,2340,2676,8527,2014,2696,1052,2243,3868,6564,1009,1015,5001,2260,1016,2260,2861,4575,4477,1011,2128,4710,2137,4382,2136,26447,6564,1009,1015,5001,2459,1015,2410,2539,8963,11503,25457,3022,8512,2522,9654,3868,5594,1009,1016,10876,2324,1014,2403,3943,4074,6415,15204,2072,12496,25378,3868,5594,1009,1016,10876,2403,1014,2321,1018,10704,17921,14906,9351,2971,5594,1009,1016,10876,1023,1014,2385,2340,14915,5795,8512,2522,9654,3868,6640,6228,2539,1014,2459,1016,28328,8945,3126,21351,2015,10625,1013,22996,3868,6255,6228,1018,1014,2324,2321,12270,11956,5232,3868,2260,6228,1021,1014,2539,1019,8473,28027,2080,2474,6371,5232,3868,2184,6228,2385,1014,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],'column_ids':[0,0,0,0,0,0,0,0,1,1,2,3,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,5,6,6,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,3,3,3,3,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,4,4,4,4,5,6,6,6,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,3,4,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,4,5,6,6,6,7,8,1,2,3,3,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,4,4,4,5,6,6,6,7,8,1,2,3,3,4,4,4,4,5,6,7,8,1,2,3,3,3,3,3,4,4,4,4,5,6,7,8,1,2,3,3,4,4,5,6,7,8,1,2,3,3,3,3,3,4,4,5,6,7,8,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],'row_ids':[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,11,11,11,11,11,11,11,11,11,11,11,11,11,11,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,12,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,14,14,14,14,14,14,14,14,14,14,14,14,14,14,14,15,15,15,15,15,15,15,15,15,15,15,15,15,16,16,16,16,16,16,16,16,16,16,16,16,17,17,17,17,17,17,17,17,17,17,17,17,17,17,17,18,18,18,18,18,18,18,18,18,18,19,19,19,19,19,19,19,19,19,19,19,19,19,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],'segment_ids':[0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]} # noqa: E231
# fmt: on
self.assertListEqual(input_ids, expected_results["input_ids"])
self.assertListEqual(segment_ids.tolist(), expected_results["segment_ids"])
self.assertListEqual(column_ids.tolist(), expected_results["column_ids"])
self.assertListEqual(row_ids.tolist(), expected_results["row_ids"])
@unittest.skip("Skip this test while all models are still to be uploaded.")
def test_pretrained_model_lists(self):
pass
@unittest.skip("Doesn't support another framework than PyTorch")
def test_np_encode_plus_sent_to_model(self):
pass
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/mlm_wwm/run_chinese_ref.py | import argparse
import json
from typing import List
from ltp import LTP
from transformers.models.bert.tokenization_bert import BertTokenizer
def _is_chinese_char(cp):
"""Checks whether CP is the codepoint of a CJK character."""
# This defines a "chinese character" as anything in the CJK Unicode block:
# https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_(Unicode_block)
#
# Note that the CJK Unicode block is NOT all Japanese and Korean characters,
# despite its name. The modern Korean Hangul alphabet is a different block,
# as is Japanese Hiragana and Katakana. Those alphabets are used to write
# space-separated words, so they are not treated specially and handled
# like the all of the other languages.
if (
(cp >= 0x4E00 and cp <= 0x9FFF)
or (cp >= 0x3400 and cp <= 0x4DBF) #
or (cp >= 0x20000 and cp <= 0x2A6DF) #
or (cp >= 0x2A700 and cp <= 0x2B73F) #
or (cp >= 0x2B740 and cp <= 0x2B81F) #
or (cp >= 0x2B820 and cp <= 0x2CEAF) #
or (cp >= 0xF900 and cp <= 0xFAFF)
or (cp >= 0x2F800 and cp <= 0x2FA1F) #
): #
return True
return False
def is_chinese(word: str):
# word like '180' or '身高' or '神'
for char in word:
char = ord(char)
if not _is_chinese_char(char):
return 0
return 1
def get_chinese_word(tokens: List[str]):
word_set = set()
for token in tokens:
chinese_word = len(token) > 1 and is_chinese(token)
if chinese_word:
word_set.add(token)
word_list = list(word_set)
return word_list
def add_sub_symbol(bert_tokens: List[str], chinese_word_set: set()):
if not chinese_word_set:
return bert_tokens
max_word_len = max([len(w) for w in chinese_word_set])
bert_word = bert_tokens
start, end = 0, len(bert_word)
while start < end:
single_word = True
if is_chinese(bert_word[start]):
l = min(end - start, max_word_len)
for i in range(l, 1, -1):
whole_word = "".join(bert_word[start : start + i])
if whole_word in chinese_word_set:
for j in range(start + 1, start + i):
bert_word[j] = "##" + bert_word[j]
start = start + i
single_word = False
break
if single_word:
start += 1
return bert_word
def prepare_ref(lines: List[str], ltp_tokenizer: LTP, bert_tokenizer: BertTokenizer):
ltp_res = []
for i in range(0, len(lines), 100):
res = ltp_tokenizer.pipeline(lines[i : i + 100], tasks=["cws"]).cws
res = [get_chinese_word(r) for r in res]
ltp_res.extend(res)
assert len(ltp_res) == len(lines)
bert_res = []
for i in range(0, len(lines), 100):
res = bert_tokenizer(lines[i : i + 100], add_special_tokens=True, truncation=True, max_length=512)
bert_res.extend(res["input_ids"])
assert len(bert_res) == len(lines)
ref_ids = []
for input_ids, chinese_word in zip(bert_res, ltp_res):
input_tokens = []
for id in input_ids:
token = bert_tokenizer._convert_id_to_token(id)
input_tokens.append(token)
input_tokens = add_sub_symbol(input_tokens, chinese_word)
ref_id = []
# We only save pos of chinese subwords start with ##, which mean is part of a whole word.
for i, token in enumerate(input_tokens):
if token[:2] == "##":
clean_token = token[2:]
# save chinese tokens' pos
if len(clean_token) == 1 and _is_chinese_char(ord(clean_token)):
ref_id.append(i)
ref_ids.append(ref_id)
assert len(ref_ids) == len(bert_res)
return ref_ids
def main(args):
# For Chinese (Ro)Bert, the best result is from : RoBERTa-wwm-ext (https://github.com/ymcui/Chinese-BERT-wwm)
# If we want to fine-tune these model, we have to use same tokenizer : LTP (https://github.com/HIT-SCIR/ltp)
with open(args.file_name, "r", encoding="utf-8") as f:
data = f.readlines()
data = [line.strip() for line in data if len(line) > 0 and not line.isspace()] # avoid delimiter like '\u2029'
ltp_tokenizer = LTP(args.ltp) # faster in GPU device
bert_tokenizer = BertTokenizer.from_pretrained(args.bert)
ref_ids = prepare_ref(data, ltp_tokenizer, bert_tokenizer)
with open(args.save_path, "w", encoding="utf-8") as f:
data = [json.dumps(ref) + "\n" for ref in ref_ids]
f.writelines(data)
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="prepare_chinese_ref")
parser.add_argument(
"--file_name",
required=False,
type=str,
default="./resources/chinese-demo.txt",
help="file need process, same as training data in lm",
)
parser.add_argument(
"--ltp",
required=False,
type=str,
default="./resources/ltp",
help="resources for LTP tokenizer, usually a path",
)
parser.add_argument(
"--bert",
required=False,
type=str,
default="./resources/robert",
help="resources for Bert tokenizer",
)
parser.add_argument(
"--save_path",
required=False,
type=str,
default="./resources/ref.txt",
help="path to save res",
)
args = parser.parse_args()
main(args)
| import argparse
import json
from typing import List
from ltp import LTP
from transformers.models.bert.tokenization_bert import BertTokenizer
def _is_chinese_char(cp):
"""Checks whether CP is the codepoint of a CJK character."""
# This defines a "chinese character" as anything in the CJK Unicode block:
# https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_(Unicode_block)
#
# Note that the CJK Unicode block is NOT all Japanese and Korean characters,
# despite its name. The modern Korean Hangul alphabet is a different block,
# as is Japanese Hiragana and Katakana. Those alphabets are used to write
# space-separated words, so they are not treated specially and handled
# like the all of the other languages.
if (
(cp >= 0x4E00 and cp <= 0x9FFF)
or (cp >= 0x3400 and cp <= 0x4DBF) #
or (cp >= 0x20000 and cp <= 0x2A6DF) #
or (cp >= 0x2A700 and cp <= 0x2B73F) #
or (cp >= 0x2B740 and cp <= 0x2B81F) #
or (cp >= 0x2B820 and cp <= 0x2CEAF) #
or (cp >= 0xF900 and cp <= 0xFAFF)
or (cp >= 0x2F800 and cp <= 0x2FA1F) #
): #
return True
return False
def is_chinese(word: str):
# word like '180' or '身高' or '神'
for char in word:
char = ord(char)
if not _is_chinese_char(char):
return 0
return 1
def get_chinese_word(tokens: List[str]):
word_set = set()
for token in tokens:
chinese_word = len(token) > 1 and is_chinese(token)
if chinese_word:
word_set.add(token)
word_list = list(word_set)
return word_list
def add_sub_symbol(bert_tokens: List[str], chinese_word_set: set()):
if not chinese_word_set:
return bert_tokens
max_word_len = max([len(w) for w in chinese_word_set])
bert_word = bert_tokens
start, end = 0, len(bert_word)
while start < end:
single_word = True
if is_chinese(bert_word[start]):
l = min(end - start, max_word_len)
for i in range(l, 1, -1):
whole_word = "".join(bert_word[start : start + i])
if whole_word in chinese_word_set:
for j in range(start + 1, start + i):
bert_word[j] = "##" + bert_word[j]
start = start + i
single_word = False
break
if single_word:
start += 1
return bert_word
def prepare_ref(lines: List[str], ltp_tokenizer: LTP, bert_tokenizer: BertTokenizer):
ltp_res = []
for i in range(0, len(lines), 100):
res = ltp_tokenizer.pipeline(lines[i : i + 100], tasks=["cws"]).cws
res = [get_chinese_word(r) for r in res]
ltp_res.extend(res)
assert len(ltp_res) == len(lines)
bert_res = []
for i in range(0, len(lines), 100):
res = bert_tokenizer(lines[i : i + 100], add_special_tokens=True, truncation=True, max_length=512)
bert_res.extend(res["input_ids"])
assert len(bert_res) == len(lines)
ref_ids = []
for input_ids, chinese_word in zip(bert_res, ltp_res):
input_tokens = []
for id in input_ids:
token = bert_tokenizer._convert_id_to_token(id)
input_tokens.append(token)
input_tokens = add_sub_symbol(input_tokens, chinese_word)
ref_id = []
# We only save pos of chinese subwords start with ##, which mean is part of a whole word.
for i, token in enumerate(input_tokens):
if token[:2] == "##":
clean_token = token[2:]
# save chinese tokens' pos
if len(clean_token) == 1 and _is_chinese_char(ord(clean_token)):
ref_id.append(i)
ref_ids.append(ref_id)
assert len(ref_ids) == len(bert_res)
return ref_ids
def main(args):
# For Chinese (Ro)Bert, the best result is from : RoBERTa-wwm-ext (https://github.com/ymcui/Chinese-BERT-wwm)
# If we want to fine-tune these model, we have to use same tokenizer : LTP (https://github.com/HIT-SCIR/ltp)
with open(args.file_name, "r", encoding="utf-8") as f:
data = f.readlines()
data = [line.strip() for line in data if len(line) > 0 and not line.isspace()] # avoid delimiter like '\u2029'
ltp_tokenizer = LTP(args.ltp) # faster in GPU device
bert_tokenizer = BertTokenizer.from_pretrained(args.bert)
ref_ids = prepare_ref(data, ltp_tokenizer, bert_tokenizer)
with open(args.save_path, "w", encoding="utf-8") as f:
data = [json.dumps(ref) + "\n" for ref in ref_ids]
f.writelines(data)
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="prepare_chinese_ref")
parser.add_argument(
"--file_name",
required=False,
type=str,
default="./resources/chinese-demo.txt",
help="file need process, same as training data in lm",
)
parser.add_argument(
"--ltp",
required=False,
type=str,
default="./resources/ltp",
help="resources for LTP tokenizer, usually a path",
)
parser.add_argument(
"--bert",
required=False,
type=str,
default="./resources/robert",
help="resources for Bert tokenizer",
)
parser.add_argument(
"--save_path",
required=False,
type=str,
default="./resources/ref.txt",
help="path to save res",
)
args = parser.parse_args()
main(args)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/en/model_doc/gpt_neo.mdx | <!--Copyright 2021 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# GPT Neo
## Overview
The GPTNeo model was released in the [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) repository by Sid
Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy. It is a GPT2 like causal language model trained on the
[Pile](https://pile.eleuther.ai/) dataset.
The architecture is similar to GPT2 except that GPT Neo uses local attention in every other layer with a window size of
256 tokens.
This model was contributed by [valhalla](https://huggingface.co/valhalla).
### Generation
The `generate()` method can be used to generate text using GPT Neo model.
```python
>>> from transformers import GPTNeoForCausalLM, GPT2Tokenizer
>>> model = GPTNeoForCausalLM.from_pretrained("EleutherAI/gpt-neo-1.3B")
>>> tokenizer = GPT2Tokenizer.from_pretrained("EleutherAI/gpt-neo-1.3B")
>>> prompt = (
... "In a shocking finding, scientists discovered a herd of unicorns living in a remote, "
... "previously unexplored valley, in the Andes Mountains. Even more surprising to the "
... "researchers was the fact that the unicorns spoke perfect English."
... )
>>> input_ids = tokenizer(prompt, return_tensors="pt").input_ids
>>> gen_tokens = model.generate(
... input_ids,
... do_sample=True,
... temperature=0.9,
... max_length=100,
... )
>>> gen_text = tokenizer.batch_decode(gen_tokens)[0]
```
## Documentation resources
- [Text classification task guide](./tasks/sequence_classification)
- [Causal language modeling task guide](./tasks/language_modeling)
## GPTNeoConfig
[[autodoc]] GPTNeoConfig
## GPTNeoModel
[[autodoc]] GPTNeoModel
- forward
## GPTNeoForCausalLM
[[autodoc]] GPTNeoForCausalLM
- forward
## GPTNeoForSequenceClassification
[[autodoc]] GPTNeoForSequenceClassification
- forward
## FlaxGPTNeoModel
[[autodoc]] FlaxGPTNeoModel
- __call__
## FlaxGPTNeoForCausalLM
[[autodoc]] FlaxGPTNeoForCausalLM
- __call__
| <!--Copyright 2021 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# GPT Neo
## Overview
The GPTNeo model was released in the [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) repository by Sid
Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy. It is a GPT2 like causal language model trained on the
[Pile](https://pile.eleuther.ai/) dataset.
The architecture is similar to GPT2 except that GPT Neo uses local attention in every other layer with a window size of
256 tokens.
This model was contributed by [valhalla](https://huggingface.co/valhalla).
### Generation
The `generate()` method can be used to generate text using GPT Neo model.
```python
>>> from transformers import GPTNeoForCausalLM, GPT2Tokenizer
>>> model = GPTNeoForCausalLM.from_pretrained("EleutherAI/gpt-neo-1.3B")
>>> tokenizer = GPT2Tokenizer.from_pretrained("EleutherAI/gpt-neo-1.3B")
>>> prompt = (
... "In a shocking finding, scientists discovered a herd of unicorns living in a remote, "
... "previously unexplored valley, in the Andes Mountains. Even more surprising to the "
... "researchers was the fact that the unicorns spoke perfect English."
... )
>>> input_ids = tokenizer(prompt, return_tensors="pt").input_ids
>>> gen_tokens = model.generate(
... input_ids,
... do_sample=True,
... temperature=0.9,
... max_length=100,
... )
>>> gen_text = tokenizer.batch_decode(gen_tokens)[0]
```
## Documentation resources
- [Text classification task guide](./tasks/sequence_classification)
- [Causal language modeling task guide](./tasks/language_modeling)
## GPTNeoConfig
[[autodoc]] GPTNeoConfig
## GPTNeoModel
[[autodoc]] GPTNeoModel
- forward
## GPTNeoForCausalLM
[[autodoc]] GPTNeoForCausalLM
- forward
## GPTNeoForSequenceClassification
[[autodoc]] GPTNeoForSequenceClassification
- forward
## FlaxGPTNeoModel
[[autodoc]] FlaxGPTNeoModel
- __call__
## FlaxGPTNeoForCausalLM
[[autodoc]] FlaxGPTNeoForCausalLM
- __call__
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/pipelines/test_pipelines_zero_shot_audio_classification.py | # Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import unittest
from datasets import load_dataset
from transformers.pipelines import pipeline
from transformers.testing_utils import nested_simplify, require_torch, slow
@require_torch
class ZeroShotAudioClassificationPipelineTests(unittest.TestCase):
# Deactivating auto tests since we don't have a good MODEL_FOR_XX mapping,
# and only CLAP would be there for now.
# model_mapping = {CLAPConfig: CLAPModel}
@require_torch
def test_small_model_pt(self):
audio_classifier = pipeline(
task="zero-shot-audio-classification", model="hf-internal-testing/tiny-clap-htsat-unfused"
)
dataset = load_dataset("ashraq/esc50")
audio = dataset["train"]["audio"][-1]["array"]
output = audio_classifier(audio, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"])
self.assertEqual(
nested_simplify(output),
[{"score": 0.501, "label": "Sound of a dog"}, {"score": 0.499, "label": "Sound of vaccum cleaner"}],
)
@unittest.skip("No models are available in TF")
def test_small_model_tf(self):
pass
@slow
@require_torch
def test_large_model_pt(self):
audio_classifier = pipeline(
task="zero-shot-audio-classification",
model="laion/clap-htsat-unfused",
)
# This is an audio of a dog
dataset = load_dataset("ashraq/esc50")
audio = dataset["train"]["audio"][-1]["array"]
output = audio_classifier(audio, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"])
self.assertEqual(
nested_simplify(output),
[
{"score": 0.999, "label": "Sound of a dog"},
{"score": 0.001, "label": "Sound of vaccum cleaner"},
],
)
output = audio_classifier([audio] * 5, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"])
self.assertEqual(
nested_simplify(output),
[
[
{"score": 0.999, "label": "Sound of a dog"},
{"score": 0.001, "label": "Sound of vaccum cleaner"},
],
]
* 5,
)
output = audio_classifier(
[audio] * 5, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"], batch_size=5
)
self.assertEqual(
nested_simplify(output),
[
[
{"score": 0.999, "label": "Sound of a dog"},
{"score": 0.001, "label": "Sound of vaccum cleaner"},
],
]
* 5,
)
@unittest.skip("No models are available in TF")
def test_large_model_tf(self):
pass
| # Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import unittest
from datasets import load_dataset
from transformers.pipelines import pipeline
from transformers.testing_utils import nested_simplify, require_torch, slow
@require_torch
class ZeroShotAudioClassificationPipelineTests(unittest.TestCase):
# Deactivating auto tests since we don't have a good MODEL_FOR_XX mapping,
# and only CLAP would be there for now.
# model_mapping = {CLAPConfig: CLAPModel}
@require_torch
def test_small_model_pt(self):
audio_classifier = pipeline(
task="zero-shot-audio-classification", model="hf-internal-testing/tiny-clap-htsat-unfused"
)
dataset = load_dataset("ashraq/esc50")
audio = dataset["train"]["audio"][-1]["array"]
output = audio_classifier(audio, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"])
self.assertEqual(
nested_simplify(output),
[{"score": 0.501, "label": "Sound of a dog"}, {"score": 0.499, "label": "Sound of vaccum cleaner"}],
)
@unittest.skip("No models are available in TF")
def test_small_model_tf(self):
pass
@slow
@require_torch
def test_large_model_pt(self):
audio_classifier = pipeline(
task="zero-shot-audio-classification",
model="laion/clap-htsat-unfused",
)
# This is an audio of a dog
dataset = load_dataset("ashraq/esc50")
audio = dataset["train"]["audio"][-1]["array"]
output = audio_classifier(audio, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"])
self.assertEqual(
nested_simplify(output),
[
{"score": 0.999, "label": "Sound of a dog"},
{"score": 0.001, "label": "Sound of vaccum cleaner"},
],
)
output = audio_classifier([audio] * 5, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"])
self.assertEqual(
nested_simplify(output),
[
[
{"score": 0.999, "label": "Sound of a dog"},
{"score": 0.001, "label": "Sound of vaccum cleaner"},
],
]
* 5,
)
output = audio_classifier(
[audio] * 5, candidate_labels=["Sound of a dog", "Sound of vaccum cleaner"], batch_size=5
)
self.assertEqual(
nested_simplify(output),
[
[
{"score": 0.999, "label": "Sound of a dog"},
{"score": 0.001, "label": "Sound of vaccum cleaner"},
],
]
* 5,
)
@unittest.skip("No models are available in TF")
def test_large_model_tf(self):
pass
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/en/model_doc/xlsr_wav2vec2.mdx | <!--Copyright 2021 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# XLSR-Wav2Vec2
## Overview
The XLSR-Wav2Vec2 model was proposed in [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael
Auli.
The abstract from the paper is the following:
*This paper presents XLSR which learns cross-lingual speech representations by pretraining a single model from the raw
waveform of speech in multiple languages. We build on wav2vec 2.0 which is trained by solving a contrastive task over
masked latent speech representations and jointly learns a quantization of the latents shared across languages. The
resulting model is fine-tuned on labeled data and experiments show that cross-lingual pretraining significantly
outperforms monolingual pretraining. On the CommonVoice benchmark, XLSR shows a relative phoneme error rate reduction
of 72% compared to the best known results. On BABEL, our approach improves word error rate by 16% relative compared to
a comparable system. Our approach enables a single multilingual speech recognition model which is competitive to strong
individual models. Analysis shows that the latent discrete speech representations are shared across languages with
increased sharing for related languages. We hope to catalyze research in low-resource speech understanding by releasing
XLSR-53, a large model pretrained in 53 languages.*
Tips:
- XLSR-Wav2Vec2 is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
- XLSR-Wav2Vec2 model was trained using connectionist temporal classification (CTC) so the model output has to be
decoded using [`Wav2Vec2CTCTokenizer`].
XLSR-Wav2Vec2's architecture is based on the Wav2Vec2 model, so one can refer to [Wav2Vec2's documentation page](wav2vec2).
The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
| <!--Copyright 2021 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# XLSR-Wav2Vec2
## Overview
The XLSR-Wav2Vec2 model was proposed in [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael
Auli.
The abstract from the paper is the following:
*This paper presents XLSR which learns cross-lingual speech representations by pretraining a single model from the raw
waveform of speech in multiple languages. We build on wav2vec 2.0 which is trained by solving a contrastive task over
masked latent speech representations and jointly learns a quantization of the latents shared across languages. The
resulting model is fine-tuned on labeled data and experiments show that cross-lingual pretraining significantly
outperforms monolingual pretraining. On the CommonVoice benchmark, XLSR shows a relative phoneme error rate reduction
of 72% compared to the best known results. On BABEL, our approach improves word error rate by 16% relative compared to
a comparable system. Our approach enables a single multilingual speech recognition model which is competitive to strong
individual models. Analysis shows that the latent discrete speech representations are shared across languages with
increased sharing for related languages. We hope to catalyze research in low-resource speech understanding by releasing
XLSR-53, a large model pretrained in 53 languages.*
Tips:
- XLSR-Wav2Vec2 is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
- XLSR-Wav2Vec2 model was trained using connectionist temporal classification (CTC) so the model output has to be
decoded using [`Wav2Vec2CTCTokenizer`].
XLSR-Wav2Vec2's architecture is based on the Wav2Vec2 model, so one can refer to [Wav2Vec2's documentation page](wav2vec2).
The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./utils/test_module/custom_processing.py | from transformers import ProcessorMixin
class CustomProcessor(ProcessorMixin):
feature_extractor_class = "AutoFeatureExtractor"
tokenizer_class = "AutoTokenizer"
| from transformers import ProcessorMixin
class CustomProcessor(ProcessorMixin):
feature_extractor_class = "AutoFeatureExtractor"
tokenizer_class = "AutoTokenizer"
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/jax-projects/HOW_TO_PROPOSE_PROJECT.md | # How to propose a Flax/JAX + Transformers project
Great that you've opened this document!
While we at 🤗 are proposing a couple of projects, we strongly
believe that the community can come up with much more **creative**, **fun**, and
**impactful** projects on their own. This being said, we are really looking forward
to seeing your project proposal!
## What a project should be about
The proposed project should fall into the machine learning fields of **Natural Language Processing (NLP)** and/or **Computer Vision (CV)** (possibly also **Speech Recognition (ASR)** depending on whether Speech Recognition models are available in Flax in due time) and aim at solving a specific task.
Possible tasks can belong to:
* text classification
* text generation
* image recognition
* image processing
* image captioning
* audio classification
* and other tasks you can think of!
The clearer a task is defined, the better your project proposal is.
*E.g.* "Using a T5 model to learn grammar correction in French" or "Adapting a pre-trained CLIP model for zero-shot image classification in Spanish" are **well-defined and clear** project proposals, while something like "Train a language model" or "Image classification" are **too vague**.
There is no limit to your creativity as long as the project is feasible and ethical.
The more creative & specific your project proposal, the more interesting it will be,
and the more likely will you find motivated team members to work on your project!
To get an idea of how to formulate your project proposals, you can browse through
existing project proposals on the [forum](https://discuss.huggingface.co/c/flax-jax-projects/22).
## How to submit a project proposal
First, you should make sure that you are [logged in](https://huggingface.co/login?sso=bm9uY2U9OTRlNjZjZmZhYjMwMmJmMWMyYjc5MmFiMTMyMzY5ODYmcmV0dXJuX3Nzb191cmw9aHR0cHMlM0ElMkYlMkZkaXNjdXNzLmh1Z2dpbmdmYWNlLmNvJTJGc2Vzc2lvbiUyRnNzb19sb2dpbg%3D%3D&sig=429ad8924bcb33c40f9823027ea749abb55d393f4f58924f36a2dba3ab0a48da) with your Hugging Face account on the forum.
Second, make sure that your project idea doesn't already exist by checking [existing projects](https://discuss.huggingface.co/c/flax-jax-projects/22).
If your project already exists - great! This means that you can comment and improve
the existing idea and join the project to form a team! If your project idea already
exists for a different language, feel free to submit the same project idea, just in
a different language.
Third, having ensured that your project doesn't exist, click on the *"New Topic"*
button on the [Flax/JAX Projects Forum category](https://discuss.huggingface.co/c/flax-jax-projects/22) to create a new project proposal.
Fourth, make sure that your project proposal includes the following information:
1. *A clear description of the project*
2. *In which language should the project be conducted?* English, German, Chinese, ...? It can also be a multi-lingual project
3. *Which model should be used?* If you want to adapt an existing model, you can add the link to one of the 4000 available checkpoints in JAX [here](https://huggingface.co/models?filter=jax) If you want to train a model from scratch, you can simply state the model architecture to be used, *e.g.* BERT, CLIP, etc. You can also base your project on a model that is not part of transformers. For an overview of libraries based on JAX, you can take a look at [awesome-jax](https://github.com/n2cholas/awesome-jax#awesome-jax-). **Note** that for a project that is not based on Transformers it will be more difficult for the 🤗 team to help you. Also have a look at the section [Quickstart Flax & Jax in Transformers](https://github.com/huggingface/transformers/tree/main/examples/research_projects/jax-projects#quickstart-flax-and-jax-in-transformers) to see what model architectures are currently supported in 🤗 Transformers.
4. *What data should be used?* It is important to state at least what kind of data you would like to use. Ideally, you can already point to publicly available data or a dataset in the 🤗 Datasets library.
5. *Are similar training scripts available in Flax/JAX?* It would be important to find similar training scripts that already exist in Flax/JAX. *E.g.* if you are working on a Seq-to-Seq task, you can make use of the [`run_summarization_flax.py`](https://github.com/huggingface/transformers/blob/main/examples/flax/summarization/run_summarization_flax.py) script which is very similar to any seq2seq training. Also have a look at the section [Quickstart Flax & Jax in Transformers](https://github.com/huggingface/transformers/tree/main/examples/research_projects/jax-projects#quickstart-flax-and-jax-in-transformers) to see what training scripts are currently supported in 🤗 Transformers.
6. *(Optionally) What are possible challenges?* List possible difficulties with your project. *E.g.* If you know that training convergence usually takes a lot of time, it is worth stating this here!
7. *(Optionally) What is the desired project outcome?* - How would you like to demo your project? One could *e.g.* create a Streamlit application.
8. *(Optionally) Links to read upon* - Can you provide any links that would help the reader to better understand your project idea?
Feel free to copy-paste the following format for your project proposal and fill out the respective sections:
```
# <FILL ME: Name of project>
<FILL ME: A clear description of the project>
## 2. Language
The model will be trained in <FILL ME: which language?>.
## 3. Model
<FILL ME: 3. Which model should be used?>
## 4. Datasets
<FILL ME: 4. Which data should be used?>
Possible links to publicly available datasets include:
- <FILL ME: Link 1 to dataset>
- <FILL ME: Link 2 to dataset>
- <FILL ME: Link 3 to dataset>
## 5. Training scripts
<FILL ME: 5. Are there publicly available training scripts that can be used/tweaked for the project?>
We can make use of <FILL ME: link to training script> to train the model.>
## 6. (Optional) Challenges
<(Optionally) FILL ME: 6. What are possible challenges?>
## 7. (Optional) Desired project outcome
<(Optionally) FILL ME: 7. What is the desired project outcome? A demo?>
## 8. (Optional) Reads
The following links can be useful to better understand the project and
what has previously been done.
- <FILL ME: Link 1 to read>
- <FILL ME: Link 2 to read>
- <FILL ME: Link 3 to read>
```
To see how a proposed project looks like, please have a look at submitted project
proposals [here](https://discuss.huggingface.co/c/flax-jax-projects/22).
## Will my project proposal be selected?
Having submitted a project proposal, you can now promote your idea in the Slack channel `#flax-jax-community-week` to try to convince other participants to join your project!
Once other people have joined your project, one of the organizers (`@Suzana, @valhalla, @osanseviero, @patrickvonplaten`) will officially create a team for your project and add your project to [this google sheet](https://docs.google.com/spreadsheets/d/1GpHebL7qrwJOc9olTpIPgjf8vOS0jNb6zR_B8x_Jtik/edit?usp=sharing).
| # How to propose a Flax/JAX + Transformers project
Great that you've opened this document!
While we at 🤗 are proposing a couple of projects, we strongly
believe that the community can come up with much more **creative**, **fun**, and
**impactful** projects on their own. This being said, we are really looking forward
to seeing your project proposal!
## What a project should be about
The proposed project should fall into the machine learning fields of **Natural Language Processing (NLP)** and/or **Computer Vision (CV)** (possibly also **Speech Recognition (ASR)** depending on whether Speech Recognition models are available in Flax in due time) and aim at solving a specific task.
Possible tasks can belong to:
* text classification
* text generation
* image recognition
* image processing
* image captioning
* audio classification
* and other tasks you can think of!
The clearer a task is defined, the better your project proposal is.
*E.g.* "Using a T5 model to learn grammar correction in French" or "Adapting a pre-trained CLIP model for zero-shot image classification in Spanish" are **well-defined and clear** project proposals, while something like "Train a language model" or "Image classification" are **too vague**.
There is no limit to your creativity as long as the project is feasible and ethical.
The more creative & specific your project proposal, the more interesting it will be,
and the more likely will you find motivated team members to work on your project!
To get an idea of how to formulate your project proposals, you can browse through
existing project proposals on the [forum](https://discuss.huggingface.co/c/flax-jax-projects/22).
## How to submit a project proposal
First, you should make sure that you are [logged in](https://huggingface.co/login?sso=bm9uY2U9OTRlNjZjZmZhYjMwMmJmMWMyYjc5MmFiMTMyMzY5ODYmcmV0dXJuX3Nzb191cmw9aHR0cHMlM0ElMkYlMkZkaXNjdXNzLmh1Z2dpbmdmYWNlLmNvJTJGc2Vzc2lvbiUyRnNzb19sb2dpbg%3D%3D&sig=429ad8924bcb33c40f9823027ea749abb55d393f4f58924f36a2dba3ab0a48da) with your Hugging Face account on the forum.
Second, make sure that your project idea doesn't already exist by checking [existing projects](https://discuss.huggingface.co/c/flax-jax-projects/22).
If your project already exists - great! This means that you can comment and improve
the existing idea and join the project to form a team! If your project idea already
exists for a different language, feel free to submit the same project idea, just in
a different language.
Third, having ensured that your project doesn't exist, click on the *"New Topic"*
button on the [Flax/JAX Projects Forum category](https://discuss.huggingface.co/c/flax-jax-projects/22) to create a new project proposal.
Fourth, make sure that your project proposal includes the following information:
1. *A clear description of the project*
2. *In which language should the project be conducted?* English, German, Chinese, ...? It can also be a multi-lingual project
3. *Which model should be used?* If you want to adapt an existing model, you can add the link to one of the 4000 available checkpoints in JAX [here](https://huggingface.co/models?filter=jax) If you want to train a model from scratch, you can simply state the model architecture to be used, *e.g.* BERT, CLIP, etc. You can also base your project on a model that is not part of transformers. For an overview of libraries based on JAX, you can take a look at [awesome-jax](https://github.com/n2cholas/awesome-jax#awesome-jax-). **Note** that for a project that is not based on Transformers it will be more difficult for the 🤗 team to help you. Also have a look at the section [Quickstart Flax & Jax in Transformers](https://github.com/huggingface/transformers/tree/main/examples/research_projects/jax-projects#quickstart-flax-and-jax-in-transformers) to see what model architectures are currently supported in 🤗 Transformers.
4. *What data should be used?* It is important to state at least what kind of data you would like to use. Ideally, you can already point to publicly available data or a dataset in the 🤗 Datasets library.
5. *Are similar training scripts available in Flax/JAX?* It would be important to find similar training scripts that already exist in Flax/JAX. *E.g.* if you are working on a Seq-to-Seq task, you can make use of the [`run_summarization_flax.py`](https://github.com/huggingface/transformers/blob/main/examples/flax/summarization/run_summarization_flax.py) script which is very similar to any seq2seq training. Also have a look at the section [Quickstart Flax & Jax in Transformers](https://github.com/huggingface/transformers/tree/main/examples/research_projects/jax-projects#quickstart-flax-and-jax-in-transformers) to see what training scripts are currently supported in 🤗 Transformers.
6. *(Optionally) What are possible challenges?* List possible difficulties with your project. *E.g.* If you know that training convergence usually takes a lot of time, it is worth stating this here!
7. *(Optionally) What is the desired project outcome?* - How would you like to demo your project? One could *e.g.* create a Streamlit application.
8. *(Optionally) Links to read upon* - Can you provide any links that would help the reader to better understand your project idea?
Feel free to copy-paste the following format for your project proposal and fill out the respective sections:
```
# <FILL ME: Name of project>
<FILL ME: A clear description of the project>
## 2. Language
The model will be trained in <FILL ME: which language?>.
## 3. Model
<FILL ME: 3. Which model should be used?>
## 4. Datasets
<FILL ME: 4. Which data should be used?>
Possible links to publicly available datasets include:
- <FILL ME: Link 1 to dataset>
- <FILL ME: Link 2 to dataset>
- <FILL ME: Link 3 to dataset>
## 5. Training scripts
<FILL ME: 5. Are there publicly available training scripts that can be used/tweaked for the project?>
We can make use of <FILL ME: link to training script> to train the model.>
## 6. (Optional) Challenges
<(Optionally) FILL ME: 6. What are possible challenges?>
## 7. (Optional) Desired project outcome
<(Optionally) FILL ME: 7. What is the desired project outcome? A demo?>
## 8. (Optional) Reads
The following links can be useful to better understand the project and
what has previously been done.
- <FILL ME: Link 1 to read>
- <FILL ME: Link 2 to read>
- <FILL ME: Link 3 to read>
```
To see how a proposed project looks like, please have a look at submitted project
proposals [here](https://discuss.huggingface.co/c/flax-jax-projects/22).
## Will my project proposal be selected?
Having submitted a project proposal, you can now promote your idea in the Slack channel `#flax-jax-community-week` to try to convince other participants to join your project!
Once other people have joined your project, one of the organizers (`@Suzana, @valhalla, @osanseviero, @patrickvonplaten`) will officially create a team for your project and add your project to [this google sheet](https://docs.google.com/spreadsheets/d/1GpHebL7qrwJOc9olTpIPgjf8vOS0jNb6zR_B8x_Jtik/edit?usp=sharing).
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/pt/tasks/token_classification.mdx | <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Classificação de tokens
<Youtube id="wVHdVlPScxA"/>
A classificação de tokens atribui um rótulo a tokens individuais em uma frase. Uma das tarefas de classificação de tokens mais comuns é o Reconhecimento de Entidade Nomeada, também chamada de NER (sigla em inglês para Named Entity Recognition). O NER tenta encontrar um rótulo para cada entidade em uma frase, como uma pessoa, local ou organização.
Este guia mostrará como realizar o fine-tuning do [DistilBERT](https://huggingface.co/distilbert-base-uncased) no conjunto de dados [WNUT 17](https://huggingface.co/datasets/wnut_17) para detectar novas entidades.
<Tip>
Consulte a [página de tarefas de classificação de tokens](https://huggingface.co/tasks/token-classification) para obter mais informações sobre outras formas de classificação de tokens e seus modelos, conjuntos de dados e métricas associadas.
</Tip>
## Carregando o conjunto de dados WNUT 17
Carregue o conjunto de dados WNUT 17 da biblioteca 🤗 Datasets:
```py
>>> from datasets import load_dataset
>>> wnut = load_dataset("wnut_17")
```
E dê uma olhada em um exemplo:
```py
>>> wnut["train"][0]
{'id': '0',
'ner_tags': [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 8, 8, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0],
'tokens': ['@paulwalk', 'It', "'s", 'the', 'view', 'from', 'where', 'I', "'m", 'living', 'for', 'two', 'weeks', '.', 'Empire', 'State', 'Building', '=', 'ESB', '.', 'Pretty', 'bad', 'storm', 'here', 'last', 'evening', '.']
}
```
Cada número em `ner_tags` representa uma entidade. Converta o número em um rótulo para obter mais informações:
```py
>>> label_list = wnut["train"].features[f"ner_tags"].feature.names
>>> label_list
[
"O",
"B-corporation",
"I-corporation",
"B-creative-work",
"I-creative-work",
"B-group",
"I-group",
"B-location",
"I-location",
"B-person",
"I-person",
"B-product",
"I-product",
]
```
O `ner_tag` descreve uma entidade, como uma organização, local ou pessoa. A letra que prefixa cada `ner_tag` indica a posição do token da entidade:
- `B-` indica o início de uma entidade.
- `I-` indica que um token está contido dentro da mesma entidade (por exemplo, o token `State` pode fazer parte de uma entidade como `Empire State Building`).
- `0` indica que o token não corresponde a nenhuma entidade.
## Pré-processamento
<Youtube id="iY2AZYdZAr0"/>
Carregue o tokenizer do DistilBERT para processar os `tokens`:
```py
>>> from transformers import AutoTokenizer
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
```
Como a entrada já foi dividida em palavras, defina `is_split_into_words=True` para tokenizar as palavras em subpalavras:
```py
>>> tokenized_input = tokenizer(example["tokens"], is_split_into_words=True)
>>> tokens = tokenizer.convert_ids_to_tokens(tokenized_input["input_ids"])
>>> tokens
['[CLS]', '@', 'paul', '##walk', 'it', "'", 's', 'the', 'view', 'from', 'where', 'i', "'", 'm', 'living', 'for', 'two', 'weeks', '.', 'empire', 'state', 'building', '=', 'es', '##b', '.', 'pretty', 'bad', 'storm', 'here', 'last', 'evening', '.', '[SEP]']
```
Ao adicionar os tokens especiais `[CLS]` e `[SEP]` e a tokenização de subpalavras uma incompatibilidade é gerada entre a entrada e os rótulos. Uma única palavra correspondente a um único rótulo pode ser dividida em duas subpalavras. Você precisará realinhar os tokens e os rótulos da seguinte forma:
1. Mapeie todos os tokens para a palavra correspondente com o método [`word_ids`](https://huggingface.co/docs/tokenizers/python/latest/api/reference.html#tokenizers.Encoding.word_ids).
2. Atribuindo o rótulo `-100` aos tokens especiais `[CLS]` e `[SEP]` para que a função de loss do PyTorch ignore eles.
3. Rotular apenas o primeiro token de uma determinada palavra. Atribuindo `-100` a outros subtokens da mesma palavra.
Aqui está como você pode criar uma função para realinhar os tokens e rótulos e truncar sequências para não serem maiores que o comprimento máximo de entrada do DistilBERT:
```py
>>> def tokenize_and_align_labels(examples):
... tokenized_inputs = tokenizer(examples["tokens"], truncation=True, is_split_into_words=True)
... labels = []
... for i, label in enumerate(examples[f"ner_tags"]):
... word_ids = tokenized_inputs.word_ids(batch_index=i) # Map tokens to their respective word.
... previous_word_idx = None
... label_ids = []
... for word_idx in word_ids: # Set the special tokens to -100.
... if word_idx is None:
... label_ids.append(-100)
... elif word_idx != previous_word_idx: # Only label the first token of a given word.
... label_ids.append(label[word_idx])
... else:
... label_ids.append(-100)
... previous_word_idx = word_idx
... labels.append(label_ids)
... tokenized_inputs["labels"] = labels
... return tokenized_inputs
```
Use a função [`map`](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.map) do 🤗 Datasets para tokenizar e alinhar os rótulos em todo o conjunto de dados. Você pode acelerar a função `map` configurando `batched=True` para processar vários elementos do conjunto de dados de uma só vez:
```py
>>> tokenized_wnut = wnut.map(tokenize_and_align_labels, batched=True)
```
Use o [`DataCollatorForTokenClassification`] para criar um batch de exemplos. Ele também *preencherá dinamicamente* seu texto e rótulos para o comprimento do elemento mais longo em seu batch, para que tenham um comprimento uniforme. Embora seja possível preencher seu texto na função `tokenizer` configurando `padding=True`, o preenchimento dinâmico é mais eficiente.
<frameworkcontent>
<pt>
```py
>>> from transformers import DataCollatorForTokenClassification
>>> data_collator = DataCollatorForTokenClassification(tokenizer=tokenizer)
```
</pt>
<tf>
```py
>>> from transformers import DataCollatorForTokenClassification
>>> data_collator = DataCollatorForTokenClassification(tokenizer=tokenizer, return_tensors="tf")
```
</tf>
</frameworkcontent>
## Treinamento
<frameworkcontent>
<pt>
Carregue o DistilBERT com o [`AutoModelForTokenClassification`] junto com o número de rótulos esperados:
```py
>>> from transformers import AutoModelForTokenClassification, TrainingArguments, Trainer
>>> model = AutoModelForTokenClassification.from_pretrained("distilbert-base-uncased", num_labels=14)
```
<Tip>
Se você não estiver familiarizado com o fine-tuning de um modelo com o [`Trainer`], dê uma olhada no tutorial básico [aqui](../training#finetune-with-trainer)!
</Tip>
Nesse ponto, restam apenas três passos:
1. Definir seus hiperparâmetros de treinamento em [`TrainingArguments`].
2. Passar os argumentos de treinamento para o [`Trainer`] junto com o modelo, conjunto de dados, tokenizador e o data collator.
3. Chamar a função [`~Trainer.train`] para executar o fine-tuning do seu modelo.
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,
... num_train_epochs=3,
... weight_decay=0.01,
... )
>>> trainer = Trainer(
... model=model,
... args=training_args,
... train_dataset=tokenized_wnut["train"],
... eval_dataset=tokenized_wnut["test"],
... tokenizer=tokenizer,
... data_collator=data_collator,
... )
>>> trainer.train()
```
</pt>
<tf>
Para executar o fine-tuning de um modelo no TensorFlow, comece convertendo seu conjunto de dados para o formato `tf.data.Dataset` com [`to_tf_dataset`](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.to_tf_dataset). Nessa execução você deverá especificar as entradas e rótulos (no parâmetro `columns`), se deseja embaralhar o conjunto de dados, o tamanho do batch e o data collator:
```py
>>> tf_train_set = tokenized_wnut["train"].to_tf_dataset(
... columns=["attention_mask", "input_ids", "labels"],
... shuffle=True,
... batch_size=16,
... collate_fn=data_collator,
... )
>>> tf_validation_set = tokenized_wnut["validation"].to_tf_dataset(
... columns=["attention_mask", "input_ids", "labels"],
... shuffle=False,
... batch_size=16,
... collate_fn=data_collator,
... )
```
<Tip>
Se você não estiver familiarizado com o fine-tuning de um modelo com o Keras, dê uma olhada no tutorial básico [aqui](training#finetune-with-keras)!
</Tip>
Configure o otimizador e alguns hiperparâmetros de treinamento:
```py
>>> from transformers import create_optimizer
>>> batch_size = 16
>>> num_train_epochs = 3
>>> num_train_steps = (len(tokenized_wnut["train"]) // batch_size) * num_train_epochs
>>> optimizer, lr_schedule = create_optimizer(
... init_lr=2e-5,
... num_train_steps=num_train_steps,
... weight_decay_rate=0.01,
... num_warmup_steps=0,
... )
```
Carregue o DistilBERT com o [`TFAutoModelForTokenClassification`] junto com o número de rótulos esperados:
```py
>>> from transformers import TFAutoModelForTokenClassification
>>> model = TFAutoModelForTokenClassification.from_pretrained("distilbert-base-uncased", num_labels=2)
```
Configure o modelo para treinamento com o método [`compile`](https://keras.io/api/models/model_training_apis/#compile-method):
```py
>>> import tensorflow as tf
>>> model.compile(optimizer=optimizer)
```
Chame o método [`fit`](https://keras.io/api/models/model_training_apis/#fit-method) para executar o fine-tuning do modelo:
```py
>>> model.fit(x=tf_train_set, validation_data=tf_validation_set, epochs=3)
```
</tf>
</frameworkcontent>
<Tip>
Para obter um exemplo mais aprofundado de como executar o fine-tuning de um modelo para classificação de tokens, dê uma olhada nesse [notebook utilizando PyTorch](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification.ipynb) ou nesse [notebook utilizando TensorFlow](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification-tf.ipynb).
</Tip> | <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Classificação de tokens
<Youtube id="wVHdVlPScxA"/>
A classificação de tokens atribui um rótulo a tokens individuais em uma frase. Uma das tarefas de classificação de tokens mais comuns é o Reconhecimento de Entidade Nomeada, também chamada de NER (sigla em inglês para Named Entity Recognition). O NER tenta encontrar um rótulo para cada entidade em uma frase, como uma pessoa, local ou organização.
Este guia mostrará como realizar o fine-tuning do [DistilBERT](https://huggingface.co/distilbert-base-uncased) no conjunto de dados [WNUT 17](https://huggingface.co/datasets/wnut_17) para detectar novas entidades.
<Tip>
Consulte a [página de tarefas de classificação de tokens](https://huggingface.co/tasks/token-classification) para obter mais informações sobre outras formas de classificação de tokens e seus modelos, conjuntos de dados e métricas associadas.
</Tip>
## Carregando o conjunto de dados WNUT 17
Carregue o conjunto de dados WNUT 17 da biblioteca 🤗 Datasets:
```py
>>> from datasets import load_dataset
>>> wnut = load_dataset("wnut_17")
```
E dê uma olhada em um exemplo:
```py
>>> wnut["train"][0]
{'id': '0',
'ner_tags': [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 8, 8, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0],
'tokens': ['@paulwalk', 'It', "'s", 'the', 'view', 'from', 'where', 'I', "'m", 'living', 'for', 'two', 'weeks', '.', 'Empire', 'State', 'Building', '=', 'ESB', '.', 'Pretty', 'bad', 'storm', 'here', 'last', 'evening', '.']
}
```
Cada número em `ner_tags` representa uma entidade. Converta o número em um rótulo para obter mais informações:
```py
>>> label_list = wnut["train"].features[f"ner_tags"].feature.names
>>> label_list
[
"O",
"B-corporation",
"I-corporation",
"B-creative-work",
"I-creative-work",
"B-group",
"I-group",
"B-location",
"I-location",
"B-person",
"I-person",
"B-product",
"I-product",
]
```
O `ner_tag` descreve uma entidade, como uma organização, local ou pessoa. A letra que prefixa cada `ner_tag` indica a posição do token da entidade:
- `B-` indica o início de uma entidade.
- `I-` indica que um token está contido dentro da mesma entidade (por exemplo, o token `State` pode fazer parte de uma entidade como `Empire State Building`).
- `0` indica que o token não corresponde a nenhuma entidade.
## Pré-processamento
<Youtube id="iY2AZYdZAr0"/>
Carregue o tokenizer do DistilBERT para processar os `tokens`:
```py
>>> from transformers import AutoTokenizer
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
```
Como a entrada já foi dividida em palavras, defina `is_split_into_words=True` para tokenizar as palavras em subpalavras:
```py
>>> tokenized_input = tokenizer(example["tokens"], is_split_into_words=True)
>>> tokens = tokenizer.convert_ids_to_tokens(tokenized_input["input_ids"])
>>> tokens
['[CLS]', '@', 'paul', '##walk', 'it', "'", 's', 'the', 'view', 'from', 'where', 'i', "'", 'm', 'living', 'for', 'two', 'weeks', '.', 'empire', 'state', 'building', '=', 'es', '##b', '.', 'pretty', 'bad', 'storm', 'here', 'last', 'evening', '.', '[SEP]']
```
Ao adicionar os tokens especiais `[CLS]` e `[SEP]` e a tokenização de subpalavras uma incompatibilidade é gerada entre a entrada e os rótulos. Uma única palavra correspondente a um único rótulo pode ser dividida em duas subpalavras. Você precisará realinhar os tokens e os rótulos da seguinte forma:
1. Mapeie todos os tokens para a palavra correspondente com o método [`word_ids`](https://huggingface.co/docs/tokenizers/python/latest/api/reference.html#tokenizers.Encoding.word_ids).
2. Atribuindo o rótulo `-100` aos tokens especiais `[CLS]` e `[SEP]` para que a função de loss do PyTorch ignore eles.
3. Rotular apenas o primeiro token de uma determinada palavra. Atribuindo `-100` a outros subtokens da mesma palavra.
Aqui está como você pode criar uma função para realinhar os tokens e rótulos e truncar sequências para não serem maiores que o comprimento máximo de entrada do DistilBERT:
```py
>>> def tokenize_and_align_labels(examples):
... tokenized_inputs = tokenizer(examples["tokens"], truncation=True, is_split_into_words=True)
... labels = []
... for i, label in enumerate(examples[f"ner_tags"]):
... word_ids = tokenized_inputs.word_ids(batch_index=i) # Map tokens to their respective word.
... previous_word_idx = None
... label_ids = []
... for word_idx in word_ids: # Set the special tokens to -100.
... if word_idx is None:
... label_ids.append(-100)
... elif word_idx != previous_word_idx: # Only label the first token of a given word.
... label_ids.append(label[word_idx])
... else:
... label_ids.append(-100)
... previous_word_idx = word_idx
... labels.append(label_ids)
... tokenized_inputs["labels"] = labels
... return tokenized_inputs
```
Use a função [`map`](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.map) do 🤗 Datasets para tokenizar e alinhar os rótulos em todo o conjunto de dados. Você pode acelerar a função `map` configurando `batched=True` para processar vários elementos do conjunto de dados de uma só vez:
```py
>>> tokenized_wnut = wnut.map(tokenize_and_align_labels, batched=True)
```
Use o [`DataCollatorForTokenClassification`] para criar um batch de exemplos. Ele também *preencherá dinamicamente* seu texto e rótulos para o comprimento do elemento mais longo em seu batch, para que tenham um comprimento uniforme. Embora seja possível preencher seu texto na função `tokenizer` configurando `padding=True`, o preenchimento dinâmico é mais eficiente.
<frameworkcontent>
<pt>
```py
>>> from transformers import DataCollatorForTokenClassification
>>> data_collator = DataCollatorForTokenClassification(tokenizer=tokenizer)
```
</pt>
<tf>
```py
>>> from transformers import DataCollatorForTokenClassification
>>> data_collator = DataCollatorForTokenClassification(tokenizer=tokenizer, return_tensors="tf")
```
</tf>
</frameworkcontent>
## Treinamento
<frameworkcontent>
<pt>
Carregue o DistilBERT com o [`AutoModelForTokenClassification`] junto com o número de rótulos esperados:
```py
>>> from transformers import AutoModelForTokenClassification, TrainingArguments, Trainer
>>> model = AutoModelForTokenClassification.from_pretrained("distilbert-base-uncased", num_labels=14)
```
<Tip>
Se você não estiver familiarizado com o fine-tuning de um modelo com o [`Trainer`], dê uma olhada no tutorial básico [aqui](../training#finetune-with-trainer)!
</Tip>
Nesse ponto, restam apenas três passos:
1. Definir seus hiperparâmetros de treinamento em [`TrainingArguments`].
2. Passar os argumentos de treinamento para o [`Trainer`] junto com o modelo, conjunto de dados, tokenizador e o data collator.
3. Chamar a função [`~Trainer.train`] para executar o fine-tuning do seu modelo.
```py
>>> training_args = TrainingArguments(
... output_dir="./results",
... evaluation_strategy="epoch",
... learning_rate=2e-5,
... per_device_train_batch_size=16,
... per_device_eval_batch_size=16,
... num_train_epochs=3,
... weight_decay=0.01,
... )
>>> trainer = Trainer(
... model=model,
... args=training_args,
... train_dataset=tokenized_wnut["train"],
... eval_dataset=tokenized_wnut["test"],
... tokenizer=tokenizer,
... data_collator=data_collator,
... )
>>> trainer.train()
```
</pt>
<tf>
Para executar o fine-tuning de um modelo no TensorFlow, comece convertendo seu conjunto de dados para o formato `tf.data.Dataset` com [`to_tf_dataset`](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.to_tf_dataset). Nessa execução você deverá especificar as entradas e rótulos (no parâmetro `columns`), se deseja embaralhar o conjunto de dados, o tamanho do batch e o data collator:
```py
>>> tf_train_set = tokenized_wnut["train"].to_tf_dataset(
... columns=["attention_mask", "input_ids", "labels"],
... shuffle=True,
... batch_size=16,
... collate_fn=data_collator,
... )
>>> tf_validation_set = tokenized_wnut["validation"].to_tf_dataset(
... columns=["attention_mask", "input_ids", "labels"],
... shuffle=False,
... batch_size=16,
... collate_fn=data_collator,
... )
```
<Tip>
Se você não estiver familiarizado com o fine-tuning de um modelo com o Keras, dê uma olhada no tutorial básico [aqui](training#finetune-with-keras)!
</Tip>
Configure o otimizador e alguns hiperparâmetros de treinamento:
```py
>>> from transformers import create_optimizer
>>> batch_size = 16
>>> num_train_epochs = 3
>>> num_train_steps = (len(tokenized_wnut["train"]) // batch_size) * num_train_epochs
>>> optimizer, lr_schedule = create_optimizer(
... init_lr=2e-5,
... num_train_steps=num_train_steps,
... weight_decay_rate=0.01,
... num_warmup_steps=0,
... )
```
Carregue o DistilBERT com o [`TFAutoModelForTokenClassification`] junto com o número de rótulos esperados:
```py
>>> from transformers import TFAutoModelForTokenClassification
>>> model = TFAutoModelForTokenClassification.from_pretrained("distilbert-base-uncased", num_labels=2)
```
Configure o modelo para treinamento com o método [`compile`](https://keras.io/api/models/model_training_apis/#compile-method):
```py
>>> import tensorflow as tf
>>> model.compile(optimizer=optimizer)
```
Chame o método [`fit`](https://keras.io/api/models/model_training_apis/#fit-method) para executar o fine-tuning do modelo:
```py
>>> model.fit(x=tf_train_set, validation_data=tf_validation_set, epochs=3)
```
</tf>
</frameworkcontent>
<Tip>
Para obter um exemplo mais aprofundado de como executar o fine-tuning de um modelo para classificação de tokens, dê uma olhada nesse [notebook utilizando PyTorch](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification.ipynb) ou nesse [notebook utilizando TensorFlow](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification-tf.ipynb).
</Tip> | -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/es/_config.py | # docstyle-ignore
INSTALL_CONTENT = """
# Transformers installation
! pip install transformers datasets
# To install from source instead of the last release, comment the command above and uncomment the following one.
# ! pip install git+https://github.com/huggingface/transformers.git
"""
notebook_first_cells = [{"type": "code", "content": INSTALL_CONTENT}]
black_avoid_patterns = {
"{processor_class}": "FakeProcessorClass",
"{model_class}": "FakeModelClass",
"{object_class}": "FakeObjectClass",
}
| # docstyle-ignore
INSTALL_CONTENT = """
# Transformers installation
! pip install transformers datasets
# To install from source instead of the last release, comment the command above and uncomment the following one.
# ! pip install git+https://github.com/huggingface/transformers.git
"""
notebook_first_cells = [{"type": "code", "content": INSTALL_CONTENT}]
black_avoid_patterns = {
"{processor_class}": "FakeProcessorClass",
"{model_class}": "FakeModelClass",
"{object_class}": "FakeObjectClass",
}
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./src/transformers/models/flava/image_processing_flava.py | # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Image processor class for Flava."""
import math
import random
from functools import lru_cache
from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
import numpy as np
from transformers.utils import is_vision_available
from transformers.utils.generic import TensorType
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import center_crop, normalize, rescale, resize, to_channel_dimension_format
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import logging
if is_vision_available():
import PIL
logger = logging.get_logger(__name__)
# These values are taken from CLIP
FLAVA_IMAGE_MEAN = OPENAI_CLIP_MEAN
FLAVA_IMAGE_STD = OPENAI_CLIP_STD
FLAVA_CODEBOOK_MEAN = [0.0, 0.0, 0.0]
FLAVA_CODEBOOK_STD = [1.0, 1.0, 1.0]
LOGIT_LAPLACE_EPS: float = 0.1
# Inspired from https://github.com/microsoft/unilm/blob/master/beit/masking_generator.py
class FlavaMaskingGenerator:
def __init__(
self,
input_size: Union[int, Tuple[int, int]] = 14,
total_mask_patches: int = 75,
mask_group_max_patches: Optional[int] = None,
mask_group_min_patches: int = 16,
mask_group_min_aspect_ratio: Optional[float] = 0.3,
mask_group_max_aspect_ratio: float = None,
):
if not isinstance(input_size, tuple):
input_size = (input_size,) * 2
self.height, self.width = input_size
self.num_patches = self.height * self.width
self.total_mask_patches = total_mask_patches
self.mask_group_min_patches = mask_group_min_patches
self.mask_group_max_patches = total_mask_patches if mask_group_max_patches is None else mask_group_max_patches
mask_group_max_aspect_ratio = mask_group_max_aspect_ratio or 1 / mask_group_min_aspect_ratio
self.log_aspect_ratio = (math.log(mask_group_min_aspect_ratio), math.log(mask_group_max_aspect_ratio))
def __repr__(self):
repr_str = "MaskingGenerator(%d, %d -> [%d ~ %d], max = %d, %.3f ~ %.3f)" % (
self.height,
self.width,
self.mask_group_min_patches,
self.mask_group_max_patches,
self.total_mask_patches,
self.log_aspect_ratio[0],
self.log_aspect_ratio[1],
)
return repr_str
def get_shape(self):
return self.height, self.width
def _mask(self, mask, max_mask_patches):
delta = 0
for _attempt in range(10):
target_area = random.uniform(self.mask_group_min_patches, max_mask_patches)
aspect_ratio = math.exp(random.uniform(*self.log_aspect_ratio))
height = int(round(math.sqrt(target_area * aspect_ratio)))
width = int(round(math.sqrt(target_area / aspect_ratio)))
if width < self.width and height < self.height:
top = random.randint(0, self.height - height)
left = random.randint(0, self.width - width)
num_masked = mask[top : top + height, left : left + width].sum()
# Overlap
if 0 < height * width - num_masked <= max_mask_patches:
for i in range(top, top + height):
for j in range(left, left + width):
if mask[i, j] == 0:
mask[i, j] = 1
delta += 1
if delta > 0:
break
return delta
def __call__(self):
mask = np.zeros(shape=self.get_shape(), dtype=int)
mask_count = 0
while mask_count < self.total_mask_patches:
max_mask_patches = self.total_mask_patches - mask_count
max_mask_patches = min(max_mask_patches, self.mask_group_max_patches)
delta = self._mask(mask, max_mask_patches)
if delta == 0:
break
else:
mask_count += delta
return mask
class FlavaImageProcessor(BaseImageProcessor):
r"""
Constructs a Flava image processor.
Args:
do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the image's (height, width) dimensions to the specified `size`. Can be overridden by the
`do_resize` parameter in `preprocess`.
size (`Dict[str, int]` *optional*, defaults to `{"height": 224, "width": 224}`):
Size of the image after resizing. Can be overridden by the `size` parameter in `preprocess`.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BICUBIC`):
Resampling filter to use if resizing the image. Can be overridden by the `resample` parameter in
`preprocess`.
do_center_crop (`bool`, *optional*, defaults to `True`):
Whether to center crop the images. Can be overridden by the `do_center_crop` parameter in `preprocess`.
crop_size (`Dict[str, int]` *optional*, defaults to `{"height": 224, "width": 224}`):
Size of image after the center crop `(crop_size["height"], crop_size["width"])`. Can be overridden by the
`crop_size` parameter in `preprocess`.
do_rescale (`bool`, *optional*, defaults to `True`):
Whether to rescale the image by the specified scale `rescale_factor`. Can be overridden by the `do_rescale`
parameter in `preprocess`.
rescale_factor (`int` or `float`, *optional*, defaults to `1/255`):
Scale factor to use if rescaling the image. Can be overridden by the `rescale_factor` parameter in
`preprocess`.
do_normalize (`bool`, *optional*, defaults to `True`):
Whether to normalize the image. Can be overridden by the `do_normalize` parameter in `preprocess`.
image_mean (`float` or `List[float]`, *optional*, defaults to `IMAGENET_STANDARD_MEAN`):
Mean to use if normalizing the image. This is a float or list of floats the length of the number of
channels in the image. Can be overridden by the `image_mean` parameter in the `preprocess` method.
image_std (`float` or `List[float]`, *optional*, defaults to `IMAGENET_STANDARD_STD`):
Standard deviation to use if normalizing the image. This is a float or list of floats the length of the
number of channels in the image. Can be overridden by the `image_std` parameter in the `preprocess` method.
return_image_mask (`bool`, *optional*, defaults to `False`):
Whether to return the image mask. Can be overridden by the `return_image_mask` parameter in `preprocess`.
input_size_patches (`int`, *optional*, defaults to 14):
Number of patches in the image in height and width direction. 14x14 = 196 total patches. Can be overridden
by the `input_size_patches` parameter in `preprocess`.
total_mask_patches (`int`, *optional*, defaults to 75):
Total number of patches that should be masked. Can be overridden by the `total_mask_patches` parameter in
`preprocess`.
mask_group_min_patches (`int`, *optional*, defaults to 16):
Minimum number of patches that should be masked. Can be overridden by the `mask_group_min_patches`
parameter in `preprocess`.
mask_group_max_patches (`int`, *optional*):
Maximum number of patches that should be masked. Can be overridden by the `mask_group_max_patches`
parameter in `preprocess`.
mask_group_min_aspect_ratio (`float`, *optional*, defaults to 0.3):
Minimum aspect ratio of the mask window. Can be overridden by the `mask_group_min_aspect_ratio` parameter
in `preprocess`.
mask_group_max_aspect_ratio (`float`, *optional*):
Maximum aspect ratio of the mask window. Can be overridden by the `mask_group_max_aspect_ratio` parameter
in `preprocess`.
codebook_do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the input for codebook to a certain. Can be overridden by the `codebook_do_resize`
parameter in `preprocess`. `codebook_size`.
codebook_size (`Dict[str, int]`, *optional*, defaults to `{"height": 224, "width": 224}`):
Resize the input for codebook to the given size. Can be overridden by the `codebook_size` parameter in
`preprocess`.
codebook_resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.LANCZOS`):
Resampling filter to use if resizing the codebook image. Can be overridden by the `codebook_resample`
parameter in `preprocess`.
codebook_do_center_crop (`bool`, *optional*, defaults to `True`):
Whether to crop the input for codebook at the center. If the input size is smaller than
`codebook_crop_size` along any edge, the image is padded with 0's and then center cropped. Can be
overridden by the `codebook_do_center_crop` parameter in `preprocess`.
codebook_crop_size (`Dict[str, int]`, *optional*, defaults to `{"height": 224, "width": 224}`):
Desired output size for codebook input when applying center-cropping. Can be overridden by the
`codebook_crop_size` parameter in `preprocess`.
codebook_do_rescale (`bool`, *optional*, defaults to `True`):
Whether to rescale the input for codebook by the specified scale `codebook_rescale_factor`. Can be
overridden by the `codebook_do_rescale` parameter in `preprocess`.
codebook_rescale_factor (`int` or `float`, *optional*, defaults to `1/255`):
Defines the scale factor to use if rescaling the codebook image. Can be overridden by the
`codebook_rescale_factor` parameter in `preprocess`.
codebook_do_map_pixels (`bool`, *optional*, defaults to `True`):
Whether to map the pixel values of the codebook input to (1 - 2e)x + e. Can be overridden by the
`codebook_do_map_pixels` parameter in `preprocess`.
codebook_do_normalize (`bool`, *optional*, defaults to `True`):
Whether or not to normalize the input for codebook with `codebook_image_mean` and `codebook_image_std`. Can
be overridden by the `codebook_do_normalize` parameter in `preprocess`.
codebook_image_mean (`Optional[Union[float, Iterable[float]]]`, *optional*, defaults to `[0, 0, 0]`):
The sequence of means for each channel, to be used when normalizing images for codebook. Can be overridden
by the `codebook_image_mean` parameter in `preprocess`.
codebook_image_std (`Optional[Union[float, Iterable[float]]]`, *optional*, defaults to `[0.5, 0.5, 0.5]`):
The sequence of standard deviations for each channel, to be used when normalizing images for codebook. Can
be overridden by the `codebook_image_std` parameter in `preprocess`.
"""
model_input_names = ["pixel_values"]
def __init__(
self,
do_resize: bool = True,
size: Dict[str, int] = None,
resample: PILImageResampling = PILImageResampling.BICUBIC,
do_center_crop: bool = True,
crop_size: Dict[str, int] = None,
do_rescale: bool = True,
rescale_factor: Union[int, float] = 1 / 255,
do_normalize: bool = True,
image_mean: Optional[Union[float, Iterable[float]]] = None,
image_std: Optional[Union[float, Iterable[float]]] = None,
# Mask related params
return_image_mask: bool = False,
input_size_patches: int = 14,
total_mask_patches: int = 75,
mask_group_min_patches: int = 16,
mask_group_max_patches: Optional[int] = None,
mask_group_min_aspect_ratio: float = 0.3,
mask_group_max_aspect_ratio: Optional[float] = None,
# Codebook related params
return_codebook_pixels: bool = False,
codebook_do_resize: bool = True,
codebook_size: bool = None,
codebook_resample: int = PILImageResampling.LANCZOS,
codebook_do_center_crop: bool = True,
codebook_crop_size: int = None,
codebook_do_rescale: bool = True,
codebook_rescale_factor: Union[int, float] = 1 / 255,
codebook_do_map_pixels: bool = True,
codebook_do_normalize: bool = True,
codebook_image_mean: Optional[Union[float, Iterable[float]]] = None,
codebook_image_std: Optional[Union[float, Iterable[float]]] = None,
**kwargs,
) -> None:
super().__init__(**kwargs)
size = size if size is not None else {"height": 224, "width": 224}
size = get_size_dict(size)
crop_size = crop_size if crop_size is not None else {"height": 224, "width": 224}
crop_size = get_size_dict(crop_size, param_name="crop_size")
codebook_size = codebook_size if codebook_size is not None else {"height": 112, "width": 112}
codebook_size = get_size_dict(codebook_size, param_name="codebook_size")
codebook_crop_size = codebook_crop_size if codebook_crop_size is not None else {"height": 112, "width": 112}
codebook_crop_size = get_size_dict(codebook_crop_size, param_name="codebook_crop_size")
self.do_resize = do_resize
self.size = size
self.resample = resample
self.do_rescale = do_rescale
self.rescale_factor = rescale_factor
self.do_center_crop = do_center_crop
self.crop_size = crop_size
self.do_normalize = do_normalize
self.image_mean = image_mean if image_mean is not None else FLAVA_IMAGE_MEAN
self.image_std = image_std if image_std is not None else FLAVA_IMAGE_STD
self.return_image_mask = return_image_mask
self.input_size_patches = input_size_patches
self.total_mask_patches = total_mask_patches
self.mask_group_min_patches = mask_group_min_patches
self.mask_group_max_patches = mask_group_max_patches
self.mask_group_min_aspect_ratio = mask_group_min_aspect_ratio
self.mask_group_max_aspect_ratio = mask_group_max_aspect_ratio
self.return_codebook_pixels = return_codebook_pixels
self.codebook_do_resize = codebook_do_resize
self.codebook_size = codebook_size
self.codebook_resample = codebook_resample
self.codebook_do_center_crop = codebook_do_center_crop
self.codebook_crop_size = codebook_crop_size
self.codebook_do_rescale = codebook_do_rescale
self.codebook_rescale_factor = codebook_rescale_factor
self.codebook_do_map_pixels = codebook_do_map_pixels
self.codebook_do_normalize = codebook_do_normalize
self.codebook_image_mean = codebook_image_mean
self.codebook_image_mean = codebook_image_mean if codebook_image_mean is not None else FLAVA_CODEBOOK_MEAN
self.codebook_image_std = codebook_image_std if codebook_image_std is not None else FLAVA_CODEBOOK_STD
@classmethod
def from_dict(cls, image_processor_dict: Dict[str, Any], **kwargs):
"""
Overrides the `from_dict` method from the base class to make sure parameters are updated if image processor is
created using from_dict and kwargs e.g. `FlavaImageProcessor.from_pretrained(checkpoint, codebook_size=600)`
"""
image_processor_dict = image_processor_dict.copy()
if "codebook_size" in kwargs:
image_processor_dict["codebook_size"] = kwargs.pop("codebook_size")
if "codebook_crop_size" in kwargs:
image_processor_dict["codebook_crop_size"] = kwargs.pop("codebook_crop_size")
return super().from_dict(image_processor_dict, **kwargs)
@lru_cache()
def masking_generator(
self,
input_size_patches,
total_mask_patches,
mask_group_min_patches,
mask_group_max_patches,
mask_group_min_aspect_ratio,
mask_group_max_aspect_ratio,
) -> FlavaMaskingGenerator:
return FlavaMaskingGenerator(
input_size=input_size_patches,
total_mask_patches=total_mask_patches,
mask_group_min_patches=mask_group_min_patches,
mask_group_max_patches=mask_group_max_patches,
mask_group_min_aspect_ratio=mask_group_min_aspect_ratio,
mask_group_max_aspect_ratio=mask_group_max_aspect_ratio,
)
def resize(
self,
image: np.ndarray,
size: Dict[str, int],
resample: PILImageResampling = PILImageResampling.BICUBIC,
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
) -> np.ndarray:
"""
Resize an image to `(size["height"], size["width"])`.
Args:
image (`np.ndarray`):
Image to resize.
size (`Dict[str, int]`):
Size of the output image.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BICUBIC`):
Resampling filter to use when resiizing the image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
size = get_size_dict(size)
if "height" not in size or "width" not in size:
raise ValueError(f"The size dictionary must contain 'height' and 'width' keys. Got {size.keys()}")
return resize(
image, size=(size["height"], size["width"]), resample=resample, data_format=data_format, **kwargs
)
def center_crop(
self,
image: np.ndarray,
size: Dict[str, int],
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
) -> np.ndarray:
"""
Center crop an image to `(size["height"], size["width"])`. If the input size is smaller than `crop_size` along
any edge, the image is padded with 0's and then center cropped.
Args:
image (`np.ndarray`):
Image to center crop.
size (`Dict[str, int]`):
Size of the output image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
size = get_size_dict(size)
if "height" not in size or "width" not in size:
raise ValueError(f"The size dictionary must contain 'height' and 'width' keys. Got {size.keys()}")
return center_crop(image, size=(size["height"], size["width"]), data_format=data_format, **kwargs)
def rescale(
self,
image: np.ndarray,
scale: Union[int, float],
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
):
"""
Rescale an image by a scale factor. image = image * scale.
Args:
image (`np.ndarray`):
Image to rescale.
scale (`int` or `float`):
Scale to apply to the image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
return rescale(image, scale=scale, data_format=data_format, **kwargs)
def normalize(
self,
image: np.ndarray,
mean: Union[float, List[float]],
std: Union[float, List[float]],
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
) -> np.ndarray:
"""
Normalize an image. image = (image - image_mean) / image_std.
Args:
image (`np.ndarray`):
Image to normalize.
image_mean (`float` or `List[float]`):
Image mean.
image_std (`float` or `List[float]`):
Image standard deviation.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
return normalize(image, mean=mean, std=std, data_format=data_format, **kwargs)
def map_pixels(self, image: np.ndarray) -> np.ndarray:
return (1 - 2 * LOGIT_LAPLACE_EPS) * image + LOGIT_LAPLACE_EPS
def _preprocess_image(
self,
image: ImageInput,
do_resize: bool = None,
size: Dict[str, int] = None,
resample: PILImageResampling = None,
do_center_crop: bool = None,
crop_size: Dict[str, int] = None,
do_rescale: bool = None,
rescale_factor: float = None,
do_normalize: bool = None,
image_mean: Optional[Union[float, List[float]]] = None,
image_std: Optional[Union[float, List[float]]] = None,
do_map_pixels: bool = None,
data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
) -> np.ndarray:
"""Preprocesses a single image."""
if do_resize and size is None or resample is None:
raise ValueError("Size and resample must be specified if do_resize is True.")
if do_rescale and rescale_factor is None:
raise ValueError("Rescale factor must be specified if do_rescale is True.")
if do_normalize and (image_mean is None or image_std is None):
raise ValueError("Image mean and std must be specified if do_normalize is True.")
# All transformations expect numpy arrays.
image = to_numpy_array(image)
if do_resize:
image = self.resize(image=image, size=size, resample=resample)
if do_center_crop:
image = self.center_crop(image=image, size=crop_size)
if do_rescale:
image = self.rescale(image=image, scale=rescale_factor)
if do_normalize:
image = self.normalize(image=image, mean=image_mean, std=image_std)
if do_map_pixels:
image = self.map_pixels(image)
if data_format is not None:
image = to_channel_dimension_format(image, data_format)
return image
def preprocess(
self,
images: ImageInput,
do_resize: Optional[bool] = None,
size: Dict[str, int] = None,
resample: PILImageResampling = None,
do_center_crop: Optional[bool] = None,
crop_size: Optional[Dict[str, int]] = None,
do_rescale: Optional[bool] = None,
rescale_factor: Optional[float] = None,
do_normalize: Optional[bool] = None,
image_mean: Optional[Union[float, List[float]]] = None,
image_std: Optional[Union[float, List[float]]] = None,
# Mask related params
return_image_mask: Optional[bool] = None,
input_size_patches: Optional[int] = None,
total_mask_patches: Optional[int] = None,
mask_group_min_patches: Optional[int] = None,
mask_group_max_patches: Optional[int] = None,
mask_group_min_aspect_ratio: Optional[float] = None,
mask_group_max_aspect_ratio: Optional[float] = None,
# Codebook related params
return_codebook_pixels: Optional[bool] = None,
codebook_do_resize: Optional[bool] = None,
codebook_size: Optional[Dict[str, int]] = None,
codebook_resample: Optional[int] = None,
codebook_do_center_crop: Optional[bool] = None,
codebook_crop_size: Optional[Dict[str, int]] = None,
codebook_do_rescale: Optional[bool] = None,
codebook_rescale_factor: Optional[float] = None,
codebook_do_map_pixels: Optional[bool] = None,
codebook_do_normalize: Optional[bool] = None,
codebook_image_mean: Optional[Iterable[float]] = None,
codebook_image_std: Optional[Iterable[float]] = None,
return_tensors: Optional[Union[str, TensorType]] = None,
data_format: ChannelDimension = ChannelDimension.FIRST,
**kwargs,
) -> PIL.Image.Image:
"""
Preprocess an image or batch of images.
Args:
images (`ImageInput`):
Image to preprocess.
do_resize (`bool`, *optional*, defaults to `self.do_resize`):
Whether to resize the image.
size (`Dict[str, int]`, *optional*, defaults to `self.size`):
Size of the image.
resample (`int`, *optional*, defaults to `self.resample`):
Resampling filter to use if resizing the image. This can be one of the enum `PILImageResampling`, Only
has an effect if `do_resize` is set to `True`.
do_center_crop (`bool`, *optional*, defaults to `self.do_center_crop`):
Whether to center crop the image.
crop_size (`Dict[str, int]`, *optional*, defaults to `self.crop_size`):
Size of the center crop. Only has an effect if `do_center_crop` is set to `True`.
do_rescale (`bool`, *optional*, defaults to `self.do_rescale`):
Whether to rescale the image values between [0 - 1].
rescale_factor (`float`, *optional*, defaults to `self.rescale_factor`):
Rescale factor to rescale the image by if `do_rescale` is set to `True`.
do_normalize (`bool`, *optional*, defaults to `self.do_normalize`):
Whether to normalize the image.
image_mean (`float` or `List[float]`, *optional*, defaults to `self.image_mean`):
Image mean.
image_std (`float` or `List[float]`, *optional*, defaults to `self.image_std`):
Image standard deviation.
return_image_mask (`bool`, *optional*, defaults to `self.return_image_mask`):
Whether to return the image mask.
input_size_patches (`int`, *optional*, defaults to `self.input_size_patches`):
Size of the patches to extract from the image.
total_mask_patches (`int`, *optional*, defaults to `self.total_mask_patches`):
Total number of patches to extract from the image.
mask_group_min_patches (`int`, *optional*, defaults to `self.mask_group_min_patches`):
Minimum number of patches to extract from the image.
mask_group_max_patches (`int`, *optional*, defaults to `self.mask_group_max_patches`):
Maximum number of patches to extract from the image.
mask_group_min_aspect_ratio (`float`, *optional*, defaults to `self.mask_group_min_aspect_ratio`):
Minimum aspect ratio of the patches to extract from the image.
mask_group_max_aspect_ratio (`float`, *optional*, defaults to `self.mask_group_max_aspect_ratio`):
Maximum aspect ratio of the patches to extract from the image.
return_codebook_pixels (`bool`, *optional*, defaults to `self.return_codebook_pixels`):
Whether to return the codebook pixels.
codebook_do_resize (`bool`, *optional*, defaults to `self.codebook_do_resize`):
Whether to resize the codebook pixels.
codebook_size (`Dict[str, int]`, *optional*, defaults to `self.codebook_size`):
Size of the codebook pixels.
codebook_resample (`int`, *optional*, defaults to `self.codebook_resample`):
Resampling filter to use if resizing the codebook pixels. This can be one of the enum
`PILImageResampling`, Only has an effect if `codebook_do_resize` is set to `True`.
codebook_do_center_crop (`bool`, *optional*, defaults to `self.codebook_do_center_crop`):
Whether to center crop the codebook pixels.
codebook_crop_size (`Dict[str, int]`, *optional*, defaults to `self.codebook_crop_size`):
Size of the center crop of the codebook pixels. Only has an effect if `codebook_do_center_crop` is set
to `True`.
codebook_do_rescale (`bool`, *optional*, defaults to `self.codebook_do_rescale`):
Whether to rescale the codebook pixels values between [0 - 1].
codebook_rescale_factor (`float`, *optional*, defaults to `self.codebook_rescale_factor`):
Rescale factor to rescale the codebook pixels by if `codebook_do_rescale` is set to `True`.
codebook_do_map_pixels (`bool`, *optional*, defaults to `self.codebook_do_map_pixels`):
Whether to map the codebook pixels values.
codebook_do_normalize (`bool`, *optional*, defaults to `self.codebook_do_normalize`):
Whether to normalize the codebook pixels.
codebook_image_mean (`float` or `List[float]`, *optional*, defaults to `self.codebook_image_mean`):
Codebook pixels mean to normalize the codebook pixels by if `codebook_do_normalize` is set to `True`.
codebook_image_std (`float` or `List[float]`, *optional*, defaults to `self.codebook_image_std`):
Codebook pixels standard deviation to normalize the codebook pixels by if `codebook_do_normalize` is
set to `True`.
return_tensors (`str` or `TensorType`, *optional*):
The type of tensors to return. Can be one of:
- Unset: Return a list of `np.ndarray`.
- `TensorType.TENSORFLOW` or `'tf'`: Return a batch of type `tf.Tensor`.
- `TensorType.PYTORCH` or `'pt'`: Return a batch of type `torch.Tensor`.
- `TensorType.NUMPY` or `'np'`: Return a batch of type `np.ndarray`.
- `TensorType.JAX` or `'jax'`: Return a batch of type `jax.numpy.ndarray`.
data_format (`ChannelDimension` or `str`, *optional*, defaults to `ChannelDimension.FIRST`):
The channel dimension format for the output image. Can be one of:
- `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
- `ChannelDimension.LAST`: image in (height, width, num_channels) format.
"""
do_resize = do_resize if do_resize is not None else self.do_resize
size = size if size is not None else self.size
size = get_size_dict(size)
resample = resample if resample is not None else self.resample
do_center_crop = do_center_crop if do_center_crop is not None else self.do_center_crop
crop_size = crop_size if crop_size is not None else self.crop_size
crop_size = get_size_dict(crop_size, param_name="crop_size")
do_rescale = do_rescale if do_rescale is not None else self.do_rescale
rescale_factor = rescale_factor if rescale_factor is not None else self.rescale_factor
do_normalize = do_normalize if do_normalize is not None else self.do_normalize
image_mean = image_mean if image_mean is not None else self.image_mean
image_std = image_std if image_std is not None else self.image_std
return_image_mask = return_image_mask if return_image_mask is not None else self.return_image_mask
input_size_patches = input_size_patches if input_size_patches is not None else self.input_size_patches
total_mask_patches = total_mask_patches if total_mask_patches is not None else self.total_mask_patches
mask_group_min_patches = (
mask_group_min_patches if mask_group_min_patches is not None else self.mask_group_min_patches
)
mask_group_max_patches = (
mask_group_max_patches if mask_group_max_patches is not None else self.mask_group_max_patches
)
mask_group_min_aspect_ratio = (
mask_group_min_aspect_ratio
if mask_group_min_aspect_ratio is not None
else self.mask_group_min_aspect_ratio
)
mask_group_max_aspect_ratio = (
mask_group_max_aspect_ratio
if mask_group_max_aspect_ratio is not None
else self.mask_group_max_aspect_ratio
)
return_codebook_pixels = (
return_codebook_pixels if return_codebook_pixels is not None else self.return_codebook_pixels
)
codebook_do_resize = codebook_do_resize if codebook_do_resize is not None else self.codebook_do_resize
codebook_size = codebook_size if codebook_size is not None else self.codebook_size
codebook_size = get_size_dict(codebook_size, param_name="codebook_size")
codebook_resample = codebook_resample if codebook_resample is not None else self.codebook_resample
codebook_do_rescale = codebook_do_rescale if codebook_do_rescale is not None else self.codebook_do_rescale
codebook_rescale_factor = (
codebook_rescale_factor if codebook_rescale_factor is not None else self.codebook_rescale_factor
)
codebook_do_center_crop = (
codebook_do_center_crop if codebook_do_center_crop is not None else self.codebook_do_center_crop
)
codebook_crop_size = codebook_crop_size if codebook_crop_size is not None else self.codebook_crop_size
codebook_crop_size = get_size_dict(codebook_crop_size, param_name="codebook_crop_size")
codebook_do_map_pixels = (
codebook_do_map_pixels if codebook_do_map_pixels is not None else self.codebook_do_map_pixels
)
codebook_do_normalize = (
codebook_do_normalize if codebook_do_normalize is not None else self.codebook_do_normalize
)
codebook_image_mean = codebook_image_mean if codebook_image_mean is not None else self.codebook_image_mean
codebook_image_std = codebook_image_std if codebook_image_std is not None else self.codebook_image_std
images = make_list_of_images(images)
if not valid_images(images):
raise ValueError(
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
"torch.Tensor, tf.Tensor or jax.ndarray."
)
processed_images = [
self._preprocess_image(
image=img,
do_resize=do_resize,
size=size,
resample=resample,
do_center_crop=do_center_crop,
crop_size=crop_size,
do_rescale=do_rescale,
rescale_factor=rescale_factor,
do_normalize=do_normalize,
image_mean=image_mean,
image_std=image_std,
do_map_pixels=False,
data_format=data_format,
)
for img in images
]
data = {"pixel_values": processed_images}
if return_codebook_pixels:
codebook_images = [
self._preprocess_image(
image=img,
do_resize=codebook_do_resize,
size=codebook_size,
resample=codebook_resample,
do_center_crop=codebook_do_center_crop,
crop_size=codebook_crop_size,
do_rescale=codebook_do_rescale,
rescale_factor=codebook_rescale_factor,
do_normalize=codebook_do_normalize,
image_mean=codebook_image_mean,
image_std=codebook_image_std,
do_map_pixels=codebook_do_map_pixels,
data_format=data_format,
)
for img in images
]
data["codebook_pixel_values"] = codebook_images
if return_image_mask:
mask_generator = self.masking_generator(
input_size_patches=input_size_patches,
total_mask_patches=total_mask_patches,
mask_group_min_patches=mask_group_min_patches,
mask_group_max_patches=mask_group_max_patches,
mask_group_min_aspect_ratio=mask_group_min_aspect_ratio,
mask_group_max_aspect_ratio=mask_group_max_aspect_ratio,
)
masks = [mask_generator() for _ in images]
data["bool_masked_pos"] = masks
return BatchFeature(data=data, tensor_type=return_tensors)
| # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Image processor class for Flava."""
import math
import random
from functools import lru_cache
from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
import numpy as np
from transformers.utils import is_vision_available
from transformers.utils.generic import TensorType
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import center_crop, normalize, rescale, resize, to_channel_dimension_format
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import logging
if is_vision_available():
import PIL
logger = logging.get_logger(__name__)
# These values are taken from CLIP
FLAVA_IMAGE_MEAN = OPENAI_CLIP_MEAN
FLAVA_IMAGE_STD = OPENAI_CLIP_STD
FLAVA_CODEBOOK_MEAN = [0.0, 0.0, 0.0]
FLAVA_CODEBOOK_STD = [1.0, 1.0, 1.0]
LOGIT_LAPLACE_EPS: float = 0.1
# Inspired from https://github.com/microsoft/unilm/blob/master/beit/masking_generator.py
class FlavaMaskingGenerator:
def __init__(
self,
input_size: Union[int, Tuple[int, int]] = 14,
total_mask_patches: int = 75,
mask_group_max_patches: Optional[int] = None,
mask_group_min_patches: int = 16,
mask_group_min_aspect_ratio: Optional[float] = 0.3,
mask_group_max_aspect_ratio: float = None,
):
if not isinstance(input_size, tuple):
input_size = (input_size,) * 2
self.height, self.width = input_size
self.num_patches = self.height * self.width
self.total_mask_patches = total_mask_patches
self.mask_group_min_patches = mask_group_min_patches
self.mask_group_max_patches = total_mask_patches if mask_group_max_patches is None else mask_group_max_patches
mask_group_max_aspect_ratio = mask_group_max_aspect_ratio or 1 / mask_group_min_aspect_ratio
self.log_aspect_ratio = (math.log(mask_group_min_aspect_ratio), math.log(mask_group_max_aspect_ratio))
def __repr__(self):
repr_str = "MaskingGenerator(%d, %d -> [%d ~ %d], max = %d, %.3f ~ %.3f)" % (
self.height,
self.width,
self.mask_group_min_patches,
self.mask_group_max_patches,
self.total_mask_patches,
self.log_aspect_ratio[0],
self.log_aspect_ratio[1],
)
return repr_str
def get_shape(self):
return self.height, self.width
def _mask(self, mask, max_mask_patches):
delta = 0
for _attempt in range(10):
target_area = random.uniform(self.mask_group_min_patches, max_mask_patches)
aspect_ratio = math.exp(random.uniform(*self.log_aspect_ratio))
height = int(round(math.sqrt(target_area * aspect_ratio)))
width = int(round(math.sqrt(target_area / aspect_ratio)))
if width < self.width and height < self.height:
top = random.randint(0, self.height - height)
left = random.randint(0, self.width - width)
num_masked = mask[top : top + height, left : left + width].sum()
# Overlap
if 0 < height * width - num_masked <= max_mask_patches:
for i in range(top, top + height):
for j in range(left, left + width):
if mask[i, j] == 0:
mask[i, j] = 1
delta += 1
if delta > 0:
break
return delta
def __call__(self):
mask = np.zeros(shape=self.get_shape(), dtype=int)
mask_count = 0
while mask_count < self.total_mask_patches:
max_mask_patches = self.total_mask_patches - mask_count
max_mask_patches = min(max_mask_patches, self.mask_group_max_patches)
delta = self._mask(mask, max_mask_patches)
if delta == 0:
break
else:
mask_count += delta
return mask
class FlavaImageProcessor(BaseImageProcessor):
r"""
Constructs a Flava image processor.
Args:
do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the image's (height, width) dimensions to the specified `size`. Can be overridden by the
`do_resize` parameter in `preprocess`.
size (`Dict[str, int]` *optional*, defaults to `{"height": 224, "width": 224}`):
Size of the image after resizing. Can be overridden by the `size` parameter in `preprocess`.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BICUBIC`):
Resampling filter to use if resizing the image. Can be overridden by the `resample` parameter in
`preprocess`.
do_center_crop (`bool`, *optional*, defaults to `True`):
Whether to center crop the images. Can be overridden by the `do_center_crop` parameter in `preprocess`.
crop_size (`Dict[str, int]` *optional*, defaults to `{"height": 224, "width": 224}`):
Size of image after the center crop `(crop_size["height"], crop_size["width"])`. Can be overridden by the
`crop_size` parameter in `preprocess`.
do_rescale (`bool`, *optional*, defaults to `True`):
Whether to rescale the image by the specified scale `rescale_factor`. Can be overridden by the `do_rescale`
parameter in `preprocess`.
rescale_factor (`int` or `float`, *optional*, defaults to `1/255`):
Scale factor to use if rescaling the image. Can be overridden by the `rescale_factor` parameter in
`preprocess`.
do_normalize (`bool`, *optional*, defaults to `True`):
Whether to normalize the image. Can be overridden by the `do_normalize` parameter in `preprocess`.
image_mean (`float` or `List[float]`, *optional*, defaults to `IMAGENET_STANDARD_MEAN`):
Mean to use if normalizing the image. This is a float or list of floats the length of the number of
channels in the image. Can be overridden by the `image_mean` parameter in the `preprocess` method.
image_std (`float` or `List[float]`, *optional*, defaults to `IMAGENET_STANDARD_STD`):
Standard deviation to use if normalizing the image. This is a float or list of floats the length of the
number of channels in the image. Can be overridden by the `image_std` parameter in the `preprocess` method.
return_image_mask (`bool`, *optional*, defaults to `False`):
Whether to return the image mask. Can be overridden by the `return_image_mask` parameter in `preprocess`.
input_size_patches (`int`, *optional*, defaults to 14):
Number of patches in the image in height and width direction. 14x14 = 196 total patches. Can be overridden
by the `input_size_patches` parameter in `preprocess`.
total_mask_patches (`int`, *optional*, defaults to 75):
Total number of patches that should be masked. Can be overridden by the `total_mask_patches` parameter in
`preprocess`.
mask_group_min_patches (`int`, *optional*, defaults to 16):
Minimum number of patches that should be masked. Can be overridden by the `mask_group_min_patches`
parameter in `preprocess`.
mask_group_max_patches (`int`, *optional*):
Maximum number of patches that should be masked. Can be overridden by the `mask_group_max_patches`
parameter in `preprocess`.
mask_group_min_aspect_ratio (`float`, *optional*, defaults to 0.3):
Minimum aspect ratio of the mask window. Can be overridden by the `mask_group_min_aspect_ratio` parameter
in `preprocess`.
mask_group_max_aspect_ratio (`float`, *optional*):
Maximum aspect ratio of the mask window. Can be overridden by the `mask_group_max_aspect_ratio` parameter
in `preprocess`.
codebook_do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the input for codebook to a certain. Can be overridden by the `codebook_do_resize`
parameter in `preprocess`. `codebook_size`.
codebook_size (`Dict[str, int]`, *optional*, defaults to `{"height": 224, "width": 224}`):
Resize the input for codebook to the given size. Can be overridden by the `codebook_size` parameter in
`preprocess`.
codebook_resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.LANCZOS`):
Resampling filter to use if resizing the codebook image. Can be overridden by the `codebook_resample`
parameter in `preprocess`.
codebook_do_center_crop (`bool`, *optional*, defaults to `True`):
Whether to crop the input for codebook at the center. If the input size is smaller than
`codebook_crop_size` along any edge, the image is padded with 0's and then center cropped. Can be
overridden by the `codebook_do_center_crop` parameter in `preprocess`.
codebook_crop_size (`Dict[str, int]`, *optional*, defaults to `{"height": 224, "width": 224}`):
Desired output size for codebook input when applying center-cropping. Can be overridden by the
`codebook_crop_size` parameter in `preprocess`.
codebook_do_rescale (`bool`, *optional*, defaults to `True`):
Whether to rescale the input for codebook by the specified scale `codebook_rescale_factor`. Can be
overridden by the `codebook_do_rescale` parameter in `preprocess`.
codebook_rescale_factor (`int` or `float`, *optional*, defaults to `1/255`):
Defines the scale factor to use if rescaling the codebook image. Can be overridden by the
`codebook_rescale_factor` parameter in `preprocess`.
codebook_do_map_pixels (`bool`, *optional*, defaults to `True`):
Whether to map the pixel values of the codebook input to (1 - 2e)x + e. Can be overridden by the
`codebook_do_map_pixels` parameter in `preprocess`.
codebook_do_normalize (`bool`, *optional*, defaults to `True`):
Whether or not to normalize the input for codebook with `codebook_image_mean` and `codebook_image_std`. Can
be overridden by the `codebook_do_normalize` parameter in `preprocess`.
codebook_image_mean (`Optional[Union[float, Iterable[float]]]`, *optional*, defaults to `[0, 0, 0]`):
The sequence of means for each channel, to be used when normalizing images for codebook. Can be overridden
by the `codebook_image_mean` parameter in `preprocess`.
codebook_image_std (`Optional[Union[float, Iterable[float]]]`, *optional*, defaults to `[0.5, 0.5, 0.5]`):
The sequence of standard deviations for each channel, to be used when normalizing images for codebook. Can
be overridden by the `codebook_image_std` parameter in `preprocess`.
"""
model_input_names = ["pixel_values"]
def __init__(
self,
do_resize: bool = True,
size: Dict[str, int] = None,
resample: PILImageResampling = PILImageResampling.BICUBIC,
do_center_crop: bool = True,
crop_size: Dict[str, int] = None,
do_rescale: bool = True,
rescale_factor: Union[int, float] = 1 / 255,
do_normalize: bool = True,
image_mean: Optional[Union[float, Iterable[float]]] = None,
image_std: Optional[Union[float, Iterable[float]]] = None,
# Mask related params
return_image_mask: bool = False,
input_size_patches: int = 14,
total_mask_patches: int = 75,
mask_group_min_patches: int = 16,
mask_group_max_patches: Optional[int] = None,
mask_group_min_aspect_ratio: float = 0.3,
mask_group_max_aspect_ratio: Optional[float] = None,
# Codebook related params
return_codebook_pixels: bool = False,
codebook_do_resize: bool = True,
codebook_size: bool = None,
codebook_resample: int = PILImageResampling.LANCZOS,
codebook_do_center_crop: bool = True,
codebook_crop_size: int = None,
codebook_do_rescale: bool = True,
codebook_rescale_factor: Union[int, float] = 1 / 255,
codebook_do_map_pixels: bool = True,
codebook_do_normalize: bool = True,
codebook_image_mean: Optional[Union[float, Iterable[float]]] = None,
codebook_image_std: Optional[Union[float, Iterable[float]]] = None,
**kwargs,
) -> None:
super().__init__(**kwargs)
size = size if size is not None else {"height": 224, "width": 224}
size = get_size_dict(size)
crop_size = crop_size if crop_size is not None else {"height": 224, "width": 224}
crop_size = get_size_dict(crop_size, param_name="crop_size")
codebook_size = codebook_size if codebook_size is not None else {"height": 112, "width": 112}
codebook_size = get_size_dict(codebook_size, param_name="codebook_size")
codebook_crop_size = codebook_crop_size if codebook_crop_size is not None else {"height": 112, "width": 112}
codebook_crop_size = get_size_dict(codebook_crop_size, param_name="codebook_crop_size")
self.do_resize = do_resize
self.size = size
self.resample = resample
self.do_rescale = do_rescale
self.rescale_factor = rescale_factor
self.do_center_crop = do_center_crop
self.crop_size = crop_size
self.do_normalize = do_normalize
self.image_mean = image_mean if image_mean is not None else FLAVA_IMAGE_MEAN
self.image_std = image_std if image_std is not None else FLAVA_IMAGE_STD
self.return_image_mask = return_image_mask
self.input_size_patches = input_size_patches
self.total_mask_patches = total_mask_patches
self.mask_group_min_patches = mask_group_min_patches
self.mask_group_max_patches = mask_group_max_patches
self.mask_group_min_aspect_ratio = mask_group_min_aspect_ratio
self.mask_group_max_aspect_ratio = mask_group_max_aspect_ratio
self.return_codebook_pixels = return_codebook_pixels
self.codebook_do_resize = codebook_do_resize
self.codebook_size = codebook_size
self.codebook_resample = codebook_resample
self.codebook_do_center_crop = codebook_do_center_crop
self.codebook_crop_size = codebook_crop_size
self.codebook_do_rescale = codebook_do_rescale
self.codebook_rescale_factor = codebook_rescale_factor
self.codebook_do_map_pixels = codebook_do_map_pixels
self.codebook_do_normalize = codebook_do_normalize
self.codebook_image_mean = codebook_image_mean
self.codebook_image_mean = codebook_image_mean if codebook_image_mean is not None else FLAVA_CODEBOOK_MEAN
self.codebook_image_std = codebook_image_std if codebook_image_std is not None else FLAVA_CODEBOOK_STD
@classmethod
def from_dict(cls, image_processor_dict: Dict[str, Any], **kwargs):
"""
Overrides the `from_dict` method from the base class to make sure parameters are updated if image processor is
created using from_dict and kwargs e.g. `FlavaImageProcessor.from_pretrained(checkpoint, codebook_size=600)`
"""
image_processor_dict = image_processor_dict.copy()
if "codebook_size" in kwargs:
image_processor_dict["codebook_size"] = kwargs.pop("codebook_size")
if "codebook_crop_size" in kwargs:
image_processor_dict["codebook_crop_size"] = kwargs.pop("codebook_crop_size")
return super().from_dict(image_processor_dict, **kwargs)
@lru_cache()
def masking_generator(
self,
input_size_patches,
total_mask_patches,
mask_group_min_patches,
mask_group_max_patches,
mask_group_min_aspect_ratio,
mask_group_max_aspect_ratio,
) -> FlavaMaskingGenerator:
return FlavaMaskingGenerator(
input_size=input_size_patches,
total_mask_patches=total_mask_patches,
mask_group_min_patches=mask_group_min_patches,
mask_group_max_patches=mask_group_max_patches,
mask_group_min_aspect_ratio=mask_group_min_aspect_ratio,
mask_group_max_aspect_ratio=mask_group_max_aspect_ratio,
)
def resize(
self,
image: np.ndarray,
size: Dict[str, int],
resample: PILImageResampling = PILImageResampling.BICUBIC,
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
) -> np.ndarray:
"""
Resize an image to `(size["height"], size["width"])`.
Args:
image (`np.ndarray`):
Image to resize.
size (`Dict[str, int]`):
Size of the output image.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BICUBIC`):
Resampling filter to use when resiizing the image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
size = get_size_dict(size)
if "height" not in size or "width" not in size:
raise ValueError(f"The size dictionary must contain 'height' and 'width' keys. Got {size.keys()}")
return resize(
image, size=(size["height"], size["width"]), resample=resample, data_format=data_format, **kwargs
)
def center_crop(
self,
image: np.ndarray,
size: Dict[str, int],
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
) -> np.ndarray:
"""
Center crop an image to `(size["height"], size["width"])`. If the input size is smaller than `crop_size` along
any edge, the image is padded with 0's and then center cropped.
Args:
image (`np.ndarray`):
Image to center crop.
size (`Dict[str, int]`):
Size of the output image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
size = get_size_dict(size)
if "height" not in size or "width" not in size:
raise ValueError(f"The size dictionary must contain 'height' and 'width' keys. Got {size.keys()}")
return center_crop(image, size=(size["height"], size["width"]), data_format=data_format, **kwargs)
def rescale(
self,
image: np.ndarray,
scale: Union[int, float],
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
):
"""
Rescale an image by a scale factor. image = image * scale.
Args:
image (`np.ndarray`):
Image to rescale.
scale (`int` or `float`):
Scale to apply to the image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
return rescale(image, scale=scale, data_format=data_format, **kwargs)
def normalize(
self,
image: np.ndarray,
mean: Union[float, List[float]],
std: Union[float, List[float]],
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs,
) -> np.ndarray:
"""
Normalize an image. image = (image - image_mean) / image_std.
Args:
image (`np.ndarray`):
Image to normalize.
image_mean (`float` or `List[float]`):
Image mean.
image_std (`float` or `List[float]`):
Image standard deviation.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
return normalize(image, mean=mean, std=std, data_format=data_format, **kwargs)
def map_pixels(self, image: np.ndarray) -> np.ndarray:
return (1 - 2 * LOGIT_LAPLACE_EPS) * image + LOGIT_LAPLACE_EPS
def _preprocess_image(
self,
image: ImageInput,
do_resize: bool = None,
size: Dict[str, int] = None,
resample: PILImageResampling = None,
do_center_crop: bool = None,
crop_size: Dict[str, int] = None,
do_rescale: bool = None,
rescale_factor: float = None,
do_normalize: bool = None,
image_mean: Optional[Union[float, List[float]]] = None,
image_std: Optional[Union[float, List[float]]] = None,
do_map_pixels: bool = None,
data_format: Optional[ChannelDimension] = ChannelDimension.FIRST,
) -> np.ndarray:
"""Preprocesses a single image."""
if do_resize and size is None or resample is None:
raise ValueError("Size and resample must be specified if do_resize is True.")
if do_rescale and rescale_factor is None:
raise ValueError("Rescale factor must be specified if do_rescale is True.")
if do_normalize and (image_mean is None or image_std is None):
raise ValueError("Image mean and std must be specified if do_normalize is True.")
# All transformations expect numpy arrays.
image = to_numpy_array(image)
if do_resize:
image = self.resize(image=image, size=size, resample=resample)
if do_center_crop:
image = self.center_crop(image=image, size=crop_size)
if do_rescale:
image = self.rescale(image=image, scale=rescale_factor)
if do_normalize:
image = self.normalize(image=image, mean=image_mean, std=image_std)
if do_map_pixels:
image = self.map_pixels(image)
if data_format is not None:
image = to_channel_dimension_format(image, data_format)
return image
def preprocess(
self,
images: ImageInput,
do_resize: Optional[bool] = None,
size: Dict[str, int] = None,
resample: PILImageResampling = None,
do_center_crop: Optional[bool] = None,
crop_size: Optional[Dict[str, int]] = None,
do_rescale: Optional[bool] = None,
rescale_factor: Optional[float] = None,
do_normalize: Optional[bool] = None,
image_mean: Optional[Union[float, List[float]]] = None,
image_std: Optional[Union[float, List[float]]] = None,
# Mask related params
return_image_mask: Optional[bool] = None,
input_size_patches: Optional[int] = None,
total_mask_patches: Optional[int] = None,
mask_group_min_patches: Optional[int] = None,
mask_group_max_patches: Optional[int] = None,
mask_group_min_aspect_ratio: Optional[float] = None,
mask_group_max_aspect_ratio: Optional[float] = None,
# Codebook related params
return_codebook_pixels: Optional[bool] = None,
codebook_do_resize: Optional[bool] = None,
codebook_size: Optional[Dict[str, int]] = None,
codebook_resample: Optional[int] = None,
codebook_do_center_crop: Optional[bool] = None,
codebook_crop_size: Optional[Dict[str, int]] = None,
codebook_do_rescale: Optional[bool] = None,
codebook_rescale_factor: Optional[float] = None,
codebook_do_map_pixels: Optional[bool] = None,
codebook_do_normalize: Optional[bool] = None,
codebook_image_mean: Optional[Iterable[float]] = None,
codebook_image_std: Optional[Iterable[float]] = None,
return_tensors: Optional[Union[str, TensorType]] = None,
data_format: ChannelDimension = ChannelDimension.FIRST,
**kwargs,
) -> PIL.Image.Image:
"""
Preprocess an image or batch of images.
Args:
images (`ImageInput`):
Image to preprocess.
do_resize (`bool`, *optional*, defaults to `self.do_resize`):
Whether to resize the image.
size (`Dict[str, int]`, *optional*, defaults to `self.size`):
Size of the image.
resample (`int`, *optional*, defaults to `self.resample`):
Resampling filter to use if resizing the image. This can be one of the enum `PILImageResampling`, Only
has an effect if `do_resize` is set to `True`.
do_center_crop (`bool`, *optional*, defaults to `self.do_center_crop`):
Whether to center crop the image.
crop_size (`Dict[str, int]`, *optional*, defaults to `self.crop_size`):
Size of the center crop. Only has an effect if `do_center_crop` is set to `True`.
do_rescale (`bool`, *optional*, defaults to `self.do_rescale`):
Whether to rescale the image values between [0 - 1].
rescale_factor (`float`, *optional*, defaults to `self.rescale_factor`):
Rescale factor to rescale the image by if `do_rescale` is set to `True`.
do_normalize (`bool`, *optional*, defaults to `self.do_normalize`):
Whether to normalize the image.
image_mean (`float` or `List[float]`, *optional*, defaults to `self.image_mean`):
Image mean.
image_std (`float` or `List[float]`, *optional*, defaults to `self.image_std`):
Image standard deviation.
return_image_mask (`bool`, *optional*, defaults to `self.return_image_mask`):
Whether to return the image mask.
input_size_patches (`int`, *optional*, defaults to `self.input_size_patches`):
Size of the patches to extract from the image.
total_mask_patches (`int`, *optional*, defaults to `self.total_mask_patches`):
Total number of patches to extract from the image.
mask_group_min_patches (`int`, *optional*, defaults to `self.mask_group_min_patches`):
Minimum number of patches to extract from the image.
mask_group_max_patches (`int`, *optional*, defaults to `self.mask_group_max_patches`):
Maximum number of patches to extract from the image.
mask_group_min_aspect_ratio (`float`, *optional*, defaults to `self.mask_group_min_aspect_ratio`):
Minimum aspect ratio of the patches to extract from the image.
mask_group_max_aspect_ratio (`float`, *optional*, defaults to `self.mask_group_max_aspect_ratio`):
Maximum aspect ratio of the patches to extract from the image.
return_codebook_pixels (`bool`, *optional*, defaults to `self.return_codebook_pixels`):
Whether to return the codebook pixels.
codebook_do_resize (`bool`, *optional*, defaults to `self.codebook_do_resize`):
Whether to resize the codebook pixels.
codebook_size (`Dict[str, int]`, *optional*, defaults to `self.codebook_size`):
Size of the codebook pixels.
codebook_resample (`int`, *optional*, defaults to `self.codebook_resample`):
Resampling filter to use if resizing the codebook pixels. This can be one of the enum
`PILImageResampling`, Only has an effect if `codebook_do_resize` is set to `True`.
codebook_do_center_crop (`bool`, *optional*, defaults to `self.codebook_do_center_crop`):
Whether to center crop the codebook pixels.
codebook_crop_size (`Dict[str, int]`, *optional*, defaults to `self.codebook_crop_size`):
Size of the center crop of the codebook pixels. Only has an effect if `codebook_do_center_crop` is set
to `True`.
codebook_do_rescale (`bool`, *optional*, defaults to `self.codebook_do_rescale`):
Whether to rescale the codebook pixels values between [0 - 1].
codebook_rescale_factor (`float`, *optional*, defaults to `self.codebook_rescale_factor`):
Rescale factor to rescale the codebook pixels by if `codebook_do_rescale` is set to `True`.
codebook_do_map_pixels (`bool`, *optional*, defaults to `self.codebook_do_map_pixels`):
Whether to map the codebook pixels values.
codebook_do_normalize (`bool`, *optional*, defaults to `self.codebook_do_normalize`):
Whether to normalize the codebook pixels.
codebook_image_mean (`float` or `List[float]`, *optional*, defaults to `self.codebook_image_mean`):
Codebook pixels mean to normalize the codebook pixels by if `codebook_do_normalize` is set to `True`.
codebook_image_std (`float` or `List[float]`, *optional*, defaults to `self.codebook_image_std`):
Codebook pixels standard deviation to normalize the codebook pixels by if `codebook_do_normalize` is
set to `True`.
return_tensors (`str` or `TensorType`, *optional*):
The type of tensors to return. Can be one of:
- Unset: Return a list of `np.ndarray`.
- `TensorType.TENSORFLOW` or `'tf'`: Return a batch of type `tf.Tensor`.
- `TensorType.PYTORCH` or `'pt'`: Return a batch of type `torch.Tensor`.
- `TensorType.NUMPY` or `'np'`: Return a batch of type `np.ndarray`.
- `TensorType.JAX` or `'jax'`: Return a batch of type `jax.numpy.ndarray`.
data_format (`ChannelDimension` or `str`, *optional*, defaults to `ChannelDimension.FIRST`):
The channel dimension format for the output image. Can be one of:
- `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
- `ChannelDimension.LAST`: image in (height, width, num_channels) format.
"""
do_resize = do_resize if do_resize is not None else self.do_resize
size = size if size is not None else self.size
size = get_size_dict(size)
resample = resample if resample is not None else self.resample
do_center_crop = do_center_crop if do_center_crop is not None else self.do_center_crop
crop_size = crop_size if crop_size is not None else self.crop_size
crop_size = get_size_dict(crop_size, param_name="crop_size")
do_rescale = do_rescale if do_rescale is not None else self.do_rescale
rescale_factor = rescale_factor if rescale_factor is not None else self.rescale_factor
do_normalize = do_normalize if do_normalize is not None else self.do_normalize
image_mean = image_mean if image_mean is not None else self.image_mean
image_std = image_std if image_std is not None else self.image_std
return_image_mask = return_image_mask if return_image_mask is not None else self.return_image_mask
input_size_patches = input_size_patches if input_size_patches is not None else self.input_size_patches
total_mask_patches = total_mask_patches if total_mask_patches is not None else self.total_mask_patches
mask_group_min_patches = (
mask_group_min_patches if mask_group_min_patches is not None else self.mask_group_min_patches
)
mask_group_max_patches = (
mask_group_max_patches if mask_group_max_patches is not None else self.mask_group_max_patches
)
mask_group_min_aspect_ratio = (
mask_group_min_aspect_ratio
if mask_group_min_aspect_ratio is not None
else self.mask_group_min_aspect_ratio
)
mask_group_max_aspect_ratio = (
mask_group_max_aspect_ratio
if mask_group_max_aspect_ratio is not None
else self.mask_group_max_aspect_ratio
)
return_codebook_pixels = (
return_codebook_pixels if return_codebook_pixels is not None else self.return_codebook_pixels
)
codebook_do_resize = codebook_do_resize if codebook_do_resize is not None else self.codebook_do_resize
codebook_size = codebook_size if codebook_size is not None else self.codebook_size
codebook_size = get_size_dict(codebook_size, param_name="codebook_size")
codebook_resample = codebook_resample if codebook_resample is not None else self.codebook_resample
codebook_do_rescale = codebook_do_rescale if codebook_do_rescale is not None else self.codebook_do_rescale
codebook_rescale_factor = (
codebook_rescale_factor if codebook_rescale_factor is not None else self.codebook_rescale_factor
)
codebook_do_center_crop = (
codebook_do_center_crop if codebook_do_center_crop is not None else self.codebook_do_center_crop
)
codebook_crop_size = codebook_crop_size if codebook_crop_size is not None else self.codebook_crop_size
codebook_crop_size = get_size_dict(codebook_crop_size, param_name="codebook_crop_size")
codebook_do_map_pixels = (
codebook_do_map_pixels if codebook_do_map_pixels is not None else self.codebook_do_map_pixels
)
codebook_do_normalize = (
codebook_do_normalize if codebook_do_normalize is not None else self.codebook_do_normalize
)
codebook_image_mean = codebook_image_mean if codebook_image_mean is not None else self.codebook_image_mean
codebook_image_std = codebook_image_std if codebook_image_std is not None else self.codebook_image_std
images = make_list_of_images(images)
if not valid_images(images):
raise ValueError(
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
"torch.Tensor, tf.Tensor or jax.ndarray."
)
processed_images = [
self._preprocess_image(
image=img,
do_resize=do_resize,
size=size,
resample=resample,
do_center_crop=do_center_crop,
crop_size=crop_size,
do_rescale=do_rescale,
rescale_factor=rescale_factor,
do_normalize=do_normalize,
image_mean=image_mean,
image_std=image_std,
do_map_pixels=False,
data_format=data_format,
)
for img in images
]
data = {"pixel_values": processed_images}
if return_codebook_pixels:
codebook_images = [
self._preprocess_image(
image=img,
do_resize=codebook_do_resize,
size=codebook_size,
resample=codebook_resample,
do_center_crop=codebook_do_center_crop,
crop_size=codebook_crop_size,
do_rescale=codebook_do_rescale,
rescale_factor=codebook_rescale_factor,
do_normalize=codebook_do_normalize,
image_mean=codebook_image_mean,
image_std=codebook_image_std,
do_map_pixels=codebook_do_map_pixels,
data_format=data_format,
)
for img in images
]
data["codebook_pixel_values"] = codebook_images
if return_image_mask:
mask_generator = self.masking_generator(
input_size_patches=input_size_patches,
total_mask_patches=total_mask_patches,
mask_group_min_patches=mask_group_min_patches,
mask_group_max_patches=mask_group_max_patches,
mask_group_min_aspect_ratio=mask_group_min_aspect_ratio,
mask_group_max_aspect_ratio=mask_group_max_aspect_ratio,
)
masks = [mask_generator() for _ in images]
data["bool_masked_pos"] = masks
return BatchFeature(data=data, tensor_type=return_tensors)
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/swinv2/__init__.py | -1 |
||
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./tests/models/audio_spectrogram_transformer/test_modeling_audio_spectrogram_transformer.py | # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch Audio Spectrogram Transformer (AST) model. """
import inspect
import unittest
from huggingface_hub import hf_hub_download
from transformers import ASTConfig
from transformers.testing_utils import require_torch, require_torchaudio, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_torchaudio_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import ASTForAudioClassification, ASTModel
from transformers.models.audio_spectrogram_transformer.modeling_audio_spectrogram_transformer import (
AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
)
if is_torchaudio_available():
import torchaudio
from transformers import ASTFeatureExtractor
class ASTModelTester:
def __init__(
self,
parent,
batch_size=13,
patch_size=2,
max_length=24,
num_mel_bins=16,
is_training=True,
use_labels=True,
hidden_size=32,
num_hidden_layers=5,
num_attention_heads=4,
intermediate_size=37,
hidden_act="gelu",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
type_sequence_label_size=10,
initializer_range=0.02,
scope=None,
frequency_stride=2,
time_stride=2,
):
self.parent = parent
self.batch_size = batch_size
self.patch_size = patch_size
self.max_length = max_length
self.num_mel_bins = num_mel_bins
self.is_training = is_training
self.use_labels = use_labels
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_act = hidden_act
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.type_sequence_label_size = type_sequence_label_size
self.initializer_range = initializer_range
self.scope = scope
self.frequency_stride = frequency_stride
self.time_stride = time_stride
# in AST, the seq length equals the number of patches + 2 (we add 2 for the [CLS] and distillation tokens)
frequency_out_dimension = (self.num_mel_bins - self.patch_size) // self.frequency_stride + 1
time_out_dimension = (self.max_length - self.patch_size) // self.time_stride + 1
num_patches = frequency_out_dimension * time_out_dimension
self.seq_length = num_patches + 2
def prepare_config_and_inputs(self):
input_values = floats_tensor([self.batch_size, self.max_length, self.num_mel_bins])
labels = None
if self.use_labels:
labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
config = self.get_config()
return config, input_values, labels
def get_config(self):
return ASTConfig(
patch_size=self.patch_size,
max_length=self.max_length,
num_mel_bins=self.num_mel_bins,
hidden_size=self.hidden_size,
num_hidden_layers=self.num_hidden_layers,
num_attention_heads=self.num_attention_heads,
intermediate_size=self.intermediate_size,
hidden_act=self.hidden_act,
hidden_dropout_prob=self.hidden_dropout_prob,
attention_probs_dropout_prob=self.attention_probs_dropout_prob,
is_decoder=False,
initializer_range=self.initializer_range,
frequency_stride=self.frequency_stride,
time_stride=self.time_stride,
)
def create_and_check_model(self, config, input_values, labels):
model = ASTModel(config=config)
model.to(torch_device)
model.eval()
result = model(input_values)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
(
config,
input_values,
labels,
) = config_and_inputs
inputs_dict = {"input_values": input_values}
return config, inputs_dict
@require_torch
class ASTModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
"""
Here we also overwrite some of the tests of test_modeling_common.py, as AST does not use input_ids, inputs_embeds,
attention_mask and seq_length.
"""
all_model_classes = (
(
ASTModel,
ASTForAudioClassification,
)
if is_torch_available()
else ()
)
pipeline_model_mapping = (
{"audio-classification": ASTForAudioClassification, "feature-extraction": ASTModel}
if is_torch_available()
else {}
)
fx_compatible = False
test_pruning = False
test_resize_embeddings = False
test_head_masking = False
def setUp(self):
self.model_tester = ASTModelTester(self)
self.config_tester = ConfigTester(self, config_class=ASTConfig, has_text_modality=False, hidden_size=37)
def test_config(self):
self.config_tester.run_common_tests()
@unittest.skip(reason="AST does not use inputs_embeds")
def test_inputs_embeds(self):
pass
def test_model_common_attributes(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
self.assertIsInstance(model.get_input_embeddings(), (nn.Module))
x = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(x, nn.Linear))
def test_forward_signature(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
signature = inspect.signature(model.forward)
# signature.parameters is an OrderedDict => so arg_names order is deterministic
arg_names = [*signature.parameters.keys()]
expected_arg_names = ["input_values"]
self.assertListEqual(arg_names[:1], expected_arg_names)
def test_model(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*config_and_inputs)
@slow
def test_model_from_pretrained(self):
for model_name in AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
model = ASTModel.from_pretrained(model_name)
self.assertIsNotNone(model)
# We will verify our results on some audio from AudioSet
def prepare_audio():
filepath = hf_hub_download(
repo_id="nielsr/audio-spectogram-transformer-checkpoint", filename="sample_audio.flac", repo_type="dataset"
)
audio, sampling_rate = torchaudio.load(filepath)
return audio, sampling_rate
@require_torch
@require_torchaudio
class ASTModelIntegrationTest(unittest.TestCase):
@cached_property
def default_feature_extractor(self):
return (
ASTFeatureExtractor.from_pretrained("MIT/ast-finetuned-audioset-10-10-0.4593")
if is_torchaudio_available()
else None
)
@slow
def test_inference_audio_classification(self):
feature_extractor = self.default_feature_extractor
model = ASTForAudioClassification.from_pretrained("MIT/ast-finetuned-audioset-10-10-0.4593").to(torch_device)
feature_extractor = self.default_feature_extractor
audio, sampling_rate = prepare_audio()
audio = audio.squeeze().numpy()
inputs = feature_extractor(audio, sampling_rate=sampling_rate, return_tensors="pt").to(torch_device)
# forward pass
with torch.no_grad():
outputs = model(**inputs)
# verify the logits
expected_shape = torch.Size((1, 527))
self.assertEqual(outputs.logits.shape, expected_shape)
expected_slice = torch.tensor([-0.8760, -7.0042, -8.6602]).to(torch_device)
self.assertTrue(torch.allclose(outputs.logits[0, :3], expected_slice, atol=1e-4))
| # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Testing suite for the PyTorch Audio Spectrogram Transformer (AST) model. """
import inspect
import unittest
from huggingface_hub import hf_hub_download
from transformers import ASTConfig
from transformers.testing_utils import require_torch, require_torchaudio, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_torchaudio_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import ASTForAudioClassification, ASTModel
from transformers.models.audio_spectrogram_transformer.modeling_audio_spectrogram_transformer import (
AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
)
if is_torchaudio_available():
import torchaudio
from transformers import ASTFeatureExtractor
class ASTModelTester:
def __init__(
self,
parent,
batch_size=13,
patch_size=2,
max_length=24,
num_mel_bins=16,
is_training=True,
use_labels=True,
hidden_size=32,
num_hidden_layers=5,
num_attention_heads=4,
intermediate_size=37,
hidden_act="gelu",
hidden_dropout_prob=0.1,
attention_probs_dropout_prob=0.1,
type_sequence_label_size=10,
initializer_range=0.02,
scope=None,
frequency_stride=2,
time_stride=2,
):
self.parent = parent
self.batch_size = batch_size
self.patch_size = patch_size
self.max_length = max_length
self.num_mel_bins = num_mel_bins
self.is_training = is_training
self.use_labels = use_labels
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.intermediate_size = intermediate_size
self.hidden_act = hidden_act
self.hidden_dropout_prob = hidden_dropout_prob
self.attention_probs_dropout_prob = attention_probs_dropout_prob
self.type_sequence_label_size = type_sequence_label_size
self.initializer_range = initializer_range
self.scope = scope
self.frequency_stride = frequency_stride
self.time_stride = time_stride
# in AST, the seq length equals the number of patches + 2 (we add 2 for the [CLS] and distillation tokens)
frequency_out_dimension = (self.num_mel_bins - self.patch_size) // self.frequency_stride + 1
time_out_dimension = (self.max_length - self.patch_size) // self.time_stride + 1
num_patches = frequency_out_dimension * time_out_dimension
self.seq_length = num_patches + 2
def prepare_config_and_inputs(self):
input_values = floats_tensor([self.batch_size, self.max_length, self.num_mel_bins])
labels = None
if self.use_labels:
labels = ids_tensor([self.batch_size], self.type_sequence_label_size)
config = self.get_config()
return config, input_values, labels
def get_config(self):
return ASTConfig(
patch_size=self.patch_size,
max_length=self.max_length,
num_mel_bins=self.num_mel_bins,
hidden_size=self.hidden_size,
num_hidden_layers=self.num_hidden_layers,
num_attention_heads=self.num_attention_heads,
intermediate_size=self.intermediate_size,
hidden_act=self.hidden_act,
hidden_dropout_prob=self.hidden_dropout_prob,
attention_probs_dropout_prob=self.attention_probs_dropout_prob,
is_decoder=False,
initializer_range=self.initializer_range,
frequency_stride=self.frequency_stride,
time_stride=self.time_stride,
)
def create_and_check_model(self, config, input_values, labels):
model = ASTModel(config=config)
model.to(torch_device)
model.eval()
result = model(input_values)
self.parent.assertEqual(result.last_hidden_state.shape, (self.batch_size, self.seq_length, self.hidden_size))
def prepare_config_and_inputs_for_common(self):
config_and_inputs = self.prepare_config_and_inputs()
(
config,
input_values,
labels,
) = config_and_inputs
inputs_dict = {"input_values": input_values}
return config, inputs_dict
@require_torch
class ASTModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
"""
Here we also overwrite some of the tests of test_modeling_common.py, as AST does not use input_ids, inputs_embeds,
attention_mask and seq_length.
"""
all_model_classes = (
(
ASTModel,
ASTForAudioClassification,
)
if is_torch_available()
else ()
)
pipeline_model_mapping = (
{"audio-classification": ASTForAudioClassification, "feature-extraction": ASTModel}
if is_torch_available()
else {}
)
fx_compatible = False
test_pruning = False
test_resize_embeddings = False
test_head_masking = False
def setUp(self):
self.model_tester = ASTModelTester(self)
self.config_tester = ConfigTester(self, config_class=ASTConfig, has_text_modality=False, hidden_size=37)
def test_config(self):
self.config_tester.run_common_tests()
@unittest.skip(reason="AST does not use inputs_embeds")
def test_inputs_embeds(self):
pass
def test_model_common_attributes(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
self.assertIsInstance(model.get_input_embeddings(), (nn.Module))
x = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(x, nn.Linear))
def test_forward_signature(self):
config, _ = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
model = model_class(config)
signature = inspect.signature(model.forward)
# signature.parameters is an OrderedDict => so arg_names order is deterministic
arg_names = [*signature.parameters.keys()]
expected_arg_names = ["input_values"]
self.assertListEqual(arg_names[:1], expected_arg_names)
def test_model(self):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*config_and_inputs)
@slow
def test_model_from_pretrained(self):
for model_name in AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
model = ASTModel.from_pretrained(model_name)
self.assertIsNotNone(model)
# We will verify our results on some audio from AudioSet
def prepare_audio():
filepath = hf_hub_download(
repo_id="nielsr/audio-spectogram-transformer-checkpoint", filename="sample_audio.flac", repo_type="dataset"
)
audio, sampling_rate = torchaudio.load(filepath)
return audio, sampling_rate
@require_torch
@require_torchaudio
class ASTModelIntegrationTest(unittest.TestCase):
@cached_property
def default_feature_extractor(self):
return (
ASTFeatureExtractor.from_pretrained("MIT/ast-finetuned-audioset-10-10-0.4593")
if is_torchaudio_available()
else None
)
@slow
def test_inference_audio_classification(self):
feature_extractor = self.default_feature_extractor
model = ASTForAudioClassification.from_pretrained("MIT/ast-finetuned-audioset-10-10-0.4593").to(torch_device)
feature_extractor = self.default_feature_extractor
audio, sampling_rate = prepare_audio()
audio = audio.squeeze().numpy()
inputs = feature_extractor(audio, sampling_rate=sampling_rate, return_tensors="pt").to(torch_device)
# forward pass
with torch.no_grad():
outputs = model(**inputs)
# verify the logits
expected_shape = torch.Size((1, 527))
self.assertEqual(outputs.logits.shape, expected_shape)
expected_slice = torch.tensor([-0.8760, -7.0042, -8.6602]).to(torch_device)
self.assertTrue(torch.allclose(outputs.logits[0, :3], expected_slice, atol=1e-4))
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./docs/source/pt/training.mdx | <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Fine-tuning de um modelo pré-treinado
[[open-in-colab]]
O uso de um modelo pré-treinado tem importantes vantagens. Redução do custo computacional, a pegada de carbono, e te
permite utilizar modelos de última geração sem ter que treinar um novo desde o início.
O 🤗 Transformers proporciona acesso a milhares de modelos pré-treinados numa ampla gama de tarefas.
Quando utilizar um modelo pré-treinado, treine-o com um dataset específico para a sua tarefa.
Isto é chamado de fine-tuning, uma técnica de treinamento incrivelmente poderosa. Neste tutorial faremos o fine-tuning
de um modelo pré-treinado com um framework de Deep Learning da sua escolha:
* Fine-tuning de um modelo pré-treinado com o 🤗 Transformers [`Trainer`].
* Fine-tuning de um modelo pré-treinado no TensorFlow com o Keras.
* Fine-tuning de um modelo pré-treinado em PyTorch nativo.
<a id='data-processing'></a>
## Preparando um dataset
<Youtube id="_BZearw7f0w"/>
Antes de aplicar o fine-tuning a um modelo pré-treinado, baixe um dataset e prepare-o para o treinamento.
O tutorial anterior ensinará a processar os dados para o treinamento, e então poderá ter a oportunidade de testar
esse novo conhecimento em algo prático.
Comece carregando o dataset [Yelp Reviews](https://huggingface.co/datasets/yelp_review_full):
```py
>>> from datasets import load_dataset
>>> dataset = load_dataset("yelp_review_full")
>>> dataset[100]
{'label': 0,
'text': 'My expectations for McDonalds are t rarely high. But for one to still fail so spectacularly...that takes something special!\\nThe cashier took my friends\'s order, then promptly ignored me. I had to force myself in front of a cashier who opened his register to wait on the person BEHIND me. I waited over five minutes for a gigantic order that included precisely one kid\'s meal. After watching two people who ordered after me be handed their food, I asked where mine was. The manager started yelling at the cashiers for \\"serving off their orders\\" when they didn\'t have their food. But neither cashier was anywhere near those controls, and the manager was the one serving food to customers and clearing the boards.\\nThe manager was rude when giving me my order. She didn\'t make sure that I had everything ON MY RECEIPT, and never even had the decency to apologize that I felt I was getting poor service.\\nI\'ve eaten at various McDonalds restaurants for over 30 years. I\'ve worked at more than one location. I expect bad days, bad moods, and the occasional mistake. But I have yet to have a decent experience at this store. It will remain a place I avoid unless someone in my party needs to avoid illness from low blood sugar. Perhaps I should go back to the racially biased service of Steak n Shake instead!'}
```
Como já sabe, é necessário ter um tokenizador para processar o texto e incluir uma estratégia de padding e truncamento,
para manejar qualquer tamanho varíavel de sequência. Para processar o seu dataset em apenas um passo, utilize o método de
🤗 Datasets [`map`](https://huggingface.co/docs/datasets/process.html#map) para aplicar uma função de preprocessamento sobre
todo o dataset.
```py
>>> from transformers import AutoTokenizer
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
>>> def tokenize_function(examples):
... return tokenizer(examples["text"], padding="max_length", truncation=True)
>>> tokenized_datasets = dataset.map(tokenize_function, batched=True)
```
Se desejar, é possível criar um subconjunto menor do dataset completo para aplicar o fine-tuning e assim reduzir o tempo necessário.
```py
>>> small_train_dataset = tokenized_datasets["train"].shuffle(seed=42).select(range(1000))
>>> small_eval_dataset = tokenized_datasets["test"].shuffle(seed=42).select(range(1000))
```
<a id='trainer'></a>
## Fine-tuning com o `Trainer`
<Youtube id="nvBXf7s7vTI"/>
O 🤗 Transformers proporciona uma classe [`Trainer`] otimizada para o treinamento de modelos de 🤗 Transformers,
facilitando os primeiros passos do treinamento sem a necessidade de escrever manualmente o seu próprio ciclo.
A API do [`Trainer`] suporta um grande conjunto de opções de treinamento e funcionalidades, como o logging,
o gradient accumulation e o mixed precision.
Comece carregando seu modelo e especifique o número de labels de previsão.
A partir do [Card Dataset](https://huggingface.co/datasets/yelp_review_full#data-fields) do Yelp Reveiw, que ja
sabemos ter 5 labels usamos o seguinte código:
```py
>>> from transformers import AutoModelForSequenceClassification
>>> model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
```
<Tip>
Você verá um alerta sobre alguns pesos pré-treinados que não estão sendo utilizados e que alguns pesos estão
sendo inicializados aleatoriamente. Não se preocupe, essa mensagem é completamente normal.
O header/cabeçário pré-treinado do modelo BERT é descartado e substitui-se por um header de classificação
inicializado aleatoriamente. Assim, pode aplicar o fine-tuning a este novo header do modelo em sua tarefa
de classificação de sequências fazendo um transfer learning do modelo pré-treinado.
</Tip>
### Hiperparâmetros de treinamento
Em seguida, crie uma classe [`TrainingArguments`] que contenha todos os hiperparâmetros que possam ser ajustados, assim
como os indicadores para ativar as diferentes opções de treinamento. Para este tutorial, você pode começar o treinamento
usando os [hiperparámetros](https://huggingface.co/docs/transformers/main_classes/trainer#transformers.TrainingArguments) padrão,
porém, sinta-se livre para experimentar com eles e encontrar uma configuração ótima.
Especifique onde salvar os checkpoints do treinamento:
```py
>>> from transformers import TrainingArguments
>>> training_args = TrainingArguments(output_dir="test_trainer")
```
### Métricas
O [`Trainer`] não avalia automaticamente o rendimento do modelo durante o treinamento. Será necessário passar ao
[`Trainer`] uma função para calcular e fazer um diagnóstico sobre as métricas. A biblioteca 🤗 Datasets proporciona
uma função de [`accuracy`](https://huggingface.co/metrics/accuracy) simples que pode ser carregada com a função
`load_metric` (ver este [tutorial](https://huggingface.co/docs/datasets/metrics.html) para mais informações):
```py
>>> import numpy as np
>>> from datasets import load_metric
>>> metric = load_metric("accuracy")
```
Defina a função `compute` dentro de `metric` para calcular a precisão das suas predições.
Antes de passar as suas predições ao `compute`, é necessário converter as predições à logits (lembre-se que
todos os modelos de 🤗 Transformers retornam logits).
```py
>>> def compute_metrics(eval_pred):
... logits, labels = eval_pred
... predictions = np.argmax(logits, axis=-1)
... return metric.compute(predictions=predictions, references=labels)
```
Se quiser controlar as suas métricas de avaliação durante o fine-tuning, especifique o parâmetro `evaluation_strategy`
nos seus argumentos de treinamento para que o modelo considere a métrica de avaliação ao final de cada época:
```py
>>> from transformers import TrainingArguments
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
```
### Trainer
Crie um objeto [`Trainer`] com o seu modelo, argumentos de treinamento, conjuntos de dados de treinamento e de teste, e a sua função de avaliação:
```py
>>> trainer = Trainer(
... model=model,
... args=training_args,
... train_dataset=small_train_dataset,
... eval_dataset=small_eval_dataset,
... compute_metrics=compute_metrics,
... )
```
Em seguida, aplique o fine-tuning a seu modelo chamado [`~transformers.Trainer.train`]:
```py
>>> trainer.train()
```
<a id='keras'></a>
## Fine-tuning com Keras
<Youtube id="rnTGBy2ax1c"/>
Os modelos de 🤗 Transformers também permitem realizar o treinamento com o TensorFlow com a API do Keras.
Contudo, será necessário fazer algumas mudanças antes de realizar o fine-tuning.
### Conversão do dataset ao formato do TensorFlow
O [`DefaultDataCollator`] junta os tensores em um batch para que o modelo possa ser treinado em cima deles.
Assegure-se de especificar os `return_tensors` para retornar os tensores do TensorFlow:
```py
>>> from transformers import DefaultDataCollator
>>> data_collator = DefaultDataCollator(return_tensors="tf")
```
<Tip>
O [`Trainer`] utiliza [`DataCollatorWithPadding`] por padrão, então você não precisa especificar explicitamente um
colador de dados (data collator).
</Tip>
Em seguida, converta os datasets tokenizados em datasets do TensorFlow com o método
[`to_tf_dataset`](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.to_tf_dataset).
Especifique suas entradas em `columns` e seu rótulo em `label_cols`:
```py
>>> tf_train_dataset = small_train_dataset.to_tf_dataset(
... columns=["attention_mask", "input_ids", "token_type_ids"],
... label_cols=["labels"],
... shuffle=True,
... collate_fn=data_collator,
... batch_size=8,
... )
>>> tf_validation_dataset = small_eval_dataset.to_tf_dataset(
... columns=["attention_mask", "input_ids", "token_type_ids"],
... label_cols=["labels"],
... shuffle=False,
... collate_fn=data_collator,
... batch_size=8,
... )
```
### Compilação e ajustes
Carregue um modelo do TensorFlow com o número esperado de rótulos:
```py
>>> import tensorflow as tf
>>> from transformers import TFAutoModelForSequenceClassification
>>> model = TFAutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
```
A seguir, compile e ajuste o fine-tuning a seu modelo com [`fit`](https://keras.io/api/models/model_training_apis/) como
faria com qualquer outro modelo do Keras:
```py
>>> model.compile(
... optimizer=tf.keras.optimizers.Adam(learning_rate=5e-5),
... loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
... metrics=tf.metrics.SparseCategoricalAccuracy(),
... )
>>> model.fit(tf_train_dataset, validation_data=tf_validation_dataset, epochs=3)
```
<a id='pytorch_native'></a>
## Fine-tune em PyTorch nativo
<Youtube id="Dh9CL8fyG80"/>
O [`Trainer`] se encarrega do ciclo de treinamento e permite aplicar o fine-tuning a um modelo em uma linha de código apenas.
Para os usuários que preferirem escrever o seu próprio ciclo de treinamento, também é possível aplicar o fine-tuning a um
modelo de 🤗 Transformers em PyTorch nativo.
Neste momento, talvez ocorra a necessidade de reinicar seu notebook ou executar a seguinte linha de código para liberar
memória:
```py
del model
del pytorch_model
del trainer
torch.cuda.empty_cache()
```
Em sequência, faremos um post-processing manual do `tokenized_dataset` e assim prepará-lo para o treinamento.
1. Apague a coluna de `text` porque o modelo não aceita texto cru como entrada:
```py
>>> tokenized_datasets = tokenized_datasets.remove_columns(["text"])
```
2. Troque o nome da coluna `label` para `labels`, pois o modelo espera um argumento de mesmo nome:
```py
>>> tokenized_datasets = tokenized_datasets.rename_column("label", "labels")
```
3. Defina o formato do dataset para retornar tensores do PyTorch no lugar de listas:
```py
>>> tokenized_datasets.set_format("torch")
```
Em sequência, crie um subconjunto menor do dataset, como foi mostrado anteriormente, para acelerá-lo o fine-tuning.
```py
>>> small_train_dataset = tokenized_datasets["train"].shuffle(seed=42).select(range(1000))
>>> small_eval_dataset = tokenized_datasets["test"].shuffle(seed=42).select(range(1000))
```
### DataLoader
Crie um `DataLoader` para os seus datasets de treinamento e de teste para poder iterar sobre batches de dados:
```py
>>> from torch.utils.data import DataLoader
>>> train_dataloader = DataLoader(small_train_dataset, shuffle=True, batch_size=8)
>>> eval_dataloader = DataLoader(small_eval_dataset, batch_size=8)
```
Carregue seu modelo com o número de labels esperados:
```py
>>> from transformers import AutoModelForSequenceClassification
>>> model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
```
### Otimização e configuração do Learning Rate
Crie um otimizador e um learning rate para aplicar o fine-tuning ao modelo.
Iremos utilizar o otimizador [`AdamW`](https://pytorch.org/docs/stable/generated/torch.optim.AdamW.html) do PyTorch:
```py
>>> from torch.optim import AdamW
>>> optimizer = AdamW(model.parameters(), lr=5e-5)
```
Defina o learning rate do [`Trainer`]:
```py
>>> from transformers import get_scheduler
>>> num_epochs = 3
>>> num_training_steps = num_epochs * len(train_dataloader)
>>> lr_scheduler = get_scheduler(
... name="linear", optimizer=optimizer, num_warmup_steps=0, num_training_steps=num_training_steps
... )
```
Por último, especifique o `device` do ambiente para utilizar uma GPU se tiver acesso à alguma. Caso contrário, o treinamento
em uma CPU pode acabar levando várias horas em vez de minutos.
```py
>>> import torch
>>> device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
>>> model.to(device)
```
<Tip>
Se necessário, você pode obter o acesso gratuito a uma GPU na núvem por meio de um notebook no
[Colaboratory](https://colab.research.google.com/) ou [SageMaker StudioLab](https://studiolab.sagemaker.aws/)
se não tiver esse recurso de forma local.
</Tip>
Perfeito, agora estamos prontos para começar o treinamento! 🥳
### Ciclo de treinamento
Para visualizar melhor o processo de treinamento, utilize a biblioteca [tqdm](https://tqdm.github.io/) para adicionar
uma barra de progresso sobre o número de passos percorridos no treinamento atual:
```py
>>> from tqdm.auto import tqdm
>>> progress_bar = tqdm(range(num_training_steps))
>>> model.train()
>>> for epoch in range(num_epochs):
... for batch in train_dataloader:
... batch = {k: v.to(device) for k, v in batch.items()}
... outputs = model(**batch)
... loss = outputs.loss
... loss.backward()
... optimizer.step()
... lr_scheduler.step()
... optimizer.zero_grad()
... progress_bar.update(1)
```
### Métricas
Da mesma forma que é necessário adicionar uma função de avaliação ao [`Trainer`], é necessário fazer o mesmo quando
escrevendo o próprio ciclo de treinamento. Contudo, em vez de calcular e retornar a métrica final de cada época,
você deverá adicionar todos os batches com [`add_batch`](https://huggingface.co/docs/datasets/package_reference/main_classes.html?highlight=add_batch#datasets.Metric.add_batch)
e calcular a métrica apenas no final.
```py
>>> metric = load_metric("accuracy")
>>> model.eval()
>>> for batch in eval_dataloader:
... batch = {k: v.to(device) for k, v in batch.items()}
... with torch.no_grad():
... outputs = model(**batch)
... logits = outputs.logits
... predictions = torch.argmax(logits, dim=-1)
... metric.add_batch(predictions=predictions, references=batch["labels"])
>>> metric.compute()
```
<a id='additional-resources'></a>
## Recursos adicionais
Para mais exemplos de fine-tuning acesse:
- [🤗 Transformers Examples](https://github.com/huggingface/transformers/tree/main/examples) inclui scripts
para treinas tarefas comuns de NLP em PyTorch e TensorFlow.
- [🤗 Transformers Notebooks](notebooks) contém vários notebooks sobre como aplicar o fine-tuning a um modelo
para tarefas específicas no PyTorch e TensorFlow.
| <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Fine-tuning de um modelo pré-treinado
[[open-in-colab]]
O uso de um modelo pré-treinado tem importantes vantagens. Redução do custo computacional, a pegada de carbono, e te
permite utilizar modelos de última geração sem ter que treinar um novo desde o início.
O 🤗 Transformers proporciona acesso a milhares de modelos pré-treinados numa ampla gama de tarefas.
Quando utilizar um modelo pré-treinado, treine-o com um dataset específico para a sua tarefa.
Isto é chamado de fine-tuning, uma técnica de treinamento incrivelmente poderosa. Neste tutorial faremos o fine-tuning
de um modelo pré-treinado com um framework de Deep Learning da sua escolha:
* Fine-tuning de um modelo pré-treinado com o 🤗 Transformers [`Trainer`].
* Fine-tuning de um modelo pré-treinado no TensorFlow com o Keras.
* Fine-tuning de um modelo pré-treinado em PyTorch nativo.
<a id='data-processing'></a>
## Preparando um dataset
<Youtube id="_BZearw7f0w"/>
Antes de aplicar o fine-tuning a um modelo pré-treinado, baixe um dataset e prepare-o para o treinamento.
O tutorial anterior ensinará a processar os dados para o treinamento, e então poderá ter a oportunidade de testar
esse novo conhecimento em algo prático.
Comece carregando o dataset [Yelp Reviews](https://huggingface.co/datasets/yelp_review_full):
```py
>>> from datasets import load_dataset
>>> dataset = load_dataset("yelp_review_full")
>>> dataset[100]
{'label': 0,
'text': 'My expectations for McDonalds are t rarely high. But for one to still fail so spectacularly...that takes something special!\\nThe cashier took my friends\'s order, then promptly ignored me. I had to force myself in front of a cashier who opened his register to wait on the person BEHIND me. I waited over five minutes for a gigantic order that included precisely one kid\'s meal. After watching two people who ordered after me be handed their food, I asked where mine was. The manager started yelling at the cashiers for \\"serving off their orders\\" when they didn\'t have their food. But neither cashier was anywhere near those controls, and the manager was the one serving food to customers and clearing the boards.\\nThe manager was rude when giving me my order. She didn\'t make sure that I had everything ON MY RECEIPT, and never even had the decency to apologize that I felt I was getting poor service.\\nI\'ve eaten at various McDonalds restaurants for over 30 years. I\'ve worked at more than one location. I expect bad days, bad moods, and the occasional mistake. But I have yet to have a decent experience at this store. It will remain a place I avoid unless someone in my party needs to avoid illness from low blood sugar. Perhaps I should go back to the racially biased service of Steak n Shake instead!'}
```
Como já sabe, é necessário ter um tokenizador para processar o texto e incluir uma estratégia de padding e truncamento,
para manejar qualquer tamanho varíavel de sequência. Para processar o seu dataset em apenas um passo, utilize o método de
🤗 Datasets [`map`](https://huggingface.co/docs/datasets/process.html#map) para aplicar uma função de preprocessamento sobre
todo o dataset.
```py
>>> from transformers import AutoTokenizer
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
>>> def tokenize_function(examples):
... return tokenizer(examples["text"], padding="max_length", truncation=True)
>>> tokenized_datasets = dataset.map(tokenize_function, batched=True)
```
Se desejar, é possível criar um subconjunto menor do dataset completo para aplicar o fine-tuning e assim reduzir o tempo necessário.
```py
>>> small_train_dataset = tokenized_datasets["train"].shuffle(seed=42).select(range(1000))
>>> small_eval_dataset = tokenized_datasets["test"].shuffle(seed=42).select(range(1000))
```
<a id='trainer'></a>
## Fine-tuning com o `Trainer`
<Youtube id="nvBXf7s7vTI"/>
O 🤗 Transformers proporciona uma classe [`Trainer`] otimizada para o treinamento de modelos de 🤗 Transformers,
facilitando os primeiros passos do treinamento sem a necessidade de escrever manualmente o seu próprio ciclo.
A API do [`Trainer`] suporta um grande conjunto de opções de treinamento e funcionalidades, como o logging,
o gradient accumulation e o mixed precision.
Comece carregando seu modelo e especifique o número de labels de previsão.
A partir do [Card Dataset](https://huggingface.co/datasets/yelp_review_full#data-fields) do Yelp Reveiw, que ja
sabemos ter 5 labels usamos o seguinte código:
```py
>>> from transformers import AutoModelForSequenceClassification
>>> model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
```
<Tip>
Você verá um alerta sobre alguns pesos pré-treinados que não estão sendo utilizados e que alguns pesos estão
sendo inicializados aleatoriamente. Não se preocupe, essa mensagem é completamente normal.
O header/cabeçário pré-treinado do modelo BERT é descartado e substitui-se por um header de classificação
inicializado aleatoriamente. Assim, pode aplicar o fine-tuning a este novo header do modelo em sua tarefa
de classificação de sequências fazendo um transfer learning do modelo pré-treinado.
</Tip>
### Hiperparâmetros de treinamento
Em seguida, crie uma classe [`TrainingArguments`] que contenha todos os hiperparâmetros que possam ser ajustados, assim
como os indicadores para ativar as diferentes opções de treinamento. Para este tutorial, você pode começar o treinamento
usando os [hiperparámetros](https://huggingface.co/docs/transformers/main_classes/trainer#transformers.TrainingArguments) padrão,
porém, sinta-se livre para experimentar com eles e encontrar uma configuração ótima.
Especifique onde salvar os checkpoints do treinamento:
```py
>>> from transformers import TrainingArguments
>>> training_args = TrainingArguments(output_dir="test_trainer")
```
### Métricas
O [`Trainer`] não avalia automaticamente o rendimento do modelo durante o treinamento. Será necessário passar ao
[`Trainer`] uma função para calcular e fazer um diagnóstico sobre as métricas. A biblioteca 🤗 Datasets proporciona
uma função de [`accuracy`](https://huggingface.co/metrics/accuracy) simples que pode ser carregada com a função
`load_metric` (ver este [tutorial](https://huggingface.co/docs/datasets/metrics.html) para mais informações):
```py
>>> import numpy as np
>>> from datasets import load_metric
>>> metric = load_metric("accuracy")
```
Defina a função `compute` dentro de `metric` para calcular a precisão das suas predições.
Antes de passar as suas predições ao `compute`, é necessário converter as predições à logits (lembre-se que
todos os modelos de 🤗 Transformers retornam logits).
```py
>>> def compute_metrics(eval_pred):
... logits, labels = eval_pred
... predictions = np.argmax(logits, axis=-1)
... return metric.compute(predictions=predictions, references=labels)
```
Se quiser controlar as suas métricas de avaliação durante o fine-tuning, especifique o parâmetro `evaluation_strategy`
nos seus argumentos de treinamento para que o modelo considere a métrica de avaliação ao final de cada época:
```py
>>> from transformers import TrainingArguments
>>> training_args = TrainingArguments(output_dir="test_trainer", evaluation_strategy="epoch")
```
### Trainer
Crie um objeto [`Trainer`] com o seu modelo, argumentos de treinamento, conjuntos de dados de treinamento e de teste, e a sua função de avaliação:
```py
>>> trainer = Trainer(
... model=model,
... args=training_args,
... train_dataset=small_train_dataset,
... eval_dataset=small_eval_dataset,
... compute_metrics=compute_metrics,
... )
```
Em seguida, aplique o fine-tuning a seu modelo chamado [`~transformers.Trainer.train`]:
```py
>>> trainer.train()
```
<a id='keras'></a>
## Fine-tuning com Keras
<Youtube id="rnTGBy2ax1c"/>
Os modelos de 🤗 Transformers também permitem realizar o treinamento com o TensorFlow com a API do Keras.
Contudo, será necessário fazer algumas mudanças antes de realizar o fine-tuning.
### Conversão do dataset ao formato do TensorFlow
O [`DefaultDataCollator`] junta os tensores em um batch para que o modelo possa ser treinado em cima deles.
Assegure-se de especificar os `return_tensors` para retornar os tensores do TensorFlow:
```py
>>> from transformers import DefaultDataCollator
>>> data_collator = DefaultDataCollator(return_tensors="tf")
```
<Tip>
O [`Trainer`] utiliza [`DataCollatorWithPadding`] por padrão, então você não precisa especificar explicitamente um
colador de dados (data collator).
</Tip>
Em seguida, converta os datasets tokenizados em datasets do TensorFlow com o método
[`to_tf_dataset`](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.to_tf_dataset).
Especifique suas entradas em `columns` e seu rótulo em `label_cols`:
```py
>>> tf_train_dataset = small_train_dataset.to_tf_dataset(
... columns=["attention_mask", "input_ids", "token_type_ids"],
... label_cols=["labels"],
... shuffle=True,
... collate_fn=data_collator,
... batch_size=8,
... )
>>> tf_validation_dataset = small_eval_dataset.to_tf_dataset(
... columns=["attention_mask", "input_ids", "token_type_ids"],
... label_cols=["labels"],
... shuffle=False,
... collate_fn=data_collator,
... batch_size=8,
... )
```
### Compilação e ajustes
Carregue um modelo do TensorFlow com o número esperado de rótulos:
```py
>>> import tensorflow as tf
>>> from transformers import TFAutoModelForSequenceClassification
>>> model = TFAutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
```
A seguir, compile e ajuste o fine-tuning a seu modelo com [`fit`](https://keras.io/api/models/model_training_apis/) como
faria com qualquer outro modelo do Keras:
```py
>>> model.compile(
... optimizer=tf.keras.optimizers.Adam(learning_rate=5e-5),
... loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
... metrics=tf.metrics.SparseCategoricalAccuracy(),
... )
>>> model.fit(tf_train_dataset, validation_data=tf_validation_dataset, epochs=3)
```
<a id='pytorch_native'></a>
## Fine-tune em PyTorch nativo
<Youtube id="Dh9CL8fyG80"/>
O [`Trainer`] se encarrega do ciclo de treinamento e permite aplicar o fine-tuning a um modelo em uma linha de código apenas.
Para os usuários que preferirem escrever o seu próprio ciclo de treinamento, também é possível aplicar o fine-tuning a um
modelo de 🤗 Transformers em PyTorch nativo.
Neste momento, talvez ocorra a necessidade de reinicar seu notebook ou executar a seguinte linha de código para liberar
memória:
```py
del model
del pytorch_model
del trainer
torch.cuda.empty_cache()
```
Em sequência, faremos um post-processing manual do `tokenized_dataset` e assim prepará-lo para o treinamento.
1. Apague a coluna de `text` porque o modelo não aceita texto cru como entrada:
```py
>>> tokenized_datasets = tokenized_datasets.remove_columns(["text"])
```
2. Troque o nome da coluna `label` para `labels`, pois o modelo espera um argumento de mesmo nome:
```py
>>> tokenized_datasets = tokenized_datasets.rename_column("label", "labels")
```
3. Defina o formato do dataset para retornar tensores do PyTorch no lugar de listas:
```py
>>> tokenized_datasets.set_format("torch")
```
Em sequência, crie um subconjunto menor do dataset, como foi mostrado anteriormente, para acelerá-lo o fine-tuning.
```py
>>> small_train_dataset = tokenized_datasets["train"].shuffle(seed=42).select(range(1000))
>>> small_eval_dataset = tokenized_datasets["test"].shuffle(seed=42).select(range(1000))
```
### DataLoader
Crie um `DataLoader` para os seus datasets de treinamento e de teste para poder iterar sobre batches de dados:
```py
>>> from torch.utils.data import DataLoader
>>> train_dataloader = DataLoader(small_train_dataset, shuffle=True, batch_size=8)
>>> eval_dataloader = DataLoader(small_eval_dataset, batch_size=8)
```
Carregue seu modelo com o número de labels esperados:
```py
>>> from transformers import AutoModelForSequenceClassification
>>> model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
```
### Otimização e configuração do Learning Rate
Crie um otimizador e um learning rate para aplicar o fine-tuning ao modelo.
Iremos utilizar o otimizador [`AdamW`](https://pytorch.org/docs/stable/generated/torch.optim.AdamW.html) do PyTorch:
```py
>>> from torch.optim import AdamW
>>> optimizer = AdamW(model.parameters(), lr=5e-5)
```
Defina o learning rate do [`Trainer`]:
```py
>>> from transformers import get_scheduler
>>> num_epochs = 3
>>> num_training_steps = num_epochs * len(train_dataloader)
>>> lr_scheduler = get_scheduler(
... name="linear", optimizer=optimizer, num_warmup_steps=0, num_training_steps=num_training_steps
... )
```
Por último, especifique o `device` do ambiente para utilizar uma GPU se tiver acesso à alguma. Caso contrário, o treinamento
em uma CPU pode acabar levando várias horas em vez de minutos.
```py
>>> import torch
>>> device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
>>> model.to(device)
```
<Tip>
Se necessário, você pode obter o acesso gratuito a uma GPU na núvem por meio de um notebook no
[Colaboratory](https://colab.research.google.com/) ou [SageMaker StudioLab](https://studiolab.sagemaker.aws/)
se não tiver esse recurso de forma local.
</Tip>
Perfeito, agora estamos prontos para começar o treinamento! 🥳
### Ciclo de treinamento
Para visualizar melhor o processo de treinamento, utilize a biblioteca [tqdm](https://tqdm.github.io/) para adicionar
uma barra de progresso sobre o número de passos percorridos no treinamento atual:
```py
>>> from tqdm.auto import tqdm
>>> progress_bar = tqdm(range(num_training_steps))
>>> model.train()
>>> for epoch in range(num_epochs):
... for batch in train_dataloader:
... batch = {k: v.to(device) for k, v in batch.items()}
... outputs = model(**batch)
... loss = outputs.loss
... loss.backward()
... optimizer.step()
... lr_scheduler.step()
... optimizer.zero_grad()
... progress_bar.update(1)
```
### Métricas
Da mesma forma que é necessário adicionar uma função de avaliação ao [`Trainer`], é necessário fazer o mesmo quando
escrevendo o próprio ciclo de treinamento. Contudo, em vez de calcular e retornar a métrica final de cada época,
você deverá adicionar todos os batches com [`add_batch`](https://huggingface.co/docs/datasets/package_reference/main_classes.html?highlight=add_batch#datasets.Metric.add_batch)
e calcular a métrica apenas no final.
```py
>>> metric = load_metric("accuracy")
>>> model.eval()
>>> for batch in eval_dataloader:
... batch = {k: v.to(device) for k, v in batch.items()}
... with torch.no_grad():
... outputs = model(**batch)
... logits = outputs.logits
... predictions = torch.argmax(logits, dim=-1)
... metric.add_batch(predictions=predictions, references=batch["labels"])
>>> metric.compute()
```
<a id='additional-resources'></a>
## Recursos adicionais
Para mais exemplos de fine-tuning acesse:
- [🤗 Transformers Examples](https://github.com/huggingface/transformers/tree/main/examples) inclui scripts
para treinas tarefas comuns de NLP em PyTorch e TensorFlow.
- [🤗 Transformers Notebooks](notebooks) contém vários notebooks sobre como aplicar o fine-tuning a um modelo
para tarefas específicas no PyTorch e TensorFlow.
| -1 |
huggingface/transformers | 20,211 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch" | # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ArthurZucker | "2022-11-14T15:30:34Z" | "2023-03-01T09:49:22Z" | b29e2dcaff114762e65eaea739ba1076fc5d1c84 | 44e3e3fb4930298f092f336c2b7add3ebf051928 | prepare for "__floordiv__ is deprecated and its behavior will change in a future version of pytorch". # What does this PR do?
Should adress the `__floordiv__` warnings mentionned in #19934. Divinding torch tensor using `//` is no longer supported and has to be done via `torch.div`.
| ./examples/research_projects/rag-end2end-retriever/callbacks_rag.py | import logging
from pathlib import Path
import numpy as np
import pytorch_lightning as pl
import torch
from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint
from pytorch_lightning.utilities import rank_zero_only
from utils_rag import save_json
def count_trainable_parameters(model):
model_parameters = filter(lambda p: p.requires_grad, model.parameters())
params = sum([np.prod(p.size()) for p in model_parameters])
return params
logger = logging.getLogger(__name__)
def get_checkpoint_callback(output_dir, metric):
"""Saves the best model by validation EM score."""
if metric == "rouge2":
exp = "{val_avg_rouge2:.4f}-{step_count}"
elif metric == "bleu":
exp = "{val_avg_bleu:.4f}-{step_count}"
elif metric == "em":
exp = "{val_avg_em:.4f}-{step_count}"
elif metric == "loss":
exp = "{val_avg_loss:.4f}-{step_count}"
else:
raise NotImplementedError(
f"seq2seq callbacks only support rouge2 and bleu, got {metric}, You can make your own by adding to this"
" function."
)
checkpoint_callback = ModelCheckpoint(
dirpath=output_dir,
filename=exp,
monitor=f"val_{metric}",
mode="max",
save_top_k=1,
every_n_epochs=1, # works only with PL > 1.3
)
return checkpoint_callback
def get_early_stopping_callback(metric, patience):
return EarlyStopping(
monitor=f"val_{metric}", # does this need avg?
mode="min" if "loss" in metric else "max",
patience=patience,
verbose=True,
)
class Seq2SeqLoggingCallback(pl.Callback):
def on_batch_end(self, trainer, pl_module):
lrs = {f"lr_group_{i}": param["lr"] for i, param in enumerate(pl_module.trainer.optimizers[0].param_groups)}
pl_module.logger.log_metrics(lrs)
@rank_zero_only
def _write_logs(
self, trainer: pl.Trainer, pl_module: pl.LightningModule, type_path: str, save_generations=True
) -> None:
logger.info(f"***** {type_path} results at step {trainer.global_step:05d} *****")
metrics = trainer.callback_metrics
trainer.logger.log_metrics({k: v for k, v in metrics.items() if k not in ["log", "progress_bar", "preds"]})
# Log results
od = Path(pl_module.hparams.output_dir)
if type_path == "test":
results_file = od / "test_results.txt"
generations_file = od / "test_generations.txt"
else:
# this never gets hit. I prefer not to save intermediate generations, and results are in metrics.json
# If people want this it will be easy enough to add back.
results_file = od / f"{type_path}_results/{trainer.global_step:05d}.txt"
generations_file = od / f"{type_path}_generations/{trainer.global_step:05d}.txt"
results_file.parent.mkdir(exist_ok=True)
generations_file.parent.mkdir(exist_ok=True)
with open(results_file, "a+") as writer:
for key in sorted(metrics):
if key in ["log", "progress_bar", "preds"]:
continue
val = metrics[key]
if isinstance(val, torch.Tensor):
val = val.item()
msg = f"{key}: {val:.6f}\n"
writer.write(msg)
if not save_generations:
return
if "preds" in metrics:
content = "\n".join(metrics["preds"])
generations_file.open("w+").write(content)
@rank_zero_only
def on_train_start(self, trainer, pl_module):
try:
npars = pl_module.model.model.num_parameters()
except AttributeError:
npars = pl_module.model.num_parameters()
n_trainable_pars = count_trainable_parameters(pl_module)
# mp stands for million parameters
trainer.logger.log_metrics({"n_params": npars, "mp": npars / 1e6, "grad_mp": n_trainable_pars / 1e6})
@rank_zero_only
def on_test_end(self, trainer: pl.Trainer, pl_module: pl.LightningModule):
save_json(pl_module.metrics, pl_module.metrics_save_path)
return self._write_logs(trainer, pl_module, "test")
@rank_zero_only
def on_validation_end(self, trainer: pl.Trainer, pl_module):
save_json(pl_module.metrics, pl_module.metrics_save_path)
# Uncommenting this will save val generations
# return self._write_logs(trainer, pl_module, "valid")
| import logging
from pathlib import Path
import numpy as np
import pytorch_lightning as pl
import torch
from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint
from pytorch_lightning.utilities import rank_zero_only
from utils_rag import save_json
def count_trainable_parameters(model):
model_parameters = filter(lambda p: p.requires_grad, model.parameters())
params = sum([np.prod(p.size()) for p in model_parameters])
return params
logger = logging.getLogger(__name__)
def get_checkpoint_callback(output_dir, metric):
"""Saves the best model by validation EM score."""
if metric == "rouge2":
exp = "{val_avg_rouge2:.4f}-{step_count}"
elif metric == "bleu":
exp = "{val_avg_bleu:.4f}-{step_count}"
elif metric == "em":
exp = "{val_avg_em:.4f}-{step_count}"
elif metric == "loss":
exp = "{val_avg_loss:.4f}-{step_count}"
else:
raise NotImplementedError(
f"seq2seq callbacks only support rouge2 and bleu, got {metric}, You can make your own by adding to this"
" function."
)
checkpoint_callback = ModelCheckpoint(
dirpath=output_dir,
filename=exp,
monitor=f"val_{metric}",
mode="max",
save_top_k=1,
every_n_epochs=1, # works only with PL > 1.3
)
return checkpoint_callback
def get_early_stopping_callback(metric, patience):
return EarlyStopping(
monitor=f"val_{metric}", # does this need avg?
mode="min" if "loss" in metric else "max",
patience=patience,
verbose=True,
)
class Seq2SeqLoggingCallback(pl.Callback):
def on_batch_end(self, trainer, pl_module):
lrs = {f"lr_group_{i}": param["lr"] for i, param in enumerate(pl_module.trainer.optimizers[0].param_groups)}
pl_module.logger.log_metrics(lrs)
@rank_zero_only
def _write_logs(
self, trainer: pl.Trainer, pl_module: pl.LightningModule, type_path: str, save_generations=True
) -> None:
logger.info(f"***** {type_path} results at step {trainer.global_step:05d} *****")
metrics = trainer.callback_metrics
trainer.logger.log_metrics({k: v for k, v in metrics.items() if k not in ["log", "progress_bar", "preds"]})
# Log results
od = Path(pl_module.hparams.output_dir)
if type_path == "test":
results_file = od / "test_results.txt"
generations_file = od / "test_generations.txt"
else:
# this never gets hit. I prefer not to save intermediate generations, and results are in metrics.json
# If people want this it will be easy enough to add back.
results_file = od / f"{type_path}_results/{trainer.global_step:05d}.txt"
generations_file = od / f"{type_path}_generations/{trainer.global_step:05d}.txt"
results_file.parent.mkdir(exist_ok=True)
generations_file.parent.mkdir(exist_ok=True)
with open(results_file, "a+") as writer:
for key in sorted(metrics):
if key in ["log", "progress_bar", "preds"]:
continue
val = metrics[key]
if isinstance(val, torch.Tensor):
val = val.item()
msg = f"{key}: {val:.6f}\n"
writer.write(msg)
if not save_generations:
return
if "preds" in metrics:
content = "\n".join(metrics["preds"])
generations_file.open("w+").write(content)
@rank_zero_only
def on_train_start(self, trainer, pl_module):
try:
npars = pl_module.model.model.num_parameters()
except AttributeError:
npars = pl_module.model.num_parameters()
n_trainable_pars = count_trainable_parameters(pl_module)
# mp stands for million parameters
trainer.logger.log_metrics({"n_params": npars, "mp": npars / 1e6, "grad_mp": n_trainable_pars / 1e6})
@rank_zero_only
def on_test_end(self, trainer: pl.Trainer, pl_module: pl.LightningModule):
save_json(pl_module.metrics, pl_module.metrics_save_path)
return self._write_logs(trainer, pl_module, "test")
@rank_zero_only
def on_validation_end(self, trainer: pl.Trainer, pl_module):
save_json(pl_module.metrics, pl_module.metrics_save_path)
# Uncommenting this will save val generations
# return self._write_logs(trainer, pl_module, "valid")
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<b>English</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>State-of-the-art Machine Learning for JAX, PyTorch and TensorFlow</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers provides thousands of pretrained models to perform tasks on different modalities such as text, vision, and audio.
These models can be applied on:
* 📝 Text, for tasks like text classification, information extraction, question answering, summarization, translation, text generation, in over 100 languages.
* 🖼️ Images, for tasks like image classification, object detection, and segmentation.
* 🗣️ Audio, for tasks like speech recognition and audio classification.
Transformer models can also perform tasks on **several modalities combined**, such as table question answering, optical character recognition, information extraction from scanned documents, video classification, and visual question answering.
🤗 Transformers provides APIs to quickly download and use those pretrained models on a given text, fine-tune them on your own datasets and then share them with the community on our [model hub](https://huggingface.co/models). At the same time, each python module defining an architecture is fully standalone and can be modified to enable quick research experiments.
🤗 Transformers is backed by the three most popular deep learning libraries — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — with a seamless integration between them. It's straightforward to train your models with one before loading them for inference with the other.
## Online demos
You can test most of our models directly on their pages from the [model hub](https://huggingface.co/models). We also offer [private model hosting, versioning, & an inference API](https://huggingface.co/pricing) for public and private models.
Here are a few examples:
In Natural Language Processing:
- [Masked word completion with BERT](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Name Entity Recognition with Electra](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [Text generation with GPT-2](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [Natural Language Inference with RoBERTa](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [Summarization with BART](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [Question answering with DistilBERT](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [Translation with T5](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
In Computer Vision:
- [Image classification with ViT](https://huggingface.co/google/vit-base-patch16-224)
- [Object Detection with DETR](https://huggingface.co/facebook/detr-resnet-50)
- [Semantic Segmentation with SegFormer](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
- [Panoptic Segmentation with DETR](https://huggingface.co/facebook/detr-resnet-50-panoptic)
In Audio:
- [Automatic Speech Recognition with Wav2Vec2](https://huggingface.co/facebook/wav2vec2-base-960h)
- [Keyword Spotting with Wav2Vec2](https://huggingface.co/superb/wav2vec2-base-superb-ks)
In Multimodal tasks:
- [Visual Question Answering with ViLT](https://huggingface.co/dandelin/vilt-b32-finetuned-vqa)
**[Write With Transformer](https://transformer.huggingface.co)**, built by the Hugging Face team, is the official demo of this repo’s text generation capabilities.
## If you are looking for custom support from the Hugging Face team
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## Quick tour
To immediately use a model on a given input (text, image, audio, ...), we provide the `pipeline` API. Pipelines group together a pretrained model with the preprocessing that was used during that model's training. Here is how to quickly use a pipeline to classify positive versus negative texts:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
The second line of code downloads and caches the pretrained model used by the pipeline, while the third evaluates it on the given text. Here the answer is "positive" with a confidence of 99.97%.
Many tasks have a pre-trained `pipeline` ready to go, in NLP but also in computer vision and speech. For example, we can easily extract detected objects in an image:
``` python
>>> import requests
>>> from PIL import Image
>>> from transformers import pipeline
# Download an image with cute cats
>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
>>> image_data = requests.get(url, stream=True).raw
>>> image = Image.open(image_data)
# Allocate a pipeline for object detection
>>> object_detector = pipeline('object-detection')
>>> object_detector(image)
[{'score': 0.9982201457023621,
'label': 'remote',
'box': {'xmin': 40, 'ymin': 70, 'xmax': 175, 'ymax': 117}},
{'score': 0.9960021376609802,
'label': 'remote',
'box': {'xmin': 333, 'ymin': 72, 'xmax': 368, 'ymax': 187}},
{'score': 0.9954745173454285,
'label': 'couch',
'box': {'xmin': 0, 'ymin': 1, 'xmax': 639, 'ymax': 473}},
{'score': 0.9988006353378296,
'label': 'cat',
'box': {'xmin': 13, 'ymin': 52, 'xmax': 314, 'ymax': 470}},
{'score': 0.9986783862113953,
'label': 'cat',
'box': {'xmin': 345, 'ymin': 23, 'xmax': 640, 'ymax': 368}}]
```
Here we get a list of objects detected in the image, with a box surrounding the object and a confidence score. Here is the original image on the left, with the predictions displayed on the right:
<h3 align="center">
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png" width="400"></a>
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample_post_processed.png" width="400"></a>
</h3>
You can learn more about the tasks supported by the `pipeline` API in [this tutorial](https://huggingface.co/docs/transformers/task_summary).
In addition to `pipeline`, to download and use any of the pretrained models on your given task, all it takes is three lines of code. Here is the PyTorch version:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
And here is the equivalent code for TensorFlow:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
The tokenizer is responsible for all the preprocessing the pretrained model expects, and can be called directly on a single string (as in the above examples) or a list. It will output a dictionary that you can use in downstream code or simply directly pass to your model using the ** argument unpacking operator.
The model itself is a regular [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) or a [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) (depending on your backend) which you can use as usual. [This tutorial](https://huggingface.co/docs/transformers/training) explains how to integrate such a model into a classic PyTorch or TensorFlow training loop, or how to use our `Trainer` API to quickly fine-tune on a new dataset.
## Why should I use transformers?
1. Easy-to-use state-of-the-art models:
- High performance on natural language understanding & generation, computer vision, and audio tasks.
- Low barrier to entry for educators and practitioners.
- Few user-facing abstractions with just three classes to learn.
- A unified API for using all our pretrained models.
1. Lower compute costs, smaller carbon footprint:
- Researchers can share trained models instead of always retraining.
- Practitioners can reduce compute time and production costs.
- Dozens of architectures with over 60,000 pretrained models across all modalities.
1. Choose the right framework for every part of a model's lifetime:
- Train state-of-the-art models in 3 lines of code.
- Move a single model between TF2.0/PyTorch/JAX frameworks at will.
- Seamlessly pick the right framework for training, evaluation and production.
1. Easily customize a model or an example to your needs:
- We provide examples for each architecture to reproduce the results published by its original authors.
- Model internals are exposed as consistently as possible.
- Model files can be used independently of the library for quick experiments.
## Why shouldn't I use transformers?
- This library is not a modular toolbox of building blocks for neural nets. The code in the model files is not refactored with additional abstractions on purpose, so that researchers can quickly iterate on each of the models without diving into additional abstractions/files.
- The training API is not intended to work on any model but is optimized to work with the models provided by the library. For generic machine learning loops, you should use another library (possibly, [Accelerate](https://huggingface.co/docs/accelerate)).
- While we strive to present as many use cases as possible, the scripts in our [examples folder](https://github.com/huggingface/transformers/tree/main/examples) are just that: examples. It is expected that they won't work out-of-the box on your specific problem and that you will be required to change a few lines of code to adapt them to your needs.
## Installation
### With pip
This repository is tested on Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+ and TensorFlow 2.3+.
You should install 🤗 Transformers in a [virtual environment](https://docs.python.org/3/library/venv.html). If you're unfamiliar with Python virtual environments, check out the [user guide](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
First, create a virtual environment with the version of Python you're going to use and activate it.
Then, you will need to install at least one of Flax, PyTorch or TensorFlow.
Please refer to [TensorFlow installation page](https://www.tensorflow.org/install/), [PyTorch installation page](https://pytorch.org/get-started/locally/#start-locally) and/or [Flax](https://github.com/google/flax#quick-install) and [Jax](https://github.com/google/jax#installation) installation pages regarding the specific installation command for your platform.
When one of those backends has been installed, 🤗 Transformers can be installed using pip as follows:
```bash
pip install transformers
```
If you'd like to play with the examples or need the bleeding edge of the code and can't wait for a new release, you must [install the library from source](https://huggingface.co/docs/transformers/installation#installing-from-source).
### With conda
Since Transformers version v4.0.0, we now have a conda channel: `huggingface`.
🤗 Transformers can be installed using conda as follows:
```shell script
conda install -c huggingface transformers
```
Follow the installation pages of Flax, PyTorch or TensorFlow to see how to install them with conda.
> **_NOTE:_** On Windows, you may be prompted to activate Developer Mode in order to benefit from caching. If this is not an option for you, please let us know in [this issue](https://github.com/huggingface/huggingface_hub/issues/1062).
## Model architectures
**[All the model checkpoints](https://huggingface.co/models)** provided by 🤗 Transformers are seamlessly integrated from the huggingface.co [model hub](https://huggingface.co) where they are uploaded directly by [users](https://huggingface.co/users) and [organizations](https://huggingface.co/organizations).
Current number of checkpoints: 
🤗 Transformers currently provides the following architectures (see [here](https://huggingface.co/docs/transformers/model_summary) for a high-level summary of each them):
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT): General Visual Representation Learning](https://arxiv.org/abs/1912.11370) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2 and ESMFold** were released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. Want to contribute a new model? We have added a **detailed guide and templates** to guide you in the process of adding a new model. You can find them in the [`templates`](./templates) folder of the repository. Be sure to check the [contributing guidelines](./CONTRIBUTING.md) and contact the maintainers or open an issue to collect feedbacks before starting your PR.
To check if each model has an implementation in Flax, PyTorch or TensorFlow, or has an associated tokenizer backed by the 🤗 Tokenizers library, refer to [this table](https://huggingface.co/docs/transformers/index#supported-frameworks).
These implementations have been tested on several datasets (see the example scripts) and should match the performance of the original implementations. You can find more details on performance in the Examples section of the [documentation](https://github.com/huggingface/transformers/tree/main/examples).
## Learn more
| Section | Description |
|-|-|
| [Documentation](https://huggingface.co/docs/transformers/) | Full API documentation and tutorials |
| [Task summary](https://huggingface.co/docs/transformers/task_summary) | Tasks supported by 🤗 Transformers |
| [Preprocessing tutorial](https://huggingface.co/docs/transformers/preprocessing) | Using the `Tokenizer` class to prepare data for the models |
| [Training and fine-tuning](https://huggingface.co/docs/transformers/training) | Using the models provided by 🤗 Transformers in a PyTorch/TensorFlow training loop and the `Trainer` API |
| [Quick tour: Fine-tuning/usage scripts](https://github.com/huggingface/transformers/tree/main/examples) | Example scripts for fine-tuning models on a wide range of tasks |
| [Model sharing and uploading](https://huggingface.co/docs/transformers/model_sharing) | Upload and share your fine-tuned models with the community |
| [Migration](https://huggingface.co/docs/transformers/migration) | Migrate to 🤗 Transformers from `pytorch-transformers` or `pytorch-pretrained-bert` |
## Citation
We now have a [paper](https://www.aclweb.org/anthology/2020.emnlp-demos.6/) you can cite for the 🤗 Transformers library:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<b>English</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>State-of-the-art Machine Learning for JAX, PyTorch and TensorFlow</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers provides thousands of pretrained models to perform tasks on different modalities such as text, vision, and audio.
These models can be applied on:
* 📝 Text, for tasks like text classification, information extraction, question answering, summarization, translation, text generation, in over 100 languages.
* 🖼️ Images, for tasks like image classification, object detection, and segmentation.
* 🗣️ Audio, for tasks like speech recognition and audio classification.
Transformer models can also perform tasks on **several modalities combined**, such as table question answering, optical character recognition, information extraction from scanned documents, video classification, and visual question answering.
🤗 Transformers provides APIs to quickly download and use those pretrained models on a given text, fine-tune them on your own datasets and then share them with the community on our [model hub](https://huggingface.co/models). At the same time, each python module defining an architecture is fully standalone and can be modified to enable quick research experiments.
🤗 Transformers is backed by the three most popular deep learning libraries — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — with a seamless integration between them. It's straightforward to train your models with one before loading them for inference with the other.
## Online demos
You can test most of our models directly on their pages from the [model hub](https://huggingface.co/models). We also offer [private model hosting, versioning, & an inference API](https://huggingface.co/pricing) for public and private models.
Here are a few examples:
In Natural Language Processing:
- [Masked word completion with BERT](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Name Entity Recognition with Electra](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [Text generation with GPT-2](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [Natural Language Inference with RoBERTa](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [Summarization with BART](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [Question answering with DistilBERT](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [Translation with T5](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
In Computer Vision:
- [Image classification with ViT](https://huggingface.co/google/vit-base-patch16-224)
- [Object Detection with DETR](https://huggingface.co/facebook/detr-resnet-50)
- [Semantic Segmentation with SegFormer](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
- [Panoptic Segmentation with DETR](https://huggingface.co/facebook/detr-resnet-50-panoptic)
In Audio:
- [Automatic Speech Recognition with Wav2Vec2](https://huggingface.co/facebook/wav2vec2-base-960h)
- [Keyword Spotting with Wav2Vec2](https://huggingface.co/superb/wav2vec2-base-superb-ks)
In Multimodal tasks:
- [Visual Question Answering with ViLT](https://huggingface.co/dandelin/vilt-b32-finetuned-vqa)
**[Write With Transformer](https://transformer.huggingface.co)**, built by the Hugging Face team, is the official demo of this repo’s text generation capabilities.
## If you are looking for custom support from the Hugging Face team
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## Quick tour
To immediately use a model on a given input (text, image, audio, ...), we provide the `pipeline` API. Pipelines group together a pretrained model with the preprocessing that was used during that model's training. Here is how to quickly use a pipeline to classify positive versus negative texts:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
The second line of code downloads and caches the pretrained model used by the pipeline, while the third evaluates it on the given text. Here the answer is "positive" with a confidence of 99.97%.
Many tasks have a pre-trained `pipeline` ready to go, in NLP but also in computer vision and speech. For example, we can easily extract detected objects in an image:
``` python
>>> import requests
>>> from PIL import Image
>>> from transformers import pipeline
# Download an image with cute cats
>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
>>> image_data = requests.get(url, stream=True).raw
>>> image = Image.open(image_data)
# Allocate a pipeline for object detection
>>> object_detector = pipeline('object-detection')
>>> object_detector(image)
[{'score': 0.9982201457023621,
'label': 'remote',
'box': {'xmin': 40, 'ymin': 70, 'xmax': 175, 'ymax': 117}},
{'score': 0.9960021376609802,
'label': 'remote',
'box': {'xmin': 333, 'ymin': 72, 'xmax': 368, 'ymax': 187}},
{'score': 0.9954745173454285,
'label': 'couch',
'box': {'xmin': 0, 'ymin': 1, 'xmax': 639, 'ymax': 473}},
{'score': 0.9988006353378296,
'label': 'cat',
'box': {'xmin': 13, 'ymin': 52, 'xmax': 314, 'ymax': 470}},
{'score': 0.9986783862113953,
'label': 'cat',
'box': {'xmin': 345, 'ymin': 23, 'xmax': 640, 'ymax': 368}}]
```
Here we get a list of objects detected in the image, with a box surrounding the object and a confidence score. Here is the original image on the left, with the predictions displayed on the right:
<h3 align="center">
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png" width="400"></a>
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample_post_processed.png" width="400"></a>
</h3>
You can learn more about the tasks supported by the `pipeline` API in [this tutorial](https://huggingface.co/docs/transformers/task_summary).
In addition to `pipeline`, to download and use any of the pretrained models on your given task, all it takes is three lines of code. Here is the PyTorch version:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
And here is the equivalent code for TensorFlow:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
The tokenizer is responsible for all the preprocessing the pretrained model expects, and can be called directly on a single string (as in the above examples) or a list. It will output a dictionary that you can use in downstream code or simply directly pass to your model using the ** argument unpacking operator.
The model itself is a regular [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) or a [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) (depending on your backend) which you can use as usual. [This tutorial](https://huggingface.co/docs/transformers/training) explains how to integrate such a model into a classic PyTorch or TensorFlow training loop, or how to use our `Trainer` API to quickly fine-tune on a new dataset.
## Why should I use transformers?
1. Easy-to-use state-of-the-art models:
- High performance on natural language understanding & generation, computer vision, and audio tasks.
- Low barrier to entry for educators and practitioners.
- Few user-facing abstractions with just three classes to learn.
- A unified API for using all our pretrained models.
1. Lower compute costs, smaller carbon footprint:
- Researchers can share trained models instead of always retraining.
- Practitioners can reduce compute time and production costs.
- Dozens of architectures with over 60,000 pretrained models across all modalities.
1. Choose the right framework for every part of a model's lifetime:
- Train state-of-the-art models in 3 lines of code.
- Move a single model between TF2.0/PyTorch/JAX frameworks at will.
- Seamlessly pick the right framework for training, evaluation and production.
1. Easily customize a model or an example to your needs:
- We provide examples for each architecture to reproduce the results published by its original authors.
- Model internals are exposed as consistently as possible.
- Model files can be used independently of the library for quick experiments.
## Why shouldn't I use transformers?
- This library is not a modular toolbox of building blocks for neural nets. The code in the model files is not refactored with additional abstractions on purpose, so that researchers can quickly iterate on each of the models without diving into additional abstractions/files.
- The training API is not intended to work on any model but is optimized to work with the models provided by the library. For generic machine learning loops, you should use another library (possibly, [Accelerate](https://huggingface.co/docs/accelerate)).
- While we strive to present as many use cases as possible, the scripts in our [examples folder](https://github.com/huggingface/transformers/tree/main/examples) are just that: examples. It is expected that they won't work out-of-the box on your specific problem and that you will be required to change a few lines of code to adapt them to your needs.
## Installation
### With pip
This repository is tested on Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+ and TensorFlow 2.3+.
You should install 🤗 Transformers in a [virtual environment](https://docs.python.org/3/library/venv.html). If you're unfamiliar with Python virtual environments, check out the [user guide](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
First, create a virtual environment with the version of Python you're going to use and activate it.
Then, you will need to install at least one of Flax, PyTorch or TensorFlow.
Please refer to [TensorFlow installation page](https://www.tensorflow.org/install/), [PyTorch installation page](https://pytorch.org/get-started/locally/#start-locally) and/or [Flax](https://github.com/google/flax#quick-install) and [Jax](https://github.com/google/jax#installation) installation pages regarding the specific installation command for your platform.
When one of those backends has been installed, 🤗 Transformers can be installed using pip as follows:
```bash
pip install transformers
```
If you'd like to play with the examples or need the bleeding edge of the code and can't wait for a new release, you must [install the library from source](https://huggingface.co/docs/transformers/installation#installing-from-source).
### With conda
Since Transformers version v4.0.0, we now have a conda channel: `huggingface`.
🤗 Transformers can be installed using conda as follows:
```shell script
conda install -c huggingface transformers
```
Follow the installation pages of Flax, PyTorch or TensorFlow to see how to install them with conda.
> **_NOTE:_** On Windows, you may be prompted to activate Developer Mode in order to benefit from caching. If this is not an option for you, please let us know in [this issue](https://github.com/huggingface/huggingface_hub/issues/1062).
## Model architectures
**[All the model checkpoints](https://huggingface.co/models)** provided by 🤗 Transformers are seamlessly integrated from the huggingface.co [model hub](https://huggingface.co) where they are uploaded directly by [users](https://huggingface.co/users) and [organizations](https://huggingface.co/organizations).
Current number of checkpoints: 
🤗 Transformers currently provides the following architectures (see [here](https://huggingface.co/docs/transformers/model_summary) for a high-level summary of each them):
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT): General Visual Representation Learning](https://arxiv.org/abs/1912.11370) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2 and ESMFold** were released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. Want to contribute a new model? We have added a **detailed guide and templates** to guide you in the process of adding a new model. You can find them in the [`templates`](./templates) folder of the repository. Be sure to check the [contributing guidelines](./CONTRIBUTING.md) and contact the maintainers or open an issue to collect feedbacks before starting your PR.
To check if each model has an implementation in Flax, PyTorch or TensorFlow, or has an associated tokenizer backed by the 🤗 Tokenizers library, refer to [this table](https://huggingface.co/docs/transformers/index#supported-frameworks).
These implementations have been tested on several datasets (see the example scripts) and should match the performance of the original implementations. You can find more details on performance in the Examples section of the [documentation](https://github.com/huggingface/transformers/tree/main/examples).
## Learn more
| Section | Description |
|-|-|
| [Documentation](https://huggingface.co/docs/transformers/) | Full API documentation and tutorials |
| [Task summary](https://huggingface.co/docs/transformers/task_summary) | Tasks supported by 🤗 Transformers |
| [Preprocessing tutorial](https://huggingface.co/docs/transformers/preprocessing) | Using the `Tokenizer` class to prepare data for the models |
| [Training and fine-tuning](https://huggingface.co/docs/transformers/training) | Using the models provided by 🤗 Transformers in a PyTorch/TensorFlow training loop and the `Trainer` API |
| [Quick tour: Fine-tuning/usage scripts](https://github.com/huggingface/transformers/tree/main/examples) | Example scripts for fine-tuning models on a wide range of tasks |
| [Model sharing and uploading](https://huggingface.co/docs/transformers/model_sharing) | Upload and share your fine-tuned models with the community |
| [Migration](https://huggingface.co/docs/transformers/migration) | Migrate to 🤗 Transformers from `pytorch-transformers` or `pytorch-pretrained-bert` |
## Citation
We now have a [paper](https://www.aclweb.org/anthology/2020.emnlp-demos.6/) you can cite for the 🤗 Transformers library:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README_es.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<b>Español</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>Lo último de Machine Learning para JAX, PyTorch y TensorFlow</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers aporta miles de modelos preentrenados Para realizar tareas en diferentes modalidades como texto, vision, y audio.
Estos modelos pueden ser aplicados en:
* 📝 Texto, Para tareas como clasificación de texto, extracción de información, responder preguntas, resumir, traducir, generación de texto, en más de 100 idiomas.
* 🖼️ Imágenes, para tareas como clasificación de imágenes, detección the objetos, y segmentación.
* 🗣️ Audio, para tareas como reconocimiento de voz y clasificación de audio.
Los modelos de Transformer también pueden realizar tareas en **muchas modalidades combinadas**, como responder pregunstas, reconocimiento de carácteres ópticos,extracción de información de documentos escaneados, clasificación de video, y respuesta de preguntas visuales.
🤗 Transformers aporta APIs para descargar rápidamente y usar estos modelos preentrenados en un texto dado, afinarlos en tus propios sets de datos y compartirlos con la comunidad en nuestro [centro de modelos](https://huggingface.co/models). Al mismo tiempo, cada módulo de Python que define una arquitectura es completamente independiente y se puede modificar para permitir experimentos de investigación rápidos.
🤗 Transformers está respaldado por las tres bibliotecas de deep learning más populares — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) y [TensorFlow](https://www.tensorflow.org/) — con una perfecta integración entre ellos. Es sencillo entrenar sus modelos con uno antes de cargarlos para la inferencia con el otro.
## Demostraciones en línea
Puedes probar la mayoría de nuestros modelos directamente en sus páginas desde el [centro de modelos](https://huggingface.co/models). También ofrecemos [alojamiento de modelos privados, control de versiones y una API de inferencia](https://huggingface.co/pricing) para modelos públicos y privados.
Aquí hay algunos ejemplos:
En procesamiento del lenguaje natural:
- [Terminación de palabras enmascaradas con BERT](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Reconocimiento del nombre de la entidad con Electra](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [Generación de texto con GPT-2](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [Inferencia del lenguaje natural con RoBERTa](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [Resumen con BART](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [Responder a preguntas con DistilBERT](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [Traducción con T5](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
En visión de ordenador:
- [Clasificación de imágenes con ViT](https://huggingface.co/google/vit-base-patch16-224)
- [Detección de objetos con DETR](https://huggingface.co/facebook/detr-resnet-50)
- [Segmentación semántica con SegFormer](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
- [Segmentación panóptica con DETR](https://huggingface.co/facebook/detr-resnet-50-panoptic)
En Audio:
- [Reconocimiento de voz automático con Wav2Vec2](https://huggingface.co/facebook/wav2vec2-base-960h)
- [Detección de palabras clave con Wav2Vec2](https://huggingface.co/superb/wav2vec2-base-superb-ks)
En tareas multimodales:
- [Respuesta visual a preguntas con ViLT](https://huggingface.co/dandelin/vilt-b32-finetuned-vqa)
**[Escribe con Transformer](https://transformer.huggingface.co)**, construido por el equipo de Hugging Face, es la demostración oficial de las capacidades de generación de texto de este repositorio.
## Si está buscando soporte personalizado del equipo de Hugging Face
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## Tour rápido
Para usar inmediatamente un modelo en una entrada determinada (texto, imagen, audio, ...), proporcionamos la API de `pipeline`. Los pipelines agrupan un modelo previamente entrenado con el preprocesamiento que se usó durante el entrenamiento de ese modelo. Aquí se explica cómo usar rápidamente un pipeline para clasificar textos positivos frente a negativos:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
La segunda línea de código descarga y almacena en caché el modelo previamente entrenado que usa la canalización, mientras que la tercera lo evalúa en el texto dado. Aquí la respuesta es "positiva" con una confianza del 99,97%.
Muchas tareas tienen un `pipeline` preentrenado listo para funcionar, en NLP pero también en visión por ordenador y habla. Por ejemplo, podemos extraer fácilmente los objetos detectados en una imagen:
``` python
>>> import requests
>>> from PIL import Image
>>> from transformers import pipeline
# Download an image with cute cats
>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
>>> image_data = requests.get(url, stream=True).raw
>>> image = Image.open(image_data)
# Allocate a pipeline for object detection
>>> object_detector = pipeline('object_detection')
>>> object_detector(image)
[{'score': 0.9982201457023621,
'label': 'remote',
'box': {'xmin': 40, 'ymin': 70, 'xmax': 175, 'ymax': 117}},
{'score': 0.9960021376609802,
'label': 'remote',
'box': {'xmin': 333, 'ymin': 72, 'xmax': 368, 'ymax': 187}},
{'score': 0.9954745173454285,
'label': 'couch',
'box': {'xmin': 0, 'ymin': 1, 'xmax': 639, 'ymax': 473}},
{'score': 0.9988006353378296,
'label': 'cat',
'box': {'xmin': 13, 'ymin': 52, 'xmax': 314, 'ymax': 470}},
{'score': 0.9986783862113953,
'label': 'cat',
'box': {'xmin': 345, 'ymin': 23, 'xmax': 640, 'ymax': 368}}]
```
Aquí obtenemos una lista de objetos detectados en la imagen, con un cuadro que rodea el objeto y una puntuación de confianza. Aquí está la imagen original a la derecha, con las predicciones mostradas a la izquierda:
<h3 align="center">
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png" width="400"></a>
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample_post_processed.png" width="400"></a>
</h3>
Puedes obtener más información sobre las tareas admitidas por la API de `pipeline` en [este tutorial](https://huggingface.co/docs/transformers/task_summary).
Además de `pipeline`, para descargar y usar cualquiera de los modelos previamente entrenados en su tarea dada, todo lo que necesita son tres líneas de código. Aquí está la versión de PyTorch:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
Y aquí está el código equivalente para TensorFlow:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
El tokenizador es responsable de todo el preprocesamiento que espera el modelo preentrenado y se puede llamar directamente en una sola cadena (como en los ejemplos anteriores) o en una lista. Dará como resultado un diccionario que puedes usar en el código descendente o simplemente pasarlo directamente a su modelo usando el operador de desempaquetado de argumento **.
El modelo en si es un [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) normal o un [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) (dependiendo De tu backend) que puedes usar de forma habitual. [Este tutorial](https://huggingface.co/docs/transformers/training) explica cómo integrar un modelo de este tipo en un ciclo de entrenamiento PyTorch o TensorFlow clásico, o como usar nuestra API `Trainer` para ajustar rápidamente un nuevo conjunto de datos.
## ¿Por qué debo usar transformers?
1. Modelos de última generación fáciles de usar:
- Alto rendimiento en comprensión y generación de lenguaje natural, visión artificial y tareas de audio.
- Baja barrera de entrada para educadores y profesionales.
- Pocas abstracciones de cara al usuario con solo tres clases para aprender.
- Una API unificada para usar todos nuestros modelos preentrenados.
1. Menores costes de cómputo, menor huella de carbono:
- Los investigadores pueden compartir modelos entrenados en lugar de siempre volver a entrenar.
- Los profesionales pueden reducir el tiempo de cómputo y los costos de producción.
- Docenas de arquitecturas con más de 60 000 modelos preentrenados en todas las modalidades.
1. Elija el marco adecuado para cada parte de la vida útil de un modelo:
- Entrene modelos de última generación en 3 líneas de código.
- Mueva un solo modelo entre los marcos TF2.0/PyTorch/JAX a voluntad.
- Elija sin problemas el marco adecuado para la formación, la evaluación y la producción.
1. Personalice fácilmente un modelo o un ejemplo según sus necesidades:
- Proporcionamos ejemplos de cada arquitectura para reproducir los resultados publicados por sus autores originales..
- Los internos del modelo están expuestos lo más consistentemente posible..
- Los archivos modelo se pueden usar independientemente de la biblioteca para experimentos rápidos.
## ¿Por qué no debería usar transformers?
- Esta biblioteca no es una caja de herramientas modular de bloques de construcción para redes neuronales. El código en los archivos del modelo no se refactoriza con abstracciones adicionales a propósito, de modo que los investigadores puedan iterar rápidamente en cada uno de los modelos sin sumergirse en abstracciones/archivos adicionales.
- La API de entrenamiento no está diseñada para funcionar en ningún modelo, pero está optimizada para funcionar con los modelos proporcionados por la biblioteca. Para bucles genéricos de aprendizaje automático, debe usar otra biblioteca (posiblemente, [Accelerate](https://huggingface.co/docs/accelerate)).
- Si bien nos esforzamos por presentar tantos casos de uso como sea posible, los scripts en nuestra [carpeta de ejemplos](https://github.com/huggingface/transformers/tree/main/examples) son solo eso: ejemplos. Se espera que no funcionen de forma inmediata en su problema específico y que deba cambiar algunas líneas de código para adaptarlas a sus necesidades.
## Instalación
### Con pip
Este repositorio está probado en Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+ y TensorFlow 2.3+.
Deberías instalar 🤗 Transformers en un [ambiente virtual](https://docs.python.org/3/library/venv.html). Si no estas familiarizado con los entornos virtuales de Python, consulta la [guía de usuario](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
Primero, crea un entorno virtual con la versión de Python que vas a usar y actívalo.
Luego, deberás instalar al menos uno de Flax, PyTorch o TensorFlow.
Por favor, ve a la [página de instalación de TensorFlow](https://www.tensorflow.org/install/), [página de instalación de PyTorch](https://pytorch.org/get-started/locally/#start-locally) y/o las páginas de instalación de [Flax](https://github.com/google/flax#quick-install) y [Jax](https://github.com/google/jax#installation) con respecto al comando de instalación específico para tu plataforma.
Cuando se ha instalado uno de esos backends, los 🤗 Transformers se pueden instalar usando pip de la siguiente manera:
```bash
pip install transformers
```
Si deseas jugar con los ejemplos o necesitas la última versión del código y no puedes esperar a una nueva versión, tienes que [instalar la librería de la fuente](https://huggingface.co/docs/transformers/installation#installing-from-source).
### Con conda
Desde la versión v4.0.0 de Transformers, ahora tenemos un canal conda: `huggingface`.
🤗 Transformers se puede instalar usando conda de la siguiente manera:
```shell script
conda install -c huggingface transformers
```
Sigue las páginas de instalación de Flax, PyTorch o TensorFlow para ver cómo instalarlos con conda.
> **_NOTA:_** En Windows, es posible que se le pida que active el modo de desarrollador para beneficiarse del almacenamiento en caché. Si esta no es una opción para usted, háganoslo saber en [esta issue](https://github.com/huggingface/huggingface_hub/issues/1062).
## Arquitecturas modelo
**[Todos los puntos de control del modelo](https://huggingface.co/models)** aportados por 🤗 Transformers están perfectamente integrados desde huggingface.co [Centro de modelos](https://huggingface.co) donde son subidos directamente por los [usuarios](https://huggingface.co/users) y [organizaciones](https://huggingface.co/organizations).
Número actual de puntos de control: 
🤗 Transformers actualmente proporciona las siguientes arquitecturas (ver [aquí](https://huggingface.co/docs/transformers/model_summary) para un resumen de alto nivel de cada uno de ellas.):
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. ¿Quieres aportar un nuevo modelo? Hemos agregado una **guía detallada y plantillas** para guiarte en el proceso de agregar un nuevo modelo. Puedes encontrarlos en la carpeta de [`templates`](./templates) del repositorio. Asegúrate de revisar las [pautas de contribución](./CONTRIBUTING.md) y comunícate con los mantenedores o abra un problema para recopilar comentarios antes de comenzar su PR.
Para comprobar si cada modelo tiene una implementación en Flax, PyTorch o TensorFlow, o tiene un tokenizador asociado respaldado por la librería 🤗 Tokenizers , ve a [esta tabla](https://huggingface.co/docs/transformers/index#supported-frameworks).
Estas implementaciones se han probado en varios conjuntos de datos (consulte los scripts de ejemplo) y deberían coincidir con el rendimiento de las implementaciones originales. Puede encontrar más detalles sobre el rendimiento en la sección Examples de la [documentación](https://github.com/huggingface/transformers/tree/main/examples).
## Aprender más
| Sección | Descripción |
|-|-|
| [Documentación](https://huggingface.co/docs/transformers/) | Toda la documentación de la API y tutoriales |
| [Resumen de tareas](https://huggingface.co/docs/transformers/task_summary) | Tareas soportadas 🤗 Transformers |
| [Tutorial de preprocesAmiento](https://huggingface.co/docs/transformers/preprocessing) | Usando la clase `Tokenizer` para preparar datos para los modelos |
| [Entrenamiento y puesta a punto](https://huggingface.co/docs/transformers/training) | Usando los modelos aportados por 🤗 Transformers en un bucle de entreno de PyTorch/TensorFlow y la API de `Trainer` |
| [Recorrido rápido: secuencias de comandos de ajuste/uso](https://github.com/huggingface/transformers/tree/main/examples) | Scripts de ejemplo para ajustar modelos en una amplia gama de tareas |
| [Compartir y subir modelos](https://huggingface.co/docs/transformers/model_sharing) | Carga y comparte tus modelos perfeccionados con la comunidad |
| [Migración](https://huggingface.co/docs/transformers/migration) | Migra a 🤗 Transformers desde `pytorch-transformers` o `pytorch-pretrained-bert` |
## Citación
Ahora nosotros tenemos un [papel](https://www.aclweb.org/anthology/2020.emnlp-demos.6/) que puedes citar para la librería de 🤗 Transformers:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<b>Español</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>Lo último de Machine Learning para JAX, PyTorch y TensorFlow</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers aporta miles de modelos preentrenados Para realizar tareas en diferentes modalidades como texto, vision, y audio.
Estos modelos pueden ser aplicados en:
* 📝 Texto, Para tareas como clasificación de texto, extracción de información, responder preguntas, resumir, traducir, generación de texto, en más de 100 idiomas.
* 🖼️ Imágenes, para tareas como clasificación de imágenes, detección the objetos, y segmentación.
* 🗣️ Audio, para tareas como reconocimiento de voz y clasificación de audio.
Los modelos de Transformer también pueden realizar tareas en **muchas modalidades combinadas**, como responder pregunstas, reconocimiento de carácteres ópticos,extracción de información de documentos escaneados, clasificación de video, y respuesta de preguntas visuales.
🤗 Transformers aporta APIs para descargar rápidamente y usar estos modelos preentrenados en un texto dado, afinarlos en tus propios sets de datos y compartirlos con la comunidad en nuestro [centro de modelos](https://huggingface.co/models). Al mismo tiempo, cada módulo de Python que define una arquitectura es completamente independiente y se puede modificar para permitir experimentos de investigación rápidos.
🤗 Transformers está respaldado por las tres bibliotecas de deep learning más populares — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) y [TensorFlow](https://www.tensorflow.org/) — con una perfecta integración entre ellos. Es sencillo entrenar sus modelos con uno antes de cargarlos para la inferencia con el otro.
## Demostraciones en línea
Puedes probar la mayoría de nuestros modelos directamente en sus páginas desde el [centro de modelos](https://huggingface.co/models). También ofrecemos [alojamiento de modelos privados, control de versiones y una API de inferencia](https://huggingface.co/pricing) para modelos públicos y privados.
Aquí hay algunos ejemplos:
En procesamiento del lenguaje natural:
- [Terminación de palabras enmascaradas con BERT](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Reconocimiento del nombre de la entidad con Electra](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [Generación de texto con GPT-2](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [Inferencia del lenguaje natural con RoBERTa](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [Resumen con BART](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [Responder a preguntas con DistilBERT](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [Traducción con T5](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
En visión de ordenador:
- [Clasificación de imágenes con ViT](https://huggingface.co/google/vit-base-patch16-224)
- [Detección de objetos con DETR](https://huggingface.co/facebook/detr-resnet-50)
- [Segmentación semántica con SegFormer](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
- [Segmentación panóptica con DETR](https://huggingface.co/facebook/detr-resnet-50-panoptic)
En Audio:
- [Reconocimiento de voz automático con Wav2Vec2](https://huggingface.co/facebook/wav2vec2-base-960h)
- [Detección de palabras clave con Wav2Vec2](https://huggingface.co/superb/wav2vec2-base-superb-ks)
En tareas multimodales:
- [Respuesta visual a preguntas con ViLT](https://huggingface.co/dandelin/vilt-b32-finetuned-vqa)
**[Escribe con Transformer](https://transformer.huggingface.co)**, construido por el equipo de Hugging Face, es la demostración oficial de las capacidades de generación de texto de este repositorio.
## Si está buscando soporte personalizado del equipo de Hugging Face
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## Tour rápido
Para usar inmediatamente un modelo en una entrada determinada (texto, imagen, audio, ...), proporcionamos la API de `pipeline`. Los pipelines agrupan un modelo previamente entrenado con el preprocesamiento que se usó durante el entrenamiento de ese modelo. Aquí se explica cómo usar rápidamente un pipeline para clasificar textos positivos frente a negativos:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
La segunda línea de código descarga y almacena en caché el modelo previamente entrenado que usa la canalización, mientras que la tercera lo evalúa en el texto dado. Aquí la respuesta es "positiva" con una confianza del 99,97%.
Muchas tareas tienen un `pipeline` preentrenado listo para funcionar, en NLP pero también en visión por ordenador y habla. Por ejemplo, podemos extraer fácilmente los objetos detectados en una imagen:
``` python
>>> import requests
>>> from PIL import Image
>>> from transformers import pipeline
# Download an image with cute cats
>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
>>> image_data = requests.get(url, stream=True).raw
>>> image = Image.open(image_data)
# Allocate a pipeline for object detection
>>> object_detector = pipeline('object_detection')
>>> object_detector(image)
[{'score': 0.9982201457023621,
'label': 'remote',
'box': {'xmin': 40, 'ymin': 70, 'xmax': 175, 'ymax': 117}},
{'score': 0.9960021376609802,
'label': 'remote',
'box': {'xmin': 333, 'ymin': 72, 'xmax': 368, 'ymax': 187}},
{'score': 0.9954745173454285,
'label': 'couch',
'box': {'xmin': 0, 'ymin': 1, 'xmax': 639, 'ymax': 473}},
{'score': 0.9988006353378296,
'label': 'cat',
'box': {'xmin': 13, 'ymin': 52, 'xmax': 314, 'ymax': 470}},
{'score': 0.9986783862113953,
'label': 'cat',
'box': {'xmin': 345, 'ymin': 23, 'xmax': 640, 'ymax': 368}}]
```
Aquí obtenemos una lista de objetos detectados en la imagen, con un cuadro que rodea el objeto y una puntuación de confianza. Aquí está la imagen original a la derecha, con las predicciones mostradas a la izquierda:
<h3 align="center">
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png" width="400"></a>
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample_post_processed.png" width="400"></a>
</h3>
Puedes obtener más información sobre las tareas admitidas por la API de `pipeline` en [este tutorial](https://huggingface.co/docs/transformers/task_summary).
Además de `pipeline`, para descargar y usar cualquiera de los modelos previamente entrenados en su tarea dada, todo lo que necesita son tres líneas de código. Aquí está la versión de PyTorch:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
Y aquí está el código equivalente para TensorFlow:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
El tokenizador es responsable de todo el preprocesamiento que espera el modelo preentrenado y se puede llamar directamente en una sola cadena (como en los ejemplos anteriores) o en una lista. Dará como resultado un diccionario que puedes usar en el código descendente o simplemente pasarlo directamente a su modelo usando el operador de desempaquetado de argumento **.
El modelo en si es un [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) normal o un [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) (dependiendo De tu backend) que puedes usar de forma habitual. [Este tutorial](https://huggingface.co/docs/transformers/training) explica cómo integrar un modelo de este tipo en un ciclo de entrenamiento PyTorch o TensorFlow clásico, o como usar nuestra API `Trainer` para ajustar rápidamente un nuevo conjunto de datos.
## ¿Por qué debo usar transformers?
1. Modelos de última generación fáciles de usar:
- Alto rendimiento en comprensión y generación de lenguaje natural, visión artificial y tareas de audio.
- Baja barrera de entrada para educadores y profesionales.
- Pocas abstracciones de cara al usuario con solo tres clases para aprender.
- Una API unificada para usar todos nuestros modelos preentrenados.
1. Menores costes de cómputo, menor huella de carbono:
- Los investigadores pueden compartir modelos entrenados en lugar de siempre volver a entrenar.
- Los profesionales pueden reducir el tiempo de cómputo y los costos de producción.
- Docenas de arquitecturas con más de 60 000 modelos preentrenados en todas las modalidades.
1. Elija el marco adecuado para cada parte de la vida útil de un modelo:
- Entrene modelos de última generación en 3 líneas de código.
- Mueva un solo modelo entre los marcos TF2.0/PyTorch/JAX a voluntad.
- Elija sin problemas el marco adecuado para la formación, la evaluación y la producción.
1. Personalice fácilmente un modelo o un ejemplo según sus necesidades:
- Proporcionamos ejemplos de cada arquitectura para reproducir los resultados publicados por sus autores originales..
- Los internos del modelo están expuestos lo más consistentemente posible..
- Los archivos modelo se pueden usar independientemente de la biblioteca para experimentos rápidos.
## ¿Por qué no debería usar transformers?
- Esta biblioteca no es una caja de herramientas modular de bloques de construcción para redes neuronales. El código en los archivos del modelo no se refactoriza con abstracciones adicionales a propósito, de modo que los investigadores puedan iterar rápidamente en cada uno de los modelos sin sumergirse en abstracciones/archivos adicionales.
- La API de entrenamiento no está diseñada para funcionar en ningún modelo, pero está optimizada para funcionar con los modelos proporcionados por la biblioteca. Para bucles genéricos de aprendizaje automático, debe usar otra biblioteca (posiblemente, [Accelerate](https://huggingface.co/docs/accelerate)).
- Si bien nos esforzamos por presentar tantos casos de uso como sea posible, los scripts en nuestra [carpeta de ejemplos](https://github.com/huggingface/transformers/tree/main/examples) son solo eso: ejemplos. Se espera que no funcionen de forma inmediata en su problema específico y que deba cambiar algunas líneas de código para adaptarlas a sus necesidades.
## Instalación
### Con pip
Este repositorio está probado en Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+ y TensorFlow 2.3+.
Deberías instalar 🤗 Transformers en un [ambiente virtual](https://docs.python.org/3/library/venv.html). Si no estas familiarizado con los entornos virtuales de Python, consulta la [guía de usuario](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
Primero, crea un entorno virtual con la versión de Python que vas a usar y actívalo.
Luego, deberás instalar al menos uno de Flax, PyTorch o TensorFlow.
Por favor, ve a la [página de instalación de TensorFlow](https://www.tensorflow.org/install/), [página de instalación de PyTorch](https://pytorch.org/get-started/locally/#start-locally) y/o las páginas de instalación de [Flax](https://github.com/google/flax#quick-install) y [Jax](https://github.com/google/jax#installation) con respecto al comando de instalación específico para tu plataforma.
Cuando se ha instalado uno de esos backends, los 🤗 Transformers se pueden instalar usando pip de la siguiente manera:
```bash
pip install transformers
```
Si deseas jugar con los ejemplos o necesitas la última versión del código y no puedes esperar a una nueva versión, tienes que [instalar la librería de la fuente](https://huggingface.co/docs/transformers/installation#installing-from-source).
### Con conda
Desde la versión v4.0.0 de Transformers, ahora tenemos un canal conda: `huggingface`.
🤗 Transformers se puede instalar usando conda de la siguiente manera:
```shell script
conda install -c huggingface transformers
```
Sigue las páginas de instalación de Flax, PyTorch o TensorFlow para ver cómo instalarlos con conda.
> **_NOTA:_** En Windows, es posible que se le pida que active el modo de desarrollador para beneficiarse del almacenamiento en caché. Si esta no es una opción para usted, háganoslo saber en [esta issue](https://github.com/huggingface/huggingface_hub/issues/1062).
## Arquitecturas modelo
**[Todos los puntos de control del modelo](https://huggingface.co/models)** aportados por 🤗 Transformers están perfectamente integrados desde huggingface.co [Centro de modelos](https://huggingface.co) donde son subidos directamente por los [usuarios](https://huggingface.co/users) y [organizaciones](https://huggingface.co/organizations).
Número actual de puntos de control: 
🤗 Transformers actualmente proporciona las siguientes arquitecturas (ver [aquí](https://huggingface.co/docs/transformers/model_summary) para un resumen de alto nivel de cada uno de ellas.):
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. ¿Quieres aportar un nuevo modelo? Hemos agregado una **guía detallada y plantillas** para guiarte en el proceso de agregar un nuevo modelo. Puedes encontrarlos en la carpeta de [`templates`](./templates) del repositorio. Asegúrate de revisar las [pautas de contribución](./CONTRIBUTING.md) y comunícate con los mantenedores o abra un problema para recopilar comentarios antes de comenzar su PR.
Para comprobar si cada modelo tiene una implementación en Flax, PyTorch o TensorFlow, o tiene un tokenizador asociado respaldado por la librería 🤗 Tokenizers , ve a [esta tabla](https://huggingface.co/docs/transformers/index#supported-frameworks).
Estas implementaciones se han probado en varios conjuntos de datos (consulte los scripts de ejemplo) y deberían coincidir con el rendimiento de las implementaciones originales. Puede encontrar más detalles sobre el rendimiento en la sección Examples de la [documentación](https://github.com/huggingface/transformers/tree/main/examples).
## Aprender más
| Sección | Descripción |
|-|-|
| [Documentación](https://huggingface.co/docs/transformers/) | Toda la documentación de la API y tutoriales |
| [Resumen de tareas](https://huggingface.co/docs/transformers/task_summary) | Tareas soportadas 🤗 Transformers |
| [Tutorial de preprocesAmiento](https://huggingface.co/docs/transformers/preprocessing) | Usando la clase `Tokenizer` para preparar datos para los modelos |
| [Entrenamiento y puesta a punto](https://huggingface.co/docs/transformers/training) | Usando los modelos aportados por 🤗 Transformers en un bucle de entreno de PyTorch/TensorFlow y la API de `Trainer` |
| [Recorrido rápido: secuencias de comandos de ajuste/uso](https://github.com/huggingface/transformers/tree/main/examples) | Scripts de ejemplo para ajustar modelos en una amplia gama de tareas |
| [Compartir y subir modelos](https://huggingface.co/docs/transformers/model_sharing) | Carga y comparte tus modelos perfeccionados con la comunidad |
| [Migración](https://huggingface.co/docs/transformers/migration) | Migra a 🤗 Transformers desde `pytorch-transformers` o `pytorch-pretrained-bert` |
## Citación
Ahora nosotros tenemos un [papel](https://www.aclweb.org/anthology/2020.emnlp-demos.6/) que puedes citar para la librería de 🤗 Transformers:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README_hd.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Hindi translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Hindi characters. E.g., कुल मिलाकर 100 से अधिक भाषाएँ; ट्रांसफॉर्मर लाइब्रेरी का उपयोग करता है।
- वर्गाकार उद्धरणों का प्रयोग करें, जैसे, "उद्धरण"
Dictionary
Hugging Face: गले लगाओ चेहरा
token: शब्द (और मूल अंग्रेजी को कोष्ठक में चिह्नित करें）
tokenize: टोकननाइज़ करें (और मूल अंग्रेज़ी को चिह्नित करने के लिए कोष्ठक का उपयोग करें)
tokenizer: Tokenizer (मूल अंग्रेजी में कोष्ठक के साथ)
transformer: transformer
pipeline: समनुक्रम
API: API (अनुवाद के बिना)
inference: विचार
Trainer: प्रशिक्षक। कक्षा के नाम के रूप में प्रस्तुत किए जाने पर अनुवादित नहीं किया गया।
pretrained/pretrain: पूर्व प्रशिक्षण
finetune: फ़ाइन ट्यूनिंग
community: समुदाय
example: जब विशिष्ट गोदाम example कैटलॉग करते समय "केस केस" के रूप में अनुवादित
Python data structures (e.g., list, set, dict): मूल अंग्रेजी को चिह्नित करने के लिए सूचियों, सेटों, शब्दकोशों में अनुवाद करें और कोष्ठक का उपयोग करें
NLP/Natural Language Processing: द्वारा NLP अनुवाद के बिना प्रकट होते हैं Natural Language Processing प्रस्तुत किए जाने पर प्राकृतिक भाषा संसाधन में अनुवाद करें
checkpoint: जाँच बिंदु
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<b>हिन्दी</b> |
<p>
</h4>
<h3 align="center">
<p>Jax, PyTorch और TensorFlow के लिए उन्नत मशीन लर्निंग</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 100 से अधिक भाषाओं में पाठ वर्गीकरण, सूचना निष्कर्षण, प्रश्न उत्तर, सारांशीकरण, अनुवाद, पाठ निर्माण का समर्थन करने के लिए हजारों पूर्व-प्रशिक्षित मॉडल प्रदान करता है। इसका उद्देश्य सबसे उन्नत एनएलपी तकनीक को सभी के लिए सुलभ बनाना है।
🤗 Transformers त्वरित डाउनलोड और उपयोग के लिए एक एपीआई प्रदान करता है, जिससे आप किसी दिए गए पाठ पर एक पूर्व-प्रशिक्षित मॉडल ले सकते हैं, इसे अपने डेटासेट पर ठीक कर सकते हैं और इसे [मॉडल हब] (https://huggingface.co/models) के माध्यम से समुदाय के साथ साझा कर सकते हैं। ) . इसी समय, प्रत्येक परिभाषित पायथन मॉड्यूल पूरी तरह से स्वतंत्र है, जो संशोधन और तेजी से अनुसंधान प्रयोगों के लिए सुविधाजनक है।
🤗 Transformers तीन सबसे लोकप्रिय गहन शिक्षण पुस्तकालयों का समर्थन करता है： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — और इसके साथ निर्बाध रूप से एकीकृत होता है। आप अपने मॉडल को सीधे एक ढांचे के साथ प्रशिक्षित कर सकते हैं और दूसरे के साथ लोड और अनुमान लगा सकते हैं।
## ऑनलाइन डेमो
आप सबसे सीधे मॉडल पृष्ठ पर परीक्षण कर सकते हैं [model hub](https://huggingface.co/models) मॉडल पर। हम [निजी मॉडल होस्टिंग, मॉडल संस्करण, और अनुमान एपीआई] भी प्रदान करते हैं।(https://huggingface.co/pricing)。
यहाँ कुछ उदाहरण हैं：
- [शब्द को भरने के लिए मास्क के रूप में BERT का प्रयोग करें](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [इलेक्ट्रा के साथ नामित इकाई पहचान](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [जीपीटी-2 के साथ टेक्स्ट जनरेशन](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [रॉबर्टा के साथ प्राकृतिक भाषा निष्कर्ष](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [बार्ट के साथ पाठ सारांश](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [डिस्टिलबर्ट के साथ प्रश्नोत्तर](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [अनुवाद के लिए T5 का प्रयोग करें](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，हगिंग फेस टीम द्वारा बनाया गया, यह एक आधिकारिक पाठ पीढ़ी है demo。
## यदि आप हगिंग फेस टीम से बीस्पोक समर्थन की तलाश कर रहे हैं
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## जल्दी शुरू करें
हम त्वरित उपयोग के लिए मॉडल प्रदान करते हैं `pipeline` (पाइपलाइन) एपीआई। पाइपलाइन पूर्व-प्रशिक्षित मॉडल और संबंधित पाठ प्रीप्रोसेसिंग को एकत्रित करती है। सकारात्मक और नकारात्मक भावना को निर्धारित करने के लिए पाइपलाइनों का उपयोग करने का एक त्वरित उदाहरण यहां दिया गया है:
```python
>>> from transformers import pipeline
# भावना विश्लेषण पाइपलाइन का उपयोग करना
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
कोड की दूसरी पंक्ति पाइपलाइन द्वारा उपयोग किए गए पूर्व-प्रशिक्षित मॉडल को डाउनलोड और कैश करती है, जबकि कोड की तीसरी पंक्ति दिए गए पाठ पर मूल्यांकन करती है। यहां उत्तर 99 आत्मविश्वास के स्तर के साथ "सकारात्मक" है।
कई एनएलपी कार्यों में आउट ऑफ़ द बॉक्स पाइपलाइनों का पूर्व-प्रशिक्षण होता है। उदाहरण के लिए, हम किसी दिए गए पाठ से किसी प्रश्न का उत्तर आसानी से निकाल सकते हैं:
``` python
>>> from transformers import pipeline
# प्रश्नोत्तर पाइपलाइन का उपयोग करना
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
उत्तर देने के अलावा, पूर्व-प्रशिक्षित मॉडल संगत आत्मविश्वास स्कोर भी देता है, जहां उत्तर टोकनयुक्त पाठ में शुरू और समाप्त होता है। आप [इस ट्यूटोरियल](https://huggingface.co/docs/transformers/task_summary) से पाइपलाइन एपीआई द्वारा समर्थित कार्यों के बारे में अधिक जान सकते हैं।
अपने कार्य पर किसी भी पूर्व-प्रशिक्षित मॉडल को डाउनलोड करना और उसका उपयोग करना भी कोड की तीन पंक्तियों की तरह सरल है। यहाँ PyTorch संस्करण के लिए एक उदाहरण दिया गया है:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
यहाँ समकक्ष है TensorFlow कोड:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
टोकननाइज़र सभी पूर्व-प्रशिक्षित मॉडलों के लिए प्रीप्रोसेसिंग प्रदान करता है और इसे सीधे एक स्ट्रिंग (जैसे ऊपर दिए गए उदाहरण) या किसी सूची पर बुलाया जा सकता है। यह एक डिक्शनरी (तानाशाही) को आउटपुट करता है जिसे आप डाउनस्ट्रीम कोड में उपयोग कर सकते हैं या `**` अनपैकिंग एक्सप्रेशन के माध्यम से सीधे मॉडल को पास कर सकते हैं।
मॉडल स्वयं एक नियमित [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) या [TensorFlow `tf.keras.Model`](https ://pytorch.org/docs/stable/nn.html#torch.nn.Module) ://www.tensorflow.org/api_docs/python/tf/keras/Model) (आपके बैकएंड के आधार पर), जो हो सकता है सामान्य तरीके से उपयोग किया जाता है। [यह ट्यूटोरियल](https://huggingface.co/transformers/training.html) बताता है कि इस तरह के मॉडल को क्लासिक PyTorch या TensorFlow प्रशिक्षण लूप में कैसे एकीकृत किया जाए, या हमारे `ट्रेनर` एपीआई का उपयोग कैसे करें ताकि इसे जल्दी से फ़ाइन ट्यून किया जा सके।एक नया डेटासेट पे।
## ट्रांसफार्मर का उपयोग क्यों करें?
1. उपयोग में आसानी के लिए उन्नत मॉडल:
- एनएलयू और एनएलजी पर बेहतर प्रदर्शन
- प्रवेश के लिए कम बाधाओं के साथ शिक्षण और अभ्यास के अनुकूल
- उपयोगकर्ता-सामना करने वाले सार तत्व, केवल तीन वर्गों को जानने की जरूरत है
- सभी मॉडलों के लिए एकीकृत एपीआई
1. कम कम्प्यूटेशनल ओवरहेड और कम कार्बन उत्सर्जन:
- शोधकर्ता हर बार नए सिरे से प्रशिक्षण देने के बजाय प्रशिक्षित मॉडल साझा कर सकते हैं
- इंजीनियर गणना समय और उत्पादन ओवरहेड को कम कर सकते हैं
- दर्जनों मॉडल आर्किटेक्चर, 2,000 से अधिक पूर्व-प्रशिक्षित मॉडल, 100 से अधिक भाषाओं का समर्थन
1.मॉडल जीवनचक्र के हर हिस्से को शामिल करता है:
- कोड की केवल 3 पंक्तियों में उन्नत मॉडलों को प्रशिक्षित करें
- मॉडल को मनमाने ढंग से विभिन्न डीप लर्निंग फ्रेमवर्क के बीच स्थानांतरित किया जा सकता है, जैसा आप चाहते हैं
- निर्बाध रूप से प्रशिक्षण, मूल्यांकन और उत्पादन के लिए सबसे उपयुक्त ढांचा चुनें
1. आसानी से अनन्य मॉडल को अनुकूलित करें और अपनी आवश्यकताओं के लिए मामलों का उपयोग करें:
- हम मूल पेपर परिणामों को पुन: पेश करने के लिए प्रत्येक मॉडल आर्किटेक्चर के लिए कई उपयोग के मामले प्रदान करते हैं
- मॉडल की आंतरिक संरचना पारदर्शी और सुसंगत रहती है
- मॉडल फ़ाइल को अलग से इस्तेमाल किया जा सकता है, जो संशोधन और त्वरित प्रयोग के लिए सुविधाजनक है
## मुझे ट्रांसफॉर्मर का उपयोग कब नहीं करना चाहिए?
- यह लाइब्रेरी मॉड्यूलर न्यूरल नेटवर्क टूलबॉक्स नहीं है। मॉडल फ़ाइल में कोड जानबूझकर अल्पविकसित है, बिना अतिरिक्त सार इनकैप्सुलेशन के, ताकि शोधकर्ता अमूर्तता और फ़ाइल जंपिंग में शामिल हुए जल्दी से पुनरावृति कर सकें।
- `ट्रेनर` एपीआई किसी भी मॉडल के साथ संगत नहीं है, यह केवल इस पुस्तकालय के मॉडल के लिए अनुकूलित है। यदि आप सामान्य मशीन लर्निंग के लिए उपयुक्त प्रशिक्षण लूप कार्यान्वयन की तलाश में हैं, तो कहीं और देखें।
- हमारे सर्वोत्तम प्रयासों के बावजूद, [उदाहरण निर्देशिका] (https://github.com/huggingface/transformers/tree/main/examples) में स्क्रिप्ट केवल उपयोग के मामले हैं। आपकी विशिष्ट समस्या के लिए, वे जरूरी नहीं कि बॉक्स से बाहर काम करें, और आपको कोड की कुछ पंक्तियों को सूट करने की आवश्यकता हो सकती है।
## स्थापित करना
### पिप का उपयोग करना
इस रिपॉजिटरी का परीक्षण Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+ और TensorFlow 2.3+ के तहत किया गया है।
आप [वर्चुअल एनवायरनमेंट] (https://docs.python.org/3/library/venv.html) में 🤗 ट्रांसफॉर्मर इंस्टॉल कर सकते हैं। यदि आप अभी तक पायथन के वर्चुअल एनवायरनमेंट से परिचित नहीं हैं, तो कृपया इसे [उपयोगकर्ता निर्देश] (https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/) पढ़ें।
सबसे पहले, पायथन के उस संस्करण के साथ एक आभासी वातावरण बनाएं जिसका आप उपयोग करने और उसे सक्रिय करने की योजना बना रहे हैं।
फिर, आपको Flax, PyTorch या TensorFlow में से किसी एक को स्थापित करने की आवश्यकता है। अपने प्लेटफ़ॉर्म पर इन फ़्रेमवर्क को स्थापित करने के लिए, [TensorFlow स्थापना पृष्ठ](https://www.tensorflow.org/install/), [PyTorch स्थापना पृष्ठ](https://pytorch.org/get-started /locally/# देखें) start-locally) या [Flax स्थापना पृष्ठ](https://github.com/google/flax#quick-install).
जब इनमें से कोई एक बैकएंड सफलतापूर्वक स्थापित हो जाता है, तो ट्रांसफॉर्मर निम्नानुसार स्थापित किए जा सकते हैं:
```bash
pip install transformers
```
यदि आप उपयोग के मामलों को आज़माना चाहते हैं या आधिकारिक रिलीज़ से पहले नवीनतम इन-डेवलपमेंट कोड का उपयोग करना चाहते हैं, तो आपको [सोर्स से इंस्टॉल करना होगा](https://huggingface.co/docs/transformers/installation#installing-from- स्रोत)।
### कोंडा का उपयोग करना
ट्रांसफॉर्मर संस्करण 4.0.0 के बाद से, हमारे पास एक कोंडा चैनल है: `हगिंगफेस`।
ट्रांसफॉर्मर कोंडा के माध्यम से निम्नानुसार स्थापित किया जा सकता है:
```shell script
conda install -c huggingface transformers
```
कोंडा के माध्यम से Flax, PyTorch, या TensorFlow में से किसी एक को स्थापित करने के लिए, निर्देशों के लिए उनके संबंधित स्थापना पृष्ठ देखें।
## मॉडल आर्किटेक्चर
[उपयोगकर्ता](https://huggingface.co/users) और [organization](https://huggingface.co) द्वारा ट्रांसफॉर्मर समर्थित [**सभी मॉडल चौकियों**](https://huggingface.co/models) /users) हगिंगफेस.को/ऑर्गनाइजेशन), सभी को बिना किसी बाधा के हगिंगफेस.को [मॉडल हब](https://huggingface.co) के साथ एकीकृत किया गया है।
चौकियों की वर्तमान संख्या: 
🤗 ट्रांसफॉर्मर वर्तमान में निम्नलिखित आर्किटेक्चर का समर्थन करते हैं (मॉडल के अवलोकन के लिए [यहां] देखें (https://huggingface.co/docs/transformers/model_summary))：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (Google Research and the Toyota Technological Institute at Chicago) साथ थीसिस [ALBERT: A Lite BERT for Self-supervised भाषा प्रतिनिधित्व सीखना](https://arxiv.org/abs/1909.11942), झेंझोंग लैन, मिंगदा चेन, सेबेस्टियन गुडमैन, केविन गिम्पेल, पीयूष शर्मा, राडू सोरिकट
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (फेसबुक) साथ थीसिस [बार्ट: प्राकृतिक भाषा निर्माण, अनुवाद के लिए अनुक्रम-से-अनुक्रम पूर्व प्रशिक्षण , और समझ] (https://arxiv.org/pdf/1910.13461.pdf) पर निर्भर माइक लुईस, यिनहान लियू, नमन गोयल, मार्जन ग़ज़विनिनेजाद, अब्देलरहमान मोहम्मद, ओमर लेवी, वेस स्टोयानोव और ल्यूक ज़ेटलमॉयर
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (से École polytechnique) साथ थीसिस [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) पर निर्भर Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis रिहाई।
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (VinAI Research से) साथ में पेपर [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701)गुयेन लुओंग ट्रान, डुओंग मिन्ह ले और डाट क्वोक गुयेन द्वारा पोस्ट किया गया।
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (Microsoft से) साथ में कागज [BEiT: BERT इमेज ट्रांसफॉर्मर्स का प्री-ट्रेनिंग](https://arxiv.org/abs/2106.08254) Hangbo Bao, Li Dong, Furu Wei द्वारा।
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (गूगल से) साथ वाला पेपर [बीईआरटी: प्री-ट्रेनिंग ऑफ डीप बिडायरेक्शनल ट्रांसफॉर्मर्स फॉर लैंग्वेज अंडरस्टैंडिंग](https://arxiv.org/abs/1810.04805) जैकब डेवलिन, मिंग-वेई चांग, केंटन ली और क्रिस्टीना टौटानोवा द्वारा प्रकाशित किया गया था। .
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (गूगल से) साथ देने वाला पेपर [सीक्वेंस जेनरेशन टास्क के लिए प्री-ट्रेंड चेकपॉइंट का इस्तेमाल करना](https ://arxiv.org/abs/1907.12461) साशा रोठे, शशि नारायण, अलियाक्सि सेवेरिन द्वारा।
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (VinAI Research से) साथ में पेपर [BERTweet: अंग्रेजी ट्वीट्स के लिए एक पूर्व-प्रशिक्षित भाषा मॉडल] (https://aclanthology.org/2020.emnlp-demos.2/) डाट क्वोक गुयेन, थान वु और अन्ह तुआन गुयेन द्वारा प्रकाशित।
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (गूगल रिसर्च से) साथ वाला पेपर [बिग बर्ड: ट्रांसफॉर्मर्स फॉर लॉन्गर सीक्वेंस](https://arxiv .org/abs/2007.14062) मंज़िल ज़हीर, गुरु गुरुगणेश, अविनावा दुबे, जोशुआ आइंस्ली, क्रिस अल्बर्टी, सैंटियागो ओंटानोन, फिलिप फाम, अनिरुद्ध रावुला, किफ़ान वांग, ली यांग, अमर अहमद द्वारा।
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (गूगल रिसर्च से) साथ में पेपर [बिग बर्ड: ट्रांसफॉर्मर्स फॉर लॉन्गर सीक्वेंस](https://arxiv.org/abs/2007.14062) मंज़िल ज़हीर, गुरु गुरुगणेश, अविनावा दुबे, जोशुआ आइंस्ली, क्रिस अल्बर्टी, सैंटियागो ओंटानन, फिलिप फाम द्वारा , अनिरुद्ध रावुला, किफ़ान वांग, ली यांग, अमर अहमद द्वारा पोस्ट किया गया।
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (फेसबुक से) साथ में कागज [एक ओपन-डोमेन चैटबॉट बनाने की विधि](https://arxiv.org /abs/2004.13637) स्टीफन रोलर, एमिली दीनन, नमन गोयल, दा जू, मैरी विलियमसन, यिनहान लियू, जिंग जू, मायल ओट, कर्ट शस्टर, एरिक एम। स्मिथ, वाई-लैन बॉरो, जेसन वेस्टन द्वारा।
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (फेसबुक से) साथ में पेपर [एक ओपन-डोमेन चैटबॉट बनाने की रेसिपी](https://arxiv .org/abs/2004.13637) स्टीफन रोलर, एमिली दीनन, नमन गोयल, दा जू, मैरी विलियमसन, यिनहान लियू, जिंग जू, मायल ओट, कर्ट शस्टर, एरिक एम स्मिथ, वाई-लैन बॉरो, जेसन वेस्टन द्वारा।
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigSicence Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (एलेक्सा से) कागज के साथ [बीईआरटी के लिए ऑप्टिमल सबआर्किटेक्चर एक्सट्रैक्शन](https://arxiv.org/abs/ 2010.10499) एड्रियन डी विंटर और डैनियल जे पेरी द्वारा।
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (Google अनुसंधान से) साथ में कागज [ByT5: पूर्व-प्रशिक्षित बाइट-टू-बाइट मॉडल के साथ एक टोकन-मुक्त भविष्य की ओर] (https://arxiv.org/abs/2105.13626) Linting Xue, Aditya Barua, Noah Constant, रामी अल-रफू, शरण नारंग, मिहिर काले, एडम रॉबर्ट्स, कॉलिन रैफेल द्वारा पोस्ट किया गया।
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (इनरिया/फेसबुक/सोरबोन से) साथ में कागज [CamemBERT: एक टेस्टी फ्रेंच लैंग्वेज मॉडल](https:// arxiv.org/abs/1911.03894) लुई मार्टिन*, बेंजामिन मुलर*, पेड्रो जेवियर ऑर्टिज़ सुआरेज़*, योआन ड्यूपॉन्ट, लॉरेंट रोमरी, एरिक विलेमोन्टे डे ला क्लर्जरी, जैमे सेडाह और बेनोइट सगोट द्वारा।
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (Google रिसर्च से) साथ में दिया गया पेपर [कैनाइन: प्री-ट्रेनिंग ए एफिशिएंट टोकनाइजेशन-फ्री एनकोडर फॉर लैंग्वेज रिप्रेजेंटेशन]( https://arxiv.org/abs/2103.06874) जोनाथन एच क्लार्क, डैन गैरेट, यूलिया टर्क, जॉन विएटिंग द्वारा।
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI से) साथ वाला पेपर [लर्निंग ट्रांसफरेबल विजुअल मॉडल फ्रॉम नेचुरल लैंग्वेज सुपरविजन](https://arxiv.org /abs/2103.00020) एलेक रैडफोर्ड, जोंग वूक किम, क्रिस हैलासी, आदित्य रमेश, गेब्रियल गोह, संध्या अग्रवाल, गिरीश शास्त्री, अमांडा एस्केल, पामेला मिश्किन, जैक क्लार्क, ग्रेचेन क्रुएगर, इल्या सुत्स्केवर द्वारा।
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (सेल्सफोर्स से) साथ में पेपर [प्रोग्राम सिंथेसिस के लिए एक संवादात्मक प्रतिमान](https://arxiv.org/abs/2203.13474) एरिक निजकैंप, बो पैंग, हिरोआकी हयाशी, लिफू तू, हुआन वांग, यिंगबो झोउ, सिल्वियो सावरेस, कैमिंग जिओंग रिलीज।
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (माइक्रोसॉफ्ट रिसर्च एशिया से) कागज के साथ [फास्ट ट्रेनिंग कन्वर्जेंस के लिए सशर्त डीईटीआर](https://arxiv. org/abs/2108.06152) डेपू मेंग, ज़ियाओकांग चेन, ज़ेजिया फैन, गैंग ज़ेंग, होउकियांग ली, युहुई युआन, लेई सन, जिंगडोंग वांग द्वारा।
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (YituTech से) साथ में कागज [ConvBERT: स्पैन-आधारित डायनेमिक कनवल्शन के साथ BERT में सुधार](https://arxiv .org/abs/2008.02496) जिहांग जियांग, वीहाओ यू, डाकान झोउ, युनपेंग चेन, जियाशी फेंग, शुइचेंग यान द्वारा।
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (Facebook AI से) साथ वाला पेपर [A ConvNet for the 2020s](https://arxiv.org/abs /2201.03545) ज़ुआंग लियू, हेंज़ी माओ, चाओ-युआन वू, क्रिस्टोफ़ फीचटेनहोफ़र, ट्रेवर डेरेल, सैनिंग ज़ी द्वारा।
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (सिंघुआ यूनिवर्सिटी से) साथ में पेपर [सीपीएम: ए लार्ज-स्केल जेनेरेटिव चाइनीज प्री-ट्रेंड लैंग्वेज मॉडल](https : //arxiv.org/abs/2012.00413) झेंग्यान झांग, जू हान, हाओ झोउ, पेई के, युक्सियन गु, डेमिंग ये, युजिया किन, युशेंग सु, हाओझे जी, जियान गुआन, फैंचाओ क्यूई, ज़ियाओझी वांग, यानान झेंग द्वारा , गुओयांग ज़ेंग, हुआनकी काओ, शेंगकी चेन, डाइक्सुआन ली, ज़ेनबो सन, ज़ियुआन लियू, मिनली हुआंग, वेंटाओ हान, जी तांग, जुआनज़ी ली, ज़ियाओयान झू, माओसोंग सन।
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (सेल्सफोर्स से) साथ में पेपर [CTRL: ए कंडिशनल ट्रांसफॉर्मर लैंग्वेज मॉडल फॉर कंट्रोलेबल जेनरेशन](https://arxiv.org/abs/1909.05858) नीतीश शिरीष केसकर*, ब्रायन मैककैन*, लव आर. वार्ष्णेय, कैमिंग जिओंग और रिचर्ड द्वारा सोचर द्वारा जारी किया गया।
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (Microsoft से) साथ में दिया गया पेपर [CvT: इंट्रोड्यूसिंग कनवॉल्यूशन टू विजन ट्रांसफॉर्मर्स](https://arxiv.org/ एब्स/2103.15808) हैपिंग वू, बिन जिओ, नोएल कोडेला, मेंगचेन लियू, जियांग दाई, लू युआन, लेई झांग द्वारा।
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (फेसबुक से) साथ में कागज [Data2Vec: भाषण, दृष्टि और भाषा में स्व-पर्यवेक्षित सीखने के लिए एक सामान्य ढांचा] (https://arxiv.org/abs/2202.03555) एलेक्सी बाएव्स्की, वेई-निंग सू, कियानटोंग जू, अरुण बाबू, जियाताओ गु, माइकल औली द्वारा पोस्ट किया गया।
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (Microsoft से) साथ में दिया गया पेपर [DeBERta: डिकोडिंग-एन्हांस्ड BERT विद डिसेंटैंगल्ड अटेंशन](https://arxiv. org/abs/2006.03654) पेंगचेंग हे, ज़ियाओडोंग लियू, जियानफेंग गाओ, वीज़ू चेन द्वारा।
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (Microsoft से) साथ में दिया गया पेपर [DeBERTa: डिकोडिंग-एन्हांस्ड BERT विथ डिसेंन्गल्ड अटेंशन](https: //arxiv.org/abs/2006.03654) पेंगचेंग हे, ज़ियाओडोंग लियू, जियानफेंग गाओ, वीज़ू चेन द्वारा पोस्ट किया गया।
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (बर्कले/फेसबुक/गूगल से) पेपर के साथ [डिसीजन ट्रांसफॉर्मर: रीनफोर्समेंट लर्निंग वाया सीक्वेंस मॉडलिंग](https : //arxiv.org/abs/2106.01345) लिली चेन, केविन लू, अरविंद राजेश्वरन, किमिन ली, आदित्य ग्रोवर, माइकल लास्किन, पीटर एबील, अरविंद श्रीनिवास, इगोर मोर्डच द्वारा पोस्ट किया गया।
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (सेंसटाइम रिसर्च से) साथ में पेपर [डिफॉर्मेबल डीईटीआर: डिफॉर्मेबल ट्रांसफॉर्मर्स फॉर एंड-टू-एंड ऑब्जेक्ट डिटेक्शन] (https://arxiv.org/abs/2010.04159) Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, जिफेंग दाई द्वारा पोस्ट किया गया।
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (फेसबुक से) साथ में पेपर [ट्रेनिंग डेटा-एफिशिएंट इमेज ट्रांसफॉर्मर और डिस्टिलेशन थ्रू अटेंशन](https://arxiv .org/abs/2012.12877) ह्यूगो टौव्रोन, मैथ्यू कॉर्ड, मैथिज्स डूज़, फ़्रांसिस्को मस्सा, एलेक्ज़ेंडर सबलेरोल्स, हर्वे जेगौ द्वारा।
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (फेसबुक से) साथ में कागज [ट्रांसफॉर्मर्स के साथ एंड-टू-एंड ऑब्जेक्ट डिटेक्शन](https://arxiv. org/abs/2005.12872) निकोलस कैरियन, फ़्रांसिस्को मस्सा, गेब्रियल सिनेव, निकोलस उसुनियर, अलेक्जेंडर किरिलोव, सर्गेई ज़ागोरुयको द्वारा।
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [DialoGPT: बड़े पैमाने पर जनरेटिव प्री-ट्रेनिंग फॉर कन्वर्सेशनल रिस्पांस जेनरेशन](https ://arxiv.org/abs/1911.00536) यिज़े झांग, सिकी सन, मिशेल गैली, येन-चुन चेन, क्रिस ब्रोकेट, जियांग गाओ, जियानफेंग गाओ, जिंगजिंग लियू, बिल डोलन द्वारा।
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (हगिंगफेस से), साथ में कागज [डिस्टिलबर्ट, बीईआरटी का डिस्टिल्ड वर्जन: छोटा, तेज, सस्ता और हल्का] (https://arxiv.org/abs/1910.01108) विक्टर सनह, लिसांड्रे डेब्यू और थॉमस वुल्फ द्वारा पोस्ट किया गया। यही तरीका GPT-2 को [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERta से [DistilRoBERta](https://github.com) पर कंप्रेस करने के लिए भी लागू किया जाता है। / हगिंगफेस/ट्रांसफॉर्मर्स/ट्री/मेन/उदाहरण/डिस्टिलेशन), बहुभाषी BERT से [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) और डिस्टिलबर्ट का जर्मन संस्करण।
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [DiT: सेल्फ सुपरवाइज्ड प्री-ट्रेनिंग फॉर डॉक्यूमेंट इमेज ट्रांसफॉर्मर](https://arxiv.org/abs/2203.02378) जुनलॉन्ग ली, यिहेंग जू, टेंगचाओ लव, लेई कुई, चा झांग द्वारा फुरु वेई द्वारा पोस्ट किया गया।
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (NAVER से) साथ में कागज [OCR-मुक्त डॉक्यूमेंट अंडरस्टैंडिंग ट्रांसफॉर्मर](https://arxiv.org/abs /2111.15664) गीवूक किम, टीकग्यू होंग, मूनबिन यिम, जियोंग्योन नाम, जिनयॉन्ग पार्क, जिनयॉन्ग यिम, वोनसेओक ह्वांग, सांगडू यूं, डोंगयून हान, सेउंग्युन पार्क द्वारा।
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (फेसबुक से) साथ में पेपर [ओपन-डोमेन क्वेश्चन आंसरिंग के लिए डेंस पैसेज रिट्रीवल](https://arxiv. org/abs/2004.04906) व्लादिमीर करपुखिन, बरलास ओज़ुज़, सेवन मिन, पैट्रिक लुईस, लेडेल वू, सर्गेई एडुनोव, डैनकी चेन, और वेन-ताऊ यिह द्वारा।
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (इंटेल लैब्स से) साथ में कागज [विज़न ट्रांसफॉर्मर्स फॉर डेंस प्रेडिक्शन](https://arxiv.org /abs/2103.13413) रेने रैनफ्टल, एलेक्सी बोचकोवस्की, व्लादलेन कोल्टन द्वारा।
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (Google रिसर्च/स्टैनफोर्ड यूनिवर्सिटी से) साथ में दिया गया पेपर [इलेक्ट्रा: जेनरेटर के बजाय भेदभाव करने वाले के रूप में टेक्स्ट एन्कोडर्स का पूर्व-प्रशिक्षण] (https://arxiv.org/abs/2003.10555) केविन क्लार्क, मिन्ह-थांग लुओंग, क्वोक वी. ले, क्रिस्टोफर डी. मैनिंग द्वारा पोस्ट किया गया।
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (Google रिसर्च से) साथ में दिया गया पेपर [सीक्वेंस जेनरेशन टास्क के लिए प्री-ट्रेंड चेकपॉइंट का इस्तेमाल करना](https:/ /arxiv.org/abs/1907.12461) साशा रोठे, शशि नारायण, अलियाक्सि सेवेरिन द्वारा।
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)**(Baidu से) साथ देने वाला पेपर [ERNIE: एन्हांस्ड रिप्रेजेंटेशन थ्रू नॉलेज इंटीग्रेशन](https://arxiv.org/abs/1904.09223) यू सन, शुओहुआन वांग, युकुन ली, शिकुन फेंग, ज़ुई चेन, हान झांग, शिन तियान, डैनक्सियांग झू, हाओ तियान, हुआ वू द्वारा पोस्ट किया गया।
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (मेटा AI से) ट्रांसफॉर्मर प्रोटीन भाषा मॉडल हैं। **ESM-1b** पेपर के साथ जारी किया गया था [ अलेक्जेंडर राइव्स, जोशुआ मेयर, टॉम सर्कु, सिद्धार्थ गोयल, ज़ेमिंग लिन द्वारा जैविक संरचना और कार्य असुरक्षित सीखने को 250 मिलियन प्रोटीन अनुक्रमों तक स्केल करने से उभरता है] (https://www.pnas.org/content/118/15/e2016239118) जेसन लियू, डेमी गुओ, मायल ओट, सी. लॉरेंस ज़िटनिक, जेरी मा और रॉब फर्गस। **ESM-1v** को पेपर के साथ जारी किया गया था [भाषा मॉडल प्रोटीन फ़ंक्शन पर उत्परिवर्तन के प्रभावों की शून्य-शॉट भविष्यवाणी को सक्षम करते हैं] (https://doi.org/10.1101/2021.07.09.450648) जोशुआ मेयर, रोशन राव, रॉबर्ट वेरकुइल, जेसन लियू, टॉम सर्कु और अलेक्जेंडर राइव्स द्वारा। **ESM-2** को पेपर के साथ जारी किया गया था [भाषा मॉडल विकास के पैमाने पर प्रोटीन अनुक्रम सटीक संरचना भविष्यवाणी को सक्षम करते हैं](https://doi.org/10.1101/2022.07.20.500902) ज़ेमिंग लिन, हलील अकिन, रोशन राव, ब्रायन ही, झोंगकाई झू, वेंटिंग लू, ए द्वारा लान डॉस सैंटोस कोस्टा, मरियम फ़ज़ल-ज़रंडी, टॉम सर्कू, साल कैंडिडो, अलेक्जेंडर राइव्स।
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (CNRS से) साथ वाला पेपर [FlauBERT: Unsupervised Language Model Pre-training for फ़्रेंच](https://arxiv .org/abs/1912.05372) Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, बेंजामिन लेकोउटेक्स, अलेक्जेंड्रे अल्लाउज़ेन, बेनोइट क्रैबे, लॉरेंट बेसेसियर, डिडिएर श्वाब द्वारा।
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (FLAVA: A फाउंडेशनल लैंग्वेज एंड विजन अलाइनमेंट मॉडल) (https://arxiv) साथ वाला पेपर .org/abs/2112.04482) अमनप्रीत सिंह, रोंगहांग हू, वेदानुज गोस्वामी, गुइल्यूम कुएरॉन, वोज्शिएक गालुबा, मार्कस रोहरबैक, और डौवे कीला द्वारा।
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (गूगल रिसर्च से) साथ वाला पेपर [FNet: मिक्सिंग टोकन विद फूरियर ट्रांसफॉर्म्स](https://arxiv.org /abs/2105.03824) जेम्स ली-थॉर्प, जोशुआ आइंस्ली, इल्या एकस्टीन, सैंटियागो ओंटानन द्वारा।
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (सीएमयू/गूगल ब्रेन से) साथ में कागज [फ़नल-ट्रांसफॉर्मर: कुशल भाषा प्रसंस्करण के लिए अनुक्रमिक अतिरेक को छानना](https://arxiv.org/abs/2006.03236) जिहांग दाई, गुओकुन लाई, यिमिंग यांग, क्वोक वी. ले द्वारा रिहाई।
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (KAIST से) साथ वाला पेपर [वर्टिकल कटडेप्थ के साथ मोनोकुलर डेप्थ एस्टीमेशन के लिए ग्लोबल-लोकल पाथ नेटवर्क्स](https:/ /arxiv.org/abs/2201.07436) डोयोन किम, वूंगह्युन गा, प्युंगवान आह, डोंगग्यू जू, सेहवान चुन, जुनमो किम द्वारा।
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (OpenAI से) साथ में दिया गया पेपर [जेनरेटिव प्री-ट्रेनिंग द्वारा भाषा की समझ में सुधार](https://blog .openai.com/language-unsupervised/) एलेक रैडफोर्ड, कार्तिक नरसिम्हन, टिम सालिमन्स और इल्या सुत्स्केवर द्वारा।
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (EleutherAI से) रिपॉजिटरी के साथ [EleutherAI/gpt-neo](https://github.com/ EleutherAI /gpt-neo) रिलीज। सिड ब्लैक, स्टेला बिडरमैन, लियो गाओ, फिल वांग और कॉनर लेही द्वारा पोस्ट किया गया।
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (EleutherAI से) पेपर के साथ जारी किया गया [GPT-NeoX-20B: एक ओपन-सोर्स ऑटोरेग्रेसिव लैंग्वेज मॉडल] (https://arxiv.org/abs/2204.06745) सिड ब्लैक, स्टेला बिडरमैन, एरिक हैलाहन, क्वेंटिन एंथोनी, लियो गाओ, लॉरेंस गोल्डिंग, होरेस हे, कॉनर लेही, काइल मैकडोनेल, जेसन फांग, माइकल पाइलर, यूएसवीएसएन साई प्रशांत द्वारा , शिवांशु पुरोहित, लारिया रेनॉल्ड्स, जोनाथन टो, बेन वांग, सैमुअल वेनबैक
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (अबेजा के जरिए) शिन्या ओटानी, ताकायोशी मकाबे, अनुज अरोड़ा, क्यो हटोरी द्वारा।
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (ओपनएआई से) साथ में पेपर [लैंग्वेज मॉडल्स अनसुपरवाइज्ड मल्टीटास्क लर्नर्स हैं](https://blog.openai.com/better-language-models/) एलेक रैडफोर्ड*, जेफरी वू*, रेवन चाइल्ड, डेविड लुआन, डारियो एमोडी* द्वारा * और इल्या सुत्सकेवर** ने पोस्ट किया।
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (EleutherAI से) साथ वाला पेपर [kingoflolz/mesh-transformer-jax](https://github. com/kingoflolz/mesh-transformer-jax/) बेन वांग और अरन कोमात्सुजाकी द्वारा।
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (UCSD, NVIDIA से) साथ में कागज [GroupViT: टेक्स्ट सुपरविजन से सिमेंटिक सेगमेंटेशन इमर्जेस](https://arxiv .org/abs/2202.11094) जियारुई जू, शालिनी डी मेलो, सिफ़ी लियू, वोनमिन बायन, थॉमस ब्रेउएल, जान कौट्ज़, ज़ियाओलोंग वांग द्वारा।
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (फेसबुक से) साथ में पेपर [ह्यूबर्ट: सेल्फ सुपरवाइज्ड स्पीच रिप्रेजेंटेशन लर्निंग बाय मास्क्ड प्रेडिक्शन ऑफ हिडन यूनिट्स](https ://arxiv.org/abs/2106.07447) वेई-निंग सू, बेंजामिन बोल्टे, याओ-हंग ह्यूबर्ट त्साई, कुशाल लखोटिया, रुस्लान सालाखुतदीनोव, अब्देलरहमान मोहम्मद द्वारा।
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (बर्कले से) साथ में कागज [I-BERT: Integer-only BERT Quantization](https:// arxiv.org/abs/2101.01321) सेहून किम, अमीर घोलमी, ज़ेवेई याओ, माइकल डब्ल्यू महोनी, कर्ट केटज़र द्वारा।
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (माइक्रोसॉफ्ट रिसर्च एशिया से) साथ देने वाला पेपर [लेआउटएलएमवी3: यूनिफाइड टेक्स्ट और इमेज मास्किंग के साथ दस्तावेज़ एआई के लिए पूर्व-प्रशिक्षण](https://arxiv.org/abs/2204.08387) युपन हुआंग, टेंगचाओ लव, लेई कुई, युटोंग लू, फुरु वेई द्वारा पोस्ट किया गया।
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (मेटा AI से) साथ वाला पेपर [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https:/ /arxiv.org/abs/2104.01136) बेन ग्राहम, अलाएल्डिन एल-नौबी, ह्यूगो टौवरन, पियरे स्टॉक, आर्मंड जौलिन, हर्वे जेगौ, मैथिज डूज़ द्वारा।
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (दक्षिण चीन प्रौद्योगिकी विश्वविद्यालय से) साथ में कागज [LiLT: एक सरल लेकिन प्रभावी भाषा-स्वतंत्र लेआउट ट्रांसफार्मर संरचित दस्तावेज़ समझ के लिए](https://arxiv.org/abs/2202.13669) जियापेंग वांग, लियानवेन जिन, काई डिंग द्वारा पोस्ट किया गया।
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (मैंडी गुओ, जोशुआ आइंस्ली, डेविड यूथस, सैंटियागो ओंटानन, जियानमो नि, यूं-हुआन सुंग, यिनफेई यांग द्वारा पोस्ट किया गया।
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (स्टूडियो औसिया से) साथ में पेपर [LUKE: डीप कॉन्टेक्स्टुअलाइज्ड एंटिटी रिप्रेजेंटेशन विद एंटिटी-अवेयर सेल्फ-अटेंशन](https ://arxiv.org/abs/2010.01057) Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto द्वारा।
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (UNC चैपल हिल से) साथ में पेपर [LXMERT: ओपन-डोमेन क्वेश्चन के लिए ट्रांसफॉर्मर से क्रॉस-मोडलिटी एनकोडर रिप्रेजेंटेशन सीखना Answering](https://arxiv.org/abs/1908.07490) हाओ टैन और मोहित बंसल द्वारा।
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (फेसबुक से) साथ देने वाला पेपर [बियॉन्ड इंग्लिश-सेंट्रिक मल्टीलिंगुअल मशीन ट्रांसलेशन](https://arxiv.org/ एब्स/2010.11125) एंजेला फैन, श्रुति भोसले, होल्गर श्वेन्क, झी मा, अहमद अल-किश्की, सिद्धार्थ गोयल, मनदीप बैनेस, ओनूर सेलेबी, गुइल्लाम वेन्जेक, विश्रव चौधरी, नमन गोयल, टॉम बर्च, विटाली लिपचिंस्की, सर्गेई एडुनोव, एडौर्ड द्वारा ग्रेव, माइकल औली, आर्मंड जौलिन द्वारा पोस्ट किया गया।
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Jörg द्वारा [OPUS](http://opus.nlpl.eu/) डेटा से प्रशिक्षित मशीनी अनुवाद मॉडल पोस्ट किया गया टाइडेमैन द्वारा। [मैरियन फ्रेमवर्क](https://marian-nmt.github.io/) माइक्रोसॉफ्ट ट्रांसलेटर टीम द्वारा विकसित।
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (माइक्रोसॉफ्ट रिसर्च एशिया से) साथ में पेपर [मार्कअपएलएम: विजुअली-रिच डॉक्यूमेंट अंडरस्टैंडिंग के लिए टेक्स्ट और मार्कअप लैंग्वेज का प्री-ट्रेनिंग] (https://arxiv.org/abs/2110.08518) जुनलॉन्ग ली, यिहेंग जू, लेई कुई, फुरु द्वारा वी द्वारा पोस्ट किया गया।
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (मेटा और UIUC से) पेपर के साथ जारी किया गया [प्रति-पिक्सेल वर्गीकरण वह सब नहीं है जिसकी आपको सिमेंटिक सेगमेंटेशन की आवश्यकता है] (https://arxiv.org/abs/2107.06278) बोवेन चेंग, अलेक्जेंडर जी. श्विंग, अलेक्जेंडर किरिलोव द्वारा >>>>>> रिबेस ठीक करें
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (फेसबुक से) साथ में पेपर [न्यूरल मशीन ट्रांसलेशन के लिए मल्टीलिंगुअल डीनोइजिंग प्री-ट्रेनिंग](https://arxiv. org/abs/2001.08210) यिनहान लियू, जियाताओ गु, नमन गोयल, जियान ली, सर्गेई एडुनोव, मार्जन ग़ज़विनिनेजाद, माइक लुईस, ल्यूक ज़ेटलमॉयर द्वारा।
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (फेसबुक से) साथ में पेपर [एक्स्टेंसिबल बहुभाषी प्रीट्रेनिंग और फाइनट्यूनिंग के साथ बहुभाषी अनुवाद](https://arxiv युकिंग टैंग, चाउ ट्रान, जियान ली, पेंग-जेन चेन, नमन गोयल, विश्रव चौधरी, जियाताओ गु, एंजेला फैन द्वारा .org/abs/2008.00401)।
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (NVIDIA से) कागज के साथ [Megatron-LM: मॉडल का उपयोग करके बहु-अरब पैरामीटर भाषा मॉडल का प्रशिक्षण Parallelism](https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा।
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA से) साथ वाला पेपर [Megatron-LM: ट्रेनिंग मल्टी-बिलियन पैरामीटर लैंग्वेज मॉडल्स यूजिंग मॉडल पैरेललिज़्म] (https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा पोस्ट किया गया।
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (फ्रॉम Studio Ousia) साथ में पेपर [mLUKE: द पावर ऑफ एंटिटी रिप्रेजेंटेशन इन मल्टीलिंगुअल प्रीट्रेन्ड लैंग्वेज मॉडल्स](https://arxiv.org/abs/2110.08151) रयोकन री, इकुया यामाडा, और योशिमासा त्सुरोका द्वारा।
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (सीएमयू/गूगल ब्रेन से) साथ में कागज [मोबाइलबर्ट: संसाधन-सीमित उपकरणों के लिए एक कॉम्पैक्ट टास्क-अज्ञेय बीईआरटी] (https://arxiv.org/abs/2004.02984) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, और Denny Zhou द्वारा पोस्ट किया गया।
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (Apple से) साथ में कागज [MobileViT: लाइट-वेट, जनरल-पर्पस, और मोबाइल-फ्रेंडली विजन ट्रांसफॉर्मर] (https://arxiv.org/abs/2110.02178) सचिन मेहता और मोहम्मद रस्तगरी द्वारा पोस्ट किया गया।
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (Google AI से) साथ वाला पेपर [mT5: एक व्यापक बहुभाषी पूर्व-प्रशिक्षित टेक्स्ट-टू-टेक्स्ट ट्रांसफॉर्मर]( https://arxiv.org/abs/2010.11934) लिंटिंग ज़ू, नोआ कॉन्सटेंट, एडम रॉबर्ट्स, मिहिर काले, रामी अल-रफू, आदित्य सिद्धांत, आदित्य बरुआ, कॉलिन रैफेल द्वारा पोस्ट किया गया।
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (हुआवेई नूह के आर्क लैब से) साथ में कागज़ [NEZHA: चीनी भाषा समझ के लिए तंत्रिका प्रासंगिक प्रतिनिधित्व](https :/ /arxiv.org/abs/1909.00204) जुन्किउ वेई, ज़ियाओज़े रेन, ज़िआओगुआंग ली, वेनयोंग हुआंग, यी लियाओ, याशेंग वांग, जियाशू लिन, शिन जियांग, जिओ चेन और कुन लियू द्वारा।
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (फ्रॉम मेटा) साथ में पेपर [नो लैंग्वेज लेफ्ट बिहाइंड: स्केलिंग ह्यूमन-सेंटेड मशीन ट्रांसलेशन] (https://arxiv.org/abs/2207.04672) एनएलएलबी टीम द्वारा प्रकाशित।
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (विस्कॉन्सिन विश्वविद्यालय - मैडिसन से) साथ में कागज [Nyströmformer: A Nyström- आधारित एल्गोरिथम आत्म-ध्यान का अनुमान लगाने के लिए ](https://arxiv.org/abs/2102.03902) युनयांग ज़िओंग, झानपेंग ज़ेंग, रुद्रसिस चक्रवर्ती, मिंगक्सिंग टैन, ग्लेन फंग, यिन ली, विकास सिंह द्वारा पोस्ट किया गया।
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (Google AI से) साथ में कागज [विज़न ट्रांसफॉर्मर्स के साथ सिंपल ओपन-वोकैबुलरी ऑब्जेक्ट डिटेक्शन](https:/ /arxiv.org/abs/2205.06230) मैथियास मिंडरर, एलेक्सी ग्रिट्सेंको, ऑस्टिन स्टोन, मैक्सिम न्यूमैन, डिर्क वीसेनबोर्न, एलेक्सी डोसोवित्स्की, अरविंद महेंद्रन, अनुराग अर्नब, मुस्तफा देहघानी, ज़ुओरन शेन, जिओ वांग, ज़ियाओहुआ झाई, थॉमस किफ़, और नील हॉल्सबी द्वारा पोस्ट किया गया।
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (Google की ओर से) साथ में दिया गया पेपर [लंबे इनपुट सारांश के लिए ट्रांसफ़ॉर्मरों को बेहतर तरीके से एक्सटेंड करना](https://arxiv .org/abs/2208.04347) जेसन फांग, याओ झाओ, पीटर जे लियू द्वारा।
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (दीपमाइंड से) साथ में पेपर [पर्सीवर आईओ: संरचित इनपुट और आउटपुट के लिए एक सामान्य वास्तुकला] (https://arxiv.org/abs/2107.14795) एंड्रयू जेगल, सेबेस्टियन बोरग्यूड, जीन-बैप्टिस्ट अलायराक, कार्ल डोर्श, कैटलिन इओनेस्कु, डेविड द्वारा डिंग, स्कंद कोप्पुला, डैनियल ज़ोरान, एंड्रयू ब्रॉक, इवान शेलहैमर, ओलिवियर हेनाफ, मैथ्यू एम। बोट्विनिक, एंड्रयू ज़िसरमैन, ओरिओल विनियल्स, जोआओ कैरेरा द्वारा पोस्ट किया गया।
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (VinAI Research से) कागज के साथ [PhoBERT: वियतनामी के लिए पूर्व-प्रशिक्षित भाषा मॉडल](https://www .aclweb.org/anthology/2020.findings-emnlp.92/) डैट क्वोक गुयेन और अन्ह तुआन गुयेन द्वारा पोस्ट किया गया।
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP से) साथ वाला पेपर [प्रोग्राम अंडरस्टैंडिंग एंड जेनरेशन के लिए यूनिफाइड प्री-ट्रेनिंग](https://arxiv .org/abs/2103.06333) वसी उद्दीन अहमद, सैकत चक्रवर्ती, बैशाखी रे, काई-वेई चांग द्वारा।
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [ProphetNet: प्रेडिक्टिंग फ्यूचर एन-ग्राम फॉर सीक्वेंस-टू-सीक्वेंस प्री-ट्रेनिंग ](https://arxiv.org/abs/2001.04063) यू यान, वीज़ेन क्यूई, येयुन गोंग, दयाहेंग लियू, नान डुआन, जिउशेंग चेन, रुओफ़ेई झांग और मिंग झोउ द्वारा पोस्ट किया गया।
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (NVIDIA से) साथ वाला पेपर [डीप लर्निंग इंफ़ेक्शन के लिए इंटीजर क्वांटिज़ेशन: प्रिंसिपल्स एंड एम्पिरिकल इवैल्यूएशन](https:// arxiv.org/abs/2004.09602) हाओ वू, पैट्रिक जुड, जिआओजी झांग, मिखाइल इसेव और पॉलियस माइकेविसियस द्वारा।
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (फेसबुक से) साथ में कागज [रिट्रीवल-ऑगमेंटेड जेनरेशन फॉर नॉलेज-इंटेंसिव एनएलपी टास्क](https://arxiv .org/abs/2005.11401) पैट्रिक लुईस, एथन पेरेज़, अलेक्जेंड्रा पिक्टस, फैबियो पेट्रोनी, व्लादिमीर कारपुखिन, नमन गोयल, हेनरिक कुटलर, माइक लुईस, वेन-ताउ यिह, टिम रॉकटाशेल, सेबस्टियन रिडेल, डौवे कीला द्वारा।
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (Google अनुसंधान से) केल्विन गु, केंटन ली, ज़ोरा तुंग, पानुपोंग पसुपत और मिंग-वेई चांग द्वारा साथ में दिया गया पेपर [REALM: रिट्रीवल-ऑगमेंटेड लैंग्वेज मॉडल प्री-ट्रेनिंग](https://arxiv.org/abs/2002.08909)।
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (META रिसर्च से) [डिज़ाइनिंग नेटवर्क डिज़ाइन स्पेस] (https://arxiv.org/) पेपर के साथ जारी किया गया एब्स/2003.13678) इलिजा राडोसावोविक, राज प्रतीक कोसाराजू, रॉस गिर्शिक, कैमिंग ही, पिओटर डॉलर द्वारा।
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (गूगल रिसर्च से) साथ वाला पेपर [पूर्व-प्रशिक्षित भाषा मॉडल में एम्बेडिंग कपलिंग पर पुनर्विचार](https://arxiv .org/pdf/2010.12821.pdf) ह्युंग वोन चुंग, थिबॉल्ट फ़ेवरी, हेनरी त्साई, एम. जॉनसन, सेबेस्टियन रुडर द्वारा।
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (माइक्रोसॉफ्ट रिसर्च से) [डीप रेसिडुअल लर्निंग फॉर इमेज रिकग्निशन] (https://arxiv. org/abs/1512.03385) कैमिंग हे, जियांग्यु झांग, शाओकिंग रेन, जियान सन द्वारा।
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (फेसबुक से), साथ में कागज [मजबूत रूप से अनुकूलित BERT प्रीट्रेनिंग दृष्टिकोण](https://arxiv.org/abs /1907.11692) यिनहान लियू, मायल ओट, नमन गोयल, जिंगफेई डू, मंदार जोशी, डैनकी चेन, ओमर लेवी, माइक लुईस, ल्यूक ज़ेटलमॉयर, वेसेलिन स्टोयानोव द्वारा।
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (झुईई टेक्नोलॉजी से), साथ में पेपर [रोफॉर्मर: रोटरी पोजिशन एंबेडिंग के साथ एन्हांस्ड ट्रांसफॉर्मर] (https://arxiv.org/pdf/2104.09864v1.pdf) जियानलिन सु और यू लू और शेंगफेंग पैन और बो वेन और युनफेंग लियू द्वारा प्रकाशित।
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP से) साथ देने वाला पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स](https ://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योव आर्टज़ी द्वारा।
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP से) साथ में पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स] (https://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योआव आर्टज़ी द्वारा पोस्ट किया गया।
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (फेसबुक से), साथ में पेपर [फेयरसेक S2T: फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग विद फेयरसेक](https: //arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया。
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (फेसबुक से) साथ में पेपर [लार्ज-स्केल सेल्फ- एंड सेमी-सुपरवाइज्ड लर्निंग फॉर स्पीच ट्रांसलेशन](https://arxiv.org/abs/2104.06678) चांगहान वांग, ऐनी वू, जुआन पिनो, एलेक्सी बेवस्की, माइकल औली, एलेक्सिस द्वारा Conneau द्वारा पोस्ट किया गया।
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (तेल अवीव यूनिवर्सिटी से) साथ में पेपर [स्पैन सिलेक्शन को प्री-ट्रेनिंग करके कुछ-शॉट क्वेश्चन आंसरिंग](https:// arxiv.org/abs/2101.00438) ओरि राम, युवल कर्स्टन, जोनाथन बेरेंट, अमीर ग्लोबर्सन, ओमर लेवी द्वारा।
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (बर्कले से) कागज के साथ [SqueezeBERT: कुशल तंत्रिका नेटवर्क के बारे में NLP को कंप्यूटर विज़न क्या सिखा सकता है?](https: //arxiv.org/abs/2006.11316) फॉरेस्ट एन. इनडोला, अल्बर्ट ई. शॉ, रवि कृष्णा, और कर्ट डब्ल्यू. केटज़र द्वारा।
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (माइक्रोसॉफ्ट से) साथ में कागज [स्वाइन ट्रांसफॉर्मर: शिफ्टेड विंडोज का उपयोग कर पदानुक्रमित विजन ट्रांसफॉर्मर](https://arxiv .org/abs/2103.14030) ज़ी लियू, युटोंग लिन, यू काओ, हान हू, यिक्सुआन वेई, झेंग झांग, स्टीफन लिन, बैनिंग गुओ द्वारा।
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (Microsoft से) साथ वाला पेपर [Swin Transformer V2: स्केलिंग अप कैपेसिटी एंड रेजोल्यूशन](https:// ज़ी लियू, हान हू, युटोंग लिन, ज़ुलिआंग याओ, ज़ेंडा ज़ी, यिक्सुआन वेई, जिया निंग, यू काओ, झेंग झांग, ली डोंग, फुरु वेई, बैनिंग गुओ द्वारा arxiv.org/abs/2111.09883।
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (来自 Google AI)कॉलिन रैफेल और नोम शज़ीर और एडम रॉबर्ट्स और कैथरीन ली और शरण नारंग और माइकल मटेना द्वारा साथ में पेपर [एक एकीकृत टेक्स्ट-टू-टेक्स्ट ट्रांसफॉर्मर के साथ स्थानांतरण सीखने की सीमा की खोज] (https://arxiv.org/abs/1910.10683) और यांकी झोउ और वेई ली और पीटर जे लियू।
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (Google AI से) साथ वाला पेपर [google-research/text-to-text-transfer- ट्रांसफॉर्मर](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) कॉलिन रैफेल और नोम शज़ीर और एडम रॉबर्ट्स और कैथरीन ली और शरण नारंग द्वारा और माइकल मटेना और यांकी झोउ और वेई ली और पीटर जे लियू।
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [पबटेबल्स-1एम: टूवर्ड्स कॉम्प्रिहेंसिव टेबल एक्सट्रैक्शन फ्रॉम अनस्ट्रक्चर्ड डॉक्यूमेंट्स ](https://arxiv.org/abs/2110.00061) ब्रैंडन स्मॉक, रोहित पेसाला, रॉबिन अब्राहम द्वारा पोस्ट किया गया।
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (Google AI से) साथ में कागज [TAPAS: पूर्व-प्रशिक्षण के माध्यम से कमजोर पर्यवेक्षण तालिका पार्सिंग](https:// arxiv.org/abs/2004.02349) जोनाथन हर्ज़िग, पावेल क्रिज़िस्तोफ़ नोवाक, थॉमस मुलर, फ्रांसेस्को पिकिन्नो और जूलियन मार्टिन ईसेन्च्लोस द्वारा।
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [TAPEX: टेबल प्री-ट्रेनिंग थ्रू लर्निंग अ न्यूरल SQL एक्ज़ीक्यूटर](https: //arxiv.org/abs/2107.07653) कियान लियू, बेई चेन, जियाकी गुओ, मोर्टेज़ा ज़ियादी, ज़ेकी लिन, वीज़ू चेन, जियान-गुआंग लू द्वारा पोस्ट किया गया।
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU की ओर से) कागज के साथ [संस्करण-एक्स: एक ब्लॉग मॉडल चौकस चौक मॉडल मॉडल] (https://arxivorg/abs/1901.02860) क्वोकोक वी. ले, रुस्लैन सलाखुतदी
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (माइक्रोसॉफ्ट रिसर्च से) साथ में दिया गया पेपर [UniSpeech: यूनिफाइड स्पीच रिप्रेजेंटेशन लर्निंग विद लेबलेड एंड अनलेबल्ड डेटा](https:/ /arxiv.org/abs/2101.07597) चेंगई वांग, यू वू, याओ कियान, केनिची कुमातानी, शुजी लियू, फुरु वेई, माइकल ज़ेंग, ज़ुएदोंग हुआंग द्वारा।
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [UNISPEECH-SAT: यूनिवर्सल स्पीच रिप्रेजेंटेशन लर्निंग विद स्पीकर अवेयर प्री-ट्रेनिंग ](https://arxiv.org/abs/2110.05752) सानयुआन चेन, यू वू, चेंग्यी वांग, झेंगयांग चेन, झूओ चेन, शुजी लियू, जियान वू, याओ कियान, फुरु वेई, जिन्यु ली, जियांगज़ान यू द्वारा पोस्ट किया गया।
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (सिंघुआ यूनिवर्सिटी और ननकाई यूनिवर्सिटी से) साथ में पेपर [विजुअल अटेंशन नेटवर्क](https://arxiv.org/ pdf/2202.09741.pdf) मेंग-हाओ गुओ, चेंग-ज़े लू, झेंग-निंग लियू, मिंग-मिंग चेंग, शि-मिन हू द्वारा।
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (मल्टीमीडिया कम्प्यूटिंग ग्रुप, नानजिंग यूनिवर्सिटी से) साथ में पेपर [वीडियोएमएई: मास्क्ड ऑटोएन्कोडर स्व-पर्यवेक्षित वीडियो प्री-ट्रेनिंग के लिए डेटा-कुशल सीखने वाले हैं] (https://arxiv.org/abs/2203.12602) ज़ान टोंग, यिबिंग सॉन्ग, जुए द्वारा वांग, लिमिन वांग द्वारा पोस्ट किया गया।
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (NAVER AI Lab/Kakao Enterprise/Kakao Brain से) साथ में कागज [ViLT: Vision-and-Language Transformer बिना कनवल्शन या रीजन सुपरविजन](https://arxiv.org/abs/2102.03334) वोनजे किम, बोक्यूंग सोन, इल्डू किम द्वारा पोस्ट किया गया।
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (गूगल एआई से) कागज के साथ [एक इमेज इज़ वर्थ 16x16 वर्ड्स: ट्रांसफॉर्मर्स फॉर इमेज रिकॉग्निशन एट स्केल](https://arxiv.org/abs/2010.11929) एलेक्सी डोसोवित्स्की, लुकास बेयर, अलेक्जेंडर कोलेसनिकोव, डिर्क वीसेनबोर्न, शियाओहुआ झाई, थॉमस अनटरथिनर, मुस्तफा देहघानी, मैथियास मिंडरर, जॉर्ज हेगोल्ड, सिल्वेन गेली, जैकब उस्ज़कोरेइट द्वारा हॉल्सबी द्वारा पोस्ट किया गया।
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (UCLA NLP से) साथ वाला पेपर [VisualBERT: A Simple and Performant Baseline for Vision and Language](https:/ /arxiv.org/pdf/1908.03557) लियुनियन हेरोल्ड ली, मार्क यात्स्कर, दा यिन, चो-जुई हसीह, काई-वेई चांग द्वारा।
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (मेटा एआई से) साथ में कागज [मास्कड ऑटोएन्कोडर स्केलेबल विजन लर्नर्स हैं](https://arxiv.org/ एब्स/2111.06377) कैमिंग हे, ज़िनेली चेन, सेनिंग ज़ी, यांगहो ली, पिओट्र डॉलर, रॉस गिर्शिक द्वारा।
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (मेटा एआई से) साथ में कागज [लेबल-कुशल सीखने के लिए मास्क्ड स्याम देश के नेटवर्क](https://arxiv. org/abs/2204.07141) महमूद असरान, मथिल्डे कैरन, ईशान मिश्रा, पियोट्र बोजानोवस्की, फ्लोरियन बोर्डेस, पास्कल विंसेंट, आर्मंड जौलिन, माइकल रब्बत, निकोलस बल्लास द्वारा।
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (फेसबुक एआई से) साथ में पेपर [wav2vec 2.0: ए फ्रेमवर्क फॉर सेल्फ-सुपरवाइज्ड लर्निंग ऑफ स्पीच रिप्रेजेंटेशन] (https://arxiv.org/abs/2006.11477) एलेक्सी बेवस्की, हेनरी झोउ, अब्देलरहमान मोहम्मद, माइकल औली द्वारा।
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (Facebook AI से) साथ वाला पेपर [FAIRSEQ S2T: FAIRSEQ के साथ फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग ](https://arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, सरव्या पोपुरी, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया।
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (Facebook AI से) साथ वाला पेपर [सरल और प्रभावी जीरो-शॉट क्रॉस-लिंगुअल फोनेम रिकॉग्निशन](https:/ /arxiv.org/abs/2109.11680) कियानटोंग जू, एलेक्सी बाएव्स्की, माइकल औली द्वारा।
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (माइक्रोसॉफ्ट रिसर्च से) पेपर के साथ जारी किया गया [WavLM: फुल स्टैक के लिए बड़े पैमाने पर स्व-पर्यवेक्षित पूर्व-प्रशिक्षण स्पीच प्रोसेसिंग] (https://arxiv.org/abs/2110.13900) सानयुआन चेन, चेंगयी वांग, झेंगयांग चेन, यू वू, शुजी लियू, ज़ुओ चेन, जिन्यु ली, नाओयुकी कांडा, ताकुया योशियोका, ज़िओंग जिओ, जियान वू, लॉन्ग झोउ, शुओ रेन, यानमिन कियान, याओ कियान, जियान वू, माइकल ज़ेंग, फुरु वेई।
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (OpenAI से) साथ में कागज [बड़े पैमाने पर कमजोर पर्यवेक्षण के माध्यम से मजबूत भाषण पहचान](https://cdn. openai.com/papers/whisper.pdf) एलेक रैडफोर्ड, जोंग वूक किम, ताओ जू, ग्रेग ब्रॉकमैन, क्रिस्टीन मैकलीवे, इल्या सुत्स्केवर द्वारा।
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [एक्सपैंडिंग लैंग्वेज-इमेज प्रीट्रेन्ड मॉडल फॉर जनरल वीडियो रिकग्निशन](https: //arxiv.org/abs/2208.02816) बोलिन नी, होउवेन पेंग, मिंगाओ चेन, सोंगयांग झांग, गाओफेंग मेंग, जियानलोंग फू, शिमिंग जियांग, हैबिन लिंग द्वारा।
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (फेसबुक से) साथ में पेपर [क्रॉस-लिंगुअल लैंग्वेज मॉडल प्रीट्रेनिंग] (https://arxiv.org/abs/1901.07291) गिलाउम लैम्पल और एलेक्सिस कोनो द्वारा।
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (माइक्रोसॉफ्ट रिसर्च से) साथ में कागज [ProphetNet: प्रेडिक्टिंग फ्यूचर एन-ग्राम फॉर सीक्वेंस-टू- सीक्वेंस प्री-ट्रेनिंग](https://arxiv.org/abs/2001.04063) यू यान, वीज़ेन क्यूई, येयुन गोंग, दयाहेंग लियू, नान डुआन, जिउशेंग चेन, रुओफ़ेई झांग और मिंग झोउ द्वारा।
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (फेसबुक एआई से), साथ में पेपर [अनसुपरवाइज्ड क्रॉस-लिंगुअल रिप्रेजेंटेशन लर्निंग एट स्केल] (https://arxiv.org/abs/1911.02116) एलेक्सिस कोन्यू*, कार्तिकेय खंडेलवाल*, नमन गोयल, विश्रव चौधरी, गिलाउम वेनज़ेक, फ्रांसिस्को गुज़मैन द्वारा , एडौर्ड ग्रेव, मायल ओट, ल्यूक ज़ेटलमॉयर और वेसेलिन स्टोयानोव द्वारा।
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (Facebook AI से) साथ में कागज [बहुभाषी नकाबपोश भाषा के लिए बड़े पैमाने पर ट्रांसफॉर्मर ] मॉडलिंग](https://arxiv.org/abs/2105.00572) नमन गोयल, जिंगफेई डू, मायल ओट, गिरि अनंतरामन, एलेक्सिस कोनो द्वारा पोस्ट किया गया।
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (Google/CMU से) साथ वाला पेपर [XLNet: जनरलाइज्ड ऑटोरेग्रेसिव प्रीट्रेनिंग फॉर लैंग्वेज अंडरस्टैंडिंग](https://arxiv ज़ीलिन यांग*, ज़िहांग दाई*, यिमिंग यांग, जैम कार्बोनेल, रुस्लान सलाखुतदीनोव, क्वोक वी. ले द्वारा .org/abs/1906.08237)।
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (Facebook AI से) साथ वाला पेपर [XLS-R: सेल्फ सुपरवाइज्ड क्रॉस-लिंगुअल स्पीच रिप्रेजेंटेशन लर्निंग एट स्केल](https://arxiv.org/abs/2111.09296) अरुण बाबू, चांगहान वांग, एंड्रोस तजंद्रा, कुशाल लखोटिया, कियानटोंग जू, नमन गोयल, कृतिका सिंह, पैट्रिक वॉन प्लैटन, याथार्थ सराफ, जुआन पिनो, एलेक्सी बेवस्की, एलेक्सिस कोन्यू, माइकल औली द्वारा पोस्ट किया गया।
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (फेसबुक एआई से) साथ में पेपर [अनसुपरवाइज्ड क्रॉस-लिंगुअल रिप्रेजेंटेशन लर्निंग फॉर स्पीच रिकग्निशन] (https://arxiv.org/abs/2006.13979) एलेक्सिस कोन्यू, एलेक्सी बेवस्की, रोनन कोलोबर्ट, अब्देलरहमान मोहम्मद, माइकल औली द्वारा।
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (हुआझोंग यूनिवर्सिटी ऑफ साइंस एंड टेक्नोलॉजी से) साथ में पेपर [यू ओनली लुक एट वन सीक्वेंस: रीथिंकिंग ट्रांसफॉर्मर इन विज़न थ्रू ऑब्जेक्ट डिटेक्शन](https://arxiv.org/abs/2106.00666) युक्सिन फेंग, बेनचेंग लियाओ, जिंगगैंग वांग, जेमिन फेंग, जियांग क्यूई, रुई वू, जियानवेई नीयू, वेन्यू लियू द्वारा पोस्ट किया गया।
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (विस्कॉन्सिन विश्वविद्यालय - मैडिसन से) साथ में पेपर [यू ओनली सैंपल (लगभग) ज़ानपेंग ज़ेंग, युनयांग ज़िओंग द्वारा , सत्य एन. रवि, शैलेश आचार्य, ग्लेन फंग, विकास सिंह द्वारा पोस्ट किया गया।
1. एक नए मॉडल में योगदान देना चाहते हैं? नए मॉडल जोड़ने में आपका मार्गदर्शन करने के लिए हमारे पास एक **विस्तृत मार्गदर्शिका और टेम्प्लेट** है। आप उन्हें [`टेम्पलेट्स`](./templates) निर्देशिका में पा सकते हैं। पीआर शुरू करने से पहले [योगदान दिशानिर्देश] (./CONTRIBUTING.md) देखना और अनुरक्षकों से संपर्क करना या प्रतिक्रिया प्राप्त करने के लिए एक नया मुद्दा खोलना याद रखें।
यह जांचने के लिए कि क्या किसी मॉडल में पहले से ही Flax, PyTorch या TensorFlow का कार्यान्वयन है, या यदि उसके पास Tokenizers लाइब्रेरी में संबंधित टोकन है, तो [यह तालिका] (https://huggingface.co/ docs/transformers/index#supported) देखें। -फ्रेमवर्क)।
इन कार्यान्वयनों का परीक्षण कई डेटासेट पर किया गया है (देखें केस स्क्रिप्ट का उपयोग करें) और वैनिला कार्यान्वयन के लिए तुलनात्मक रूप से प्रदर्शन करना चाहिए। आप उपयोग के मामले के दस्तावेज़ [इस अनुभाग](https://huggingface.co/docs/transformers/examples) में व्यवहार का विवरण पढ़ सकते हैं।
## अधिक समझें
|अध्याय | विवरण |
|-|-|
| [दस्तावेज़ीकरण](https://huggingface.co/transformers/) | पूरा एपीआई दस्तावेज़ीकरण और ट्यूटोरियल |
| [कार्य सारांश](https://huggingface.co/docs/transformers/task_summary) | ट्रांसफॉर्मर समर्थित कार्य |
| [प्रीप्रोसेसिंग ट्यूटोरियल](https://huggingface.co/docs/transformers/preprocessing) | मॉडल के लिए डेटा तैयार करने के लिए `टोकनाइज़र` का उपयोग करना |
| [प्रशिक्षण और फाइन-ट्यूनिंग](https://huggingface.co/docs/transformers/training) | PyTorch/TensorFlow के ट्रेनिंग लूप या `ट्रेनर` API में ट्रांसफॉर्मर द्वारा दिए गए मॉडल का उपयोग करें |
| [क्विक स्टार्ट: ट्वीकिंग एंड यूज़ केस स्क्रिप्ट्स](https://github.com/huggingface/transformers/tree/main/examples) | विभिन्न कार्यों के लिए केस स्क्रिप्ट का उपयोग करें |
| [मॉडल साझा करना और अपलोड करना](https://huggingface.co/docs/transformers/model_sharing) | समुदाय के साथ अपने फाइन टूनड मॉडल अपलोड और साझा करें |
| [माइग्रेशन](https://huggingface.co/docs/transformers/migration) | `पाइटोरच-ट्रांसफॉर्मर्स` या `पाइटोरच-प्रीट्रेनड-बर्ट` से ट्रांसफॉर्मर में माइग्रेट करना |
## उद्धरण
हमने आधिकारिक तौर पर इस लाइब्रेरी का [पेपर](https://www.aclweb.org/anthology/2020.emnlp-demos.6/) प्रकाशित किया है, अगर आप ट्रान्सफ़ॉर्मर्स लाइब्रेरी का उपयोग करते हैं, तो कृपया उद्धृत करें:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Hindi translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Hindi characters. E.g., कुल मिलाकर 100 से अधिक भाषाएँ; ट्रांसफॉर्मर लाइब्रेरी का उपयोग करता है।
- वर्गाकार उद्धरणों का प्रयोग करें, जैसे, "उद्धरण"
Dictionary
Hugging Face: गले लगाओ चेहरा
token: शब्द (और मूल अंग्रेजी को कोष्ठक में चिह्नित करें）
tokenize: टोकननाइज़ करें (और मूल अंग्रेज़ी को चिह्नित करने के लिए कोष्ठक का उपयोग करें)
tokenizer: Tokenizer (मूल अंग्रेजी में कोष्ठक के साथ)
transformer: transformer
pipeline: समनुक्रम
API: API (अनुवाद के बिना)
inference: विचार
Trainer: प्रशिक्षक। कक्षा के नाम के रूप में प्रस्तुत किए जाने पर अनुवादित नहीं किया गया।
pretrained/pretrain: पूर्व प्रशिक्षण
finetune: फ़ाइन ट्यूनिंग
community: समुदाय
example: जब विशिष्ट गोदाम example कैटलॉग करते समय "केस केस" के रूप में अनुवादित
Python data structures (e.g., list, set, dict): मूल अंग्रेजी को चिह्नित करने के लिए सूचियों, सेटों, शब्दकोशों में अनुवाद करें और कोष्ठक का उपयोग करें
NLP/Natural Language Processing: द्वारा NLP अनुवाद के बिना प्रकट होते हैं Natural Language Processing प्रस्तुत किए जाने पर प्राकृतिक भाषा संसाधन में अनुवाद करें
checkpoint: जाँच बिंदु
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<b>हिन्दी</b> |
<p>
</h4>
<h3 align="center">
<p>Jax, PyTorch और TensorFlow के लिए उन्नत मशीन लर्निंग</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 100 से अधिक भाषाओं में पाठ वर्गीकरण, सूचना निष्कर्षण, प्रश्न उत्तर, सारांशीकरण, अनुवाद, पाठ निर्माण का समर्थन करने के लिए हजारों पूर्व-प्रशिक्षित मॉडल प्रदान करता है। इसका उद्देश्य सबसे उन्नत एनएलपी तकनीक को सभी के लिए सुलभ बनाना है।
🤗 Transformers त्वरित डाउनलोड और उपयोग के लिए एक एपीआई प्रदान करता है, जिससे आप किसी दिए गए पाठ पर एक पूर्व-प्रशिक्षित मॉडल ले सकते हैं, इसे अपने डेटासेट पर ठीक कर सकते हैं और इसे [मॉडल हब] (https://huggingface.co/models) के माध्यम से समुदाय के साथ साझा कर सकते हैं। ) . इसी समय, प्रत्येक परिभाषित पायथन मॉड्यूल पूरी तरह से स्वतंत्र है, जो संशोधन और तेजी से अनुसंधान प्रयोगों के लिए सुविधाजनक है।
🤗 Transformers तीन सबसे लोकप्रिय गहन शिक्षण पुस्तकालयों का समर्थन करता है： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — और इसके साथ निर्बाध रूप से एकीकृत होता है। आप अपने मॉडल को सीधे एक ढांचे के साथ प्रशिक्षित कर सकते हैं और दूसरे के साथ लोड और अनुमान लगा सकते हैं।
## ऑनलाइन डेमो
आप सबसे सीधे मॉडल पृष्ठ पर परीक्षण कर सकते हैं [model hub](https://huggingface.co/models) मॉडल पर। हम [निजी मॉडल होस्टिंग, मॉडल संस्करण, और अनुमान एपीआई] भी प्रदान करते हैं।(https://huggingface.co/pricing)。
यहाँ कुछ उदाहरण हैं：
- [शब्द को भरने के लिए मास्क के रूप में BERT का प्रयोग करें](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [इलेक्ट्रा के साथ नामित इकाई पहचान](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [जीपीटी-2 के साथ टेक्स्ट जनरेशन](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [रॉबर्टा के साथ प्राकृतिक भाषा निष्कर्ष](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [बार्ट के साथ पाठ सारांश](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [डिस्टिलबर्ट के साथ प्रश्नोत्तर](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [अनुवाद के लिए T5 का प्रयोग करें](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，हगिंग फेस टीम द्वारा बनाया गया, यह एक आधिकारिक पाठ पीढ़ी है demo。
## यदि आप हगिंग फेस टीम से बीस्पोक समर्थन की तलाश कर रहे हैं
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## जल्दी शुरू करें
हम त्वरित उपयोग के लिए मॉडल प्रदान करते हैं `pipeline` (पाइपलाइन) एपीआई। पाइपलाइन पूर्व-प्रशिक्षित मॉडल और संबंधित पाठ प्रीप्रोसेसिंग को एकत्रित करती है। सकारात्मक और नकारात्मक भावना को निर्धारित करने के लिए पाइपलाइनों का उपयोग करने का एक त्वरित उदाहरण यहां दिया गया है:
```python
>>> from transformers import pipeline
# भावना विश्लेषण पाइपलाइन का उपयोग करना
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
कोड की दूसरी पंक्ति पाइपलाइन द्वारा उपयोग किए गए पूर्व-प्रशिक्षित मॉडल को डाउनलोड और कैश करती है, जबकि कोड की तीसरी पंक्ति दिए गए पाठ पर मूल्यांकन करती है। यहां उत्तर 99 आत्मविश्वास के स्तर के साथ "सकारात्मक" है।
कई एनएलपी कार्यों में आउट ऑफ़ द बॉक्स पाइपलाइनों का पूर्व-प्रशिक्षण होता है। उदाहरण के लिए, हम किसी दिए गए पाठ से किसी प्रश्न का उत्तर आसानी से निकाल सकते हैं:
``` python
>>> from transformers import pipeline
# प्रश्नोत्तर पाइपलाइन का उपयोग करना
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
उत्तर देने के अलावा, पूर्व-प्रशिक्षित मॉडल संगत आत्मविश्वास स्कोर भी देता है, जहां उत्तर टोकनयुक्त पाठ में शुरू और समाप्त होता है। आप [इस ट्यूटोरियल](https://huggingface.co/docs/transformers/task_summary) से पाइपलाइन एपीआई द्वारा समर्थित कार्यों के बारे में अधिक जान सकते हैं।
अपने कार्य पर किसी भी पूर्व-प्रशिक्षित मॉडल को डाउनलोड करना और उसका उपयोग करना भी कोड की तीन पंक्तियों की तरह सरल है। यहाँ PyTorch संस्करण के लिए एक उदाहरण दिया गया है:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
यहाँ समकक्ष है TensorFlow कोड:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
टोकननाइज़र सभी पूर्व-प्रशिक्षित मॉडलों के लिए प्रीप्रोसेसिंग प्रदान करता है और इसे सीधे एक स्ट्रिंग (जैसे ऊपर दिए गए उदाहरण) या किसी सूची पर बुलाया जा सकता है। यह एक डिक्शनरी (तानाशाही) को आउटपुट करता है जिसे आप डाउनस्ट्रीम कोड में उपयोग कर सकते हैं या `**` अनपैकिंग एक्सप्रेशन के माध्यम से सीधे मॉडल को पास कर सकते हैं।
मॉडल स्वयं एक नियमित [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) या [TensorFlow `tf.keras.Model`](https ://pytorch.org/docs/stable/nn.html#torch.nn.Module) ://www.tensorflow.org/api_docs/python/tf/keras/Model) (आपके बैकएंड के आधार पर), जो हो सकता है सामान्य तरीके से उपयोग किया जाता है। [यह ट्यूटोरियल](https://huggingface.co/transformers/training.html) बताता है कि इस तरह के मॉडल को क्लासिक PyTorch या TensorFlow प्रशिक्षण लूप में कैसे एकीकृत किया जाए, या हमारे `ट्रेनर` एपीआई का उपयोग कैसे करें ताकि इसे जल्दी से फ़ाइन ट्यून किया जा सके।एक नया डेटासेट पे।
## ट्रांसफार्मर का उपयोग क्यों करें?
1. उपयोग में आसानी के लिए उन्नत मॉडल:
- एनएलयू और एनएलजी पर बेहतर प्रदर्शन
- प्रवेश के लिए कम बाधाओं के साथ शिक्षण और अभ्यास के अनुकूल
- उपयोगकर्ता-सामना करने वाले सार तत्व, केवल तीन वर्गों को जानने की जरूरत है
- सभी मॉडलों के लिए एकीकृत एपीआई
1. कम कम्प्यूटेशनल ओवरहेड और कम कार्बन उत्सर्जन:
- शोधकर्ता हर बार नए सिरे से प्रशिक्षण देने के बजाय प्रशिक्षित मॉडल साझा कर सकते हैं
- इंजीनियर गणना समय और उत्पादन ओवरहेड को कम कर सकते हैं
- दर्जनों मॉडल आर्किटेक्चर, 2,000 से अधिक पूर्व-प्रशिक्षित मॉडल, 100 से अधिक भाषाओं का समर्थन
1.मॉडल जीवनचक्र के हर हिस्से को शामिल करता है:
- कोड की केवल 3 पंक्तियों में उन्नत मॉडलों को प्रशिक्षित करें
- मॉडल को मनमाने ढंग से विभिन्न डीप लर्निंग फ्रेमवर्क के बीच स्थानांतरित किया जा सकता है, जैसा आप चाहते हैं
- निर्बाध रूप से प्रशिक्षण, मूल्यांकन और उत्पादन के लिए सबसे उपयुक्त ढांचा चुनें
1. आसानी से अनन्य मॉडल को अनुकूलित करें और अपनी आवश्यकताओं के लिए मामलों का उपयोग करें:
- हम मूल पेपर परिणामों को पुन: पेश करने के लिए प्रत्येक मॉडल आर्किटेक्चर के लिए कई उपयोग के मामले प्रदान करते हैं
- मॉडल की आंतरिक संरचना पारदर्शी और सुसंगत रहती है
- मॉडल फ़ाइल को अलग से इस्तेमाल किया जा सकता है, जो संशोधन और त्वरित प्रयोग के लिए सुविधाजनक है
## मुझे ट्रांसफॉर्मर का उपयोग कब नहीं करना चाहिए?
- यह लाइब्रेरी मॉड्यूलर न्यूरल नेटवर्क टूलबॉक्स नहीं है। मॉडल फ़ाइल में कोड जानबूझकर अल्पविकसित है, बिना अतिरिक्त सार इनकैप्सुलेशन के, ताकि शोधकर्ता अमूर्तता और फ़ाइल जंपिंग में शामिल हुए जल्दी से पुनरावृति कर सकें।
- `ट्रेनर` एपीआई किसी भी मॉडल के साथ संगत नहीं है, यह केवल इस पुस्तकालय के मॉडल के लिए अनुकूलित है। यदि आप सामान्य मशीन लर्निंग के लिए उपयुक्त प्रशिक्षण लूप कार्यान्वयन की तलाश में हैं, तो कहीं और देखें।
- हमारे सर्वोत्तम प्रयासों के बावजूद, [उदाहरण निर्देशिका] (https://github.com/huggingface/transformers/tree/main/examples) में स्क्रिप्ट केवल उपयोग के मामले हैं। आपकी विशिष्ट समस्या के लिए, वे जरूरी नहीं कि बॉक्स से बाहर काम करें, और आपको कोड की कुछ पंक्तियों को सूट करने की आवश्यकता हो सकती है।
## स्थापित करना
### पिप का उपयोग करना
इस रिपॉजिटरी का परीक्षण Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+ और TensorFlow 2.3+ के तहत किया गया है।
आप [वर्चुअल एनवायरनमेंट] (https://docs.python.org/3/library/venv.html) में 🤗 ट्रांसफॉर्मर इंस्टॉल कर सकते हैं। यदि आप अभी तक पायथन के वर्चुअल एनवायरनमेंट से परिचित नहीं हैं, तो कृपया इसे [उपयोगकर्ता निर्देश] (https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/) पढ़ें।
सबसे पहले, पायथन के उस संस्करण के साथ एक आभासी वातावरण बनाएं जिसका आप उपयोग करने और उसे सक्रिय करने की योजना बना रहे हैं।
फिर, आपको Flax, PyTorch या TensorFlow में से किसी एक को स्थापित करने की आवश्यकता है। अपने प्लेटफ़ॉर्म पर इन फ़्रेमवर्क को स्थापित करने के लिए, [TensorFlow स्थापना पृष्ठ](https://www.tensorflow.org/install/), [PyTorch स्थापना पृष्ठ](https://pytorch.org/get-started /locally/# देखें) start-locally) या [Flax स्थापना पृष्ठ](https://github.com/google/flax#quick-install).
जब इनमें से कोई एक बैकएंड सफलतापूर्वक स्थापित हो जाता है, तो ट्रांसफॉर्मर निम्नानुसार स्थापित किए जा सकते हैं:
```bash
pip install transformers
```
यदि आप उपयोग के मामलों को आज़माना चाहते हैं या आधिकारिक रिलीज़ से पहले नवीनतम इन-डेवलपमेंट कोड का उपयोग करना चाहते हैं, तो आपको [सोर्स से इंस्टॉल करना होगा](https://huggingface.co/docs/transformers/installation#installing-from- स्रोत)।
### कोंडा का उपयोग करना
ट्रांसफॉर्मर संस्करण 4.0.0 के बाद से, हमारे पास एक कोंडा चैनल है: `हगिंगफेस`।
ट्रांसफॉर्मर कोंडा के माध्यम से निम्नानुसार स्थापित किया जा सकता है:
```shell script
conda install -c huggingface transformers
```
कोंडा के माध्यम से Flax, PyTorch, या TensorFlow में से किसी एक को स्थापित करने के लिए, निर्देशों के लिए उनके संबंधित स्थापना पृष्ठ देखें।
## मॉडल आर्किटेक्चर
[उपयोगकर्ता](https://huggingface.co/users) और [organization](https://huggingface.co) द्वारा ट्रांसफॉर्मर समर्थित [**सभी मॉडल चौकियों**](https://huggingface.co/models) /users) हगिंगफेस.को/ऑर्गनाइजेशन), सभी को बिना किसी बाधा के हगिंगफेस.को [मॉडल हब](https://huggingface.co) के साथ एकीकृत किया गया है।
चौकियों की वर्तमान संख्या: 
🤗 ट्रांसफॉर्मर वर्तमान में निम्नलिखित आर्किटेक्चर का समर्थन करते हैं (मॉडल के अवलोकन के लिए [यहां] देखें (https://huggingface.co/docs/transformers/model_summary))：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (Google Research and the Toyota Technological Institute at Chicago) साथ थीसिस [ALBERT: A Lite BERT for Self-supervised भाषा प्रतिनिधित्व सीखना](https://arxiv.org/abs/1909.11942), झेंझोंग लैन, मिंगदा चेन, सेबेस्टियन गुडमैन, केविन गिम्पेल, पीयूष शर्मा, राडू सोरिकट
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (फेसबुक) साथ थीसिस [बार्ट: प्राकृतिक भाषा निर्माण, अनुवाद के लिए अनुक्रम-से-अनुक्रम पूर्व प्रशिक्षण , और समझ] (https://arxiv.org/pdf/1910.13461.pdf) पर निर्भर माइक लुईस, यिनहान लियू, नमन गोयल, मार्जन ग़ज़विनिनेजाद, अब्देलरहमान मोहम्मद, ओमर लेवी, वेस स्टोयानोव और ल्यूक ज़ेटलमॉयर
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (से École polytechnique) साथ थीसिस [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) पर निर्भर Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis रिहाई।
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (VinAI Research से) साथ में पेपर [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701)गुयेन लुओंग ट्रान, डुओंग मिन्ह ले और डाट क्वोक गुयेन द्वारा पोस्ट किया गया।
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (Microsoft से) साथ में कागज [BEiT: BERT इमेज ट्रांसफॉर्मर्स का प्री-ट्रेनिंग](https://arxiv.org/abs/2106.08254) Hangbo Bao, Li Dong, Furu Wei द्वारा।
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (गूगल से) साथ वाला पेपर [बीईआरटी: प्री-ट्रेनिंग ऑफ डीप बिडायरेक्शनल ट्रांसफॉर्मर्स फॉर लैंग्वेज अंडरस्टैंडिंग](https://arxiv.org/abs/1810.04805) जैकब डेवलिन, मिंग-वेई चांग, केंटन ली और क्रिस्टीना टौटानोवा द्वारा प्रकाशित किया गया था। .
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (गूगल से) साथ देने वाला पेपर [सीक्वेंस जेनरेशन टास्क के लिए प्री-ट्रेंड चेकपॉइंट का इस्तेमाल करना](https ://arxiv.org/abs/1907.12461) साशा रोठे, शशि नारायण, अलियाक्सि सेवेरिन द्वारा।
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (VinAI Research से) साथ में पेपर [BERTweet: अंग्रेजी ट्वीट्स के लिए एक पूर्व-प्रशिक्षित भाषा मॉडल] (https://aclanthology.org/2020.emnlp-demos.2/) डाट क्वोक गुयेन, थान वु और अन्ह तुआन गुयेन द्वारा प्रकाशित।
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (गूगल रिसर्च से) साथ वाला पेपर [बिग बर्ड: ट्रांसफॉर्मर्स फॉर लॉन्गर सीक्वेंस](https://arxiv .org/abs/2007.14062) मंज़िल ज़हीर, गुरु गुरुगणेश, अविनावा दुबे, जोशुआ आइंस्ली, क्रिस अल्बर्टी, सैंटियागो ओंटानोन, फिलिप फाम, अनिरुद्ध रावुला, किफ़ान वांग, ली यांग, अमर अहमद द्वारा।
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (गूगल रिसर्च से) साथ में पेपर [बिग बर्ड: ट्रांसफॉर्मर्स फॉर लॉन्गर सीक्वेंस](https://arxiv.org/abs/2007.14062) मंज़िल ज़हीर, गुरु गुरुगणेश, अविनावा दुबे, जोशुआ आइंस्ली, क्रिस अल्बर्टी, सैंटियागो ओंटानन, फिलिप फाम द्वारा , अनिरुद्ध रावुला, किफ़ान वांग, ली यांग, अमर अहमद द्वारा पोस्ट किया गया।
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (फेसबुक से) साथ में कागज [एक ओपन-डोमेन चैटबॉट बनाने की विधि](https://arxiv.org /abs/2004.13637) स्टीफन रोलर, एमिली दीनन, नमन गोयल, दा जू, मैरी विलियमसन, यिनहान लियू, जिंग जू, मायल ओट, कर्ट शस्टर, एरिक एम। स्मिथ, वाई-लैन बॉरो, जेसन वेस्टन द्वारा।
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (फेसबुक से) साथ में पेपर [एक ओपन-डोमेन चैटबॉट बनाने की रेसिपी](https://arxiv .org/abs/2004.13637) स्टीफन रोलर, एमिली दीनन, नमन गोयल, दा जू, मैरी विलियमसन, यिनहान लियू, जिंग जू, मायल ओट, कर्ट शस्टर, एरिक एम स्मिथ, वाई-लैन बॉरो, जेसन वेस्टन द्वारा।
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigSicence Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (एलेक्सा से) कागज के साथ [बीईआरटी के लिए ऑप्टिमल सबआर्किटेक्चर एक्सट्रैक्शन](https://arxiv.org/abs/ 2010.10499) एड्रियन डी विंटर और डैनियल जे पेरी द्वारा।
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (Google अनुसंधान से) साथ में कागज [ByT5: पूर्व-प्रशिक्षित बाइट-टू-बाइट मॉडल के साथ एक टोकन-मुक्त भविष्य की ओर] (https://arxiv.org/abs/2105.13626) Linting Xue, Aditya Barua, Noah Constant, रामी अल-रफू, शरण नारंग, मिहिर काले, एडम रॉबर्ट्स, कॉलिन रैफेल द्वारा पोस्ट किया गया।
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (इनरिया/फेसबुक/सोरबोन से) साथ में कागज [CamemBERT: एक टेस्टी फ्रेंच लैंग्वेज मॉडल](https:// arxiv.org/abs/1911.03894) लुई मार्टिन*, बेंजामिन मुलर*, पेड्रो जेवियर ऑर्टिज़ सुआरेज़*, योआन ड्यूपॉन्ट, लॉरेंट रोमरी, एरिक विलेमोन्टे डे ला क्लर्जरी, जैमे सेडाह और बेनोइट सगोट द्वारा।
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (Google रिसर्च से) साथ में दिया गया पेपर [कैनाइन: प्री-ट्रेनिंग ए एफिशिएंट टोकनाइजेशन-फ्री एनकोडर फॉर लैंग्वेज रिप्रेजेंटेशन]( https://arxiv.org/abs/2103.06874) जोनाथन एच क्लार्क, डैन गैरेट, यूलिया टर्क, जॉन विएटिंग द्वारा।
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI से) साथ वाला पेपर [लर्निंग ट्रांसफरेबल विजुअल मॉडल फ्रॉम नेचुरल लैंग्वेज सुपरविजन](https://arxiv.org /abs/2103.00020) एलेक रैडफोर्ड, जोंग वूक किम, क्रिस हैलासी, आदित्य रमेश, गेब्रियल गोह, संध्या अग्रवाल, गिरीश शास्त्री, अमांडा एस्केल, पामेला मिश्किन, जैक क्लार्क, ग्रेचेन क्रुएगर, इल्या सुत्स्केवर द्वारा।
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (सेल्सफोर्स से) साथ में पेपर [प्रोग्राम सिंथेसिस के लिए एक संवादात्मक प्रतिमान](https://arxiv.org/abs/2203.13474) एरिक निजकैंप, बो पैंग, हिरोआकी हयाशी, लिफू तू, हुआन वांग, यिंगबो झोउ, सिल्वियो सावरेस, कैमिंग जिओंग रिलीज।
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (माइक्रोसॉफ्ट रिसर्च एशिया से) कागज के साथ [फास्ट ट्रेनिंग कन्वर्जेंस के लिए सशर्त डीईटीआर](https://arxiv. org/abs/2108.06152) डेपू मेंग, ज़ियाओकांग चेन, ज़ेजिया फैन, गैंग ज़ेंग, होउकियांग ली, युहुई युआन, लेई सन, जिंगडोंग वांग द्वारा।
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (YituTech से) साथ में कागज [ConvBERT: स्पैन-आधारित डायनेमिक कनवल्शन के साथ BERT में सुधार](https://arxiv .org/abs/2008.02496) जिहांग जियांग, वीहाओ यू, डाकान झोउ, युनपेंग चेन, जियाशी फेंग, शुइचेंग यान द्वारा।
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (Facebook AI से) साथ वाला पेपर [A ConvNet for the 2020s](https://arxiv.org/abs /2201.03545) ज़ुआंग लियू, हेंज़ी माओ, चाओ-युआन वू, क्रिस्टोफ़ फीचटेनहोफ़र, ट्रेवर डेरेल, सैनिंग ज़ी द्वारा।
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (सिंघुआ यूनिवर्सिटी से) साथ में पेपर [सीपीएम: ए लार्ज-स्केल जेनेरेटिव चाइनीज प्री-ट्रेंड लैंग्वेज मॉडल](https : //arxiv.org/abs/2012.00413) झेंग्यान झांग, जू हान, हाओ झोउ, पेई के, युक्सियन गु, डेमिंग ये, युजिया किन, युशेंग सु, हाओझे जी, जियान गुआन, फैंचाओ क्यूई, ज़ियाओझी वांग, यानान झेंग द्वारा , गुओयांग ज़ेंग, हुआनकी काओ, शेंगकी चेन, डाइक्सुआन ली, ज़ेनबो सन, ज़ियुआन लियू, मिनली हुआंग, वेंटाओ हान, जी तांग, जुआनज़ी ली, ज़ियाओयान झू, माओसोंग सन।
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (सेल्सफोर्स से) साथ में पेपर [CTRL: ए कंडिशनल ट्रांसफॉर्मर लैंग्वेज मॉडल फॉर कंट्रोलेबल जेनरेशन](https://arxiv.org/abs/1909.05858) नीतीश शिरीष केसकर*, ब्रायन मैककैन*, लव आर. वार्ष्णेय, कैमिंग जिओंग और रिचर्ड द्वारा सोचर द्वारा जारी किया गया।
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (Microsoft से) साथ में दिया गया पेपर [CvT: इंट्रोड्यूसिंग कनवॉल्यूशन टू विजन ट्रांसफॉर्मर्स](https://arxiv.org/ एब्स/2103.15808) हैपिंग वू, बिन जिओ, नोएल कोडेला, मेंगचेन लियू, जियांग दाई, लू युआन, लेई झांग द्वारा।
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (फेसबुक से) साथ में कागज [Data2Vec: भाषण, दृष्टि और भाषा में स्व-पर्यवेक्षित सीखने के लिए एक सामान्य ढांचा] (https://arxiv.org/abs/2202.03555) एलेक्सी बाएव्स्की, वेई-निंग सू, कियानटोंग जू, अरुण बाबू, जियाताओ गु, माइकल औली द्वारा पोस्ट किया गया।
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (Microsoft से) साथ में दिया गया पेपर [DeBERta: डिकोडिंग-एन्हांस्ड BERT विद डिसेंटैंगल्ड अटेंशन](https://arxiv. org/abs/2006.03654) पेंगचेंग हे, ज़ियाओडोंग लियू, जियानफेंग गाओ, वीज़ू चेन द्वारा।
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (Microsoft से) साथ में दिया गया पेपर [DeBERTa: डिकोडिंग-एन्हांस्ड BERT विथ डिसेंन्गल्ड अटेंशन](https: //arxiv.org/abs/2006.03654) पेंगचेंग हे, ज़ियाओडोंग लियू, जियानफेंग गाओ, वीज़ू चेन द्वारा पोस्ट किया गया।
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (बर्कले/फेसबुक/गूगल से) पेपर के साथ [डिसीजन ट्रांसफॉर्मर: रीनफोर्समेंट लर्निंग वाया सीक्वेंस मॉडलिंग](https : //arxiv.org/abs/2106.01345) लिली चेन, केविन लू, अरविंद राजेश्वरन, किमिन ली, आदित्य ग्रोवर, माइकल लास्किन, पीटर एबील, अरविंद श्रीनिवास, इगोर मोर्डच द्वारा पोस्ट किया गया।
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (सेंसटाइम रिसर्च से) साथ में पेपर [डिफॉर्मेबल डीईटीआर: डिफॉर्मेबल ट्रांसफॉर्मर्स फॉर एंड-टू-एंड ऑब्जेक्ट डिटेक्शन] (https://arxiv.org/abs/2010.04159) Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, जिफेंग दाई द्वारा पोस्ट किया गया।
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (फेसबुक से) साथ में पेपर [ट्रेनिंग डेटा-एफिशिएंट इमेज ट्रांसफॉर्मर और डिस्टिलेशन थ्रू अटेंशन](https://arxiv .org/abs/2012.12877) ह्यूगो टौव्रोन, मैथ्यू कॉर्ड, मैथिज्स डूज़, फ़्रांसिस्को मस्सा, एलेक्ज़ेंडर सबलेरोल्स, हर्वे जेगौ द्वारा।
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (फेसबुक से) साथ में कागज [ट्रांसफॉर्मर्स के साथ एंड-टू-एंड ऑब्जेक्ट डिटेक्शन](https://arxiv. org/abs/2005.12872) निकोलस कैरियन, फ़्रांसिस्को मस्सा, गेब्रियल सिनेव, निकोलस उसुनियर, अलेक्जेंडर किरिलोव, सर्गेई ज़ागोरुयको द्वारा।
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [DialoGPT: बड़े पैमाने पर जनरेटिव प्री-ट्रेनिंग फॉर कन्वर्सेशनल रिस्पांस जेनरेशन](https ://arxiv.org/abs/1911.00536) यिज़े झांग, सिकी सन, मिशेल गैली, येन-चुन चेन, क्रिस ब्रोकेट, जियांग गाओ, जियानफेंग गाओ, जिंगजिंग लियू, बिल डोलन द्वारा।
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (हगिंगफेस से), साथ में कागज [डिस्टिलबर्ट, बीईआरटी का डिस्टिल्ड वर्जन: छोटा, तेज, सस्ता और हल्का] (https://arxiv.org/abs/1910.01108) विक्टर सनह, लिसांड्रे डेब्यू और थॉमस वुल्फ द्वारा पोस्ट किया गया। यही तरीका GPT-2 को [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERta से [DistilRoBERta](https://github.com) पर कंप्रेस करने के लिए भी लागू किया जाता है। / हगिंगफेस/ट्रांसफॉर्मर्स/ट्री/मेन/उदाहरण/डिस्टिलेशन), बहुभाषी BERT से [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) और डिस्टिलबर्ट का जर्मन संस्करण।
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [DiT: सेल्फ सुपरवाइज्ड प्री-ट्रेनिंग फॉर डॉक्यूमेंट इमेज ट्रांसफॉर्मर](https://arxiv.org/abs/2203.02378) जुनलॉन्ग ली, यिहेंग जू, टेंगचाओ लव, लेई कुई, चा झांग द्वारा फुरु वेई द्वारा पोस्ट किया गया।
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (NAVER से) साथ में कागज [OCR-मुक्त डॉक्यूमेंट अंडरस्टैंडिंग ट्रांसफॉर्मर](https://arxiv.org/abs /2111.15664) गीवूक किम, टीकग्यू होंग, मूनबिन यिम, जियोंग्योन नाम, जिनयॉन्ग पार्क, जिनयॉन्ग यिम, वोनसेओक ह्वांग, सांगडू यूं, डोंगयून हान, सेउंग्युन पार्क द्वारा।
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (फेसबुक से) साथ में पेपर [ओपन-डोमेन क्वेश्चन आंसरिंग के लिए डेंस पैसेज रिट्रीवल](https://arxiv. org/abs/2004.04906) व्लादिमीर करपुखिन, बरलास ओज़ुज़, सेवन मिन, पैट्रिक लुईस, लेडेल वू, सर्गेई एडुनोव, डैनकी चेन, और वेन-ताऊ यिह द्वारा।
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (इंटेल लैब्स से) साथ में कागज [विज़न ट्रांसफॉर्मर्स फॉर डेंस प्रेडिक्शन](https://arxiv.org /abs/2103.13413) रेने रैनफ्टल, एलेक्सी बोचकोवस्की, व्लादलेन कोल्टन द्वारा।
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (Google रिसर्च/स्टैनफोर्ड यूनिवर्सिटी से) साथ में दिया गया पेपर [इलेक्ट्रा: जेनरेटर के बजाय भेदभाव करने वाले के रूप में टेक्स्ट एन्कोडर्स का पूर्व-प्रशिक्षण] (https://arxiv.org/abs/2003.10555) केविन क्लार्क, मिन्ह-थांग लुओंग, क्वोक वी. ले, क्रिस्टोफर डी. मैनिंग द्वारा पोस्ट किया गया।
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (Google रिसर्च से) साथ में दिया गया पेपर [सीक्वेंस जेनरेशन टास्क के लिए प्री-ट्रेंड चेकपॉइंट का इस्तेमाल करना](https:/ /arxiv.org/abs/1907.12461) साशा रोठे, शशि नारायण, अलियाक्सि सेवेरिन द्वारा।
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)**(Baidu से) साथ देने वाला पेपर [ERNIE: एन्हांस्ड रिप्रेजेंटेशन थ्रू नॉलेज इंटीग्रेशन](https://arxiv.org/abs/1904.09223) यू सन, शुओहुआन वांग, युकुन ली, शिकुन फेंग, ज़ुई चेन, हान झांग, शिन तियान, डैनक्सियांग झू, हाओ तियान, हुआ वू द्वारा पोस्ट किया गया।
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (मेटा AI से) ट्रांसफॉर्मर प्रोटीन भाषा मॉडल हैं। **ESM-1b** पेपर के साथ जारी किया गया था [ अलेक्जेंडर राइव्स, जोशुआ मेयर, टॉम सर्कु, सिद्धार्थ गोयल, ज़ेमिंग लिन द्वारा जैविक संरचना और कार्य असुरक्षित सीखने को 250 मिलियन प्रोटीन अनुक्रमों तक स्केल करने से उभरता है] (https://www.pnas.org/content/118/15/e2016239118) जेसन लियू, डेमी गुओ, मायल ओट, सी. लॉरेंस ज़िटनिक, जेरी मा और रॉब फर्गस। **ESM-1v** को पेपर के साथ जारी किया गया था [भाषा मॉडल प्रोटीन फ़ंक्शन पर उत्परिवर्तन के प्रभावों की शून्य-शॉट भविष्यवाणी को सक्षम करते हैं] (https://doi.org/10.1101/2021.07.09.450648) जोशुआ मेयर, रोशन राव, रॉबर्ट वेरकुइल, जेसन लियू, टॉम सर्कु और अलेक्जेंडर राइव्स द्वारा। **ESM-2** को पेपर के साथ जारी किया गया था [भाषा मॉडल विकास के पैमाने पर प्रोटीन अनुक्रम सटीक संरचना भविष्यवाणी को सक्षम करते हैं](https://doi.org/10.1101/2022.07.20.500902) ज़ेमिंग लिन, हलील अकिन, रोशन राव, ब्रायन ही, झोंगकाई झू, वेंटिंग लू, ए द्वारा लान डॉस सैंटोस कोस्टा, मरियम फ़ज़ल-ज़रंडी, टॉम सर्कू, साल कैंडिडो, अलेक्जेंडर राइव्स।
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (CNRS से) साथ वाला पेपर [FlauBERT: Unsupervised Language Model Pre-training for फ़्रेंच](https://arxiv .org/abs/1912.05372) Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, बेंजामिन लेकोउटेक्स, अलेक्जेंड्रे अल्लाउज़ेन, बेनोइट क्रैबे, लॉरेंट बेसेसियर, डिडिएर श्वाब द्वारा।
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (FLAVA: A फाउंडेशनल लैंग्वेज एंड विजन अलाइनमेंट मॉडल) (https://arxiv) साथ वाला पेपर .org/abs/2112.04482) अमनप्रीत सिंह, रोंगहांग हू, वेदानुज गोस्वामी, गुइल्यूम कुएरॉन, वोज्शिएक गालुबा, मार्कस रोहरबैक, और डौवे कीला द्वारा।
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (गूगल रिसर्च से) साथ वाला पेपर [FNet: मिक्सिंग टोकन विद फूरियर ट्रांसफॉर्म्स](https://arxiv.org /abs/2105.03824) जेम्स ली-थॉर्प, जोशुआ आइंस्ली, इल्या एकस्टीन, सैंटियागो ओंटानन द्वारा।
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (सीएमयू/गूगल ब्रेन से) साथ में कागज [फ़नल-ट्रांसफॉर्मर: कुशल भाषा प्रसंस्करण के लिए अनुक्रमिक अतिरेक को छानना](https://arxiv.org/abs/2006.03236) जिहांग दाई, गुओकुन लाई, यिमिंग यांग, क्वोक वी. ले द्वारा रिहाई।
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (KAIST से) साथ वाला पेपर [वर्टिकल कटडेप्थ के साथ मोनोकुलर डेप्थ एस्टीमेशन के लिए ग्लोबल-लोकल पाथ नेटवर्क्स](https:/ /arxiv.org/abs/2201.07436) डोयोन किम, वूंगह्युन गा, प्युंगवान आह, डोंगग्यू जू, सेहवान चुन, जुनमो किम द्वारा।
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (OpenAI से) साथ में दिया गया पेपर [जेनरेटिव प्री-ट्रेनिंग द्वारा भाषा की समझ में सुधार](https://blog .openai.com/language-unsupervised/) एलेक रैडफोर्ड, कार्तिक नरसिम्हन, टिम सालिमन्स और इल्या सुत्स्केवर द्वारा।
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (EleutherAI से) रिपॉजिटरी के साथ [EleutherAI/gpt-neo](https://github.com/ EleutherAI /gpt-neo) रिलीज। सिड ब्लैक, स्टेला बिडरमैन, लियो गाओ, फिल वांग और कॉनर लेही द्वारा पोस्ट किया गया।
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (EleutherAI से) पेपर के साथ जारी किया गया [GPT-NeoX-20B: एक ओपन-सोर्स ऑटोरेग्रेसिव लैंग्वेज मॉडल] (https://arxiv.org/abs/2204.06745) सिड ब्लैक, स्टेला बिडरमैन, एरिक हैलाहन, क्वेंटिन एंथोनी, लियो गाओ, लॉरेंस गोल्डिंग, होरेस हे, कॉनर लेही, काइल मैकडोनेल, जेसन फांग, माइकल पाइलर, यूएसवीएसएन साई प्रशांत द्वारा , शिवांशु पुरोहित, लारिया रेनॉल्ड्स, जोनाथन टो, बेन वांग, सैमुअल वेनबैक
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (अबेजा के जरिए) शिन्या ओटानी, ताकायोशी मकाबे, अनुज अरोड़ा, क्यो हटोरी द्वारा।
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (ओपनएआई से) साथ में पेपर [लैंग्वेज मॉडल्स अनसुपरवाइज्ड मल्टीटास्क लर्नर्स हैं](https://blog.openai.com/better-language-models/) एलेक रैडफोर्ड*, जेफरी वू*, रेवन चाइल्ड, डेविड लुआन, डारियो एमोडी* द्वारा * और इल्या सुत्सकेवर** ने पोस्ट किया।
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (EleutherAI से) साथ वाला पेपर [kingoflolz/mesh-transformer-jax](https://github. com/kingoflolz/mesh-transformer-jax/) बेन वांग और अरन कोमात्सुजाकी द्वारा।
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (UCSD, NVIDIA से) साथ में कागज [GroupViT: टेक्स्ट सुपरविजन से सिमेंटिक सेगमेंटेशन इमर्जेस](https://arxiv .org/abs/2202.11094) जियारुई जू, शालिनी डी मेलो, सिफ़ी लियू, वोनमिन बायन, थॉमस ब्रेउएल, जान कौट्ज़, ज़ियाओलोंग वांग द्वारा।
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (फेसबुक से) साथ में पेपर [ह्यूबर्ट: सेल्फ सुपरवाइज्ड स्पीच रिप्रेजेंटेशन लर्निंग बाय मास्क्ड प्रेडिक्शन ऑफ हिडन यूनिट्स](https ://arxiv.org/abs/2106.07447) वेई-निंग सू, बेंजामिन बोल्टे, याओ-हंग ह्यूबर्ट त्साई, कुशाल लखोटिया, रुस्लान सालाखुतदीनोव, अब्देलरहमान मोहम्मद द्वारा।
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (बर्कले से) साथ में कागज [I-BERT: Integer-only BERT Quantization](https:// arxiv.org/abs/2101.01321) सेहून किम, अमीर घोलमी, ज़ेवेई याओ, माइकल डब्ल्यू महोनी, कर्ट केटज़र द्वारा।
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (माइक्रोसॉफ्ट रिसर्च एशिया से) साथ देने वाला पेपर [लेआउटएलएमवी3: यूनिफाइड टेक्स्ट और इमेज मास्किंग के साथ दस्तावेज़ एआई के लिए पूर्व-प्रशिक्षण](https://arxiv.org/abs/2204.08387) युपन हुआंग, टेंगचाओ लव, लेई कुई, युटोंग लू, फुरु वेई द्वारा पोस्ट किया गया।
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (मेटा AI से) साथ वाला पेपर [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https:/ /arxiv.org/abs/2104.01136) बेन ग्राहम, अलाएल्डिन एल-नौबी, ह्यूगो टौवरन, पियरे स्टॉक, आर्मंड जौलिन, हर्वे जेगौ, मैथिज डूज़ द्वारा।
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (दक्षिण चीन प्रौद्योगिकी विश्वविद्यालय से) साथ में कागज [LiLT: एक सरल लेकिन प्रभावी भाषा-स्वतंत्र लेआउट ट्रांसफार्मर संरचित दस्तावेज़ समझ के लिए](https://arxiv.org/abs/2202.13669) जियापेंग वांग, लियानवेन जिन, काई डिंग द्वारा पोस्ट किया गया।
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (मैंडी गुओ, जोशुआ आइंस्ली, डेविड यूथस, सैंटियागो ओंटानन, जियानमो नि, यूं-हुआन सुंग, यिनफेई यांग द्वारा पोस्ट किया गया।
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (स्टूडियो औसिया से) साथ में पेपर [LUKE: डीप कॉन्टेक्स्टुअलाइज्ड एंटिटी रिप्रेजेंटेशन विद एंटिटी-अवेयर सेल्फ-अटेंशन](https ://arxiv.org/abs/2010.01057) Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto द्वारा।
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (UNC चैपल हिल से) साथ में पेपर [LXMERT: ओपन-डोमेन क्वेश्चन के लिए ट्रांसफॉर्मर से क्रॉस-मोडलिटी एनकोडर रिप्रेजेंटेशन सीखना Answering](https://arxiv.org/abs/1908.07490) हाओ टैन और मोहित बंसल द्वारा।
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (फेसबुक से) साथ देने वाला पेपर [बियॉन्ड इंग्लिश-सेंट्रिक मल्टीलिंगुअल मशीन ट्रांसलेशन](https://arxiv.org/ एब्स/2010.11125) एंजेला फैन, श्रुति भोसले, होल्गर श्वेन्क, झी मा, अहमद अल-किश्की, सिद्धार्थ गोयल, मनदीप बैनेस, ओनूर सेलेबी, गुइल्लाम वेन्जेक, विश्रव चौधरी, नमन गोयल, टॉम बर्च, विटाली लिपचिंस्की, सर्गेई एडुनोव, एडौर्ड द्वारा ग्रेव, माइकल औली, आर्मंड जौलिन द्वारा पोस्ट किया गया।
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Jörg द्वारा [OPUS](http://opus.nlpl.eu/) डेटा से प्रशिक्षित मशीनी अनुवाद मॉडल पोस्ट किया गया टाइडेमैन द्वारा। [मैरियन फ्रेमवर्क](https://marian-nmt.github.io/) माइक्रोसॉफ्ट ट्रांसलेटर टीम द्वारा विकसित।
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (माइक्रोसॉफ्ट रिसर्च एशिया से) साथ में पेपर [मार्कअपएलएम: विजुअली-रिच डॉक्यूमेंट अंडरस्टैंडिंग के लिए टेक्स्ट और मार्कअप लैंग्वेज का प्री-ट्रेनिंग] (https://arxiv.org/abs/2110.08518) जुनलॉन्ग ली, यिहेंग जू, लेई कुई, फुरु द्वारा वी द्वारा पोस्ट किया गया।
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (मेटा और UIUC से) पेपर के साथ जारी किया गया [प्रति-पिक्सेल वर्गीकरण वह सब नहीं है जिसकी आपको सिमेंटिक सेगमेंटेशन की आवश्यकता है] (https://arxiv.org/abs/2107.06278) बोवेन चेंग, अलेक्जेंडर जी. श्विंग, अलेक्जेंडर किरिलोव द्वारा >>>>>> रिबेस ठीक करें
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (फेसबुक से) साथ में पेपर [न्यूरल मशीन ट्रांसलेशन के लिए मल्टीलिंगुअल डीनोइजिंग प्री-ट्रेनिंग](https://arxiv. org/abs/2001.08210) यिनहान लियू, जियाताओ गु, नमन गोयल, जियान ली, सर्गेई एडुनोव, मार्जन ग़ज़विनिनेजाद, माइक लुईस, ल्यूक ज़ेटलमॉयर द्वारा।
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (फेसबुक से) साथ में पेपर [एक्स्टेंसिबल बहुभाषी प्रीट्रेनिंग और फाइनट्यूनिंग के साथ बहुभाषी अनुवाद](https://arxiv युकिंग टैंग, चाउ ट्रान, जियान ली, पेंग-जेन चेन, नमन गोयल, विश्रव चौधरी, जियाताओ गु, एंजेला फैन द्वारा .org/abs/2008.00401)।
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (NVIDIA से) कागज के साथ [Megatron-LM: मॉडल का उपयोग करके बहु-अरब पैरामीटर भाषा मॉडल का प्रशिक्षण Parallelism](https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा।
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA से) साथ वाला पेपर [Megatron-LM: ट्रेनिंग मल्टी-बिलियन पैरामीटर लैंग्वेज मॉडल्स यूजिंग मॉडल पैरेललिज़्म] (https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा पोस्ट किया गया।
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (फ्रॉम Studio Ousia) साथ में पेपर [mLUKE: द पावर ऑफ एंटिटी रिप्रेजेंटेशन इन मल्टीलिंगुअल प्रीट्रेन्ड लैंग्वेज मॉडल्स](https://arxiv.org/abs/2110.08151) रयोकन री, इकुया यामाडा, और योशिमासा त्सुरोका द्वारा।
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (सीएमयू/गूगल ब्रेन से) साथ में कागज [मोबाइलबर्ट: संसाधन-सीमित उपकरणों के लिए एक कॉम्पैक्ट टास्क-अज्ञेय बीईआरटी] (https://arxiv.org/abs/2004.02984) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, और Denny Zhou द्वारा पोस्ट किया गया।
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (Apple से) साथ में कागज [MobileViT: लाइट-वेट, जनरल-पर्पस, और मोबाइल-फ्रेंडली विजन ट्रांसफॉर्मर] (https://arxiv.org/abs/2110.02178) सचिन मेहता और मोहम्मद रस्तगरी द्वारा पोस्ट किया गया।
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (Google AI से) साथ वाला पेपर [mT5: एक व्यापक बहुभाषी पूर्व-प्रशिक्षित टेक्स्ट-टू-टेक्स्ट ट्रांसफॉर्मर]( https://arxiv.org/abs/2010.11934) लिंटिंग ज़ू, नोआ कॉन्सटेंट, एडम रॉबर्ट्स, मिहिर काले, रामी अल-रफू, आदित्य सिद्धांत, आदित्य बरुआ, कॉलिन रैफेल द्वारा पोस्ट किया गया।
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (हुआवेई नूह के आर्क लैब से) साथ में कागज़ [NEZHA: चीनी भाषा समझ के लिए तंत्रिका प्रासंगिक प्रतिनिधित्व](https :/ /arxiv.org/abs/1909.00204) जुन्किउ वेई, ज़ियाओज़े रेन, ज़िआओगुआंग ली, वेनयोंग हुआंग, यी लियाओ, याशेंग वांग, जियाशू लिन, शिन जियांग, जिओ चेन और कुन लियू द्वारा।
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (फ्रॉम मेटा) साथ में पेपर [नो लैंग्वेज लेफ्ट बिहाइंड: स्केलिंग ह्यूमन-सेंटेड मशीन ट्रांसलेशन] (https://arxiv.org/abs/2207.04672) एनएलएलबी टीम द्वारा प्रकाशित।
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (विस्कॉन्सिन विश्वविद्यालय - मैडिसन से) साथ में कागज [Nyströmformer: A Nyström- आधारित एल्गोरिथम आत्म-ध्यान का अनुमान लगाने के लिए ](https://arxiv.org/abs/2102.03902) युनयांग ज़िओंग, झानपेंग ज़ेंग, रुद्रसिस चक्रवर्ती, मिंगक्सिंग टैन, ग्लेन फंग, यिन ली, विकास सिंह द्वारा पोस्ट किया गया।
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (Google AI से) साथ में कागज [विज़न ट्रांसफॉर्मर्स के साथ सिंपल ओपन-वोकैबुलरी ऑब्जेक्ट डिटेक्शन](https:/ /arxiv.org/abs/2205.06230) मैथियास मिंडरर, एलेक्सी ग्रिट्सेंको, ऑस्टिन स्टोन, मैक्सिम न्यूमैन, डिर्क वीसेनबोर्न, एलेक्सी डोसोवित्स्की, अरविंद महेंद्रन, अनुराग अर्नब, मुस्तफा देहघानी, ज़ुओरन शेन, जिओ वांग, ज़ियाओहुआ झाई, थॉमस किफ़, और नील हॉल्सबी द्वारा पोस्ट किया गया।
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (Google की ओर से) साथ में दिया गया पेपर [लंबे इनपुट सारांश के लिए ट्रांसफ़ॉर्मरों को बेहतर तरीके से एक्सटेंड करना](https://arxiv .org/abs/2208.04347) जेसन फांग, याओ झाओ, पीटर जे लियू द्वारा।
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (दीपमाइंड से) साथ में पेपर [पर्सीवर आईओ: संरचित इनपुट और आउटपुट के लिए एक सामान्य वास्तुकला] (https://arxiv.org/abs/2107.14795) एंड्रयू जेगल, सेबेस्टियन बोरग्यूड, जीन-बैप्टिस्ट अलायराक, कार्ल डोर्श, कैटलिन इओनेस्कु, डेविड द्वारा डिंग, स्कंद कोप्पुला, डैनियल ज़ोरान, एंड्रयू ब्रॉक, इवान शेलहैमर, ओलिवियर हेनाफ, मैथ्यू एम। बोट्विनिक, एंड्रयू ज़िसरमैन, ओरिओल विनियल्स, जोआओ कैरेरा द्वारा पोस्ट किया गया।
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (VinAI Research से) कागज के साथ [PhoBERT: वियतनामी के लिए पूर्व-प्रशिक्षित भाषा मॉडल](https://www .aclweb.org/anthology/2020.findings-emnlp.92/) डैट क्वोक गुयेन और अन्ह तुआन गुयेन द्वारा पोस्ट किया गया।
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP से) साथ वाला पेपर [प्रोग्राम अंडरस्टैंडिंग एंड जेनरेशन के लिए यूनिफाइड प्री-ट्रेनिंग](https://arxiv .org/abs/2103.06333) वसी उद्दीन अहमद, सैकत चक्रवर्ती, बैशाखी रे, काई-वेई चांग द्वारा।
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [ProphetNet: प्रेडिक्टिंग फ्यूचर एन-ग्राम फॉर सीक्वेंस-टू-सीक्वेंस प्री-ट्रेनिंग ](https://arxiv.org/abs/2001.04063) यू यान, वीज़ेन क्यूई, येयुन गोंग, दयाहेंग लियू, नान डुआन, जिउशेंग चेन, रुओफ़ेई झांग और मिंग झोउ द्वारा पोस्ट किया गया।
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (NVIDIA से) साथ वाला पेपर [डीप लर्निंग इंफ़ेक्शन के लिए इंटीजर क्वांटिज़ेशन: प्रिंसिपल्स एंड एम्पिरिकल इवैल्यूएशन](https:// arxiv.org/abs/2004.09602) हाओ वू, पैट्रिक जुड, जिआओजी झांग, मिखाइल इसेव और पॉलियस माइकेविसियस द्वारा।
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (फेसबुक से) साथ में कागज [रिट्रीवल-ऑगमेंटेड जेनरेशन फॉर नॉलेज-इंटेंसिव एनएलपी टास्क](https://arxiv .org/abs/2005.11401) पैट्रिक लुईस, एथन पेरेज़, अलेक्जेंड्रा पिक्टस, फैबियो पेट्रोनी, व्लादिमीर कारपुखिन, नमन गोयल, हेनरिक कुटलर, माइक लुईस, वेन-ताउ यिह, टिम रॉकटाशेल, सेबस्टियन रिडेल, डौवे कीला द्वारा।
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (Google अनुसंधान से) केल्विन गु, केंटन ली, ज़ोरा तुंग, पानुपोंग पसुपत और मिंग-वेई चांग द्वारा साथ में दिया गया पेपर [REALM: रिट्रीवल-ऑगमेंटेड लैंग्वेज मॉडल प्री-ट्रेनिंग](https://arxiv.org/abs/2002.08909)।
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (META रिसर्च से) [डिज़ाइनिंग नेटवर्क डिज़ाइन स्पेस] (https://arxiv.org/) पेपर के साथ जारी किया गया एब्स/2003.13678) इलिजा राडोसावोविक, राज प्रतीक कोसाराजू, रॉस गिर्शिक, कैमिंग ही, पिओटर डॉलर द्वारा।
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (गूगल रिसर्च से) साथ वाला पेपर [पूर्व-प्रशिक्षित भाषा मॉडल में एम्बेडिंग कपलिंग पर पुनर्विचार](https://arxiv .org/pdf/2010.12821.pdf) ह्युंग वोन चुंग, थिबॉल्ट फ़ेवरी, हेनरी त्साई, एम. जॉनसन, सेबेस्टियन रुडर द्वारा।
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (माइक्रोसॉफ्ट रिसर्च से) [डीप रेसिडुअल लर्निंग फॉर इमेज रिकग्निशन] (https://arxiv. org/abs/1512.03385) कैमिंग हे, जियांग्यु झांग, शाओकिंग रेन, जियान सन द्वारा।
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (फेसबुक से), साथ में कागज [मजबूत रूप से अनुकूलित BERT प्रीट्रेनिंग दृष्टिकोण](https://arxiv.org/abs /1907.11692) यिनहान लियू, मायल ओट, नमन गोयल, जिंगफेई डू, मंदार जोशी, डैनकी चेन, ओमर लेवी, माइक लुईस, ल्यूक ज़ेटलमॉयर, वेसेलिन स्टोयानोव द्वारा।
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (झुईई टेक्नोलॉजी से), साथ में पेपर [रोफॉर्मर: रोटरी पोजिशन एंबेडिंग के साथ एन्हांस्ड ट्रांसफॉर्मर] (https://arxiv.org/pdf/2104.09864v1.pdf) जियानलिन सु और यू लू और शेंगफेंग पैन और बो वेन और युनफेंग लियू द्वारा प्रकाशित।
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP से) साथ देने वाला पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स](https ://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योव आर्टज़ी द्वारा।
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (ASAPP से) साथ में पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स] (https://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योआव आर्टज़ी द्वारा पोस्ट किया गया।
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (फेसबुक से), साथ में पेपर [फेयरसेक S2T: फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग विद फेयरसेक](https: //arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया。
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (फेसबुक से) साथ में पेपर [लार्ज-स्केल सेल्फ- एंड सेमी-सुपरवाइज्ड लर्निंग फॉर स्पीच ट्रांसलेशन](https://arxiv.org/abs/2104.06678) चांगहान वांग, ऐनी वू, जुआन पिनो, एलेक्सी बेवस्की, माइकल औली, एलेक्सिस द्वारा Conneau द्वारा पोस्ट किया गया।
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (तेल अवीव यूनिवर्सिटी से) साथ में पेपर [स्पैन सिलेक्शन को प्री-ट्रेनिंग करके कुछ-शॉट क्वेश्चन आंसरिंग](https:// arxiv.org/abs/2101.00438) ओरि राम, युवल कर्स्टन, जोनाथन बेरेंट, अमीर ग्लोबर्सन, ओमर लेवी द्वारा।
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (बर्कले से) कागज के साथ [SqueezeBERT: कुशल तंत्रिका नेटवर्क के बारे में NLP को कंप्यूटर विज़न क्या सिखा सकता है?](https: //arxiv.org/abs/2006.11316) फॉरेस्ट एन. इनडोला, अल्बर्ट ई. शॉ, रवि कृष्णा, और कर्ट डब्ल्यू. केटज़र द्वारा।
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (माइक्रोसॉफ्ट से) साथ में कागज [स्वाइन ट्रांसफॉर्मर: शिफ्टेड विंडोज का उपयोग कर पदानुक्रमित विजन ट्रांसफॉर्मर](https://arxiv .org/abs/2103.14030) ज़ी लियू, युटोंग लिन, यू काओ, हान हू, यिक्सुआन वेई, झेंग झांग, स्टीफन लिन, बैनिंग गुओ द्वारा।
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (Microsoft से) साथ वाला पेपर [Swin Transformer V2: स्केलिंग अप कैपेसिटी एंड रेजोल्यूशन](https:// ज़ी लियू, हान हू, युटोंग लिन, ज़ुलिआंग याओ, ज़ेंडा ज़ी, यिक्सुआन वेई, जिया निंग, यू काओ, झेंग झांग, ली डोंग, फुरु वेई, बैनिंग गुओ द्वारा arxiv.org/abs/2111.09883।
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (来自 Google AI)कॉलिन रैफेल और नोम शज़ीर और एडम रॉबर्ट्स और कैथरीन ली और शरण नारंग और माइकल मटेना द्वारा साथ में पेपर [एक एकीकृत टेक्स्ट-टू-टेक्स्ट ट्रांसफॉर्मर के साथ स्थानांतरण सीखने की सीमा की खोज] (https://arxiv.org/abs/1910.10683) और यांकी झोउ और वेई ली और पीटर जे लियू।
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (Google AI से) साथ वाला पेपर [google-research/text-to-text-transfer- ट्रांसफॉर्मर](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) कॉलिन रैफेल और नोम शज़ीर और एडम रॉबर्ट्स और कैथरीन ली और शरण नारंग द्वारा और माइकल मटेना और यांकी झोउ और वेई ली और पीटर जे लियू।
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [पबटेबल्स-1एम: टूवर्ड्स कॉम्प्रिहेंसिव टेबल एक्सट्रैक्शन फ्रॉम अनस्ट्रक्चर्ड डॉक्यूमेंट्स ](https://arxiv.org/abs/2110.00061) ब्रैंडन स्मॉक, रोहित पेसाला, रॉबिन अब्राहम द्वारा पोस्ट किया गया।
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (Google AI से) साथ में कागज [TAPAS: पूर्व-प्रशिक्षण के माध्यम से कमजोर पर्यवेक्षण तालिका पार्सिंग](https:// arxiv.org/abs/2004.02349) जोनाथन हर्ज़िग, पावेल क्रिज़िस्तोफ़ नोवाक, थॉमस मुलर, फ्रांसेस्को पिकिन्नो और जूलियन मार्टिन ईसेन्च्लोस द्वारा।
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (माइक्रोसॉफ्ट रिसर्च से) साथ में पेपर [TAPEX: टेबल प्री-ट्रेनिंग थ्रू लर्निंग अ न्यूरल SQL एक्ज़ीक्यूटर](https: //arxiv.org/abs/2107.07653) कियान लियू, बेई चेन, जियाकी गुओ, मोर्टेज़ा ज़ियादी, ज़ेकी लिन, वीज़ू चेन, जियान-गुआंग लू द्वारा पोस्ट किया गया।
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU की ओर से) कागज के साथ [संस्करण-एक्स: एक ब्लॉग मॉडल चौकस चौक मॉडल मॉडल] (https://arxivorg/abs/1901.02860) क्वोकोक वी. ले, रुस्लैन सलाखुतदी
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (माइक्रोसॉफ्ट रिसर्च से) साथ में दिया गया पेपर [UniSpeech: यूनिफाइड स्पीच रिप्रेजेंटेशन लर्निंग विद लेबलेड एंड अनलेबल्ड डेटा](https:/ /arxiv.org/abs/2101.07597) चेंगई वांग, यू वू, याओ कियान, केनिची कुमातानी, शुजी लियू, फुरु वेई, माइकल ज़ेंग, ज़ुएदोंग हुआंग द्वारा।
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [UNISPEECH-SAT: यूनिवर्सल स्पीच रिप्रेजेंटेशन लर्निंग विद स्पीकर अवेयर प्री-ट्रेनिंग ](https://arxiv.org/abs/2110.05752) सानयुआन चेन, यू वू, चेंग्यी वांग, झेंगयांग चेन, झूओ चेन, शुजी लियू, जियान वू, याओ कियान, फुरु वेई, जिन्यु ली, जियांगज़ान यू द्वारा पोस्ट किया गया।
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (सिंघुआ यूनिवर्सिटी और ननकाई यूनिवर्सिटी से) साथ में पेपर [विजुअल अटेंशन नेटवर्क](https://arxiv.org/ pdf/2202.09741.pdf) मेंग-हाओ गुओ, चेंग-ज़े लू, झेंग-निंग लियू, मिंग-मिंग चेंग, शि-मिन हू द्वारा।
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (मल्टीमीडिया कम्प्यूटिंग ग्रुप, नानजिंग यूनिवर्सिटी से) साथ में पेपर [वीडियोएमएई: मास्क्ड ऑटोएन्कोडर स्व-पर्यवेक्षित वीडियो प्री-ट्रेनिंग के लिए डेटा-कुशल सीखने वाले हैं] (https://arxiv.org/abs/2203.12602) ज़ान टोंग, यिबिंग सॉन्ग, जुए द्वारा वांग, लिमिन वांग द्वारा पोस्ट किया गया।
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (NAVER AI Lab/Kakao Enterprise/Kakao Brain से) साथ में कागज [ViLT: Vision-and-Language Transformer बिना कनवल्शन या रीजन सुपरविजन](https://arxiv.org/abs/2102.03334) वोनजे किम, बोक्यूंग सोन, इल्डू किम द्वारा पोस्ट किया गया।
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (गूगल एआई से) कागज के साथ [एक इमेज इज़ वर्थ 16x16 वर्ड्स: ट्रांसफॉर्मर्स फॉर इमेज रिकॉग्निशन एट स्केल](https://arxiv.org/abs/2010.11929) एलेक्सी डोसोवित्स्की, लुकास बेयर, अलेक्जेंडर कोलेसनिकोव, डिर्क वीसेनबोर्न, शियाओहुआ झाई, थॉमस अनटरथिनर, मुस्तफा देहघानी, मैथियास मिंडरर, जॉर्ज हेगोल्ड, सिल्वेन गेली, जैकब उस्ज़कोरेइट द्वारा हॉल्सबी द्वारा पोस्ट किया गया।
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (UCLA NLP से) साथ वाला पेपर [VisualBERT: A Simple and Performant Baseline for Vision and Language](https:/ /arxiv.org/pdf/1908.03557) लियुनियन हेरोल्ड ली, मार्क यात्स्कर, दा यिन, चो-जुई हसीह, काई-वेई चांग द्वारा।
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (मेटा एआई से) साथ में कागज [मास्कड ऑटोएन्कोडर स्केलेबल विजन लर्नर्स हैं](https://arxiv.org/ एब्स/2111.06377) कैमिंग हे, ज़िनेली चेन, सेनिंग ज़ी, यांगहो ली, पिओट्र डॉलर, रॉस गिर्शिक द्वारा।
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (मेटा एआई से) साथ में कागज [लेबल-कुशल सीखने के लिए मास्क्ड स्याम देश के नेटवर्क](https://arxiv. org/abs/2204.07141) महमूद असरान, मथिल्डे कैरन, ईशान मिश्रा, पियोट्र बोजानोवस्की, फ्लोरियन बोर्डेस, पास्कल विंसेंट, आर्मंड जौलिन, माइकल रब्बत, निकोलस बल्लास द्वारा।
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (फेसबुक एआई से) साथ में पेपर [wav2vec 2.0: ए फ्रेमवर्क फॉर सेल्फ-सुपरवाइज्ड लर्निंग ऑफ स्पीच रिप्रेजेंटेशन] (https://arxiv.org/abs/2006.11477) एलेक्सी बेवस्की, हेनरी झोउ, अब्देलरहमान मोहम्मद, माइकल औली द्वारा।
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (Facebook AI से) साथ वाला पेपर [FAIRSEQ S2T: FAIRSEQ के साथ फास्ट स्पीच-टू-टेक्स्ट मॉडलिंग ](https://arxiv.org/abs/2010.05171) चांगहान वांग, यूं तांग, जुताई मा, ऐनी वू, सरव्या पोपुरी, दिमित्रो ओखोनको, जुआन पिनो द्वारा पोस्ट किया गया।
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (Facebook AI से) साथ वाला पेपर [सरल और प्रभावी जीरो-शॉट क्रॉस-लिंगुअल फोनेम रिकॉग्निशन](https:/ /arxiv.org/abs/2109.11680) कियानटोंग जू, एलेक्सी बाएव्स्की, माइकल औली द्वारा।
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (माइक्रोसॉफ्ट रिसर्च से) पेपर के साथ जारी किया गया [WavLM: फुल स्टैक के लिए बड़े पैमाने पर स्व-पर्यवेक्षित पूर्व-प्रशिक्षण स्पीच प्रोसेसिंग] (https://arxiv.org/abs/2110.13900) सानयुआन चेन, चेंगयी वांग, झेंगयांग चेन, यू वू, शुजी लियू, ज़ुओ चेन, जिन्यु ली, नाओयुकी कांडा, ताकुया योशियोका, ज़िओंग जिओ, जियान वू, लॉन्ग झोउ, शुओ रेन, यानमिन कियान, याओ कियान, जियान वू, माइकल ज़ेंग, फुरु वेई।
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (OpenAI से) साथ में कागज [बड़े पैमाने पर कमजोर पर्यवेक्षण के माध्यम से मजबूत भाषण पहचान](https://cdn. openai.com/papers/whisper.pdf) एलेक रैडफोर्ड, जोंग वूक किम, ताओ जू, ग्रेग ब्रॉकमैन, क्रिस्टीन मैकलीवे, इल्या सुत्स्केवर द्वारा।
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [एक्सपैंडिंग लैंग्वेज-इमेज प्रीट्रेन्ड मॉडल फॉर जनरल वीडियो रिकग्निशन](https: //arxiv.org/abs/2208.02816) बोलिन नी, होउवेन पेंग, मिंगाओ चेन, सोंगयांग झांग, गाओफेंग मेंग, जियानलोंग फू, शिमिंग जियांग, हैबिन लिंग द्वारा।
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (फेसबुक से) साथ में पेपर [क्रॉस-लिंगुअल लैंग्वेज मॉडल प्रीट्रेनिंग] (https://arxiv.org/abs/1901.07291) गिलाउम लैम्पल और एलेक्सिस कोनो द्वारा।
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (माइक्रोसॉफ्ट रिसर्च से) साथ में कागज [ProphetNet: प्रेडिक्टिंग फ्यूचर एन-ग्राम फॉर सीक्वेंस-टू- सीक्वेंस प्री-ट्रेनिंग](https://arxiv.org/abs/2001.04063) यू यान, वीज़ेन क्यूई, येयुन गोंग, दयाहेंग लियू, नान डुआन, जिउशेंग चेन, रुओफ़ेई झांग और मिंग झोउ द्वारा।
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (फेसबुक एआई से), साथ में पेपर [अनसुपरवाइज्ड क्रॉस-लिंगुअल रिप्रेजेंटेशन लर्निंग एट स्केल] (https://arxiv.org/abs/1911.02116) एलेक्सिस कोन्यू*, कार्तिकेय खंडेलवाल*, नमन गोयल, विश्रव चौधरी, गिलाउम वेनज़ेक, फ्रांसिस्को गुज़मैन द्वारा , एडौर्ड ग्रेव, मायल ओट, ल्यूक ज़ेटलमॉयर और वेसेलिन स्टोयानोव द्वारा।
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (Facebook AI से) साथ में कागज [बहुभाषी नकाबपोश भाषा के लिए बड़े पैमाने पर ट्रांसफॉर्मर ] मॉडलिंग](https://arxiv.org/abs/2105.00572) नमन गोयल, जिंगफेई डू, मायल ओट, गिरि अनंतरामन, एलेक्सिस कोनो द्वारा पोस्ट किया गया।
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (Google/CMU से) साथ वाला पेपर [XLNet: जनरलाइज्ड ऑटोरेग्रेसिव प्रीट्रेनिंग फॉर लैंग्वेज अंडरस्टैंडिंग](https://arxiv ज़ीलिन यांग*, ज़िहांग दाई*, यिमिंग यांग, जैम कार्बोनेल, रुस्लान सलाखुतदीनोव, क्वोक वी. ले द्वारा .org/abs/1906.08237)।
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (Facebook AI से) साथ वाला पेपर [XLS-R: सेल्फ सुपरवाइज्ड क्रॉस-लिंगुअल स्पीच रिप्रेजेंटेशन लर्निंग एट स्केल](https://arxiv.org/abs/2111.09296) अरुण बाबू, चांगहान वांग, एंड्रोस तजंद्रा, कुशाल लखोटिया, कियानटोंग जू, नमन गोयल, कृतिका सिंह, पैट्रिक वॉन प्लैटन, याथार्थ सराफ, जुआन पिनो, एलेक्सी बेवस्की, एलेक्सिस कोन्यू, माइकल औली द्वारा पोस्ट किया गया।
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (फेसबुक एआई से) साथ में पेपर [अनसुपरवाइज्ड क्रॉस-लिंगुअल रिप्रेजेंटेशन लर्निंग फॉर स्पीच रिकग्निशन] (https://arxiv.org/abs/2006.13979) एलेक्सिस कोन्यू, एलेक्सी बेवस्की, रोनन कोलोबर्ट, अब्देलरहमान मोहम्मद, माइकल औली द्वारा।
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (हुआझोंग यूनिवर्सिटी ऑफ साइंस एंड टेक्नोलॉजी से) साथ में पेपर [यू ओनली लुक एट वन सीक्वेंस: रीथिंकिंग ट्रांसफॉर्मर इन विज़न थ्रू ऑब्जेक्ट डिटेक्शन](https://arxiv.org/abs/2106.00666) युक्सिन फेंग, बेनचेंग लियाओ, जिंगगैंग वांग, जेमिन फेंग, जियांग क्यूई, रुई वू, जियानवेई नीयू, वेन्यू लियू द्वारा पोस्ट किया गया।
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (विस्कॉन्सिन विश्वविद्यालय - मैडिसन से) साथ में पेपर [यू ओनली सैंपल (लगभग) ज़ानपेंग ज़ेंग, युनयांग ज़िओंग द्वारा , सत्य एन. रवि, शैलेश आचार्य, ग्लेन फंग, विकास सिंह द्वारा पोस्ट किया गया।
1. एक नए मॉडल में योगदान देना चाहते हैं? नए मॉडल जोड़ने में आपका मार्गदर्शन करने के लिए हमारे पास एक **विस्तृत मार्गदर्शिका और टेम्प्लेट** है। आप उन्हें [`टेम्पलेट्स`](./templates) निर्देशिका में पा सकते हैं। पीआर शुरू करने से पहले [योगदान दिशानिर्देश] (./CONTRIBUTING.md) देखना और अनुरक्षकों से संपर्क करना या प्रतिक्रिया प्राप्त करने के लिए एक नया मुद्दा खोलना याद रखें।
यह जांचने के लिए कि क्या किसी मॉडल में पहले से ही Flax, PyTorch या TensorFlow का कार्यान्वयन है, या यदि उसके पास Tokenizers लाइब्रेरी में संबंधित टोकन है, तो [यह तालिका] (https://huggingface.co/ docs/transformers/index#supported) देखें। -फ्रेमवर्क)।
इन कार्यान्वयनों का परीक्षण कई डेटासेट पर किया गया है (देखें केस स्क्रिप्ट का उपयोग करें) और वैनिला कार्यान्वयन के लिए तुलनात्मक रूप से प्रदर्शन करना चाहिए। आप उपयोग के मामले के दस्तावेज़ [इस अनुभाग](https://huggingface.co/docs/transformers/examples) में व्यवहार का विवरण पढ़ सकते हैं।
## अधिक समझें
|अध्याय | विवरण |
|-|-|
| [दस्तावेज़ीकरण](https://huggingface.co/transformers/) | पूरा एपीआई दस्तावेज़ीकरण और ट्यूटोरियल |
| [कार्य सारांश](https://huggingface.co/docs/transformers/task_summary) | ट्रांसफॉर्मर समर्थित कार्य |
| [प्रीप्रोसेसिंग ट्यूटोरियल](https://huggingface.co/docs/transformers/preprocessing) | मॉडल के लिए डेटा तैयार करने के लिए `टोकनाइज़र` का उपयोग करना |
| [प्रशिक्षण और फाइन-ट्यूनिंग](https://huggingface.co/docs/transformers/training) | PyTorch/TensorFlow के ट्रेनिंग लूप या `ट्रेनर` API में ट्रांसफॉर्मर द्वारा दिए गए मॉडल का उपयोग करें |
| [क्विक स्टार्ट: ट्वीकिंग एंड यूज़ केस स्क्रिप्ट्स](https://github.com/huggingface/transformers/tree/main/examples) | विभिन्न कार्यों के लिए केस स्क्रिप्ट का उपयोग करें |
| [मॉडल साझा करना और अपलोड करना](https://huggingface.co/docs/transformers/model_sharing) | समुदाय के साथ अपने फाइन टूनड मॉडल अपलोड और साझा करें |
| [माइग्रेशन](https://huggingface.co/docs/transformers/migration) | `पाइटोरच-ट्रांसफॉर्मर्स` या `पाइटोरच-प्रीट्रेनड-बर्ट` से ट्रांसफॉर्मर में माइग्रेट करना |
## उद्धरण
हमने आधिकारिक तौर पर इस लाइब्रेरी का [पेपर](https://www.aclweb.org/anthology/2020.emnlp-demos.6/) प्रकाशित किया है, अगर आप ट्रान्सफ़ॉर्मर्स लाइब्रेरी का उपयोग करते हैं, तो कृपया उद्धृत करें:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README_ja.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Traditional Japanese translation of Hugging Face documentation
- Use square quotes, e.g.,「引用」
Dictionary
API: API(翻訳しない)
add: 追加
checkpoint: チェックポイント
code: コード
community: コミュニティ
confidence: 信頼度
dataset: データセット
documentation: ドキュメント
example: 例
finetune: 微調整
Hugging Face: Hugging Face(翻訳しない)
implementation: 実装
inference: 推論
library: ライブラリ
module: モジュール
NLP/Natural Language Processing: NLPと表示される場合は翻訳されず、Natural Language Processingと表示される場合は翻訳される
online demos: オンラインデモ
pipeline: pipeline(翻訳しない)
pretrained/pretrain: 学習済み
Python data structures (e.g., list, set, dict): リスト、セット、ディクショナリと訳され、括弧内は原文英語
repository: repository(翻訳しない)
summary: 概要
token-: token-(翻訳しない)
Trainer: Trainer(翻訳しない)
transformer: transformer(翻訳しない)
tutorial: チュートリアル
user: ユーザ
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<b>日本語</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>JAX、PyTorch、TensorFlowのための最先端機械学習</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗Transformersは、テキスト、視覚、音声などの異なるモダリティに対してタスクを実行するために、事前に学習させた数千のモデルを提供します。
これらのモデルは次のような場合に適用できます:
* 📝 テキストは、テキストの分類、情報抽出、質問応答、要約、翻訳、テキスト生成などのタスクのために、100以上の言語に対応しています。
* 🖼️ 画像分類、物体検出、セグメンテーションなどのタスクのための画像。
* 🗣️ 音声は、音声認識や音声分類などのタスクに使用します。
トランスフォーマーモデルは、テーブル質問応答、光学文字認識、スキャン文書からの情報抽出、ビデオ分類、視覚的質問応答など、**複数のモダリティを組み合わせた**タスクも実行可能です。
🤗Transformersは、与えられたテキストに対してそれらの事前学習されたモデルを素早くダウンロードして使用し、あなた自身のデータセットでそれらを微調整し、私たちの[model hub](https://huggingface.co/models)でコミュニティと共有するためのAPIを提供します。同時に、アーキテクチャを定義する各Pythonモジュールは完全にスタンドアロンであり、迅速な研究実験を可能にするために変更することができます。
🤗Transformersは[Jax](https://jax.readthedocs.io/en/latest/)、[PyTorch](https://pytorch.org/)、[TensorFlow](https://www.tensorflow.org/)という3大ディープラーニングライブラリーに支えられ、それぞれのライブラリをシームレスに統合しています。片方でモデルを学習してから、もう片方で推論用にロードするのは簡単なことです。
## オンラインデモ
[model hub](https://huggingface.co/models)から、ほとんどのモデルのページで直接テストすることができます。また、パブリックモデル、プライベートモデルに対して、[プライベートモデルのホスティング、バージョニング、推論API](https://huggingface.co/pricing)を提供しています。
以下はその一例です:
自然言語処理にて:
- [BERTによるマスクドワード補完](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Electraによる名前実体認識](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [GPT-2によるテキスト生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [RoBERTaによる自然言語推論](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [BARTによる要約](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [DistilBERTによる質問応答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [T5による翻訳](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
コンピュータビジョンにて:
- [ViTによる画像分類](https://huggingface.co/google/vit-base-patch16-224)
- [DETRによる物体検出](https://huggingface.co/facebook/detr-resnet-50)
- [SegFormerによるセマンティックセグメンテーション](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
- [DETRによるパノプティックセグメンテーション](https://huggingface.co/facebook/detr-resnet-50-panoptic)
オーディオにて:
- [Wav2Vec2による自動音声認識](https://huggingface.co/facebook/wav2vec2-base-960h)
- [Wav2Vec2によるキーワード検索](https://huggingface.co/superb/wav2vec2-base-superb-ks)
マルチモーダルなタスクにて:
- [ViLTによる視覚的質問応答](https://huggingface.co/dandelin/vilt-b32-finetuned-vqa)
Hugging Faceチームによって作られた **[トランスフォーマーを使った書き込み](https://transformer.huggingface.co)** は、このリポジトリのテキスト生成機能の公式デモである。
## Hugging Faceチームによるカスタム・サポートをご希望の場合
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## クイックツアー
与えられた入力（テキスト、画像、音声、...）に対してすぐにモデルを使うために、我々は`pipeline`というAPIを提供しております。pipelineは、学習済みのモデルと、そのモデルの学習時に使用された前処理をグループ化したものです。以下は、肯定的なテキストと否定的なテキストを分類するためにpipelineを使用する方法です:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
2行目のコードでは、pipelineで使用される事前学習済みモデルをダウンロードしてキャッシュし、3行目では与えられたテキストに対してそのモデルを評価します。ここでは、答えは99.97%の信頼度で「ポジティブ」です。
自然言語処理だけでなく、コンピュータビジョンや音声処理においても、多くのタスクにはあらかじめ訓練された`pipeline`が用意されている。例えば、画像から検出された物体を簡単に抽出することができる:
``` python
>>> import requests
>>> from PIL import Image
>>> from transformers import pipeline
# Download an image with cute cats
>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
>>> image_data = requests.get(url, stream=True).raw
>>> image = Image.open(image_data)
# Allocate a pipeline for object detection
>>> object_detector = pipeline('object-detection')
>>> object_detector(image)
[{'score': 0.9982201457023621,
'label': 'remote',
'box': {'xmin': 40, 'ymin': 70, 'xmax': 175, 'ymax': 117}},
{'score': 0.9960021376609802,
'label': 'remote',
'box': {'xmin': 333, 'ymin': 72, 'xmax': 368, 'ymax': 187}},
{'score': 0.9954745173454285,
'label': 'couch',
'box': {'xmin': 0, 'ymin': 1, 'xmax': 639, 'ymax': 473}},
{'score': 0.9988006353378296,
'label': 'cat',
'box': {'xmin': 13, 'ymin': 52, 'xmax': 314, 'ymax': 470}},
{'score': 0.9986783862113953,
'label': 'cat',
'box': {'xmin': 345, 'ymin': 23, 'xmax': 640, 'ymax': 368}}]
```
ここでは、画像から検出されたオブジェクトのリストが得られ、オブジェクトを囲むボックスと信頼度スコアが表示されます。左側が元画像、右側が予測結果を表示したものです:
<h3 align="center">
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png" width="400"></a>
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample_post_processed.png" width="400"></a>
</h3>
[このチュートリアル](https://huggingface.co/docs/transformers/task_summary)では、`pipeline`APIでサポートされているタスクについて詳しく説明しています。
`pipeline`に加えて、与えられたタスクに学習済みのモデルをダウンロードして使用するために必要なのは、3行のコードだけです。以下はPyTorchのバージョンです:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
And here is the equivalent code for TensorFlow:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
トークナイザは学習済みモデルが期待するすべての前処理を担当し、単一の文字列 (上記の例のように) またはリストに対して直接呼び出すことができます。これは下流のコードで使用できる辞書を出力します。また、単純に ** 引数展開演算子を使用してモデルに直接渡すこともできます。
モデル自体は通常の[Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) または [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) (バックエンドによって異なる)で、通常通り使用することが可能です。[このチュートリアル](https://huggingface.co/docs/transformers/training)では、このようなモデルを従来のPyTorchやTensorFlowの学習ループに統合する方法や、私たちの`Trainer`APIを使って新しいデータセットで素早く微調整を行う方法について説明します。
## なぜtransformersを使う必要があるのでしょうか？
1. 使いやすい最新モデル:
- 自然言語理解・生成、コンピュータビジョン、オーディオの各タスクで高いパフォーマンスを発揮します。
- 教育者、実務者にとっての低い参入障壁。
- 学習するクラスは3つだけで、ユーザが直面する抽象化はほとんどありません。
- 学習済みモデルを利用するための統一されたAPI。
1. 低い計算コスト、少ないカーボンフットプリント:
- 研究者は、常に再トレーニングを行うのではなく、トレーニングされたモデルを共有することができます。
- 実務家は、計算時間や生産コストを削減することができます。
- すべてのモダリティにおいて、60,000以上の事前学習済みモデルを持つ数多くのアーキテクチャを提供します。
1. モデルのライフタイムのあらゆる部分で適切なフレームワークを選択可能:
- 3行のコードで最先端のモデルをトレーニング。
- TF2.0/PyTorch/JAXフレームワーク間で1つのモデルを自在に移動させる。
- 学習、評価、生産に適したフレームワークをシームレスに選択できます。
1. モデルやサンプルをニーズに合わせて簡単にカスタマイズ可能:
- 原著者が発表した結果を再現するために、各アーキテクチャの例を提供しています。
- モデル内部は可能な限り一貫して公開されています。
- モデルファイルはライブラリとは独立して利用することができ、迅速な実験が可能です。
## なぜtransformersを使ってはいけないのでしょうか？
- このライブラリは、ニューラルネットのためのビルディングブロックのモジュール式ツールボックスではありません。モデルファイルのコードは、研究者が追加の抽象化/ファイルに飛び込むことなく、各モデルを素早く反復できるように、意図的に追加の抽象化でリファクタリングされていません。
- 学習APIはどのようなモデルでも動作するわけではなく、ライブラリが提供するモデルで動作するように最適化されています。一般的な機械学習のループには、別のライブラリ(おそらく[Accelerate](https://huggingface.co/docs/accelerate))を使用する必要があります。
- 私たちはできるだけ多くの使用例を紹介するよう努力していますが、[examples フォルダ](https://github.com/huggingface/transformers/tree/main/examples) にあるスクリプトはあくまで例です。あなたの特定の問題に対してすぐに動作するわけではなく、あなたのニーズに合わせるために数行のコードを変更する必要があることが予想されます。
## インストール
### pipにて
このリポジトリは、Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+, TensorFlow 2.3+ でテストされています。
🤗Transformersは[仮想環境](https://docs.python.org/3/library/venv.html)にインストールする必要があります。Pythonの仮想環境に慣れていない場合は、[ユーザーガイド](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)を確認してください。
まず、使用するバージョンのPythonで仮想環境を作成し、アクティベートします。
その後、Flax, PyTorch, TensorFlowのうち少なくとも1つをインストールする必要があります。
[TensorFlowインストールページ](https://www.tensorflow.org/install/)、[PyTorchインストールページ](https://pytorch.org/get-started/locally/#start-locally)、[Flax](https://github.com/google/flax#quick-install)、[Jax](https://github.com/google/jax#installation)インストールページで、お使いのプラットフォーム別のインストールコマンドを参照してください。
これらのバックエンドのいずれかがインストールされている場合、🤗Transformersは以下のようにpipを使用してインストールすることができます:
```bash
pip install transformers
```
もしサンプルを試したい、またはコードの最先端が必要で、新しいリリースを待てない場合は、[ライブラリをソースからインストール](https://huggingface.co/docs/transformers/installation#installing-from-source)する必要があります。
### condaにて
Transformersバージョン4.0.0から、condaチャンネルを搭載しました: `huggingface`。
🤗Transformersは以下のようにcondaを使って設置することができます:
```shell script
conda install -c huggingface transformers
```
Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それぞれのインストールページに従ってください。
> **_注意:_** Windowsでは、キャッシュの恩恵を受けるために、デベロッパーモードを有効にするよう促されることがあります。このような場合は、[このissue](https://github.com/huggingface/huggingface_hub/issues/1062)でお知らせください。
## モデルアーキテクチャ
🤗Transformersが提供する **[全モデルチェックポイント](https://huggingface.co/models)** は、[ユーザー](https://huggingface.co/users)や[組織](https://huggingface.co/organizations)によって直接アップロードされるhuggingface.co [model hub](https://huggingface.co)からシームレスに統合されています。
現在のチェックポイント数: 
🤗Transformersは現在、以下のアーキテクチャを提供しています（それぞれのハイレベルな要約は[こちら](https://huggingface.co/docs/transformers/model_summary)を参照してください）:
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. 新しいモデルを投稿したいですか？新しいモデルを追加するためのガイドとして、**詳細なガイドとテンプレート**が追加されました。これらはリポジトリの[`templates`](./templates)フォルダにあります。PRを始める前に、必ず[コントリビューションガイド](./CONTRIBUTING.md)を確認し、メンテナに連絡するか、フィードバックを収集するためにissueを開いてください。
各モデルがFlax、PyTorch、TensorFlowで実装されているか、🤗Tokenizersライブラリに支えられた関連トークナイザを持っているかは、[この表](https://huggingface.co/docs/transformers/index#supported-frameworks)を参照してください。
これらの実装はいくつかのデータセットでテストされており(サンプルスクリプトを参照)、オリジナルの実装の性能と一致するはずである。性能の詳細は[documentation](https://github.com/huggingface/transformers/tree/main/examples)のExamplesセクションで見ることができます。
## さらに詳しく
| セクション | 概要 |
|-|-|
| [ドキュメント](https://huggingface.co/docs/transformers/) | 完全なAPIドキュメントとチュートリアル |
| [タスク概要](https://huggingface.co/docs/transformers/task_summary) | 🤗Transformersがサポートするタスク |
| [前処理チュートリアル](https://huggingface.co/docs/transformers/preprocessing) | モデル用のデータを準備するために`Tokenizer`クラスを使用 |
| [トレーニングと微調整](https://huggingface.co/docs/transformers/training) | PyTorch/TensorFlowの学習ループと`Trainer`APIで🤗Transformersが提供するモデルを使用 |
| [クイックツアー: 微調整/使用方法スクリプト](https://github.com/huggingface/transformers/tree/main/examples) | 様々なタスクでモデルの微調整を行うためのスクリプト例 |
| [モデルの共有とアップロード](https://huggingface.co/docs/transformers/model_sharing) | 微調整したモデルをアップロードしてコミュニティで共有する |
| [マイグレーション](https://huggingface.co/docs/transformers/migration) | `pytorch-transformers`または`pytorch-pretrained-bert`から🤗Transformers に移行する |
## 引用
🤗 トランスフォーマーライブラリに引用できる[論文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)が出来ました:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Traditional Japanese translation of Hugging Face documentation
- Use square quotes, e.g.,「引用」
Dictionary
API: API(翻訳しない)
add: 追加
checkpoint: チェックポイント
code: コード
community: コミュニティ
confidence: 信頼度
dataset: データセット
documentation: ドキュメント
example: 例
finetune: 微調整
Hugging Face: Hugging Face(翻訳しない)
implementation: 実装
inference: 推論
library: ライブラリ
module: モジュール
NLP/Natural Language Processing: NLPと表示される場合は翻訳されず、Natural Language Processingと表示される場合は翻訳される
online demos: オンラインデモ
pipeline: pipeline(翻訳しない)
pretrained/pretrain: 学習済み
Python data structures (e.g., list, set, dict): リスト、セット、ディクショナリと訳され、括弧内は原文英語
repository: repository(翻訳しない)
summary: 概要
token-: token-(翻訳しない)
Trainer: Trainer(翻訳しない)
transformer: transformer(翻訳しない)
tutorial: チュートリアル
user: ユーザ
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<b>日本語</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>JAX、PyTorch、TensorFlowのための最先端機械学習</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗Transformersは、テキスト、視覚、音声などの異なるモダリティに対してタスクを実行するために、事前に学習させた数千のモデルを提供します。
これらのモデルは次のような場合に適用できます:
* 📝 テキストは、テキストの分類、情報抽出、質問応答、要約、翻訳、テキスト生成などのタスクのために、100以上の言語に対応しています。
* 🖼️ 画像分類、物体検出、セグメンテーションなどのタスクのための画像。
* 🗣️ 音声は、音声認識や音声分類などのタスクに使用します。
トランスフォーマーモデルは、テーブル質問応答、光学文字認識、スキャン文書からの情報抽出、ビデオ分類、視覚的質問応答など、**複数のモダリティを組み合わせた**タスクも実行可能です。
🤗Transformersは、与えられたテキストに対してそれらの事前学習されたモデルを素早くダウンロードして使用し、あなた自身のデータセットでそれらを微調整し、私たちの[model hub](https://huggingface.co/models)でコミュニティと共有するためのAPIを提供します。同時に、アーキテクチャを定義する各Pythonモジュールは完全にスタンドアロンであり、迅速な研究実験を可能にするために変更することができます。
🤗Transformersは[Jax](https://jax.readthedocs.io/en/latest/)、[PyTorch](https://pytorch.org/)、[TensorFlow](https://www.tensorflow.org/)という3大ディープラーニングライブラリーに支えられ、それぞれのライブラリをシームレスに統合しています。片方でモデルを学習してから、もう片方で推論用にロードするのは簡単なことです。
## オンラインデモ
[model hub](https://huggingface.co/models)から、ほとんどのモデルのページで直接テストすることができます。また、パブリックモデル、プライベートモデルに対して、[プライベートモデルのホスティング、バージョニング、推論API](https://huggingface.co/pricing)を提供しています。
以下はその一例です:
自然言語処理にて:
- [BERTによるマスクドワード補完](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Electraによる名前実体認識](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [GPT-2によるテキスト生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [RoBERTaによる自然言語推論](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [BARTによる要約](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [DistilBERTによる質問応答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [T5による翻訳](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
コンピュータビジョンにて:
- [ViTによる画像分類](https://huggingface.co/google/vit-base-patch16-224)
- [DETRによる物体検出](https://huggingface.co/facebook/detr-resnet-50)
- [SegFormerによるセマンティックセグメンテーション](https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512)
- [DETRによるパノプティックセグメンテーション](https://huggingface.co/facebook/detr-resnet-50-panoptic)
オーディオにて:
- [Wav2Vec2による自動音声認識](https://huggingface.co/facebook/wav2vec2-base-960h)
- [Wav2Vec2によるキーワード検索](https://huggingface.co/superb/wav2vec2-base-superb-ks)
マルチモーダルなタスクにて:
- [ViLTによる視覚的質問応答](https://huggingface.co/dandelin/vilt-b32-finetuned-vqa)
Hugging Faceチームによって作られた **[トランスフォーマーを使った書き込み](https://transformer.huggingface.co)** は、このリポジトリのテキスト生成機能の公式デモである。
## Hugging Faceチームによるカスタム・サポートをご希望の場合
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## クイックツアー
与えられた入力（テキスト、画像、音声、...）に対してすぐにモデルを使うために、我々は`pipeline`というAPIを提供しております。pipelineは、学習済みのモデルと、そのモデルの学習時に使用された前処理をグループ化したものです。以下は、肯定的なテキストと否定的なテキストを分類するためにpipelineを使用する方法です:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
2行目のコードでは、pipelineで使用される事前学習済みモデルをダウンロードしてキャッシュし、3行目では与えられたテキストに対してそのモデルを評価します。ここでは、答えは99.97%の信頼度で「ポジティブ」です。
自然言語処理だけでなく、コンピュータビジョンや音声処理においても、多くのタスクにはあらかじめ訓練された`pipeline`が用意されている。例えば、画像から検出された物体を簡単に抽出することができる:
``` python
>>> import requests
>>> from PIL import Image
>>> from transformers import pipeline
# Download an image with cute cats
>>> url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png"
>>> image_data = requests.get(url, stream=True).raw
>>> image = Image.open(image_data)
# Allocate a pipeline for object detection
>>> object_detector = pipeline('object-detection')
>>> object_detector(image)
[{'score': 0.9982201457023621,
'label': 'remote',
'box': {'xmin': 40, 'ymin': 70, 'xmax': 175, 'ymax': 117}},
{'score': 0.9960021376609802,
'label': 'remote',
'box': {'xmin': 333, 'ymin': 72, 'xmax': 368, 'ymax': 187}},
{'score': 0.9954745173454285,
'label': 'couch',
'box': {'xmin': 0, 'ymin': 1, 'xmax': 639, 'ymax': 473}},
{'score': 0.9988006353378296,
'label': 'cat',
'box': {'xmin': 13, 'ymin': 52, 'xmax': 314, 'ymax': 470}},
{'score': 0.9986783862113953,
'label': 'cat',
'box': {'xmin': 345, 'ymin': 23, 'xmax': 640, 'ymax': 368}}]
```
ここでは、画像から検出されたオブジェクトのリストが得られ、オブジェクトを囲むボックスと信頼度スコアが表示されます。左側が元画像、右側が予測結果を表示したものです:
<h3 align="center">
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample.png" width="400"></a>
<a><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/coco_sample_post_processed.png" width="400"></a>
</h3>
[このチュートリアル](https://huggingface.co/docs/transformers/task_summary)では、`pipeline`APIでサポートされているタスクについて詳しく説明しています。
`pipeline`に加えて、与えられたタスクに学習済みのモデルをダウンロードして使用するために必要なのは、3行のコードだけです。以下はPyTorchのバージョンです:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
And here is the equivalent code for TensorFlow:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
トークナイザは学習済みモデルが期待するすべての前処理を担当し、単一の文字列 (上記の例のように) またはリストに対して直接呼び出すことができます。これは下流のコードで使用できる辞書を出力します。また、単純に ** 引数展開演算子を使用してモデルに直接渡すこともできます。
モデル自体は通常の[Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) または [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model) (バックエンドによって異なる)で、通常通り使用することが可能です。[このチュートリアル](https://huggingface.co/docs/transformers/training)では、このようなモデルを従来のPyTorchやTensorFlowの学習ループに統合する方法や、私たちの`Trainer`APIを使って新しいデータセットで素早く微調整を行う方法について説明します。
## なぜtransformersを使う必要があるのでしょうか？
1. 使いやすい最新モデル:
- 自然言語理解・生成、コンピュータビジョン、オーディオの各タスクで高いパフォーマンスを発揮します。
- 教育者、実務者にとっての低い参入障壁。
- 学習するクラスは3つだけで、ユーザが直面する抽象化はほとんどありません。
- 学習済みモデルを利用するための統一されたAPI。
1. 低い計算コスト、少ないカーボンフットプリント:
- 研究者は、常に再トレーニングを行うのではなく、トレーニングされたモデルを共有することができます。
- 実務家は、計算時間や生産コストを削減することができます。
- すべてのモダリティにおいて、60,000以上の事前学習済みモデルを持つ数多くのアーキテクチャを提供します。
1. モデルのライフタイムのあらゆる部分で適切なフレームワークを選択可能:
- 3行のコードで最先端のモデルをトレーニング。
- TF2.0/PyTorch/JAXフレームワーク間で1つのモデルを自在に移動させる。
- 学習、評価、生産に適したフレームワークをシームレスに選択できます。
1. モデルやサンプルをニーズに合わせて簡単にカスタマイズ可能:
- 原著者が発表した結果を再現するために、各アーキテクチャの例を提供しています。
- モデル内部は可能な限り一貫して公開されています。
- モデルファイルはライブラリとは独立して利用することができ、迅速な実験が可能です。
## なぜtransformersを使ってはいけないのでしょうか？
- このライブラリは、ニューラルネットのためのビルディングブロックのモジュール式ツールボックスではありません。モデルファイルのコードは、研究者が追加の抽象化/ファイルに飛び込むことなく、各モデルを素早く反復できるように、意図的に追加の抽象化でリファクタリングされていません。
- 学習APIはどのようなモデルでも動作するわけではなく、ライブラリが提供するモデルで動作するように最適化されています。一般的な機械学習のループには、別のライブラリ(おそらく[Accelerate](https://huggingface.co/docs/accelerate))を使用する必要があります。
- 私たちはできるだけ多くの使用例を紹介するよう努力していますが、[examples フォルダ](https://github.com/huggingface/transformers/tree/main/examples) にあるスクリプトはあくまで例です。あなたの特定の問題に対してすぐに動作するわけではなく、あなたのニーズに合わせるために数行のコードを変更する必要があることが予想されます。
## インストール
### pipにて
このリポジトリは、Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+, TensorFlow 2.3+ でテストされています。
🤗Transformersは[仮想環境](https://docs.python.org/3/library/venv.html)にインストールする必要があります。Pythonの仮想環境に慣れていない場合は、[ユーザーガイド](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)を確認してください。
まず、使用するバージョンのPythonで仮想環境を作成し、アクティベートします。
その後、Flax, PyTorch, TensorFlowのうち少なくとも1つをインストールする必要があります。
[TensorFlowインストールページ](https://www.tensorflow.org/install/)、[PyTorchインストールページ](https://pytorch.org/get-started/locally/#start-locally)、[Flax](https://github.com/google/flax#quick-install)、[Jax](https://github.com/google/jax#installation)インストールページで、お使いのプラットフォーム別のインストールコマンドを参照してください。
これらのバックエンドのいずれかがインストールされている場合、🤗Transformersは以下のようにpipを使用してインストールすることができます:
```bash
pip install transformers
```
もしサンプルを試したい、またはコードの最先端が必要で、新しいリリースを待てない場合は、[ライブラリをソースからインストール](https://huggingface.co/docs/transformers/installation#installing-from-source)する必要があります。
### condaにて
Transformersバージョン4.0.0から、condaチャンネルを搭載しました: `huggingface`。
🤗Transformersは以下のようにcondaを使って設置することができます:
```shell script
conda install -c huggingface transformers
```
Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それぞれのインストールページに従ってください。
> **_注意:_** Windowsでは、キャッシュの恩恵を受けるために、デベロッパーモードを有効にするよう促されることがあります。このような場合は、[このissue](https://github.com/huggingface/huggingface_hub/issues/1062)でお知らせください。
## モデルアーキテクチャ
🤗Transformersが提供する **[全モデルチェックポイント](https://huggingface.co/models)** は、[ユーザー](https://huggingface.co/users)や[組織](https://huggingface.co/organizations)によって直接アップロードされるhuggingface.co [model hub](https://huggingface.co)からシームレスに統合されています。
現在のチェックポイント数: 
🤗Transformersは現在、以下のアーキテクチャを提供しています（それぞれのハイレベルな要約は[こちら](https://huggingface.co/docs/transformers/model_summary)を参照してください）:
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. 新しいモデルを投稿したいですか？新しいモデルを追加するためのガイドとして、**詳細なガイドとテンプレート**が追加されました。これらはリポジトリの[`templates`](./templates)フォルダにあります。PRを始める前に、必ず[コントリビューションガイド](./CONTRIBUTING.md)を確認し、メンテナに連絡するか、フィードバックを収集するためにissueを開いてください。
各モデルがFlax、PyTorch、TensorFlowで実装されているか、🤗Tokenizersライブラリに支えられた関連トークナイザを持っているかは、[この表](https://huggingface.co/docs/transformers/index#supported-frameworks)を参照してください。
これらの実装はいくつかのデータセットでテストされており(サンプルスクリプトを参照)、オリジナルの実装の性能と一致するはずである。性能の詳細は[documentation](https://github.com/huggingface/transformers/tree/main/examples)のExamplesセクションで見ることができます。
## さらに詳しく
| セクション | 概要 |
|-|-|
| [ドキュメント](https://huggingface.co/docs/transformers/) | 完全なAPIドキュメントとチュートリアル |
| [タスク概要](https://huggingface.co/docs/transformers/task_summary) | 🤗Transformersがサポートするタスク |
| [前処理チュートリアル](https://huggingface.co/docs/transformers/preprocessing) | モデル用のデータを準備するために`Tokenizer`クラスを使用 |
| [トレーニングと微調整](https://huggingface.co/docs/transformers/training) | PyTorch/TensorFlowの学習ループと`Trainer`APIで🤗Transformersが提供するモデルを使用 |
| [クイックツアー: 微調整/使用方法スクリプト](https://github.com/huggingface/transformers/tree/main/examples) | 様々なタスクでモデルの微調整を行うためのスクリプト例 |
| [モデルの共有とアップロード](https://huggingface.co/docs/transformers/model_sharing) | 微調整したモデルをアップロードしてコミュニティで共有する |
| [マイグレーション](https://huggingface.co/docs/transformers/migration) | `pytorch-transformers`または`pytorch-pretrained-bert`から🤗Transformers に移行する |
## 引用
🤗 トランスフォーマーライブラリに引用できる[論文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)が出来ました:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README_ko.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<b>한국어</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p> Jax, Pytorch, TensorFlow를 위한 최첨단 자연어처리</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers는 분류, 정보 추출, 질문 답변, 요약, 번역, 문장 생성 등을 100개 이상의 언어로 수행할 수 있는 수천개의 사전학습된 모델을 제공합니다. 우리의 목표는 모두가 최첨단의 NLP 기술을 쉽게 사용하는 것입니다.
🤗 Transformers는 이러한 사전학습 모델을 빠르게 다운로드해 특정 텍스트에 사용하고, 원하는 데이터로 fine-tuning해 커뮤니티나 우리의 [모델 허브](https://huggingface.co/models)에 공유할 수 있도록 API를 제공합니다. 또한, 모델 구조를 정의하는 각 파이썬 모듈은 완전히 독립적이여서 연구 실험을 위해 손쉽게 수정할 수 있습니다.
🤗 Transformers는 가장 유명한 3개의 딥러닝 라이브러리를 지원합니다. 이들은 서로 완벽히 연동됩니다 — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/), [TensorFlow](https://www.tensorflow.org/). 간단하게 이 라이브러리 중 하나로 모델을 학습하고, 또 다른 라이브러리로 추론을 위해 모델을 불러올 수 있습니다.
## 온라인 데모
대부분의 모델을 [모델 허브](https://huggingface.co/models) 페이지에서 바로 테스트해볼 수 있습니다. 공개 및 비공개 모델을 위한 [비공개 모델 호스팅, 버전 관리, 추론 API](https://huggingface.co/pricing)도 제공합니다.
예시:
- [BERT로 마스킹된 단어 완성하기](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Electra를 이용한 개체명 인식](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [GPT-2로 텍스트 생성하기](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [RoBERTa로 자연어 추론하기](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [BART를 이용한 요약](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [DistilBERT를 이용한 질문 답변](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [T5로 번역하기](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Transformer와 글쓰기](https://transformer.huggingface.co)** 는 이 저장소의 텍스트 생성 능력에 관한 Hugging Face 팀의 공식 데모입니다.
## Hugging Face 팀의 커스텀 지원을 원한다면
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 퀵 투어
원하는 텍스트에 바로 모델을 사용할 수 있도록, 우리는 `pipeline` API를 제공합니다. Pipeline은 사전학습 모델과 그 모델을 학습할 때 적용한 전처리 방식을 하나로 합칩니다. 다음은 긍정적인 텍스트와 부정적인 텍스트를 분류하기 위해 pipeline을 사용한 간단한 예시입니다:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
코드의 두번째 줄은 pipeline이 사용하는 사전학습 모델을 다운로드하고 캐시로 저장합니다. 세번째 줄에선 그 모델이 주어진 텍스트를 평가합니다. 여기서 모델은 99.97%의 확률로 텍스트가 긍정적이라고 평가했습니다.
많은 NLP 과제들을 `pipeline`으로 바로 수행할 수 있습니다. 예를 들어, 질문과 문맥이 주어지면 손쉽게 답변을 추출할 수 있습니다:
``` python
>>> from transformers import pipeline
# Allocate a pipeline for question-answering
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
답변뿐만 아니라, 여기에 사용된 사전학습 모델은 확신도와 토크나이즈된 문장 속 답변의 시작점, 끝점까지 반환합니다. [이 튜토리얼](https://huggingface.co/docs/transformers/task_summary)에서 `pipeline` API가 지원하는 다양한 과제를 확인할 수 있습니다.
코드 3줄로 원하는 과제에 맞게 사전학습 모델을 다운로드 받고 사용할 수 있습니다. 다음은 PyTorch 버전입니다:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
다음은 TensorFlow 버전입니다:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
토크나이저는 사전학습 모델의 모든 전처리를 책임집니다. 그리고 (위의 예시처럼) 1개의 스트링이나 리스트도 처리할 수 있습니다. 토크나이저는 딕셔너리를 반환하는데, 이는 다운스트림 코드에 사용하거나 언패킹 연산자 ** 를 이용해 모델에 바로 전달할 수도 있습니다.
모델 자체는 일반적으로 사용되는 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module)나 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)입니다. [이 튜토리얼](https://huggingface.co/transformers/training.html)은 이러한 모델을 표준적인 PyTorch나 TensorFlow 학습 과정에서 사용하는 방법, 또는 새로운 데이터로 fine-tune하기 위해 `Trainer` API를 사용하는 방법을 설명해줍니다.
## 왜 transformers를 사용해야 할까요?
1. 손쉽게 사용할 수 있는 최첨단 모델:
- NLU와 NLG 과제에서 뛰어난 성능을 보입니다.
- 교육자 실무자에게 진입 장벽이 낮습니다.
- 3개의 클래스만 배우면 바로 사용할 수 있습니다.
- 하나의 API로 모든 사전학습 모델을 사용할 수 있습니다.
1. 더 적은 계산 비용, 더 적은 탄소 발자국:
- 연구자들은 모델을 계속 다시 학습시키는 대신 학습된 모델을 공유할 수 있습니다.
- 실무자들은 학습에 필요한 시간과 비용을 절약할 수 있습니다.
- 수십개의 모델 구조, 2,000개 이상의 사전학습 모델, 100개 이상의 언어로 학습된 모델 등.
1. 모델의 각 생애주기에 적합한 프레임워크:
- 코드 3줄로 최첨단 모델을 학습하세요.
- 자유롭게 모델을 TF2.0나 PyTorch 프레임워크로 변환하세요.
- 학습, 평가, 공개 등 각 단계에 맞는 프레임워크를 원하는대로 선택하세요.
1. 필요한 대로 모델이나 예시를 커스터마이즈하세요:
- 우리는 저자가 공개한 결과를 재현하기 위해 각 모델 구조의 예시를 제공합니다.
- 모델 내부 구조는 가능한 일관적으로 공개되어 있습니다.
- 빠른 실험을 위해 모델 파일은 라이브러리와 독립적으로 사용될 수 있습니다.
## 왜 transformers를 사용하지 말아야 할까요?
- 이 라이브러리는 신경망 블록을 만들기 위한 모듈이 아닙니다. 연구자들이 여러 파일을 살펴보지 않고 바로 각 모델을 사용할 수 있도록, 모델 파일 코드의 추상화 수준을 적정하게 유지했습니다.
- 학습 API는 모든 모델에 적용할 수 있도록 만들어지진 않았지만, 라이브러리가 제공하는 모델들에 적용할 수 있도록 최적화되었습니다. 일반적인 머신 러닝을 위해선, 다른 라이브러리를 사용하세요.
- 가능한 많은 사용 예시를 보여드리고 싶어서, [예시 폴더](https://github.com/huggingface/transformers/tree/main/examples)의 스크립트를 준비했습니다. 이 스크립트들을 수정 없이 특정한 문제에 바로 적용하지 못할 수 있습니다. 필요에 맞게 일부 코드를 수정해야 할 수 있습니다.
## 설치
### pip로 설치하기
이 저장소는 Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+, TensorFlow 2.3+에서 테스트 되었습니다.
[가상 환경](https://docs.python.org/3/library/venv.html)에 🤗 Transformers를 설치하세요. Python 가상 환경에 익숙하지 않다면, [사용자 가이드](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)를 확인하세요.
우선, 사용할 Python 버전으로 가상 환경을 만들고 실행하세요.
그 다음, Flax, PyTorch, TensorFlow 중 적어도 하나는 설치해야 합니다.
플랫폼에 맞는 설치 명령어를 확인하기 위해 [TensorFlow 설치 페이지](https://www.tensorflow.org/install/), [PyTorch 설치 페이지](https://pytorch.org/get-started/locally/#start-locally), [Flax 설치 페이지](https://github.com/google/flax#quick-install)를 확인하세요.
이들 중 적어도 하나가 설치되었다면, 🤗 Transformers는 다음과 같이 pip을 이용해 설치할 수 있습니다:
```bash
pip install transformers
```
예시들을 체험해보고 싶거나, 최최최첨단 코드를 원하거나, 새로운 버전이 나올 때까지 기다릴 수 없다면 [라이브러리를 소스에서 바로 설치](https://huggingface.co/docs/transformers/installation#installing-from-source)하셔야 합니다.
### conda로 설치하기
Transformers 버전 v4.0.0부터, conda 채널이 생겼습니다: `huggingface`.
🤗 Transformers는 다음과 같이 conda로 설치할 수 있습니다:
```shell script
conda install -c huggingface transformers
```
Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는 방법을 확인하세요.
## 모델 구조
**🤗 Transformers가 제공하는 [모든 모델 체크포인트](https://huggingface.co/models)** 는 huggingface.co [모델 허브](https://huggingface.co)에 완벽히 연동되어 있습니다. [개인](https://huggingface.co/users)과 [기관](https://huggingface.co/organizations)이 모델 허브에 직접 업로드할 수 있습니다.
현재 사용 가능한 모델 체크포인트의 개수: 
🤗 Transformers는 다음 모델들을 제공합니다 (각 모델의 요약은 [여기](https://huggingface.co/docs/transformers/model_summary)서 확인하세요):
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER) released with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper a [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI) released with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. 새로운 모델을 올리고 싶나요? 우리가 **상세한 가이드와 템플릿** 으로 새로운 모델을 올리도록 도와드릴게요. 가이드와 템플릿은 이 저장소의 [`templates`](./templates) 폴더에서 확인하실 수 있습니다. [컨트리뷰션 가이드라인](./CONTRIBUTING.md)을 꼭 확인해주시고, PR을 올리기 전에 메인테이너에게 연락하거나 이슈를 오픈해 피드백을 받으시길 바랍니다.
각 모델이 Flax, PyTorch, TensorFlow으로 구현되었는지 또는 🤗 Tokenizers 라이브러리가 지원하는 토크나이저를 사용하는지 확인하려면, [이 표](https://huggingface.co/docs/transformers/index#supported-frameworks)를 확인하세요.
이 구현은 여러 데이터로 검증되었고 (예시 스크립트를 참고하세요) 오리지널 구현의 성능과 같아야 합니다. [도큐먼트](https://huggingface.co/docs/transformers/examples)의 Examples 섹션에서 성능에 대한 자세한 설명을 확인할 수 있습니다.
## 더 알아보기
| 섹션 | 설명 |
|-|-|
| [도큐먼트](https://huggingface.co/transformers/) | 전체 API 도큐먼트와 튜토리얼 |
| [과제 요약](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers가 지원하는 과제들 |
| [전처리 튜토리얼](https://huggingface.co/docs/transformers/preprocessing) | `Tokenizer` 클래스를 이용해 모델을 위한 데이터 준비하기 |
| [학습과 fine-tuning](https://huggingface.co/docs/transformers/training) | 🤗 Transformers가 제공하는 모델 PyTorch/TensorFlow 학습 과정과 `Trainer` API에서 사용하기 |
| [퀵 투어: Fine-tuning/사용 스크립트](https://github.com/huggingface/transformers/tree/main/examples) | 다양한 과제에서 모델 fine-tuning하는 예시 스크립트 |
| [모델 공유 및 업로드](https://huggingface.co/docs/transformers/model_sharing) | 커뮤니티에 fine-tune된 모델을 업로드 및 공유하기 |
| [마이그레이션](https://huggingface.co/docs/transformers/migration) | `pytorch-transformers`나 `pytorch-pretrained-bert`에서 🤗 Transformers로 이동하기|
## 인용
🤗 Transformers 라이브러리를 인용하고 싶다면, 이 [논문](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)을 인용해 주세요:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<b>한국어</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p> Jax, Pytorch, TensorFlow를 위한 최첨단 자연어처리</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers는 분류, 정보 추출, 질문 답변, 요약, 번역, 문장 생성 등을 100개 이상의 언어로 수행할 수 있는 수천개의 사전학습된 모델을 제공합니다. 우리의 목표는 모두가 최첨단의 NLP 기술을 쉽게 사용하는 것입니다.
🤗 Transformers는 이러한 사전학습 모델을 빠르게 다운로드해 특정 텍스트에 사용하고, 원하는 데이터로 fine-tuning해 커뮤니티나 우리의 [모델 허브](https://huggingface.co/models)에 공유할 수 있도록 API를 제공합니다. 또한, 모델 구조를 정의하는 각 파이썬 모듈은 완전히 독립적이여서 연구 실험을 위해 손쉽게 수정할 수 있습니다.
🤗 Transformers는 가장 유명한 3개의 딥러닝 라이브러리를 지원합니다. 이들은 서로 완벽히 연동됩니다 — [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/), [TensorFlow](https://www.tensorflow.org/). 간단하게 이 라이브러리 중 하나로 모델을 학습하고, 또 다른 라이브러리로 추론을 위해 모델을 불러올 수 있습니다.
## 온라인 데모
대부분의 모델을 [모델 허브](https://huggingface.co/models) 페이지에서 바로 테스트해볼 수 있습니다. 공개 및 비공개 모델을 위한 [비공개 모델 호스팅, 버전 관리, 추론 API](https://huggingface.co/pricing)도 제공합니다.
예시:
- [BERT로 마스킹된 단어 완성하기](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [Electra를 이용한 개체명 인식](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [GPT-2로 텍스트 생성하기](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [RoBERTa로 자연어 추론하기](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [BART를 이용한 요약](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [DistilBERT를 이용한 질문 답변](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [T5로 번역하기](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Transformer와 글쓰기](https://transformer.huggingface.co)** 는 이 저장소의 텍스트 생성 능력에 관한 Hugging Face 팀의 공식 데모입니다.
## Hugging Face 팀의 커스텀 지원을 원한다면
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 퀵 투어
원하는 텍스트에 바로 모델을 사용할 수 있도록, 우리는 `pipeline` API를 제공합니다. Pipeline은 사전학습 모델과 그 모델을 학습할 때 적용한 전처리 방식을 하나로 합칩니다. 다음은 긍정적인 텍스트와 부정적인 텍스트를 분류하기 위해 pipeline을 사용한 간단한 예시입니다:
```python
>>> from transformers import pipeline
# Allocate a pipeline for sentiment-analysis
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
코드의 두번째 줄은 pipeline이 사용하는 사전학습 모델을 다운로드하고 캐시로 저장합니다. 세번째 줄에선 그 모델이 주어진 텍스트를 평가합니다. 여기서 모델은 99.97%의 확률로 텍스트가 긍정적이라고 평가했습니다.
많은 NLP 과제들을 `pipeline`으로 바로 수행할 수 있습니다. 예를 들어, 질문과 문맥이 주어지면 손쉽게 답변을 추출할 수 있습니다:
``` python
>>> from transformers import pipeline
# Allocate a pipeline for question-answering
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
답변뿐만 아니라, 여기에 사용된 사전학습 모델은 확신도와 토크나이즈된 문장 속 답변의 시작점, 끝점까지 반환합니다. [이 튜토리얼](https://huggingface.co/docs/transformers/task_summary)에서 `pipeline` API가 지원하는 다양한 과제를 확인할 수 있습니다.
코드 3줄로 원하는 과제에 맞게 사전학습 모델을 다운로드 받고 사용할 수 있습니다. 다음은 PyTorch 버전입니다:
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
다음은 TensorFlow 버전입니다:
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
토크나이저는 사전학습 모델의 모든 전처리를 책임집니다. 그리고 (위의 예시처럼) 1개의 스트링이나 리스트도 처리할 수 있습니다. 토크나이저는 딕셔너리를 반환하는데, 이는 다운스트림 코드에 사용하거나 언패킹 연산자 ** 를 이용해 모델에 바로 전달할 수도 있습니다.
모델 자체는 일반적으로 사용되는 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module)나 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)입니다. [이 튜토리얼](https://huggingface.co/transformers/training.html)은 이러한 모델을 표준적인 PyTorch나 TensorFlow 학습 과정에서 사용하는 방법, 또는 새로운 데이터로 fine-tune하기 위해 `Trainer` API를 사용하는 방법을 설명해줍니다.
## 왜 transformers를 사용해야 할까요?
1. 손쉽게 사용할 수 있는 최첨단 모델:
- NLU와 NLG 과제에서 뛰어난 성능을 보입니다.
- 교육자 실무자에게 진입 장벽이 낮습니다.
- 3개의 클래스만 배우면 바로 사용할 수 있습니다.
- 하나의 API로 모든 사전학습 모델을 사용할 수 있습니다.
1. 더 적은 계산 비용, 더 적은 탄소 발자국:
- 연구자들은 모델을 계속 다시 학습시키는 대신 학습된 모델을 공유할 수 있습니다.
- 실무자들은 학습에 필요한 시간과 비용을 절약할 수 있습니다.
- 수십개의 모델 구조, 2,000개 이상의 사전학습 모델, 100개 이상의 언어로 학습된 모델 등.
1. 모델의 각 생애주기에 적합한 프레임워크:
- 코드 3줄로 최첨단 모델을 학습하세요.
- 자유롭게 모델을 TF2.0나 PyTorch 프레임워크로 변환하세요.
- 학습, 평가, 공개 등 각 단계에 맞는 프레임워크를 원하는대로 선택하세요.
1. 필요한 대로 모델이나 예시를 커스터마이즈하세요:
- 우리는 저자가 공개한 결과를 재현하기 위해 각 모델 구조의 예시를 제공합니다.
- 모델 내부 구조는 가능한 일관적으로 공개되어 있습니다.
- 빠른 실험을 위해 모델 파일은 라이브러리와 독립적으로 사용될 수 있습니다.
## 왜 transformers를 사용하지 말아야 할까요?
- 이 라이브러리는 신경망 블록을 만들기 위한 모듈이 아닙니다. 연구자들이 여러 파일을 살펴보지 않고 바로 각 모델을 사용할 수 있도록, 모델 파일 코드의 추상화 수준을 적정하게 유지했습니다.
- 학습 API는 모든 모델에 적용할 수 있도록 만들어지진 않았지만, 라이브러리가 제공하는 모델들에 적용할 수 있도록 최적화되었습니다. 일반적인 머신 러닝을 위해선, 다른 라이브러리를 사용하세요.
- 가능한 많은 사용 예시를 보여드리고 싶어서, [예시 폴더](https://github.com/huggingface/transformers/tree/main/examples)의 스크립트를 준비했습니다. 이 스크립트들을 수정 없이 특정한 문제에 바로 적용하지 못할 수 있습니다. 필요에 맞게 일부 코드를 수정해야 할 수 있습니다.
## 설치
### pip로 설치하기
이 저장소는 Python 3.6+, Flax 0.3.2+, PyTorch 1.3.1+, TensorFlow 2.3+에서 테스트 되었습니다.
[가상 환경](https://docs.python.org/3/library/venv.html)에 🤗 Transformers를 설치하세요. Python 가상 환경에 익숙하지 않다면, [사용자 가이드](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)를 확인하세요.
우선, 사용할 Python 버전으로 가상 환경을 만들고 실행하세요.
그 다음, Flax, PyTorch, TensorFlow 중 적어도 하나는 설치해야 합니다.
플랫폼에 맞는 설치 명령어를 확인하기 위해 [TensorFlow 설치 페이지](https://www.tensorflow.org/install/), [PyTorch 설치 페이지](https://pytorch.org/get-started/locally/#start-locally), [Flax 설치 페이지](https://github.com/google/flax#quick-install)를 확인하세요.
이들 중 적어도 하나가 설치되었다면, 🤗 Transformers는 다음과 같이 pip을 이용해 설치할 수 있습니다:
```bash
pip install transformers
```
예시들을 체험해보고 싶거나, 최최최첨단 코드를 원하거나, 새로운 버전이 나올 때까지 기다릴 수 없다면 [라이브러리를 소스에서 바로 설치](https://huggingface.co/docs/transformers/installation#installing-from-source)하셔야 합니다.
### conda로 설치하기
Transformers 버전 v4.0.0부터, conda 채널이 생겼습니다: `huggingface`.
🤗 Transformers는 다음과 같이 conda로 설치할 수 있습니다:
```shell script
conda install -c huggingface transformers
```
Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는 방법을 확인하세요.
## 모델 구조
**🤗 Transformers가 제공하는 [모든 모델 체크포인트](https://huggingface.co/models)** 는 huggingface.co [모델 허브](https://huggingface.co)에 완벽히 연동되어 있습니다. [개인](https://huggingface.co/users)과 [기관](https://huggingface.co/organizations)이 모델 허브에 직접 업로드할 수 있습니다.
현재 사용 가능한 모델 체크포인트의 개수: 
🤗 Transformers는 다음 모델들을 제공합니다 (각 모델의 요약은 [여기](https://huggingface.co/docs/transformers/model_summary)서 확인하세요):
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER) released with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper a [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI) released with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. 새로운 모델을 올리고 싶나요? 우리가 **상세한 가이드와 템플릿** 으로 새로운 모델을 올리도록 도와드릴게요. 가이드와 템플릿은 이 저장소의 [`templates`](./templates) 폴더에서 확인하실 수 있습니다. [컨트리뷰션 가이드라인](./CONTRIBUTING.md)을 꼭 확인해주시고, PR을 올리기 전에 메인테이너에게 연락하거나 이슈를 오픈해 피드백을 받으시길 바랍니다.
각 모델이 Flax, PyTorch, TensorFlow으로 구현되었는지 또는 🤗 Tokenizers 라이브러리가 지원하는 토크나이저를 사용하는지 확인하려면, [이 표](https://huggingface.co/docs/transformers/index#supported-frameworks)를 확인하세요.
이 구현은 여러 데이터로 검증되었고 (예시 스크립트를 참고하세요) 오리지널 구현의 성능과 같아야 합니다. [도큐먼트](https://huggingface.co/docs/transformers/examples)의 Examples 섹션에서 성능에 대한 자세한 설명을 확인할 수 있습니다.
## 더 알아보기
| 섹션 | 설명 |
|-|-|
| [도큐먼트](https://huggingface.co/transformers/) | 전체 API 도큐먼트와 튜토리얼 |
| [과제 요약](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers가 지원하는 과제들 |
| [전처리 튜토리얼](https://huggingface.co/docs/transformers/preprocessing) | `Tokenizer` 클래스를 이용해 모델을 위한 데이터 준비하기 |
| [학습과 fine-tuning](https://huggingface.co/docs/transformers/training) | 🤗 Transformers가 제공하는 모델 PyTorch/TensorFlow 학습 과정과 `Trainer` API에서 사용하기 |
| [퀵 투어: Fine-tuning/사용 스크립트](https://github.com/huggingface/transformers/tree/main/examples) | 다양한 과제에서 모델 fine-tuning하는 예시 스크립트 |
| [모델 공유 및 업로드](https://huggingface.co/docs/transformers/model_sharing) | 커뮤니티에 fine-tune된 모델을 업로드 및 공유하기 |
| [마이그레이션](https://huggingface.co/docs/transformers/migration) | `pytorch-transformers`나 `pytorch-pretrained-bert`에서 🤗 Transformers로 이동하기|
## 인용
🤗 Transformers 라이브러리를 인용하고 싶다면, 이 [논문](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)을 인용해 주세요:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README_zh-hans.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Chinese translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Chinese characters. E.g., 共 100 多种语言; 使用 transformers 库。
- Use square quotes, e.g.,「引用」
Dictionary
Hugging Face: 抱抱脸
token: 词符（并用括号标注原英文）
tokenize: 词符化（并用括号标注原英文）
tokenizer: 词符化器（并用括号标注原英文）
transformer: transformer（不翻译）
pipeline: 流水线
API: API (不翻译）
inference: 推理
Trainer: 训练器。当作为类名出现时不翻译。
pretrained/pretrain: 预训练
finetune: 微调
community: 社区
example: 当特指仓库中 example 目录时翻译为「用例」
Python data structures (e.g., list, set, dict): 翻译为列表，集合，词典，并用括号标注原英文
NLP/Natural Language Processing: 以 NLP 出现时不翻译，以 Natural Language Processing 出现时翻译为自然语言处理
checkpoint: 检查点
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<b>简体中文</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>为 Jax、PyTorch 和 TensorFlow 打造的先进的自然语言处理</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 提供了数以千计的预训练模型，支持 100 多种语言的文本分类、信息抽取、问答、摘要、翻译、文本生成。它的宗旨让最先进的 NLP 技术人人易用。
🤗 Transformers 提供了便于快速下载和使用的API，让你可以把预训练模型用在给定文本、在你的数据集上微调然后通过 [model hub](https://huggingface.co/models) 与社区共享。同时，每个定义的 Python 模块均完全独立，方便修改和快速研究实验。
🤗 Transformers 支持三个最热门的深度学习库： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — 并与之无缝整合。你可以直接使用一个框架训练你的模型然后用另一个加载和推理。
## 在线演示
你可以直接在模型页面上测试大多数 [model hub](https://huggingface.co/models) 上的模型。 我们也提供了 [私有模型托管、模型版本管理以及推理API](https://huggingface.co/pricing)。
这里是一些例子：
- [用 BERT 做掩码填词](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [用 Electra 做命名实体识别](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [用 GPT-2 做文本生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [用 RoBERTa 做自然语言推理](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [用 BART 做文本摘要](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [用 DistilBERT 做问答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [用 T5 做翻译](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，由抱抱脸团队打造，是一个文本生成的官方 demo。
## 如果你在寻找由抱抱脸团队提供的定制化支持服务
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 快速上手
我们为快速使用模型提供了 `pipeline` （流水线）API。流水线聚合了预训练模型和对应的文本预处理。下面是一个快速使用流水线去判断正负面情绪的例子：
```python
>>> from transformers import pipeline
# 使用情绪分析流水线
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
第二行代码下载并缓存了流水线使用的预训练模型，而第三行代码则在给定的文本上进行了评估。这里的答案“正面” (positive) 具有 99 的置信度。
许多的 NLP 任务都有开箱即用的预训练流水线。比如说，我们可以轻松的从给定文本中抽取问题答案：
``` python
>>> from transformers import pipeline
# 使用问答流水线
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
除了给出答案，预训练模型还给出了对应的置信度分数、答案在词符化 (tokenized) 后的文本中开始和结束的位置。你可以从[这个教程](https://huggingface.co/docs/transformers/task_summary)了解更多流水线API支持的任务。
要在你的任务上下载和使用任意预训练模型也很简单，只需三行代码。这里是 PyTorch 版的示例：
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
这里是等效的 TensorFlow 代码：
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
词符化器 (tokenizer) 为所有的预训练模型提供了预处理，并可以直接对单个字符串进行调用（比如上面的例子）或对列表 (list) 调用。它会输出一个你可以在下游代码里使用或直接通过 `**` 解包表达式传给模型的词典 (dict)。
模型本身是一个常规的 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) 或 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)（取决于你的后端），可以常规方式使用。 [这个教程](https://huggingface.co/transformers/training.html)解释了如何将这样的模型整合到经典的 PyTorch 或 TensorFlow 训练循环中，或是如何使用我们的 `Trainer` 训练器）API 来在一个新的数据集上快速微调。
## 为什么要用 transformers？
1. 便于使用的先进模型：
- NLU 和 NLG 上表现优越
- 对教学和实践友好且低门槛
- 高级抽象，只需了解三个类
- 对所有模型统一的API
1. 更低计算开销，更少的碳排放：
- 研究人员可以分享已训练的模型而非每次从头开始训练
- 工程师可以减少计算用时和生产环境开销
- 数十种模型架构、两千多个预训练模型、100多种语言支持
1. 对于模型生命周期的每一个部分都面面俱到：
- 训练先进的模型，只需 3 行代码
- 模型在不同深度学习框架间任意转移，随你心意
- 为训练、评估和生产选择最适合的框架，衔接无缝
1. 为你的需求轻松定制专属模型和用例：
- 我们为每种模型架构提供了多个用例来复现原论文结果
- 模型内部结构保持透明一致
- 模型文件可单独使用，方便魔改和快速实验
## 什么情况下我不该用 transformers？
- 本库并不是模块化的神经网络工具箱。模型文件中的代码特意呈若璞玉，未经额外抽象封装，以便研究人员快速迭代魔改而不致溺于抽象和文件跳转之中。
- `Trainer` API 并非兼容任何模型，只为本库之模型优化。若是在寻找适用于通用机器学习的训练循环实现，请另觅他库。
- 尽管我们已尽力而为，[examples 目录](https://github.com/huggingface/transformers/tree/main/examples)中的脚本也仅为用例而已。对于你的特定问题，它们并不一定开箱即用，可能需要改几行代码以适之。
## 安装
### 使用 pip
这个仓库已在 Python 3.6+、Flax 0.3.2+、PyTorch 1.3.1+ 和 TensorFlow 2.3+ 下经过测试。
你可以在[虚拟环境](https://docs.python.org/3/library/venv.html)中安装 🤗 Transformers。如果你还不熟悉 Python 的虚拟环境，请阅此[用户说明](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
首先，用你打算使用的版本的 Python 创建一个虚拟环境并激活。
然后，你需要安装 Flax、PyTorch 或 TensorFlow 其中之一。关于在你使用的平台上安装这些框架，请参阅 [TensorFlow 安装页](https://www.tensorflow.org/install/), [PyTorch 安装页](https://pytorch.org/get-started/locally/#start-locally) 或 [Flax 安装页](https://github.com/google/flax#quick-install)。
当这些后端之一安装成功后， 🤗 Transformers 可依此安装：
```bash
pip install transformers
```
如果你想要试试用例或者想在正式发布前使用最新的开发中代码，你得[从源代码安装](https://huggingface.co/docs/transformers/installation#installing-from-source)。
### 使用 conda
自 Transformers 4.0.0 版始，我们有了一个 conda 频道： `huggingface`。
🤗 Transformers 可以通过 conda 依此安装：
```shell script
conda install -c huggingface transformers
```
要通过 conda 安装 Flax、PyTorch 或 TensorFlow 其中之一，请参阅它们各自安装页的说明。
## 模型架构
🤗 Transformers 支持的[**所有的模型检查点**](https://huggingface.co/models)由[用户](https://huggingface.co/users)和[组织](https://huggingface.co/organizations)上传，均与 huggingface.co [model hub](https://huggingface.co) 无缝整合。
目前的检查点数量： 
🤗 Transformers 目前支持如下的架构（模型概述请阅[这里](https://huggingface.co/docs/transformers/model_summary)）：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (来自 Google Research and the Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (来自 MIT) 伴随论文 [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) 由 Yuan Gong, Yu-An Chung, James Glass 发布。
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (来自 Facebook) 伴随论文 [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) 由 Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer 发布。
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (来自 École polytechnique) 伴随论文 [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) 由 Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis 发布。
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (来自 VinAI Research) 伴随论文 [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) 由 Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen 发布。
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (来自 Microsoft) 伴随论文 [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) 由 Hangbo Bao, Li Dong, Furu Wei 发布。
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (来自 Google) 伴随论文 [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) 由 Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova 发布。
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (来自 Google) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (来自 VinAI Research) 伴随论文 [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) 由 Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen 发布。
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (来自 Microsoft Research AI4Science) 伴随论文 [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) 由 Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu 发布。
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (来自 Google AI) 伴随论文 [Big Transfer (BiT) 由 Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby 发布。
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (来自 Alexa) 伴随论文 [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) 由 Adrian de Wynter and Daniel J. Perry 发布。
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (来自 Google Research) 伴随论文 [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) 由 Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel 发布。
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (来自 Inria/Facebook/Sorbonne) 伴随论文 [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) 由 Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot 发布。
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (来自 Google Research) 伴随论文 [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) 由 Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting 发布。
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (来自 OFA-Sys) 伴随论文 [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) 由 An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou 发布。
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (来自 Microsoft Research Asia) 伴随论文 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 由 Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 发布。
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (来自 YituTech) 伴随论文 [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) 由 Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan 发布。
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (来自 Facebook AI) 伴随论文 [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) 由 Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie 发布。
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (来自 Tsinghua University) 伴随论文 [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) 由 Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun 发布。
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (来自 Salesforce) 伴随论文 [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) 由 Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher 发布。
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (来自 Microsoft) 伴随论文 [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) 由 Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang 发布。
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (来自 Facebook) 伴随论文 [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) 由 Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli 发布。
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (来自 Berkeley/Facebook/Google) 伴随论文 [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) 由 Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch 发布。
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (来自 SenseTime Research) 伴随论文 [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) 由 Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai 发布。
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (来自 Facebook) 伴随论文 [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) 由 Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou 发布。
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (来自 Facebook) 伴随论文 [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) 由 Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko 发布。
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (来自 Microsoft Research) 伴随论文 [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) 由 Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan 发布。
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (来自 SHI Labs) 伴随论文 [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) 由 Ali Hassani and Humphrey Shi 发布。
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (来自 HuggingFace), 伴随论文 [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) 由 Victor Sanh, Lysandre Debut and Thomas Wolf 发布。 同样的方法也应用于压缩 GPT-2 到 [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa 到 [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT 到 [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) 和德语版 DistilBERT。
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (来自 Microsoft Research) 伴随论文 [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) 由 Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei 发布。
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (来自 NAVER) 伴随论文 [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) 由 Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park 发布。
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (来自 Facebook) 伴随论文 [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) 由 Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih 发布。
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (来自 Intel Labs) 伴随论文 [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) 由 René Ranftl, Alexey Bochkovskiy, Vladlen Koltun 发布。
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (来自 Google Research/Stanford University) 伴随论文 [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) 由 Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning 发布。
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (来自 Google Research) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (来自 Baidu) 伴随论文 [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu 发布。
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (来自 CNRS) 伴随论文 [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) 由 Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab 发布。
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (来自 Facebook AI) 伴随论文 [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) 由 Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela 发布。
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (来自 Google Research) 伴随论文 [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) 由 James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon 发布。
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (来自 CMU/Google Brain) 伴随论文 [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) 由 Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le 发布。
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (来自 KAIST) 伴随论文 [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) 由 Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim 发布。
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (来自 OpenAI) 伴随论文 [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) 由 Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever 发布。
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (来自 EleutherAI) 随仓库 [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) 发布。作者为 Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy 发布。
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (来自 ABEJA) 由 Shinya Otani, Takayoshi Makabe, Anuj Arora, Kyo Hattori。
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (来自 OpenAI) 伴随论文 [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) 由 Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever** 发布。
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (来自 EleutherAI) 伴随论文 [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) 由 Ben Wang and Aran Komatsuzaki 发布。
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (来自 UCSD, NVIDIA) 伴随论文 [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) 由 Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang 发布。
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (来自 Facebook) 伴随论文 [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) 由 Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed 发布。
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (来自 Berkeley) 伴随论文 [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) 由 Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer 发布。
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (来自 OpenAI) 伴随论文 [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) 由 Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever 发布。
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) 由 Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou 发布。
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) 由 Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou 发布。
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) 由 Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei 发布。
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) 由 Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei 发布。
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (来自 Meta AI) 伴随论文 [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) 由 Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze 发布。
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (来自 South China University of Technology) 伴随论文 [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) 由 Jiapeng Wang, Lianwen Jin, Kai Ding 发布。
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (来自 Google AI) released 伴随论文 [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) 由 Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang 发布。
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (来自 Studio Ousia) 伴随论文 [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) 由 Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto 发布。
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (来自 UNC Chapel Hill) 伴随论文 [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) 由 Hao Tan and Mohit Bansal 发布。
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (来自 Facebook) 伴随论文 [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) 由 Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert 发布。
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (来自 Facebook) 伴随论文 [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) 由 Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin 发布。
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** 用 [OPUS](http://opus.nlpl.eu/) 数据训练的机器翻译模型由 Jörg Tiedemann 发布。[Marian Framework](https://marian-nmt.github.io/) 由微软翻译团队开发。
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (来自 Microsoft Research Asia) 伴随论文 [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) 由 Junlong Li, Yiheng Xu, Lei Cui, Furu Wei 发布。
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov >>>>>>> Fix rebase
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) 由 Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer 发布。
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) 由 Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan 发布。
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (来自 Studio Ousia) 伴随论文 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 由 Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 发布。
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (来自 CMU/Google Brain) 伴随论文 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 由 Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 发布。
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (来自 Google Inc.) 伴随论文 [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) 由 Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam 发布。
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (来自 Google Inc.) 伴随论文 [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) 由 Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen 发布。
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (来自 Apple) 伴随论文 [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) 由 Sachin Mehta and Mohammad Rastegari 发布。
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (来自 Microsoft Research) 伴随论文 [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) 由 Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu 发布。
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (来自 Google AI) 伴随论文 [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) 由 Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel 发布。
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (来自 中国人民大学 AI Box) 伴随论文 [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) 由 Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen 发布。
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (来自 SHI Labs) 伴随论文 [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) 由 Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi 发布。
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (来自华为诺亚方舟实验室) 伴随论文 [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) 由 Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu 发布。
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (来自 Meta) 伴随论文 [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) 由 the NLLB team 发布。
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (来自 the University of Wisconsin - Madison) 伴随论文 [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) 由 Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh 发布。
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (来自 Meta AI) 伴随论文 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 由 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 发布。
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (来自 Google AI) 伴随论文 [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) 由 Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby 发布。
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (来自 Google) 伴随论文 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 由 Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 发布。
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (来自 Google) 伴随论文 [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) 由 Jason Phang, Yao Zhao, Peter J. Liu 发布。
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (来自 Deepmind) 伴随论文 [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) 由 Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira 发布。
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (来自 VinAI Research) 伴随论文 [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) 由 Dat Quoc Nguyen and Anh Tuan Nguyen 发布。
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (来自 UCLA NLP) 伴随论文 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 由 Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 发布。
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (来自 Sea AI Labs) 伴随论文 [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) 由 Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng 发布。
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (来自 NVIDIA) 伴随论文 [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) 由 Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius 发布。
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (来自 Facebook) 伴随论文 [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) 由 Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela 发布。
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (来自 Google Research) 伴随论文 [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) 由 Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang 发布。
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (来自 Google Research) 伴随论文 [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) 由 Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya 发布。
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (来自 Google Research) 伴随论文 [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) 由 Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder 发布。
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (来自 Facebook), 伴随论文 [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) 由 Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov 发布。
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (来自 WeChatAI), 伴随论文 [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) 由 HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou 发布。
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (来自 ZhuiyiTechnology), 伴随论文 [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) 由 Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu 发布。
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (来自 NVIDIA) 伴随论文 [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) 由 Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo 发布。
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (来自 Facebook), 伴随论文 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 发布。
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (来自 Facebook) 伴随论文 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 由 Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 发布。
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (来自 Tel Aviv University) 伴随论文 [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) 由 Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy 发布。
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (来自 Berkeley) 伴随论文 [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) 由 Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer 发布。
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (来自 Microsoft) 伴随论文 [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) 由 Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo 发布。
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (来自 Microsoft) 伴随论文 [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) 由 Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo 发布。
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (来自 Google AI) 伴随论文 [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (来自 Google AI) 伴随论文 [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (来自 Microsoft Research) 伴随论文 [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) 由 Brandon Smock, Rohith Pesala, Robin Abraham 发布。
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (来自 Google AI) 伴随论文 [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) 由 Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos 发布。
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (来自 Microsoft Research) 伴随论文 [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) 由 Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou 发布。
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (来自 Google/CMU) 伴随论文 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 由 Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 发布。
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (来自 Microsoft Research) 伴随论文 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 由 Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 发布。
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (来自 Tsinghua University and Nankai University) 伴随论文 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 由 Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 发布。
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (来自 Multimedia Computing Group, Nanjing University) 伴随论文 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 由 Zhan Tong, Yibing Song, Jue Wang, Limin Wang 发布。
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (来自 NAVER AI Lab/Kakao Enterprise/Kakao Brain) 伴随论文 [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) 由 Wonjae Kim, Bokyung Son, Ildoo Kim 发布。
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (来自 UCLA NLP) 伴随论文 [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) 由 Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang 发布。
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (来自 Meta AI) 伴随论文 [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) 由 Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick 发布。
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (来自 Meta AI) 伴随论文 [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas 发布.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (来自 Facebook AI) 伴随论文 [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) 由 Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli 发布。
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (来自 Facebook AI) 伴随论文 [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino 发布。
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (来自 Facebook AI) 伴随论文 [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) 由 Qiantong Xu, Alexei Baevski, Michael Auli 发布。
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (来自 OpenAI) 伴随论文 [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) 由 Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever 发布。
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (来自 Microsoft Research) 伴随论文 [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) 由 Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling 发布。
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (来自 Facebook) 伴随论文 [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) 由 Guillaume Lample and Alexis Conneau 发布。
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (来自 Facebook AI), 伴随论文 [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) 由 Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov 发布。
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (来自 Facebook AI) 伴随论文 [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) 由 Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau 发布。
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (来自 Google/CMU) 伴随论文 [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) 由 Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le 发布。
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (来自 Facebook AI) 伴随论文 [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) 由 Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli 发布。
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (来自 Facebook AI) 伴随论文 [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) 由 Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli 发布。
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (来自 Huazhong University of Science & Technology) 伴随论文 [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) 由 Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu 发布。
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (来自 the University of Wisconsin - Madison) 伴随论文 [You Only Sample (Almost) 由 Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh 发布。
1. 想要贡献新的模型？我们这里有一份**详细指引和模板**来引导你添加新的模型。你可以在 [`templates`](./templates) 目录中找到他们。记得查看 [贡献指南](./CONTRIBUTING.md) 并在开始写 PR 前联系维护人员或开一个新的 issue 来获得反馈。
要检查某个模型是否已有 Flax、PyTorch 或 TensorFlow 的实现，或其是否在 🤗 Tokenizers 库中有对应词符化器（tokenizer），敬请参阅[此表](https://huggingface.co/docs/transformers/index#supported-frameworks)。
这些实现均已于多个数据集测试（请参看用例脚本）并应于原版实现表现相当。你可以在用例文档的[此节](https://huggingface.co/docs/transformers/examples)中了解表现的细节。
## 了解更多
| 章节 | 描述 |
|-|-|
| [文档](https://huggingface.co/transformers/) | 完整的 API 文档和教程 |
| [任务总结](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers 支持的任务 |
| [预处理教程](https://huggingface.co/docs/transformers/preprocessing) | 使用 `Tokenizer` 来为模型准备数据 |
| [训练和微调](https://huggingface.co/docs/transformers/training) | 在 PyTorch/TensorFlow 的训练循环或 `Trainer` API 中使用 🤗 Transformers 提供的模型 |
| [快速上手：微调和用例脚本](https://github.com/huggingface/transformers/tree/main/examples) | 为各种任务提供的用例脚本 |
| [模型分享和上传](https://huggingface.co/docs/transformers/model_sharing) | 和社区上传和分享你微调的模型 |
| [迁移](https://huggingface.co/docs/transformers/migration) | 从 `pytorch-transformers` 或 `pytorch-pretrained-bert` 迁移到 🤗 Transformers |
## 引用
我们已将此库的[论文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)正式发表，如果你使用了 🤗 Transformers 库，请引用:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Chinese translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Chinese characters. E.g., 共 100 多种语言; 使用 transformers 库。
- Use square quotes, e.g.,「引用」
Dictionary
Hugging Face: 抱抱脸
token: 词符（并用括号标注原英文）
tokenize: 词符化（并用括号标注原英文）
tokenizer: 词符化器（并用括号标注原英文）
transformer: transformer（不翻译）
pipeline: 流水线
API: API (不翻译）
inference: 推理
Trainer: 训练器。当作为类名出现时不翻译。
pretrained/pretrain: 预训练
finetune: 微调
community: 社区
example: 当特指仓库中 example 目录时翻译为「用例」
Python data structures (e.g., list, set, dict): 翻译为列表，集合，词典，并用括号标注原英文
NLP/Natural Language Processing: 以 NLP 出现时不翻译，以 Natural Language Processing 出现时翻译为自然语言处理
checkpoint: 检查点
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<b>简体中文</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hant.md">繁體中文</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>为 Jax、PyTorch 和 TensorFlow 打造的先进的自然语言处理</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 提供了数以千计的预训练模型，支持 100 多种语言的文本分类、信息抽取、问答、摘要、翻译、文本生成。它的宗旨让最先进的 NLP 技术人人易用。
🤗 Transformers 提供了便于快速下载和使用的API，让你可以把预训练模型用在给定文本、在你的数据集上微调然后通过 [model hub](https://huggingface.co/models) 与社区共享。同时，每个定义的 Python 模块均完全独立，方便修改和快速研究实验。
🤗 Transformers 支持三个最热门的深度学习库： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) and [TensorFlow](https://www.tensorflow.org/) — 并与之无缝整合。你可以直接使用一个框架训练你的模型然后用另一个加载和推理。
## 在线演示
你可以直接在模型页面上测试大多数 [model hub](https://huggingface.co/models) 上的模型。 我们也提供了 [私有模型托管、模型版本管理以及推理API](https://huggingface.co/pricing)。
这里是一些例子：
- [用 BERT 做掩码填词](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [用 Electra 做命名实体识别](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [用 GPT-2 做文本生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [用 RoBERTa 做自然语言推理](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [用 BART 做文本摘要](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [用 DistilBERT 做问答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [用 T5 做翻译](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，由抱抱脸团队打造，是一个文本生成的官方 demo。
## 如果你在寻找由抱抱脸团队提供的定制化支持服务
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 快速上手
我们为快速使用模型提供了 `pipeline` （流水线）API。流水线聚合了预训练模型和对应的文本预处理。下面是一个快速使用流水线去判断正负面情绪的例子：
```python
>>> from transformers import pipeline
# 使用情绪分析流水线
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
第二行代码下载并缓存了流水线使用的预训练模型，而第三行代码则在给定的文本上进行了评估。这里的答案“正面” (positive) 具有 99 的置信度。
许多的 NLP 任务都有开箱即用的预训练流水线。比如说，我们可以轻松的从给定文本中抽取问题答案：
``` python
>>> from transformers import pipeline
# 使用问答流水线
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
除了给出答案，预训练模型还给出了对应的置信度分数、答案在词符化 (tokenized) 后的文本中开始和结束的位置。你可以从[这个教程](https://huggingface.co/docs/transformers/task_summary)了解更多流水线API支持的任务。
要在你的任务上下载和使用任意预训练模型也很简单，只需三行代码。这里是 PyTorch 版的示例：
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
这里是等效的 TensorFlow 代码：
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
词符化器 (tokenizer) 为所有的预训练模型提供了预处理，并可以直接对单个字符串进行调用（比如上面的例子）或对列表 (list) 调用。它会输出一个你可以在下游代码里使用或直接通过 `**` 解包表达式传给模型的词典 (dict)。
模型本身是一个常规的 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) 或 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)（取决于你的后端），可以常规方式使用。 [这个教程](https://huggingface.co/transformers/training.html)解释了如何将这样的模型整合到经典的 PyTorch 或 TensorFlow 训练循环中，或是如何使用我们的 `Trainer` 训练器）API 来在一个新的数据集上快速微调。
## 为什么要用 transformers？
1. 便于使用的先进模型：
- NLU 和 NLG 上表现优越
- 对教学和实践友好且低门槛
- 高级抽象，只需了解三个类
- 对所有模型统一的API
1. 更低计算开销，更少的碳排放：
- 研究人员可以分享已训练的模型而非每次从头开始训练
- 工程师可以减少计算用时和生产环境开销
- 数十种模型架构、两千多个预训练模型、100多种语言支持
1. 对于模型生命周期的每一个部分都面面俱到：
- 训练先进的模型，只需 3 行代码
- 模型在不同深度学习框架间任意转移，随你心意
- 为训练、评估和生产选择最适合的框架，衔接无缝
1. 为你的需求轻松定制专属模型和用例：
- 我们为每种模型架构提供了多个用例来复现原论文结果
- 模型内部结构保持透明一致
- 模型文件可单独使用，方便魔改和快速实验
## 什么情况下我不该用 transformers？
- 本库并不是模块化的神经网络工具箱。模型文件中的代码特意呈若璞玉，未经额外抽象封装，以便研究人员快速迭代魔改而不致溺于抽象和文件跳转之中。
- `Trainer` API 并非兼容任何模型，只为本库之模型优化。若是在寻找适用于通用机器学习的训练循环实现，请另觅他库。
- 尽管我们已尽力而为，[examples 目录](https://github.com/huggingface/transformers/tree/main/examples)中的脚本也仅为用例而已。对于你的特定问题，它们并不一定开箱即用，可能需要改几行代码以适之。
## 安装
### 使用 pip
这个仓库已在 Python 3.6+、Flax 0.3.2+、PyTorch 1.3.1+ 和 TensorFlow 2.3+ 下经过测试。
你可以在[虚拟环境](https://docs.python.org/3/library/venv.html)中安装 🤗 Transformers。如果你还不熟悉 Python 的虚拟环境，请阅此[用户说明](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
首先，用你打算使用的版本的 Python 创建一个虚拟环境并激活。
然后，你需要安装 Flax、PyTorch 或 TensorFlow 其中之一。关于在你使用的平台上安装这些框架，请参阅 [TensorFlow 安装页](https://www.tensorflow.org/install/), [PyTorch 安装页](https://pytorch.org/get-started/locally/#start-locally) 或 [Flax 安装页](https://github.com/google/flax#quick-install)。
当这些后端之一安装成功后， 🤗 Transformers 可依此安装：
```bash
pip install transformers
```
如果你想要试试用例或者想在正式发布前使用最新的开发中代码，你得[从源代码安装](https://huggingface.co/docs/transformers/installation#installing-from-source)。
### 使用 conda
自 Transformers 4.0.0 版始，我们有了一个 conda 频道： `huggingface`。
🤗 Transformers 可以通过 conda 依此安装：
```shell script
conda install -c huggingface transformers
```
要通过 conda 安装 Flax、PyTorch 或 TensorFlow 其中之一，请参阅它们各自安装页的说明。
## 模型架构
🤗 Transformers 支持的[**所有的模型检查点**](https://huggingface.co/models)由[用户](https://huggingface.co/users)和[组织](https://huggingface.co/organizations)上传，均与 huggingface.co [model hub](https://huggingface.co) 无缝整合。
目前的检查点数量： 
🤗 Transformers 目前支持如下的架构（模型概述请阅[这里](https://huggingface.co/docs/transformers/model_summary)）：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (来自 Google Research and the Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (来自 MIT) 伴随论文 [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) 由 Yuan Gong, Yu-An Chung, James Glass 发布。
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (来自 Facebook) 伴随论文 [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) 由 Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer 发布。
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (来自 École polytechnique) 伴随论文 [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) 由 Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis 发布。
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (来自 VinAI Research) 伴随论文 [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) 由 Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen 发布。
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (来自 Microsoft) 伴随论文 [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) 由 Hangbo Bao, Li Dong, Furu Wei 发布。
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (来自 Google) 伴随论文 [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) 由 Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova 发布。
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (来自 Google) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (来自 VinAI Research) 伴随论文 [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) 由 Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen 发布。
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (来自 Google Research) 伴随论文 [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) 由 Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed 发布。
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (来自 Microsoft Research AI4Science) 伴随论文 [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) 由 Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu 发布。
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (来自 Google AI) 伴随论文 [Big Transfer (BiT) 由 Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby 发布。
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (来自 Facebook) 伴随论文 [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) 由 Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston 发布。
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (来自 Alexa) 伴随论文 [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) 由 Adrian de Wynter and Daniel J. Perry 发布。
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (来自 Google Research) 伴随论文 [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) 由 Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel 发布。
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (来自 Inria/Facebook/Sorbonne) 伴随论文 [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) 由 Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot 发布。
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (来自 Google Research) 伴随论文 [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) 由 Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting 发布。
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (来自 OFA-Sys) 伴随论文 [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) 由 An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou 发布。
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (来自 Microsoft Research Asia) 伴随论文 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 由 Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 发布。
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (来自 YituTech) 伴随论文 [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) 由 Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan 发布。
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (来自 Facebook AI) 伴随论文 [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) 由 Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie 发布。
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (来自 Tsinghua University) 伴随论文 [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) 由 Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun 发布。
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (来自 Salesforce) 伴随论文 [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) 由 Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher 发布。
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (来自 Microsoft) 伴随论文 [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) 由 Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang 发布。
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (来自 Facebook) 伴随论文 [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) 由 Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli 发布。
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (来自 Microsoft) 伴随论文 [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) 由 Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen 发布。
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (来自 Berkeley/Facebook/Google) 伴随论文 [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) 由 Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch 发布。
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (来自 SenseTime Research) 伴随论文 [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) 由 Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai 发布。
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (来自 Facebook) 伴随论文 [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) 由 Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou 发布。
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (来自 Facebook) 伴随论文 [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) 由 Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko 发布。
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (来自 Microsoft Research) 伴随论文 [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) 由 Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan 发布。
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (来自 SHI Labs) 伴随论文 [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) 由 Ali Hassani and Humphrey Shi 发布。
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (来自 HuggingFace), 伴随论文 [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) 由 Victor Sanh, Lysandre Debut and Thomas Wolf 发布。 同样的方法也应用于压缩 GPT-2 到 [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa 到 [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT 到 [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) 和德语版 DistilBERT。
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (来自 Microsoft Research) 伴随论文 [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) 由 Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei 发布。
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (来自 NAVER) 伴随论文 [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) 由 Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park 发布。
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (来自 Facebook) 伴随论文 [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) 由 Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih 发布。
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (来自 Intel Labs) 伴随论文 [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) 由 René Ranftl, Alexey Bochkovskiy, Vladlen Koltun 发布。
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (来自 Google Research/Stanford University) 伴随论文 [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) 由 Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning 发布。
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (来自 Google Research) 伴随论文 [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) 由 Sascha Rothe, Shashi Narayan, Aliaksei Severyn 发布。
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (来自 Baidu) 伴随论文 [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu 发布。
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (来自 CNRS) 伴随论文 [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) 由 Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab 发布。
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (来自 Facebook AI) 伴随论文 [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) 由 Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela 发布。
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (来自 Google Research) 伴随论文 [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) 由 James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon 发布。
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (来自 CMU/Google Brain) 伴随论文 [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) 由 Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le 发布。
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (来自 KAIST) 伴随论文 [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) 由 Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim 发布。
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (来自 OpenAI) 伴随论文 [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) 由 Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever 发布。
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (来自 EleutherAI) 随仓库 [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) 发布。作者为 Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy 发布。
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (来自 ABEJA) 由 Shinya Otani, Takayoshi Makabe, Anuj Arora, Kyo Hattori。
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (来自 OpenAI) 伴随论文 [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) 由 Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever** 发布。
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (来自 EleutherAI) 伴随论文 [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) 由 Ben Wang and Aran Komatsuzaki 发布。
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (来自 UCSD, NVIDIA) 伴随论文 [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) 由 Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang 发布。
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (来自 Facebook) 伴随论文 [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) 由 Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed 发布。
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (来自 Berkeley) 伴随论文 [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) 由 Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer 发布。
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (来自 OpenAI) 伴随论文 [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) 由 Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever 发布。
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) 由 Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou 发布。
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) 由 Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou 发布。
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (来自 Microsoft Research Asia) 伴随论文 [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) 由 Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei 发布。
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (来自 Microsoft Research Asia) 伴随论文 [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) 由 Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei 发布。
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (来自 Meta AI) 伴随论文 [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) 由 Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze 发布。
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (来自 South China University of Technology) 伴随论文 [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) 由 Jiapeng Wang, Lianwen Jin, Kai Ding 发布。
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (来自 Google AI) released 伴随论文 [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) 由 Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang 发布。
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (来自 Studio Ousia) 伴随论文 [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) 由 Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto 发布。
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (来自 UNC Chapel Hill) 伴随论文 [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) 由 Hao Tan and Mohit Bansal 发布。
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (来自 Facebook) 伴随论文 [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) 由 Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert 发布。
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (来自 Facebook) 伴随论文 [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) 由 Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin 发布。
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** 用 [OPUS](http://opus.nlpl.eu/) 数据训练的机器翻译模型由 Jörg Tiedemann 发布。[Marian Framework](https://marian-nmt.github.io/) 由微软翻译团队开发。
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (来自 Microsoft Research Asia) 伴随论文 [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) 由 Junlong Li, Yiheng Xu, Lei Cui, Furu Wei 发布。
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov >>>>>>> Fix rebase
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) 由 Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer 发布。
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (来自 Facebook) 伴随论文 [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) 由 Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan 发布。
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (来自 Studio Ousia) 伴随论文 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 由 Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 发布。
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (来自 CMU/Google Brain) 伴随论文 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 由 Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 发布。
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (来自 Google Inc.) 伴随论文 [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) 由 Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam 发布。
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (来自 Google Inc.) 伴随论文 [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) 由 Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen 发布。
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (来自 Apple) 伴随论文 [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) 由 Sachin Mehta and Mohammad Rastegari 发布。
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (来自 Microsoft Research) 伴随论文 [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) 由 Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu 发布。
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (来自 Google AI) 伴随论文 [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) 由 Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel 发布。
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (来自 中国人民大学 AI Box) 伴随论文 [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) 由 Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen 发布。
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (来自 SHI Labs) 伴随论文 [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) 由 Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi 发布。
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (来自华为诺亚方舟实验室) 伴随论文 [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) 由 Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu 发布。
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (来自 Meta) 伴随论文 [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) 由 the NLLB team 发布。
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (来自 the University of Wisconsin - Madison) 伴随论文 [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) 由 Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh 发布。
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (来自 Meta AI) 伴随论文 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 由 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 发布。
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (来自 Google AI) 伴随论文 [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) 由 Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby 发布。
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (来自 Google) 伴随论文 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 由 Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 发布。
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (来自 Google) 伴随论文 [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) 由 Jason Phang, Yao Zhao, Peter J. Liu 发布。
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (来自 Deepmind) 伴随论文 [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) 由 Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira 发布。
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (来自 VinAI Research) 伴随论文 [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) 由 Dat Quoc Nguyen and Anh Tuan Nguyen 发布。
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (来自 UCLA NLP) 伴随论文 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 由 Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 发布。
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (来自 Sea AI Labs) 伴随论文 [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) 由 Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng 发布。
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (来自 NVIDIA) 伴随论文 [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) 由 Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius 发布。
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (来自 Facebook) 伴随论文 [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) 由 Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela 发布。
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (来自 Google Research) 伴随论文 [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) 由 Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang 发布。
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (来自 Google Research) 伴随论文 [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) 由 Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya 发布。
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (来自 Google Research) 伴随论文 [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) 由 Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder 发布。
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (来自 Facebook), 伴随论文 [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) 由 Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov 发布。
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (来自 WeChatAI), 伴随论文 [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) 由 HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou 发布。
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (来自 ZhuiyiTechnology), 伴随论文 [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) 由 Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu 发布。
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (来自 NVIDIA) 伴随论文 [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) 由 Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo 发布。
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (来自 Facebook), 伴随论文 [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino 发布。
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (来自 Facebook) 伴随论文 [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) 由 Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau 发布。
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (来自 Tel Aviv University) 伴随论文 [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) 由 Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy 发布。
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (来自 Berkeley) 伴随论文 [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) 由 Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer 发布。
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (来自 Microsoft) 伴随论文 [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) 由 Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo 发布。
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (来自 Microsoft) 伴随论文 [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) 由 Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo 发布。
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (来自 Google AI) 伴随论文 [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (来自 Google AI) 伴随论文 [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) 由 Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu 发布。
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (来自 Microsoft Research) 伴随论文 [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) 由 Brandon Smock, Rohith Pesala, Robin Abraham 发布。
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (来自 Google AI) 伴随论文 [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) 由 Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos 发布。
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (来自 Microsoft Research) 伴随论文 [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) 由 Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou 发布。
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (来自 Google/CMU) 伴随论文 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 由 Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 发布。
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (来自 Microsoft Research) 伴随论文 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 由 Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 发布。
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (来自 Tsinghua University and Nankai University) 伴随论文 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 由 Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 发布。
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (来自 Multimedia Computing Group, Nanjing University) 伴随论文 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 由 Zhan Tong, Yibing Song, Jue Wang, Limin Wang 发布。
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (来自 NAVER AI Lab/Kakao Enterprise/Kakao Brain) 伴随论文 [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) 由 Wonjae Kim, Bokyung Son, Ildoo Kim 发布。
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (来自 UCLA NLP) 伴随论文 [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) 由 Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang 发布。
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (来自 Google AI) 伴随论文 [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) 由 Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby 发布。
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (来自 Meta AI) 伴随论文 [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) 由 Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick 发布。
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (来自 Meta AI) 伴随论文 [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas 发布.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (来自 Facebook AI) 伴随论文 [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) 由 Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli 发布。
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (来自 Facebook AI) 伴随论文 [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) 由 Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino 发布。
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (来自 Facebook AI) 伴随论文 [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) 由 Qiantong Xu, Alexei Baevski, Michael Auli 发布。
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (来自 OpenAI) 伴随论文 [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) 由 Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever 发布。
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (来自 Microsoft Research) 伴随论文 [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) 由 Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling 发布。
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (来自 Facebook) 伴随论文 [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) 由 Guillaume Lample and Alexis Conneau 发布。
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (来自 Microsoft Research) 伴随论文 [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) 由 Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou 发布。
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (来自 Facebook AI), 伴随论文 [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) 由 Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov 发布。
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (来自 Facebook AI) 伴随论文 [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) 由 Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau 发布。
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (来自 Google/CMU) 伴随论文 [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) 由 Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le 发布。
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (来自 Facebook AI) 伴随论文 [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) 由 Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli 发布。
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (来自 Facebook AI) 伴随论文 [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) 由 Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli 发布。
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (来自 Huazhong University of Science & Technology) 伴随论文 [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) 由 Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu 发布。
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (来自 the University of Wisconsin - Madison) 伴随论文 [You Only Sample (Almost) 由 Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh 发布。
1. 想要贡献新的模型？我们这里有一份**详细指引和模板**来引导你添加新的模型。你可以在 [`templates`](./templates) 目录中找到他们。记得查看 [贡献指南](./CONTRIBUTING.md) 并在开始写 PR 前联系维护人员或开一个新的 issue 来获得反馈。
要检查某个模型是否已有 Flax、PyTorch 或 TensorFlow 的实现，或其是否在 🤗 Tokenizers 库中有对应词符化器（tokenizer），敬请参阅[此表](https://huggingface.co/docs/transformers/index#supported-frameworks)。
这些实现均已于多个数据集测试（请参看用例脚本）并应于原版实现表现相当。你可以在用例文档的[此节](https://huggingface.co/docs/transformers/examples)中了解表现的细节。
## 了解更多
| 章节 | 描述 |
|-|-|
| [文档](https://huggingface.co/transformers/) | 完整的 API 文档和教程 |
| [任务总结](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers 支持的任务 |
| [预处理教程](https://huggingface.co/docs/transformers/preprocessing) | 使用 `Tokenizer` 来为模型准备数据 |
| [训练和微调](https://huggingface.co/docs/transformers/training) | 在 PyTorch/TensorFlow 的训练循环或 `Trainer` API 中使用 🤗 Transformers 提供的模型 |
| [快速上手：微调和用例脚本](https://github.com/huggingface/transformers/tree/main/examples) | 为各种任务提供的用例脚本 |
| [模型分享和上传](https://huggingface.co/docs/transformers/model_sharing) | 和社区上传和分享你微调的模型 |
| [迁移](https://huggingface.co/docs/transformers/migration) | 从 `pytorch-transformers` 或 `pytorch-pretrained-bert` 迁移到 🤗 Transformers |
## 引用
我们已将此库的[论文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)正式发表，如果你使用了 🤗 Transformers 库，请引用:
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./README_zh-hant.md | <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Traditional Chinese translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Chinese characters. E.g., 共 100 多種語言; 使用 transformers 函式庫。
- Use square quotes, e.g.,「引用」
- Some of terms in the file can be found at National Academy for Educational Research (https://terms.naer.edu.tw/), an official website providing bilingual translations between English and Traditional Chinese.
Dictionary
API: API (不翻譯）
add: 加入
checkpoint: 檢查點
code: 程式碼
community: 社群
confidence: 信賴度
dataset: 資料集
documentation: 文件
example: 基本翻譯為「範例」，或依語意翻為「例子」
finetune: 微調
Hugging Face: Hugging Face（不翻譯）
implementation: 實作
inference: 推論
library: 函式庫
module: 模組
NLP/Natural Language Processing: 以 NLP 出現時不翻譯，以 Natural Language Processing 出現時翻譯為自然語言處理
online demos: 線上Demo
pipeline: pipeline（不翻譯）
pretrained/pretrain: 預訓練
Python data structures (e.g., list, set, dict): 翻譯為串列，集合，字典，並用括號標註原英文
repository: repository（不翻譯）
summary: 概覽
token-: token-（不翻譯）
Trainer: Trainer（不翻譯）
transformer: transformer（不翻譯）
tutorial: 教學
user: 使用者
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<b>繁體中文</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>為 Jax、PyTorch 以及 TensorFlow 打造的先進自然語言處理函式庫</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 提供了數以千計的預訓練模型，支援 100 多種語言的文本分類、資訊擷取、問答、摘要、翻譯、文本生成。它的宗旨是讓最先進的 NLP 技術人人易用。
🤗 Transformers 提供了便於快速下載和使用的API，讓你可以將預訓練模型用在給定文本、在你的資料集上微調然後經由 [model hub](https://huggingface.co/models) 與社群共享。同時，每個定義的 Python 模組架構均完全獨立，方便修改和快速研究實驗。
🤗 Transformers 支援三個最熱門的深度學習函式庫： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) 以及 [TensorFlow](https://www.tensorflow.org/) — 並與之完美整合。你可以直接使用其中一個框架訓練你的模型，然後用另一個載入和推論。
## 線上Demo
你可以直接在 [model hub](https://huggingface.co/models) 上測試大多數的模型。我們也提供了 [私有模型託管、模型版本管理以及推論API](https://huggingface.co/pricing)。
這裡是一些範例：
- [用 BERT 做遮蓋填詞](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [用 Electra 做專有名詞辨識](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [用 GPT-2 做文本生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [用 RoBERTa 做自然語言推論](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [用 BART 做文本摘要](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [用 DistilBERT 做問答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [用 T5 做翻譯](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，由 Hugging Face 團隊所打造，是一個文本生成的官方 demo。
## 如果你在尋找由 Hugging Face 團隊所提供的客製化支援服務
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 快速上手
我們為快速使用模型提供了 `pipeline` API。 Pipeline 包含了預訓練模型和對應的文本預處理。下面是一個快速使用 pipeline 去判斷正負面情緒的例子：
```python
>>> from transformers import pipeline
# 使用情緒分析 pipeline
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
第二行程式碼下載並快取 pipeline 使用的預訓練模型，而第三行程式碼則在給定的文本上進行了評估。這裡的答案“正面” (positive) 具有 99.97% 的信賴度。
許多的 NLP 任務都有隨選即用的預訓練 `pipeline`。例如，我們可以輕鬆地從給定文本中擷取問題答案：
``` python
>>> from transformers import pipeline
# 使用問答 pipeline
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
除了提供問題解答，預訓練模型還提供了對應的信賴度分數以及解答在 tokenized 後的文本中開始和結束的位置。你可以從[這個教學](https://huggingface.co/docs/transformers/task_summary)了解更多 `pipeline` API支援的任務。
要在你的任務中下載和使用任何預訓練模型很簡單，只需三行程式碼。這裡是 PyTorch 版的範例：
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
這裡是對應的 TensorFlow 程式碼：
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
Tokenizer 為所有的預訓練模型提供了預處理，並可以直接轉換單一字串（比如上面的例子）或串列 (list)。它會輸出一個的字典 (dict) 讓你可以在下游程式碼裡使用或直接藉由 `**` 運算式傳給模型。
模型本身是一個常規的 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) 或 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)（取決於你的後端），可依常規方式使用。 [這個教學](https://huggingface.co/transformers/training.html)解釋了如何將這樣的模型整合到一般的 PyTorch 或 TensorFlow 訓練迴圈中，或是如何使用我們的 `Trainer` API 在一個新的資料集上快速進行微調。
## 為什麼要用 transformers？
1. 便於使用的先進模型：
- NLU 和 NLG 上性能卓越
- 對教學和實作友好且低門檻
- 高度抽象，使用者只須學習 3 個類別
- 對所有模型使用的制式化API
1. 更低的運算成本，更少的碳排放：
- 研究人員可以分享已訓練的模型而非每次從頭開始訓練
- 工程師可以減少計算時間以及生產成本
- 數十種模型架構、兩千多個預訓練模型、100多種語言支援
1. 對於模型生命週期的每一個部分都面面俱到：
- 訓練先進的模型，只需 3 行程式碼
- 模型可以在不同深度學習框架之間任意轉換
- 為訓練、評估和生產選擇最適合的框架，並完美銜接
1. 為你的需求輕鬆客製化專屬模型和範例：
- 我們為每種模型架構提供了多個範例來重現原論文結果
- 一致的模型內部架構
- 模型檔案可單獨使用，便於修改和快速實驗
## 什麼情況下我不該用 transformers？
- 本函式庫並不是模組化的神經網絡工具箱。模型文件中的程式碼並未做額外的抽象封裝，以便研究人員快速地翻閱及修改程式碼，而不會深陷複雜的類別包裝之中。
- `Trainer` API 並非相容任何模型，它只為本函式庫中的模型最佳化。對於一般的機器學習用途，請使用其他函式庫。
- 儘管我們已盡力而為，[examples 目錄](https://github.com/huggingface/transformers/tree/main/examples)中的腳本也僅為範例而已。對於特定問題，它們並不一定隨選即用，可能需要修改幾行程式碼以符合需求。
## 安裝
### 使用 pip
這個 Repository 已在 Python 3.6+、Flax 0.3.2+、PyTorch 1.3.1+ 和 TensorFlow 2.3+ 下經過測試。
你可以在[虛擬環境](https://docs.python.org/3/library/venv.html)中安裝 🤗 Transformers。如果你還不熟悉 Python 的虛擬環境，請閱此[使用者指引](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
首先，用你打算使用的版本的 Python 創建一個虛擬環境並進入。
然後，你需要安裝 Flax、PyTorch 或 TensorFlow 其中之一。對於該如何在你使用的平台上安裝這些框架，請參閱 [TensorFlow 安裝頁面](https://www.tensorflow.org/install/), [PyTorch 安裝頁面](https://pytorch.org/get-started/locally/#start-locally) 或 [Flax 安裝頁面](https://github.com/google/flax#quick-install)。
當其中一個後端安裝成功後，🤗 Transformers 可依此安裝：
```bash
pip install transformers
```
如果你想要試試範例或者想在正式發布前使用最新開發中的程式碼，你必須[從原始碼安裝](https://huggingface.co/docs/transformers/installation#installing-from-source)。
### 使用 conda
自 Transformers 4.0.0 版始，我們有了一個 conda channel： `huggingface`。
🤗 Transformers 可以藉由 conda 依此安裝：
```shell script
conda install -c huggingface transformers
```
要藉由 conda 安裝 Flax、PyTorch 或 TensorFlow 其中之一，請參閱它們各自安裝頁面的說明。
## 模型架構
**🤗 Transformers 支援的[所有的模型檢查點](https://huggingface.co/models)**，由[使用者](https://huggingface.co/users)和[組織](https://huggingface.co/organizations)上傳，均與 huggingface.co [model hub](https://huggingface.co) 完美結合。
目前的檢查點數量： 
🤗 Transformers 目前支援以下的架構（模型概覽請參閱[這裡](https://huggingface.co/docs/transformers/model_summary)）：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER) released with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released with the paper [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper a [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook) released with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University) released with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released with the paper [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI) released with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. 想要貢獻新的模型？我們這裡有一份**詳細指引和模板**來引導你加入新的模型。你可以在 [`templates`](./templates) 目錄中找到它們。記得查看[貢獻指引](./CONTRIBUTING.md)並在開始寫 PR 前聯繫維護人員或開一個新的 issue 來獲得 feedbacks。
要檢查某個模型是否已有 Flax、PyTorch 或 TensorFlow 的實作，或其是否在🤗 Tokenizers 函式庫中有對應的 tokenizer，敬請參閱[此表](https://huggingface.co/docs/transformers/index#supported-frameworks)。
這些實作均已於多個資料集測試（請參閱範例腳本）並應與原版實作表現相當。你可以在範例文件的[此節](https://huggingface.co/docs/transformers/examples)中了解實作的細節。
## 了解更多
| 章節 | 描述 |
|-|-|
| [文件](https://huggingface.co/transformers/) | 完整的 API 文件和教學 |
| [任務概覽](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers 支援的任務 |
| [預處理教學](https://huggingface.co/docs/transformers/preprocessing) | 使用 `Tokenizer` 來為模型準備資料 |
| [訓練和微調](https://huggingface.co/docs/transformers/training) | 使用 PyTorch/TensorFlow 的內建的訓練方式或於 `Trainer` API 中使用 🤗 Transformers 提供的模型 |
| [快速上手：微調和範例腳本](https://github.com/huggingface/transformers/tree/main/examples) | 為各種任務提供的範例腳本 |
| [模型分享和上傳](https://huggingface.co/docs/transformers/model_sharing) | 上傳並與社群分享你微調的模型 |
| [遷移](https://huggingface.co/docs/transformers/migration) | 從 `pytorch-transformers` 或 `pytorch-pretrained-bert` 遷移到 🤗 Transformers |
## 引用
我們已將此函式庫的[論文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)正式發表。如果你使用了 🤗 Transformers 函式庫，可以引用：
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| <!---
Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<!---
A useful guide for English-Traditional Chinese translation of Hugging Face documentation
- Add space around English words and numbers when they appear between Chinese characters. E.g., 共 100 多種語言; 使用 transformers 函式庫。
- Use square quotes, e.g.,「引用」
- Some of terms in the file can be found at National Academy for Educational Research (https://terms.naer.edu.tw/), an official website providing bilingual translations between English and Traditional Chinese.
Dictionary
API: API (不翻譯）
add: 加入
checkpoint: 檢查點
code: 程式碼
community: 社群
confidence: 信賴度
dataset: 資料集
documentation: 文件
example: 基本翻譯為「範例」，或依語意翻為「例子」
finetune: 微調
Hugging Face: Hugging Face（不翻譯）
implementation: 實作
inference: 推論
library: 函式庫
module: 模組
NLP/Natural Language Processing: 以 NLP 出現時不翻譯，以 Natural Language Processing 出現時翻譯為自然語言處理
online demos: 線上Demo
pipeline: pipeline（不翻譯）
pretrained/pretrain: 預訓練
Python data structures (e.g., list, set, dict): 翻譯為串列，集合，字典，並用括號標註原英文
repository: repository（不翻譯）
summary: 概覽
token-: token-（不翻譯）
Trainer: Trainer（不翻譯）
transformer: transformer（不翻譯）
tutorial: 教學
user: 使用者
-->
<p align="center">
<br>
<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers_logo_name.png" width="400"/>
<br>
<p>
<p align="center">
<a href="https://circleci.com/gh/huggingface/transformers">
<img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/main">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/LICENSE">
<img alt="GitHub" src="https://img.shields.io/github/license/huggingface/transformers.svg?color=blue">
</a>
<a href="https://huggingface.co/docs/transformers/index">
<img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/transformers/index.svg?down_color=red&down_message=offline&up_message=online">
</a>
<a href="https://github.com/huggingface/transformers/releases">
<img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/transformers.svg">
</a>
<a href="https://github.com/huggingface/transformers/blob/main/CODE_OF_CONDUCT.md">
<img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg">
</a>
<a href="https://zenodo.org/badge/latestdoi/155220641"><img src="https://zenodo.org/badge/155220641.svg" alt="DOI"></a>
</p>
<h4 align="center">
<p>
<a href="https://github.com/huggingface/transformers/">English</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_zh-hans.md">简体中文</a> |
<b>繁體中文</b> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ko.md">한국어</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_es.md">Español</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_ja.md">日本語</a> |
<a href="https://github.com/huggingface/transformers/blob/main/README_hd.md">हिन्दी</a>
<p>
</h4>
<h3 align="center">
<p>為 Jax、PyTorch 以及 TensorFlow 打造的先進自然語言處理函式庫</p>
</h3>
<h3 align="center">
<a href="https://hf.co/course"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/course_banner.png"></a>
</h3>
🤗 Transformers 提供了數以千計的預訓練模型，支援 100 多種語言的文本分類、資訊擷取、問答、摘要、翻譯、文本生成。它的宗旨是讓最先進的 NLP 技術人人易用。
🤗 Transformers 提供了便於快速下載和使用的API，讓你可以將預訓練模型用在給定文本、在你的資料集上微調然後經由 [model hub](https://huggingface.co/models) 與社群共享。同時，每個定義的 Python 模組架構均完全獨立，方便修改和快速研究實驗。
🤗 Transformers 支援三個最熱門的深度學習函式庫： [Jax](https://jax.readthedocs.io/en/latest/), [PyTorch](https://pytorch.org/) 以及 [TensorFlow](https://www.tensorflow.org/) — 並與之完美整合。你可以直接使用其中一個框架訓練你的模型，然後用另一個載入和推論。
## 線上Demo
你可以直接在 [model hub](https://huggingface.co/models) 上測試大多數的模型。我們也提供了 [私有模型託管、模型版本管理以及推論API](https://huggingface.co/pricing)。
這裡是一些範例：
- [用 BERT 做遮蓋填詞](https://huggingface.co/bert-base-uncased?text=Paris+is+the+%5BMASK%5D+of+France)
- [用 Electra 做專有名詞辨識](https://huggingface.co/dbmdz/electra-large-discriminator-finetuned-conll03-english?text=My+name+is+Sarah+and+I+live+in+London+city)
- [用 GPT-2 做文本生成](https://huggingface.co/gpt2?text=A+long+time+ago%2C+)
- [用 RoBERTa 做自然語言推論](https://huggingface.co/roberta-large-mnli?text=The+dog+was+lost.+Nobody+lost+any+animal)
- [用 BART 做文本摘要](https://huggingface.co/facebook/bart-large-cnn?text=The+tower+is+324+metres+%281%2C063+ft%29+tall%2C+about+the+same+height+as+an+81-storey+building%2C+and+the+tallest+structure+in+Paris.+Its+base+is+square%2C+measuring+125+metres+%28410+ft%29+on+each+side.+During+its+construction%2C+the+Eiffel+Tower+surpassed+the+Washington+Monument+to+become+the+tallest+man-made+structure+in+the+world%2C+a+title+it+held+for+41+years+until+the+Chrysler+Building+in+New+York+City+was+finished+in+1930.+It+was+the+first+structure+to+reach+a+height+of+300+metres.+Due+to+the+addition+of+a+broadcasting+aerial+at+the+top+of+the+tower+in+1957%2C+it+is+now+taller+than+the+Chrysler+Building+by+5.2+metres+%2817+ft%29.+Excluding+transmitters%2C+the+Eiffel+Tower+is+the+second+tallest+free-standing+structure+in+France+after+the+Millau+Viaduct)
- [用 DistilBERT 做問答](https://huggingface.co/distilbert-base-uncased-distilled-squad?text=Which+name+is+also+used+to+describe+the+Amazon+rainforest+in+English%3F&context=The+Amazon+rainforest+%28Portuguese%3A+Floresta+Amaz%C3%B4nica+or+Amaz%C3%B4nia%3B+Spanish%3A+Selva+Amaz%C3%B3nica%2C+Amazon%C3%ADa+or+usually+Amazonia%3B+French%3A+For%C3%AAt+amazonienne%3B+Dutch%3A+Amazoneregenwoud%29%2C+also+known+in+English+as+Amazonia+or+the+Amazon+Jungle%2C+is+a+moist+broadleaf+forest+that+covers+most+of+the+Amazon+basin+of+South+America.+This+basin+encompasses+7%2C000%2C000+square+kilometres+%282%2C700%2C000+sq+mi%29%2C+of+which+5%2C500%2C000+square+kilometres+%282%2C100%2C000+sq+mi%29+are+covered+by+the+rainforest.+This+region+includes+territory+belonging+to+nine+nations.+The+majority+of+the+forest+is+contained+within+Brazil%2C+with+60%25+of+the+rainforest%2C+followed+by+Peru+with+13%25%2C+Colombia+with+10%25%2C+and+with+minor+amounts+in+Venezuela%2C+Ecuador%2C+Bolivia%2C+Guyana%2C+Suriname+and+French+Guiana.+States+or+departments+in+four+nations+contain+%22Amazonas%22+in+their+names.+The+Amazon+represents+over+half+of+the+planet%27s+remaining+rainforests%2C+and+comprises+the+largest+and+most+biodiverse+tract+of+tropical+rainforest+in+the+world%2C+with+an+estimated+390+billion+individual+trees+divided+into+16%2C000+species)
- [用 T5 做翻譯](https://huggingface.co/t5-base?text=My+name+is+Wolfgang+and+I+live+in+Berlin)
**[Write With Transformer](https://transformer.huggingface.co)**，由 Hugging Face 團隊所打造，是一個文本生成的官方 demo。
## 如果你在尋找由 Hugging Face 團隊所提供的客製化支援服務
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://huggingface.co/front/thumbnails/support.png" style="max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a><br>
## 快速上手
我們為快速使用模型提供了 `pipeline` API。 Pipeline 包含了預訓練模型和對應的文本預處理。下面是一個快速使用 pipeline 去判斷正負面情緒的例子：
```python
>>> from transformers import pipeline
# 使用情緒分析 pipeline
>>> classifier = pipeline('sentiment-analysis')
>>> classifier('We are very happy to introduce pipeline to the transformers repository.')
[{'label': 'POSITIVE', 'score': 0.9996980428695679}]
```
第二行程式碼下載並快取 pipeline 使用的預訓練模型，而第三行程式碼則在給定的文本上進行了評估。這裡的答案“正面” (positive) 具有 99.97% 的信賴度。
許多的 NLP 任務都有隨選即用的預訓練 `pipeline`。例如，我們可以輕鬆地從給定文本中擷取問題答案：
``` python
>>> from transformers import pipeline
# 使用問答 pipeline
>>> question_answerer = pipeline('question-answering')
>>> question_answerer({
... 'question': 'What is the name of the repository ?',
... 'context': 'Pipeline has been included in the huggingface/transformers repository'
... })
{'score': 0.30970096588134766, 'start': 34, 'end': 58, 'answer': 'huggingface/transformers'}
```
除了提供問題解答，預訓練模型還提供了對應的信賴度分數以及解答在 tokenized 後的文本中開始和結束的位置。你可以從[這個教學](https://huggingface.co/docs/transformers/task_summary)了解更多 `pipeline` API支援的任務。
要在你的任務中下載和使用任何預訓練模型很簡單，只需三行程式碼。這裡是 PyTorch 版的範例：
```python
>>> from transformers import AutoTokenizer, AutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = AutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="pt")
>>> outputs = model(**inputs)
```
這裡是對應的 TensorFlow 程式碼：
```python
>>> from transformers import AutoTokenizer, TFAutoModel
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> model = TFAutoModel.from_pretrained("bert-base-uncased")
>>> inputs = tokenizer("Hello world!", return_tensors="tf")
>>> outputs = model(**inputs)
```
Tokenizer 為所有的預訓練模型提供了預處理，並可以直接轉換單一字串（比如上面的例子）或串列 (list)。它會輸出一個的字典 (dict) 讓你可以在下游程式碼裡使用或直接藉由 `**` 運算式傳給模型。
模型本身是一個常規的 [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) 或 [TensorFlow `tf.keras.Model`](https://www.tensorflow.org/api_docs/python/tf/keras/Model)（取決於你的後端），可依常規方式使用。 [這個教學](https://huggingface.co/transformers/training.html)解釋了如何將這樣的模型整合到一般的 PyTorch 或 TensorFlow 訓練迴圈中，或是如何使用我們的 `Trainer` API 在一個新的資料集上快速進行微調。
## 為什麼要用 transformers？
1. 便於使用的先進模型：
- NLU 和 NLG 上性能卓越
- 對教學和實作友好且低門檻
- 高度抽象，使用者只須學習 3 個類別
- 對所有模型使用的制式化API
1. 更低的運算成本，更少的碳排放：
- 研究人員可以分享已訓練的模型而非每次從頭開始訓練
- 工程師可以減少計算時間以及生產成本
- 數十種模型架構、兩千多個預訓練模型、100多種語言支援
1. 對於模型生命週期的每一個部分都面面俱到：
- 訓練先進的模型，只需 3 行程式碼
- 模型可以在不同深度學習框架之間任意轉換
- 為訓練、評估和生產選擇最適合的框架，並完美銜接
1. 為你的需求輕鬆客製化專屬模型和範例：
- 我們為每種模型架構提供了多個範例來重現原論文結果
- 一致的模型內部架構
- 模型檔案可單獨使用，便於修改和快速實驗
## 什麼情況下我不該用 transformers？
- 本函式庫並不是模組化的神經網絡工具箱。模型文件中的程式碼並未做額外的抽象封裝，以便研究人員快速地翻閱及修改程式碼，而不會深陷複雜的類別包裝之中。
- `Trainer` API 並非相容任何模型，它只為本函式庫中的模型最佳化。對於一般的機器學習用途，請使用其他函式庫。
- 儘管我們已盡力而為，[examples 目錄](https://github.com/huggingface/transformers/tree/main/examples)中的腳本也僅為範例而已。對於特定問題，它們並不一定隨選即用，可能需要修改幾行程式碼以符合需求。
## 安裝
### 使用 pip
這個 Repository 已在 Python 3.6+、Flax 0.3.2+、PyTorch 1.3.1+ 和 TensorFlow 2.3+ 下經過測試。
你可以在[虛擬環境](https://docs.python.org/3/library/venv.html)中安裝 🤗 Transformers。如果你還不熟悉 Python 的虛擬環境，請閱此[使用者指引](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
首先，用你打算使用的版本的 Python 創建一個虛擬環境並進入。
然後，你需要安裝 Flax、PyTorch 或 TensorFlow 其中之一。對於該如何在你使用的平台上安裝這些框架，請參閱 [TensorFlow 安裝頁面](https://www.tensorflow.org/install/), [PyTorch 安裝頁面](https://pytorch.org/get-started/locally/#start-locally) 或 [Flax 安裝頁面](https://github.com/google/flax#quick-install)。
當其中一個後端安裝成功後，🤗 Transformers 可依此安裝：
```bash
pip install transformers
```
如果你想要試試範例或者想在正式發布前使用最新開發中的程式碼，你必須[從原始碼安裝](https://huggingface.co/docs/transformers/installation#installing-from-source)。
### 使用 conda
自 Transformers 4.0.0 版始，我們有了一個 conda channel： `huggingface`。
🤗 Transformers 可以藉由 conda 依此安裝：
```shell script
conda install -c huggingface transformers
```
要藉由 conda 安裝 Flax、PyTorch 或 TensorFlow 其中之一，請參閱它們各自安裝頁面的說明。
## 模型架構
**🤗 Transformers 支援的[所有的模型檢查點](https://huggingface.co/models)**，由[使用者](https://huggingface.co/users)和[組織](https://huggingface.co/organizations)上傳，均與 huggingface.co [model hub](https://huggingface.co) 完美結合。
目前的檢查點數量： 
🤗 Transformers 目前支援以下的架構（模型概覽請參閱[這裡](https://huggingface.co/docs/transformers/model_summary)）：
1. **[ALBERT](https://huggingface.co/docs/transformers/model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](https://huggingface.co/docs/transformers/model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/pdf/1910.13461.pdf) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](https://huggingface.co/docs/transformers/model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](https://huggingface.co/docs/transformers/model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](https://huggingface.co/docs/transformers/model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](https://huggingface.co/docs/transformers/model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](https://huggingface.co/docs/transformers/model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](https://huggingface.co/docs/transformers/model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](https://huggingface.co/docs/transformers/model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](https://huggingface.co/docs/transformers/model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](https://huggingface.co/docs/transformers/main/model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](https://huggingface.co/docs/transformers/main/model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](https://huggingface.co/docs/transformers/model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](https://huggingface.co/docs/transformers/model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](https://huggingface.co/docs/transformers/model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](https://huggingface.co/docs/transformers/model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](https://huggingface.co/docs/transformers/model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](https://huggingface.co/docs/transformers/model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](https://huggingface.co/docs/transformers/model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](https://huggingface.co/docs/transformers/model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](https://huggingface.co/docs/transformers/model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](https://huggingface.co/docs/transformers/model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](https://huggingface.co/docs/transformers/model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](https://huggingface.co/docs/transformers/model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](https://huggingface.co/docs/transformers/model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](https://huggingface.co/docs/transformers/model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](https://huggingface.co/docs/transformers/model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](https://huggingface.co/docs/transformers/model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](https://huggingface.co/docs/transformers/model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](https://huggingface.co/docs/transformers/model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) and a German version of DistilBERT.
1. **[DiT](https://huggingface.co/docs/transformers/model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](https://huggingface.co/docs/transformers/model_doc/donut)** (from NAVER) released with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](https://huggingface.co/docs/transformers/model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](https://huggingface.co/docs/transformers/master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](https://huggingface.co/docs/transformers/model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](https://huggingface.co/docs/transformers/model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](https://huggingface.co/docs/transformers/model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](https://huggingface.co/docs/transformers/model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2** was released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](https://huggingface.co/docs/transformers/model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](https://huggingface.co/docs/transformers/model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](https://huggingface.co/docs/transformers/model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](https://huggingface.co/docs/transformers/model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](https://huggingface.co/docs/transformers/model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](https://huggingface.co/docs/transformers/model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](https://huggingface.co/docs/transformers/model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](https://huggingface.co/docs/transformers/model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](https://huggingface.co/docs/transformers/model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)** (from EleutherAI) released with the paper [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GPT-Sw3](https://huggingface.co/docs/transformers/main/model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](https://huggingface.co/docs/transformers/model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](https://huggingface.co/docs/transformers/model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](https://huggingface.co/docs/transformers/model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](https://huggingface.co/docs/transformers/model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](https://huggingface.co/docs/transformers/model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](https://huggingface.co/docs/transformers/model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](https://huggingface.co/docs/transformers/model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](https://huggingface.co/docs/transformers/model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](https://huggingface.co/docs/transformers/model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](https://huggingface.co/docs/transformers/model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](https://huggingface.co/docs/transformers/model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](https://huggingface.co/docs/transformers/model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov
1. **[mBART](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](https://huggingface.co/docs/transformers/model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](https://huggingface.co/docs/transformers/model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](https://huggingface.co/docs/transformers/model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](https://huggingface.co/docs/transformers/model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](https://huggingface.co/docs/transformers/model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](https://huggingface.co/docs/transformers/model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](https://huggingface.co/docs/transformers/model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](https://huggingface.co/docs/transformers/model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](https://huggingface.co/docs/transformers/model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](https://huggingface.co/docs/transformers/model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](https://huggingface.co/docs/transformers/model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](https://huggingface.co/docs/transformers/model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, Peter J. Liu.
1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](https://huggingface.co/docs/transformers/model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](https://huggingface.co/docs/transformers/model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](https://huggingface.co/docs/transformers/model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](https://huggingface.co/docs/transformers/model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](https://huggingface.co/docs/transformers/model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](https://huggingface.co/docs/transformers/model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](https://huggingface.co/docs/transformers/model_doc/regnet)** (from META Research) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](https://huggingface.co/docs/transformers/model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/pdf/2010.12821.pdf) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](https://huggingface.co/docs/transformers/model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](https://huggingface.co/docs/transformers/model_doc/roberta)** (from Facebook), released together with the paper a [Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](https://huggingface.co/docs/transformers/model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper a [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](https://huggingface.co/docs/transformers/model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](https://huggingface.co/docs/transformers/model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](https://huggingface.co/docs/transformers/model_doc/speech_to_text_2)** (from Facebook) released with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](https://huggingface.co/docs/transformers/model_doc/splinter)** (from Tel Aviv University) released with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](https://huggingface.co/docs/transformers/model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](https://huggingface.co/docs/transformers/model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](https://huggingface.co/docs/transformers/model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](https://huggingface.co/docs/transformers/model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](https://huggingface.co/docs/transformers/model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](https://huggingface.co/docs/transformers/model_doc/t5v1.1)** (from Google AI) released with the paper [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](https://huggingface.co/docs/transformers/model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](https://huggingface.co/docs/transformers/model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](https://huggingface.co/docs/transformers/model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](https://huggingface.co/docs/transformers/model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](https://huggingface.co/docs/transformers/main/model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](https://huggingface.co/docs/transformers/model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](https://huggingface.co/docs/transformers/model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](https://huggingface.co/docs/transformers/model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](https://huggingface.co/docs/transformers/model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](https://huggingface.co/docs/transformers/main/model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](https://huggingface.co/docs/transformers/model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](https://huggingface.co/docs/transformers/model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](https://huggingface.co/docs/transformers/model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](https://huggingface.co/docs/transformers/model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](https://huggingface.co/docs/transformers/model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](https://huggingface.co/docs/transformers/model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](https://huggingface.co/docs/transformers/model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](https://huggingface.co/docs/transformers/model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](https://huggingface.co/docs/transformers/model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](https://huggingface.co/docs/transformers/model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](https://huggingface.co/docs/transformers/model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](https://huggingface.co/docs/transformers/model_doc/xlm-roberta-xl)** (from Facebook AI) released with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](https://huggingface.co/docs/transformers/model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](https://huggingface.co/docs/transformers/model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](https://huggingface.co/docs/transformers/model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](https://huggingface.co/docs/transformers/model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](https://huggingface.co/docs/transformers/model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
1. 想要貢獻新的模型？我們這裡有一份**詳細指引和模板**來引導你加入新的模型。你可以在 [`templates`](./templates) 目錄中找到它們。記得查看[貢獻指引](./CONTRIBUTING.md)並在開始寫 PR 前聯繫維護人員或開一個新的 issue 來獲得 feedbacks。
要檢查某個模型是否已有 Flax、PyTorch 或 TensorFlow 的實作，或其是否在🤗 Tokenizers 函式庫中有對應的 tokenizer，敬請參閱[此表](https://huggingface.co/docs/transformers/index#supported-frameworks)。
這些實作均已於多個資料集測試（請參閱範例腳本）並應與原版實作表現相當。你可以在範例文件的[此節](https://huggingface.co/docs/transformers/examples)中了解實作的細節。
## 了解更多
| 章節 | 描述 |
|-|-|
| [文件](https://huggingface.co/transformers/) | 完整的 API 文件和教學 |
| [任務概覽](https://huggingface.co/docs/transformers/task_summary) | 🤗 Transformers 支援的任務 |
| [預處理教學](https://huggingface.co/docs/transformers/preprocessing) | 使用 `Tokenizer` 來為模型準備資料 |
| [訓練和微調](https://huggingface.co/docs/transformers/training) | 使用 PyTorch/TensorFlow 的內建的訓練方式或於 `Trainer` API 中使用 🤗 Transformers 提供的模型 |
| [快速上手：微調和範例腳本](https://github.com/huggingface/transformers/tree/main/examples) | 為各種任務提供的範例腳本 |
| [模型分享和上傳](https://huggingface.co/docs/transformers/model_sharing) | 上傳並與社群分享你微調的模型 |
| [遷移](https://huggingface.co/docs/transformers/migration) | 從 `pytorch-transformers` 或 `pytorch-pretrained-bert` 遷移到 🤗 Transformers |
## 引用
我們已將此函式庫的[論文](https://www.aclweb.org/anthology/2020.emnlp-demos.6/)正式發表。如果你使用了 🤗 Transformers 函式庫，可以引用：
```bibtex
@inproceedings{wolf-etal-2020-transformers,
title = "Transformers: State-of-the-Art Natural Language Processing",
author = "Thomas Wolf and Lysandre Debut and Victor Sanh and Julien Chaumond and Clement Delangue and Anthony Moi and Pierric Cistac and Tim Rault and Rémi Louf and Morgan Funtowicz and Joe Davison and Sam Shleifer and Patrick von Platen and Clara Ma and Yacine Jernite and Julien Plu and Canwen Xu and Teven Le Scao and Sylvain Gugger and Mariama Drame and Quentin Lhoest and Alexander M. Rush",
booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.emnlp-demos.6",
pages = "38--45"
}
```
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./docs/source/en/_toctree.yml | - sections:
- local: index
title: 🤗 Transformers
- local: quicktour
title: Quick tour
- local: installation
title: Installation
title: Get started
- sections:
- local: pipeline_tutorial
title: Pipelines for inference
- local: autoclass_tutorial
title: Load pretrained instances with an AutoClass
- local: preprocessing
title: Preprocess
- local: training
title: Fine-tune a pretrained model
- local: accelerate
title: Distributed training with 🤗 Accelerate
- local: model_sharing
title: Share a model
title: Tutorials
- sections:
- sections:
- local: create_a_model
title: Create a custom architecture
- local: custom_models
title: Sharing custom models
- local: run_scripts
title: Train with a script
- local: sagemaker
title: Run training on Amazon SageMaker
- local: converting_tensorflow_models
title: Converting from TensorFlow checkpoints
- local: serialization
title: Export to ONNX
- local: torchscript
title: Export to TorchScript
- local: troubleshooting
title: Troubleshoot
title: General usage
- sections:
- local: fast_tokenizers
title: Use tokenizers from 🤗 Tokenizers
- local: multilingual
title: Inference for multilingual models
- sections:
- local: tasks/sequence_classification
title: Text classification
- local: tasks/token_classification
title: Token classification
- local: tasks/question_answering
title: Question answering
- local: tasks/language_modeling
title: Language modeling
- local: tasks/translation
title: Translation
- local: tasks/summarization
title: Summarization
- local: tasks/multiple_choice
title: Multiple choice
title: Task guides
isExpanded: false
title: Natural Language Processing
- sections:
- local: tasks/audio_classification
title: Audio classification
- local: tasks/asr
title: Automatic speech recognition
title: Audio
- sections:
- local: tasks/image_classification
title: Image classification
- local: tasks/semantic_segmentation
title: Semantic segmentation
title: Computer Vision
- sections:
- local: performance
title: Overview
- local: perf_train_gpu_one
title: Training on one GPU
- local: perf_train_gpu_many
title: Training on many GPUs
- local: perf_train_cpu
title: Training on CPU
- local: perf_train_cpu_many
title: Training on many CPUs
- local: perf_train_tpu
title: Training on TPUs
- local: perf_train_special
title: Training on Specialized Hardware
- local: perf_infer_cpu
title: Inference on CPU
- local: perf_infer_gpu_one
title: Inference on one GPU
- local: perf_infer_gpu_many
title: Inference on many GPUs
- local: perf_infer_special
title: Inference on Specialized Hardware
- local: perf_hardware
title: Custom hardware for training
- local: big_models
title: Instantiating a big model
- local: debugging
title: Debugging
- local: hpo_train
title: Hyperparameter Search using Trainer API
title: Performance and scalability
- sections:
- local: contributing
title: How to contribute to transformers?
- local: add_new_model
title: How to add a model to 🤗 Transformers?
- local: add_tensorflow_model
title: How to convert a 🤗 Transformers model to TensorFlow?
- local: add_new_pipeline
title: How to add a pipeline to 🤗 Transformers?
- local: testing
title: Testing
- local: pr_checks
title: Checks on a Pull Request
title: Contribute
- local: notebooks
title: 🤗 Transformers Notebooks
- local: community
title: Community resources
- local: benchmarks
title: Benchmarks
- local: migration
title: Migrating from previous packages
title: How-to guides
- sections:
- local: philosophy
title: Philosophy
- local: glossary
title: Glossary
- local: task_summary
title: Summary of the tasks
- local: model_summary
title: Summary of the models
- local: tokenizer_summary
title: Summary of the tokenizers
- local: pad_truncation
title: Padding and truncation
- local: bertology
title: BERTology
- local: perplexity
title: Perplexity of fixed-length models
- local: pipeline_webserver
title: Pipelines for webserver inference
title: Conceptual guides
- sections:
- sections:
- local: model_doc/auto
title: Auto Classes
- local: main_classes/callback
title: Callbacks
- local: main_classes/configuration
title: Configuration
- local: main_classes/data_collator
title: Data Collator
- local: main_classes/keras_callbacks
title: Keras callbacks
- local: main_classes/logging
title: Logging
- local: main_classes/model
title: Models
- local: main_classes/text_generation
title: Text Generation
- local: main_classes/onnx
title: ONNX
- local: main_classes/optimizer_schedules
title: Optimization
- local: main_classes/output
title: Model outputs
- local: main_classes/pipelines
title: Pipelines
- local: main_classes/processors
title: Processors
- local: main_classes/tokenizer
title: Tokenizer
- local: main_classes/trainer
title: Trainer
- local: main_classes/deepspeed
title: DeepSpeed Integration
- local: main_classes/feature_extractor
title: Feature Extractor
- local: main_classes/image_processor
title: Image Processor
title: Main Classes
- sections:
- isExpanded: false
sections:
- local: model_doc/albert
title: ALBERT
- local: model_doc/bart
title: BART
- local: model_doc/barthez
title: BARThez
- local: model_doc/bartpho
title: BARTpho
- local: model_doc/bert
title: BERT
- local: model_doc/bert-generation
title: BertGeneration
- local: model_doc/bert-japanese
title: BertJapanese
- local: model_doc/bertweet
title: Bertweet
- local: model_doc/big_bird
title: BigBird
- local: model_doc/bigbird_pegasus
title: BigBirdPegasus
- local: model_doc/biogpt
title: BioGpt
- local: model_doc/blenderbot
title: Blenderbot
- local: model_doc/blenderbot-small
title: Blenderbot Small
- local: model_doc/bloom
title: BLOOM
- local: model_doc/bort
title: BORT
- local: model_doc/byt5
title: ByT5
- local: model_doc/camembert
title: CamemBERT
- local: model_doc/canine
title: CANINE
- local: model_doc/codegen
title: CodeGen
- local: model_doc/convbert
title: ConvBERT
- local: model_doc/cpm
title: CPM
- local: model_doc/ctrl
title: CTRL
- local: model_doc/deberta
title: DeBERTa
- local: model_doc/deberta-v2
title: DeBERTa-v2
- local: model_doc/dialogpt
title: DialoGPT
- local: model_doc/distilbert
title: DistilBERT
- local: model_doc/dpr
title: DPR
- local: model_doc/electra
title: ELECTRA
- local: model_doc/encoder-decoder
title: Encoder Decoder Models
- local: model_doc/ernie
title: ERNIE
- local: model_doc/esm
title: ESM
- local: model_doc/flan-t5
title: FLAN-T5
- local: model_doc/flaubert
title: FlauBERT
- local: model_doc/fnet
title: FNet
- local: model_doc/fsmt
title: FSMT
- local: model_doc/funnel
title: Funnel Transformer
- local: model_doc/openai-gpt
title: GPT
- local: model_doc/gpt_neo
title: GPT Neo
- local: model_doc/gpt_neox
title: GPT NeoX
- local: model_doc/gpt_neox_japanese
title: GPT NeoX Japanese
- local: model_doc/gptj
title: GPT-J
- local: model_doc/gpt2
title: GPT2
- local: model_doc/herbert
title: HerBERT
- local: model_doc/ibert
title: I-BERT
- local: model_doc/jukebox
title: Jukebox
- local: model_doc/layoutlm
title: LayoutLM
- local: model_doc/led
title: LED
- local: model_doc/lilt
title: LiLT
- local: model_doc/longformer
title: Longformer
- local: model_doc/longt5
title: LongT5
- local: model_doc/luke
title: LUKE
- local: model_doc/m2m_100
title: M2M100
- local: model_doc/marian
title: MarianMT
- local: model_doc/markuplm
title: MarkupLM
- local: model_doc/mbart
title: MBart and MBart-50
- local: model_doc/megatron-bert
title: MegatronBERT
- local: model_doc/megatron_gpt2
title: MegatronGPT2
- local: model_doc/mluke
title: mLUKE
- local: model_doc/mobilebert
title: MobileBERT
- local: model_doc/mpnet
title: MPNet
- local: model_doc/mt5
title: MT5
- local: model_doc/mvp
title: MVP
- local: model_doc/nezha
title: NEZHA
- local: model_doc/nllb
title: NLLB
- local: model_doc/nystromformer
title: Nyströmformer
- local: model_doc/opt
title: OPT
- local: model_doc/pegasus
title: Pegasus
- local: model_doc/pegasus_x
title: PEGASUS-X
- local: model_doc/phobert
title: PhoBERT
- local: model_doc/plbart
title: PLBart
- local: model_doc/prophetnet
title: ProphetNet
- local: model_doc/qdqbert
title: QDQBert
- local: model_doc/rag
title: RAG
- local: model_doc/realm
title: REALM
- local: model_doc/reformer
title: Reformer
- local: model_doc/rembert
title: RemBERT
- local: model_doc/retribert
title: RetriBERT
- local: model_doc/roberta
title: RoBERTa
- local: model_doc/roc_bert
title: RoCBert
- local: model_doc/roformer
title: RoFormer
- local: model_doc/splinter
title: Splinter
- local: model_doc/squeezebert
title: SqueezeBERT
- local: model_doc/switch_transformers
title: SwitchTransformers
- local: model_doc/t5
title: T5
- local: model_doc/t5v1.1
title: T5v1.1
- local: model_doc/tapas
title: TAPAS
- local: model_doc/tapex
title: TAPEX
- local: model_doc/transfo-xl
title: Transformer XL
- local: model_doc/ul2
title: UL2
- local: model_doc/xglm
title: XGLM
- local: model_doc/xlm
title: XLM
- local: model_doc/xlm-prophetnet
title: XLM-ProphetNet
- local: model_doc/xlm-roberta
title: XLM-RoBERTa
- local: model_doc/xlm-roberta-xl
title: XLM-RoBERTa-XL
- local: model_doc/xlnet
title: XLNet
- local: model_doc/yoso
title: YOSO
title: Text models
- isExpanded: false
sections:
- local: model_doc/beit
title: BEiT
- local: model_doc/bit
title: BiT
- local: model_doc/conditional_detr
title: Conditional DETR
- local: model_doc/convnext
title: ConvNeXT
- local: model_doc/cvt
title: CvT
- local: model_doc/deformable_detr
title: Deformable DETR
- local: model_doc/deit
title: DeiT
- local: model_doc/detr
title: DETR
- local: model_doc/dinat
title: DiNAT
- local: model_doc/dit
title: DiT
- local: model_doc/dpt
title: DPT
- local: model_doc/glpn
title: GLPN
- local: model_doc/imagegpt
title: ImageGPT
- local: model_doc/levit
title: LeViT
- local: model_doc/maskformer
title: MaskFormer
- local: model_doc/mobilenet_v1
title: MobileNetV1
- local: model_doc/mobilenet_v2
title: MobileNetV2
- local: model_doc/mobilevit
title: MobileViT
- local: model_doc/nat
title: NAT
- local: model_doc/poolformer
title: PoolFormer
- local: model_doc/regnet
title: RegNet
- local: model_doc/resnet
title: ResNet
- local: model_doc/segformer
title: SegFormer
- local: model_doc/swin
title: Swin Transformer
- local: model_doc/swinv2
title: Swin Transformer V2
- local: model_doc/table-transformer
title: Table Transformer
- local: model_doc/timesformer
title: TimeSformer
- local: model_doc/van
title: VAN
- local: model_doc/videomae
title: VideoMAE
- local: model_doc/vit
title: Vision Transformer (ViT)
- local: model_doc/vit_hybrid
title: ViT Hybrid
- local: model_doc/vit_mae
title: ViTMAE
- local: model_doc/vit_msn
title: ViTMSN
- local: model_doc/yolos
title: YOLOS
title: Vision models
- isExpanded: false
sections:
- local: model_doc/audio-spectrogram-transformer
title: Audio Spectrogram Transformer
- local: model_doc/hubert
title: Hubert
- local: model_doc/mctct
title: MCTCT
- local: model_doc/sew
title: SEW
- local: model_doc/sew-d
title: SEW-D
- local: model_doc/speech_to_text
title: Speech2Text
- local: model_doc/speech_to_text_2
title: Speech2Text2
- local: model_doc/unispeech
title: UniSpeech
- local: model_doc/unispeech-sat
title: UniSpeech-SAT
- local: model_doc/wav2vec2
title: Wav2Vec2
- local: model_doc/wav2vec2-conformer
title: Wav2Vec2-Conformer
- local: model_doc/wav2vec2_phoneme
title: Wav2Vec2Phoneme
- local: model_doc/wavlm
title: WavLM
- local: model_doc/whisper
title: Whisper
- local: model_doc/xls_r
title: XLS-R
- local: model_doc/xlsr_wav2vec2
title: XLSR-Wav2Vec2
title: Audio models
- isExpanded: false
sections:
- local: model_doc/chinese_clip
title: Chinese-CLIP
- local: model_doc/clip
title: CLIP
- local: model_doc/clipseg
title: CLIPSeg
- local: model_doc/data2vec
title: Data2Vec
- local: model_doc/donut
title: Donut
- local: model_doc/flava
title: FLAVA
- local: model_doc/groupvit
title: GroupViT
- local: model_doc/layoutlmv2
title: LayoutLMV2
- local: model_doc/layoutlmv3
title: LayoutLMV3
- local: model_doc/layoutxlm
title: LayoutXLM
- local: model_doc/lxmert
title: LXMERT
- local: model_doc/owlvit
title: OWL-ViT
- local: model_doc/perceiver
title: Perceiver
- local: model_doc/speech-encoder-decoder
title: Speech Encoder Decoder Models
- local: model_doc/trocr
title: TrOCR
- local: model_doc/vilt
title: ViLT
- local: model_doc/vision-encoder-decoder
title: Vision Encoder Decoder Models
- local: model_doc/vision-text-dual-encoder
title: Vision Text Dual Encoder
- local: model_doc/visual_bert
title: VisualBERT
- local: model_doc/xclip
title: X-CLIP
title: Multimodal models
- isExpanded: false
sections:
- local: model_doc/decision_transformer
title: Decision Transformer
- local: model_doc/trajectory_transformer
title: Trajectory Transformer
title: Reinforcement learning models
- isExpanded: false
sections:
- local: model_doc/time_series_transformer
title: Time Series Transformer
title: Time series models
title: Models
- sections:
- local: internal/modeling_utils
title: Custom Layers and Utilities
- local: internal/pipelines_utils
title: Utilities for pipelines
- local: internal/tokenization_utils
title: Utilities for Tokenizers
- local: internal/trainer_utils
title: Utilities for Trainer
- local: internal/generation_utils
title: Utilities for Generation
- local: internal/image_processing_utils
title: Utilities for Image Processors
- local: internal/file_utils
title: General Utilities
title: Internal Helpers
title: API
| - sections:
- local: index
title: 🤗 Transformers
- local: quicktour
title: Quick tour
- local: installation
title: Installation
title: Get started
- sections:
- local: pipeline_tutorial
title: Pipelines for inference
- local: autoclass_tutorial
title: Load pretrained instances with an AutoClass
- local: preprocessing
title: Preprocess
- local: training
title: Fine-tune a pretrained model
- local: accelerate
title: Distributed training with 🤗 Accelerate
- local: model_sharing
title: Share a model
title: Tutorials
- sections:
- sections:
- local: create_a_model
title: Create a custom architecture
- local: custom_models
title: Sharing custom models
- local: run_scripts
title: Train with a script
- local: sagemaker
title: Run training on Amazon SageMaker
- local: converting_tensorflow_models
title: Converting from TensorFlow checkpoints
- local: serialization
title: Export to ONNX
- local: torchscript
title: Export to TorchScript
- local: troubleshooting
title: Troubleshoot
title: General usage
- sections:
- local: fast_tokenizers
title: Use tokenizers from 🤗 Tokenizers
- local: multilingual
title: Inference for multilingual models
- sections:
- local: tasks/sequence_classification
title: Text classification
- local: tasks/token_classification
title: Token classification
- local: tasks/question_answering
title: Question answering
- local: tasks/language_modeling
title: Language modeling
- local: tasks/translation
title: Translation
- local: tasks/summarization
title: Summarization
- local: tasks/multiple_choice
title: Multiple choice
title: Task guides
isExpanded: false
title: Natural Language Processing
- sections:
- local: tasks/audio_classification
title: Audio classification
- local: tasks/asr
title: Automatic speech recognition
title: Audio
- sections:
- local: tasks/image_classification
title: Image classification
- local: tasks/semantic_segmentation
title: Semantic segmentation
title: Computer Vision
- sections:
- local: performance
title: Overview
- local: perf_train_gpu_one
title: Training on one GPU
- local: perf_train_gpu_many
title: Training on many GPUs
- local: perf_train_cpu
title: Training on CPU
- local: perf_train_cpu_many
title: Training on many CPUs
- local: perf_train_tpu
title: Training on TPUs
- local: perf_train_special
title: Training on Specialized Hardware
- local: perf_infer_cpu
title: Inference on CPU
- local: perf_infer_gpu_one
title: Inference on one GPU
- local: perf_infer_gpu_many
title: Inference on many GPUs
- local: perf_infer_special
title: Inference on Specialized Hardware
- local: perf_hardware
title: Custom hardware for training
- local: big_models
title: Instantiating a big model
- local: debugging
title: Debugging
- local: hpo_train
title: Hyperparameter Search using Trainer API
title: Performance and scalability
- sections:
- local: contributing
title: How to contribute to transformers?
- local: add_new_model
title: How to add a model to 🤗 Transformers?
- local: add_tensorflow_model
title: How to convert a 🤗 Transformers model to TensorFlow?
- local: add_new_pipeline
title: How to add a pipeline to 🤗 Transformers?
- local: testing
title: Testing
- local: pr_checks
title: Checks on a Pull Request
title: Contribute
- local: notebooks
title: 🤗 Transformers Notebooks
- local: community
title: Community resources
- local: benchmarks
title: Benchmarks
- local: migration
title: Migrating from previous packages
title: How-to guides
- sections:
- local: philosophy
title: Philosophy
- local: glossary
title: Glossary
- local: task_summary
title: Summary of the tasks
- local: model_summary
title: Summary of the models
- local: tokenizer_summary
title: Summary of the tokenizers
- local: pad_truncation
title: Padding and truncation
- local: bertology
title: BERTology
- local: perplexity
title: Perplexity of fixed-length models
- local: pipeline_webserver
title: Pipelines for webserver inference
title: Conceptual guides
- sections:
- sections:
- local: model_doc/auto
title: Auto Classes
- local: main_classes/callback
title: Callbacks
- local: main_classes/configuration
title: Configuration
- local: main_classes/data_collator
title: Data Collator
- local: main_classes/keras_callbacks
title: Keras callbacks
- local: main_classes/logging
title: Logging
- local: main_classes/model
title: Models
- local: main_classes/text_generation
title: Text Generation
- local: main_classes/onnx
title: ONNX
- local: main_classes/optimizer_schedules
title: Optimization
- local: main_classes/output
title: Model outputs
- local: main_classes/pipelines
title: Pipelines
- local: main_classes/processors
title: Processors
- local: main_classes/tokenizer
title: Tokenizer
- local: main_classes/trainer
title: Trainer
- local: main_classes/deepspeed
title: DeepSpeed Integration
- local: main_classes/feature_extractor
title: Feature Extractor
- local: main_classes/image_processor
title: Image Processor
title: Main Classes
- sections:
- isExpanded: false
sections:
- local: model_doc/albert
title: ALBERT
- local: model_doc/bart
title: BART
- local: model_doc/barthez
title: BARThez
- local: model_doc/bartpho
title: BARTpho
- local: model_doc/bert
title: BERT
- local: model_doc/bert-generation
title: BertGeneration
- local: model_doc/bert-japanese
title: BertJapanese
- local: model_doc/bertweet
title: Bertweet
- local: model_doc/big_bird
title: BigBird
- local: model_doc/bigbird_pegasus
title: BigBirdPegasus
- local: model_doc/biogpt
title: BioGpt
- local: model_doc/blenderbot
title: Blenderbot
- local: model_doc/blenderbot-small
title: Blenderbot Small
- local: model_doc/bloom
title: BLOOM
- local: model_doc/bort
title: BORT
- local: model_doc/byt5
title: ByT5
- local: model_doc/camembert
title: CamemBERT
- local: model_doc/canine
title: CANINE
- local: model_doc/codegen
title: CodeGen
- local: model_doc/convbert
title: ConvBERT
- local: model_doc/cpm
title: CPM
- local: model_doc/ctrl
title: CTRL
- local: model_doc/deberta
title: DeBERTa
- local: model_doc/deberta-v2
title: DeBERTa-v2
- local: model_doc/dialogpt
title: DialoGPT
- local: model_doc/distilbert
title: DistilBERT
- local: model_doc/dpr
title: DPR
- local: model_doc/electra
title: ELECTRA
- local: model_doc/encoder-decoder
title: Encoder Decoder Models
- local: model_doc/ernie
title: ERNIE
- local: model_doc/esm
title: ESM
- local: model_doc/flan-t5
title: FLAN-T5
- local: model_doc/flaubert
title: FlauBERT
- local: model_doc/fnet
title: FNet
- local: model_doc/fsmt
title: FSMT
- local: model_doc/funnel
title: Funnel Transformer
- local: model_doc/openai-gpt
title: GPT
- local: model_doc/gpt_neo
title: GPT Neo
- local: model_doc/gpt_neox
title: GPT NeoX
- local: model_doc/gpt_neox_japanese
title: GPT NeoX Japanese
- local: model_doc/gptj
title: GPT-J
- local: model_doc/gpt2
title: GPT2
- local: model_doc/gpt-sw3
title: GPTSw3
- local: model_doc/herbert
title: HerBERT
- local: model_doc/ibert
title: I-BERT
- local: model_doc/jukebox
title: Jukebox
- local: model_doc/layoutlm
title: LayoutLM
- local: model_doc/led
title: LED
- local: model_doc/lilt
title: LiLT
- local: model_doc/longformer
title: Longformer
- local: model_doc/longt5
title: LongT5
- local: model_doc/luke
title: LUKE
- local: model_doc/m2m_100
title: M2M100
- local: model_doc/marian
title: MarianMT
- local: model_doc/markuplm
title: MarkupLM
- local: model_doc/mbart
title: MBart and MBart-50
- local: model_doc/megatron-bert
title: MegatronBERT
- local: model_doc/megatron_gpt2
title: MegatronGPT2
- local: model_doc/mluke
title: mLUKE
- local: model_doc/mobilebert
title: MobileBERT
- local: model_doc/mpnet
title: MPNet
- local: model_doc/mt5
title: MT5
- local: model_doc/mvp
title: MVP
- local: model_doc/nezha
title: NEZHA
- local: model_doc/nllb
title: NLLB
- local: model_doc/nystromformer
title: Nyströmformer
- local: model_doc/opt
title: OPT
- local: model_doc/pegasus
title: Pegasus
- local: model_doc/pegasus_x
title: PEGASUS-X
- local: model_doc/phobert
title: PhoBERT
- local: model_doc/plbart
title: PLBart
- local: model_doc/prophetnet
title: ProphetNet
- local: model_doc/qdqbert
title: QDQBert
- local: model_doc/rag
title: RAG
- local: model_doc/realm
title: REALM
- local: model_doc/reformer
title: Reformer
- local: model_doc/rembert
title: RemBERT
- local: model_doc/retribert
title: RetriBERT
- local: model_doc/roberta
title: RoBERTa
- local: model_doc/roc_bert
title: RoCBert
- local: model_doc/roformer
title: RoFormer
- local: model_doc/splinter
title: Splinter
- local: model_doc/squeezebert
title: SqueezeBERT
- local: model_doc/switch_transformers
title: SwitchTransformers
- local: model_doc/t5
title: T5
- local: model_doc/t5v1.1
title: T5v1.1
- local: model_doc/tapas
title: TAPAS
- local: model_doc/tapex
title: TAPEX
- local: model_doc/transfo-xl
title: Transformer XL
- local: model_doc/ul2
title: UL2
- local: model_doc/xglm
title: XGLM
- local: model_doc/xlm
title: XLM
- local: model_doc/xlm-prophetnet
title: XLM-ProphetNet
- local: model_doc/xlm-roberta
title: XLM-RoBERTa
- local: model_doc/xlm-roberta-xl
title: XLM-RoBERTa-XL
- local: model_doc/xlnet
title: XLNet
- local: model_doc/yoso
title: YOSO
title: Text models
- isExpanded: false
sections:
- local: model_doc/beit
title: BEiT
- local: model_doc/bit
title: BiT
- local: model_doc/conditional_detr
title: Conditional DETR
- local: model_doc/convnext
title: ConvNeXT
- local: model_doc/cvt
title: CvT
- local: model_doc/deformable_detr
title: Deformable DETR
- local: model_doc/deit
title: DeiT
- local: model_doc/detr
title: DETR
- local: model_doc/dinat
title: DiNAT
- local: model_doc/dit
title: DiT
- local: model_doc/dpt
title: DPT
- local: model_doc/glpn
title: GLPN
- local: model_doc/imagegpt
title: ImageGPT
- local: model_doc/levit
title: LeViT
- local: model_doc/maskformer
title: MaskFormer
- local: model_doc/mobilenet_v1
title: MobileNetV1
- local: model_doc/mobilenet_v2
title: MobileNetV2
- local: model_doc/mobilevit
title: MobileViT
- local: model_doc/nat
title: NAT
- local: model_doc/poolformer
title: PoolFormer
- local: model_doc/regnet
title: RegNet
- local: model_doc/resnet
title: ResNet
- local: model_doc/segformer
title: SegFormer
- local: model_doc/swin
title: Swin Transformer
- local: model_doc/swinv2
title: Swin Transformer V2
- local: model_doc/table-transformer
title: Table Transformer
- local: model_doc/timesformer
title: TimeSformer
- local: model_doc/van
title: VAN
- local: model_doc/videomae
title: VideoMAE
- local: model_doc/vit
title: Vision Transformer (ViT)
- local: model_doc/vit_hybrid
title: ViT Hybrid
- local: model_doc/vit_mae
title: ViTMAE
- local: model_doc/vit_msn
title: ViTMSN
- local: model_doc/yolos
title: YOLOS
title: Vision models
- isExpanded: false
sections:
- local: model_doc/audio-spectrogram-transformer
title: Audio Spectrogram Transformer
- local: model_doc/hubert
title: Hubert
- local: model_doc/mctct
title: MCTCT
- local: model_doc/sew
title: SEW
- local: model_doc/sew-d
title: SEW-D
- local: model_doc/speech_to_text
title: Speech2Text
- local: model_doc/speech_to_text_2
title: Speech2Text2
- local: model_doc/unispeech
title: UniSpeech
- local: model_doc/unispeech-sat
title: UniSpeech-SAT
- local: model_doc/wav2vec2
title: Wav2Vec2
- local: model_doc/wav2vec2-conformer
title: Wav2Vec2-Conformer
- local: model_doc/wav2vec2_phoneme
title: Wav2Vec2Phoneme
- local: model_doc/wavlm
title: WavLM
- local: model_doc/whisper
title: Whisper
- local: model_doc/xls_r
title: XLS-R
- local: model_doc/xlsr_wav2vec2
title: XLSR-Wav2Vec2
title: Audio models
- isExpanded: false
sections:
- local: model_doc/chinese_clip
title: Chinese-CLIP
- local: model_doc/clip
title: CLIP
- local: model_doc/clipseg
title: CLIPSeg
- local: model_doc/data2vec
title: Data2Vec
- local: model_doc/donut
title: Donut
- local: model_doc/flava
title: FLAVA
- local: model_doc/groupvit
title: GroupViT
- local: model_doc/layoutlmv2
title: LayoutLMV2
- local: model_doc/layoutlmv3
title: LayoutLMV3
- local: model_doc/layoutxlm
title: LayoutXLM
- local: model_doc/lxmert
title: LXMERT
- local: model_doc/owlvit
title: OWL-ViT
- local: model_doc/perceiver
title: Perceiver
- local: model_doc/speech-encoder-decoder
title: Speech Encoder Decoder Models
- local: model_doc/trocr
title: TrOCR
- local: model_doc/vilt
title: ViLT
- local: model_doc/vision-encoder-decoder
title: Vision Encoder Decoder Models
- local: model_doc/vision-text-dual-encoder
title: Vision Text Dual Encoder
- local: model_doc/visual_bert
title: VisualBERT
- local: model_doc/xclip
title: X-CLIP
title: Multimodal models
- isExpanded: false
sections:
- local: model_doc/decision_transformer
title: Decision Transformer
- local: model_doc/trajectory_transformer
title: Trajectory Transformer
title: Reinforcement learning models
- isExpanded: false
sections:
- local: model_doc/time_series_transformer
title: Time Series Transformer
title: Time series models
title: Models
- sections:
- local: internal/modeling_utils
title: Custom Layers and Utilities
- local: internal/pipelines_utils
title: Utilities for pipelines
- local: internal/tokenization_utils
title: Utilities for Tokenizers
- local: internal/trainer_utils
title: Utilities for Trainer
- local: internal/generation_utils
title: Utilities for Generation
- local: internal/image_processing_utils
title: Utilities for Image Processors
- local: internal/file_utils
title: General Utilities
title: Internal Helpers
title: API
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./docs/source/en/index.mdx | <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# 🤗 Transformers
State-of-the-art Machine Learning for [PyTorch](https://pytorch.org/), [TensorFlow](https://www.tensorflow.org/), and [JAX](https://jax.readthedocs.io/en/latest/).
🤗 Transformers provides APIs and tools to easily download and train state-of-the-art pretrained models. Using pretrained models can reduce your compute costs, carbon footprint, and save you the time and resources required to train a model from scratch. These models support common tasks in different modalities, such as:
📝 **Natural Language Processing**: text classification, named entity recognition, question answering, language modeling, summarization, translation, multiple choice, and text generation.<br>
🖼️ **Computer Vision**: image classification, object detection, and segmentation.<br>
🗣️ **Audio**: automatic speech recognition and audio classification.<br>
🐙 **Multimodal**: table question answering, optical character recognition, information extraction from scanned documents, video classification, and visual question answering.
🤗 Transformers support framework interoperability between PyTorch, TensorFlow, and JAX. This provides the flexibility to use a different framework at each stage of a model's life; train a model in three lines of code in one framework, and load it for inference in another. Models can also be exported to a format like ONNX and TorchScript for deployment in production environments.
Join the growing community on the [Hub](https://huggingface.co/models), [forum](https://discuss.huggingface.co/), or [Discord](https://discord.com/invite/JfAtkvEtRb) today!
## If you are looking for custom support from the Hugging Face team
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="width: 100%; max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a>
## Contents
The documentation is organized into five sections:
- **GET STARTED** provides a quick tour of the library and installation instructions to get up and running.
- **TUTORIALS** are a great place to start if you're a beginner. This section will help you gain the basic skills you need to start using the library.
- **HOW-TO GUIDES** show you how to achieve a specific goal, like finetuning a pretrained model for language modeling or how to write and share a custom model.
- **CONCEPTUAL GUIDES** offers more discussion and explanation of the underlying concepts and ideas behind models, tasks, and the design philosophy of 🤗 Transformers.
- **API** describes all classes and functions:
- **MAIN CLASSES** details the most important classes like configuration, model, tokenizer, and pipeline.
- **MODELS** details the classes and functions related to each model implemented in the library.
- **INTERNAL HELPERS** details utility classes and functions used internally.
### Supported models
<!--This list is updated automatically from the README with _make fix-copies_. Do not update manually! -->
1. **[ALBERT](model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT): General Visual Representation Learning](https://arxiv.org/abs/1912.11370) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2 and ESMFold** were released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GroupViT](model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
### Supported frameworks
The table below represents the current support in the library for each of those models, whether they have a Python
tokenizer (called "slow"). A "fast" tokenizer backed by the 🤗 Tokenizers library, whether they have support in Jax (via
Flax), PyTorch, and/or TensorFlow.
<!--This table is updated automatically from the auto modules with _make fix-copies_. Do not update manually!-->
| Model | Tokenizer slow | Tokenizer fast | PyTorch support | TensorFlow support | Flax Support |
|:-----------------------------:|:--------------:|:--------------:|:---------------:|:------------------:|:------------:|
| ALBERT | ✅ | ✅ | ✅ | ✅ | ✅ |
| Audio Spectrogram Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| BART | ✅ | ✅ | ✅ | ✅ | ✅ |
| BEiT | ❌ | ❌ | ✅ | ❌ | ✅ |
| BERT | ✅ | ✅ | ✅ | ✅ | ✅ |
| Bert Generation | ✅ | ❌ | ✅ | ❌ | ❌ |
| BigBird | ✅ | ✅ | ✅ | ❌ | ✅ |
| BigBird-Pegasus | ❌ | ❌ | ✅ | ❌ | ❌ |
| BioGpt | ✅ | ❌ | ✅ | ❌ | ❌ |
| BiT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Blenderbot | ✅ | ✅ | ✅ | ✅ | ✅ |
| BlenderbotSmall | ✅ | ✅ | ✅ | ✅ | ✅ |
| BLOOM | ❌ | ✅ | ✅ | ❌ | ❌ |
| CamemBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| CANINE | ✅ | ❌ | ✅ | ❌ | ❌ |
| Chinese-CLIP | ❌ | ❌ | ✅ | ❌ | ❌ |
| CLIP | ✅ | ✅ | ✅ | ✅ | ✅ |
| CLIPSeg | ❌ | ❌ | ✅ | ❌ | ❌ |
| CodeGen | ✅ | ✅ | ✅ | ❌ | ❌ |
| Conditional DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
| ConvBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| ConvNeXT | ❌ | ❌ | ✅ | ✅ | ❌ |
| CTRL | ✅ | ❌ | ✅ | ✅ | ❌ |
| CvT | ❌ | ❌ | ✅ | ✅ | ❌ |
| Data2VecAudio | ❌ | ❌ | ✅ | ❌ | ❌ |
| Data2VecText | ❌ | ❌ | ✅ | ❌ | ❌ |
| Data2VecVision | ❌ | ❌ | ✅ | ✅ | ❌ |
| DeBERTa | ✅ | ✅ | ✅ | ✅ | ❌ |
| DeBERTa-v2 | ✅ | ✅ | ✅ | ✅ | ❌ |
| Decision Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| Deformable DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
| DeiT | ❌ | ❌ | ✅ | ✅ | ❌ |
| DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
| DiNAT | ❌ | ❌ | ✅ | ❌ | ❌ |
| DistilBERT | ✅ | ✅ | ✅ | ✅ | ✅ |
| DonutSwin | ❌ | ❌ | ✅ | ❌ | ❌ |
| DPR | ✅ | ✅ | ✅ | ✅ | ❌ |
| DPT | ❌ | ❌ | ✅ | ❌ | ❌ |
| ELECTRA | ✅ | ✅ | ✅ | ✅ | ✅ |
| Encoder decoder | ❌ | ❌ | ✅ | ✅ | ✅ |
| ERNIE | ❌ | ❌ | ✅ | ❌ | ❌ |
| ESM | ✅ | ❌ | ✅ | ✅ | ❌ |
| FairSeq Machine-Translation | ✅ | ❌ | ✅ | ❌ | ❌ |
| FlauBERT | ✅ | ❌ | ✅ | ✅ | ❌ |
| FLAVA | ❌ | ❌ | ✅ | ❌ | ❌ |
| FNet | ✅ | ✅ | ✅ | ❌ | ❌ |
| Funnel Transformer | ✅ | ✅ | ✅ | ✅ | ❌ |
| GLPN | ❌ | ❌ | ✅ | ❌ | ❌ |
| GPT Neo | ❌ | ❌ | ✅ | ❌ | ✅ |
| GPT NeoX | ❌ | ✅ | ✅ | ❌ | ❌ |
| GPT NeoX Japanese | ✅ | ❌ | ✅ | ❌ | ❌ |
| GPT-J | ❌ | ❌ | ✅ | ✅ | ✅ |
| GroupViT | ❌ | ❌ | ✅ | ✅ | ❌ |
| Hubert | ❌ | ❌ | ✅ | ✅ | ❌ |
| I-BERT | ❌ | ❌ | ✅ | ❌ | ❌ |
| ImageGPT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Jukebox | ✅ | ❌ | ✅ | ❌ | ❌ |
| LayoutLM | ✅ | ✅ | ✅ | ✅ | ❌ |
| LayoutLMv2 | ✅ | ✅ | ✅ | ❌ | ❌ |
| LayoutLMv3 | ✅ | ✅ | ✅ | ✅ | ❌ |
| LED | ✅ | ✅ | ✅ | ✅ | ❌ |
| LeViT | ❌ | ❌ | ✅ | ❌ | ❌ |
| LiLT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Longformer | ✅ | ✅ | ✅ | ✅ | ❌ |
| LongT5 | ❌ | ❌ | ✅ | ❌ | ✅ |
| LUKE | ✅ | ❌ | ✅ | ❌ | ❌ |
| LXMERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| M-CTC-T | ❌ | ❌ | ✅ | ❌ | ❌ |
| M2M100 | ✅ | ❌ | ✅ | ❌ | ❌ |
| Marian | ✅ | ❌ | ✅ | ✅ | ✅ |
| MarkupLM | ✅ | ✅ | ✅ | ❌ | ❌ |
| MaskFormer | ❌ | ❌ | ✅ | ❌ | ❌ |
| MaskFormerSwin | ❌ | ❌ | ❌ | ❌ | ❌ |
| mBART | ✅ | ✅ | ✅ | ✅ | ✅ |
| Megatron-BERT | ❌ | ❌ | ✅ | ❌ | ❌ |
| MobileBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| MobileNetV1 | ❌ | ❌ | ✅ | ❌ | ❌ |
| MobileNetV2 | ❌ | ❌ | ✅ | ❌ | ❌ |
| MobileViT | ❌ | ❌ | ✅ | ✅ | ❌ |
| MPNet | ✅ | ✅ | ✅ | ✅ | ❌ |
| MT5 | ✅ | ✅ | ✅ | ✅ | ✅ |
| MVP | ✅ | ✅ | ✅ | ❌ | ❌ |
| NAT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Nezha | ❌ | ❌ | ✅ | ❌ | ❌ |
| Nyströmformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| OpenAI GPT | ✅ | ✅ | ✅ | ✅ | ❌ |
| OpenAI GPT-2 | ✅ | ✅ | ✅ | ✅ | ✅ |
| OPT | ❌ | ❌ | ✅ | ✅ | ✅ |
| OWL-ViT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Pegasus | ✅ | ✅ | ✅ | ✅ | ✅ |
| PEGASUS-X | ❌ | ❌ | ✅ | ❌ | ❌ |
| Perceiver | ✅ | ❌ | ✅ | ❌ | ❌ |
| PLBart | ✅ | ❌ | ✅ | ❌ | ❌ |
| PoolFormer | ❌ | ❌ | ✅ | ❌ | ❌ |
| ProphetNet | ✅ | ❌ | ✅ | ❌ | ❌ |
| QDQBert | ❌ | ❌ | ✅ | ❌ | ❌ |
| RAG | ✅ | ❌ | ✅ | ✅ | ❌ |
| REALM | ✅ | ✅ | ✅ | ❌ | ❌ |
| Reformer | ✅ | ✅ | ✅ | ❌ | ❌ |
| RegNet | ❌ | ❌ | ✅ | ✅ | ❌ |
| RemBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| ResNet | ❌ | ❌ | ✅ | ✅ | ❌ |
| RetriBERT | ✅ | ✅ | ✅ | ❌ | ❌ |
| RoBERTa | ✅ | ✅ | ✅ | ✅ | ✅ |
| RoCBert | ✅ | ❌ | ✅ | ❌ | ❌ |
| RoFormer | ✅ | ✅ | ✅ | ✅ | ✅ |
| SegFormer | ❌ | ❌ | ✅ | ✅ | ❌ |
| SEW | ❌ | ❌ | ✅ | ❌ | ❌ |
| SEW-D | ❌ | ❌ | ✅ | ❌ | ❌ |
| Speech Encoder decoder | ❌ | ❌ | ✅ | ❌ | ✅ |
| Speech2Text | ✅ | ❌ | ✅ | ✅ | ❌ |
| Speech2Text2 | ✅ | ❌ | ❌ | ❌ | ❌ |
| Splinter | ✅ | ✅ | ✅ | ❌ | ❌ |
| SqueezeBERT | ✅ | ✅ | ✅ | ❌ | ❌ |
| Swin Transformer | ❌ | ❌ | ✅ | ✅ | ❌ |
| Swin Transformer V2 | ❌ | ❌ | ✅ | ❌ | ❌ |
| SwitchTransformers | ❌ | ❌ | ✅ | ❌ | ❌ |
| T5 | ✅ | ✅ | ✅ | ✅ | ✅ |
| Table Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| TAPAS | ✅ | ❌ | ✅ | ✅ | ❌ |
| Time Series Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| TimeSformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| Trajectory Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| Transformer-XL | ✅ | ❌ | ✅ | ✅ | ❌ |
| TrOCR | ❌ | ❌ | ✅ | ❌ | ❌ |
| UniSpeech | ❌ | ❌ | ✅ | ❌ | ❌ |
| UniSpeechSat | ❌ | ❌ | ✅ | ❌ | ❌ |
| VAN | ❌ | ❌ | ✅ | ❌ | ❌ |
| VideoMAE | ❌ | ❌ | ✅ | ❌ | ❌ |
| ViLT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Vision Encoder decoder | ❌ | ❌ | ✅ | ✅ | ✅ |
| VisionTextDualEncoder | ❌ | ❌ | ✅ | ❌ | ✅ |
| VisualBERT | ❌ | ❌ | ✅ | ❌ | ❌ |
| ViT | ❌ | ❌ | ✅ | ✅ | ✅ |
| ViT Hybrid | ❌ | ❌ | ✅ | ❌ | ❌ |
| ViTMAE | ❌ | ❌ | ✅ | ✅ | ❌ |
| ViTMSN | ❌ | ❌ | ✅ | ❌ | ❌ |
| Wav2Vec2 | ✅ | ❌ | ✅ | ✅ | ✅ |
| Wav2Vec2-Conformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| WavLM | ❌ | ❌ | ✅ | ❌ | ❌ |
| Whisper | ✅ | ❌ | ✅ | ✅ | ❌ |
| X-CLIP | ❌ | ❌ | ✅ | ❌ | ❌ |
| XGLM | ✅ | ✅ | ✅ | ✅ | ✅ |
| XLM | ✅ | ❌ | ✅ | ✅ | ❌ |
| XLM-ProphetNet | ✅ | ❌ | ✅ | ❌ | ❌ |
| XLM-RoBERTa | ✅ | ✅ | ✅ | ✅ | ✅ |
| XLM-RoBERTa-XL | ❌ | ❌ | ✅ | ❌ | ❌ |
| XLNet | ✅ | ✅ | ✅ | ✅ | ❌ |
| YOLOS | ❌ | ❌ | ✅ | ❌ | ❌ |
| YOSO | ❌ | ❌ | ✅ | ❌ | ❌ |
<!-- End table--> | <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# 🤗 Transformers
State-of-the-art Machine Learning for [PyTorch](https://pytorch.org/), [TensorFlow](https://www.tensorflow.org/), and [JAX](https://jax.readthedocs.io/en/latest/).
🤗 Transformers provides APIs and tools to easily download and train state-of-the-art pretrained models. Using pretrained models can reduce your compute costs, carbon footprint, and save you the time and resources required to train a model from scratch. These models support common tasks in different modalities, such as:
📝 **Natural Language Processing**: text classification, named entity recognition, question answering, language modeling, summarization, translation, multiple choice, and text generation.<br>
🖼️ **Computer Vision**: image classification, object detection, and segmentation.<br>
🗣️ **Audio**: automatic speech recognition and audio classification.<br>
🐙 **Multimodal**: table question answering, optical character recognition, information extraction from scanned documents, video classification, and visual question answering.
🤗 Transformers support framework interoperability between PyTorch, TensorFlow, and JAX. This provides the flexibility to use a different framework at each stage of a model's life; train a model in three lines of code in one framework, and load it for inference in another. Models can also be exported to a format like ONNX and TorchScript for deployment in production environments.
Join the growing community on the [Hub](https://huggingface.co/models), [forum](https://discuss.huggingface.co/), or [Discord](https://discord.com/invite/JfAtkvEtRb) today!
## If you are looking for custom support from the Hugging Face team
<a target="_blank" href="https://huggingface.co/support">
<img alt="HuggingFace Expert Acceleration Program" src="https://cdn-media.huggingface.co/marketing/transformers/new-support-improved.png" style="width: 100%; max-width: 600px; border: 1px solid #eee; border-radius: 4px; box-shadow: 0 1px 2px 0 rgba(0, 0, 0, 0.05);">
</a>
## Contents
The documentation is organized into five sections:
- **GET STARTED** provides a quick tour of the library and installation instructions to get up and running.
- **TUTORIALS** are a great place to start if you're a beginner. This section will help you gain the basic skills you need to start using the library.
- **HOW-TO GUIDES** show you how to achieve a specific goal, like finetuning a pretrained model for language modeling or how to write and share a custom model.
- **CONCEPTUAL GUIDES** offers more discussion and explanation of the underlying concepts and ideas behind models, tasks, and the design philosophy of 🤗 Transformers.
- **API** describes all classes and functions:
- **MAIN CLASSES** details the most important classes like configuration, model, tokenizer, and pipeline.
- **MODELS** details the classes and functions related to each model implemented in the library.
- **INTERNAL HELPERS** details utility classes and functions used internally.
### Supported models
<!--This list is updated automatically from the README with _make fix-copies_. Do not update manually! -->
1. **[ALBERT](model_doc/albert)** (from Google Research and the Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.
1. **[Audio Spectrogram Transformer](model_doc/audio-spectrogram-transformer)** (from MIT) released with the paper [AST: Audio Spectrogram Transformer](https://arxiv.org/abs/2104.01778) by Yuan Gong, Yu-An Chung, James Glass.
1. **[BART](model_doc/bart)** (from Facebook) released with the paper [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://arxiv.org/abs/1910.13461) by Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov and Luke Zettlemoyer.
1. **[BARThez](model_doc/barthez)** (from École polytechnique) released with the paper [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321) by Moussa Kamal Eddine, Antoine J.-P. Tixier, Michalis Vazirgiannis.
1. **[BARTpho](model_doc/bartpho)** (from VinAI Research) released with the paper [BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese](https://arxiv.org/abs/2109.09701) by Nguyen Luong Tran, Duong Minh Le and Dat Quoc Nguyen.
1. **[BEiT](model_doc/beit)** (from Microsoft) released with the paper [BEiT: BERT Pre-Training of Image Transformers](https://arxiv.org/abs/2106.08254) by Hangbo Bao, Li Dong, Furu Wei.
1. **[BERT](model_doc/bert)** (from Google) released with the paper [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805) by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
1. **[BERT For Sequence Generation](model_doc/bert-generation)** (from Google) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[BERTweet](model_doc/bertweet)** (from VinAI Research) released with the paper [BERTweet: A pre-trained language model for English Tweets](https://aclanthology.org/2020.emnlp-demos.2/) by Dat Quoc Nguyen, Thanh Vu and Anh Tuan Nguyen.
1. **[BigBird-Pegasus](model_doc/bigbird_pegasus)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BigBird-RoBERTa](model_doc/big_bird)** (from Google Research) released with the paper [Big Bird: Transformers for Longer Sequences](https://arxiv.org/abs/2007.14062) by Manzil Zaheer, Guru Guruganesh, Avinava Dubey, Joshua Ainslie, Chris Alberti, Santiago Ontanon, Philip Pham, Anirudh Ravula, Qifan Wang, Li Yang, Amr Ahmed.
1. **[BioGpt](model_doc/biogpt)** (from Microsoft Research AI4Science) released with the paper [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu.
1. **[BiT](model_doc/bit)** (from Google AI) released with the paper [Big Transfer (BiT): General Visual Representation Learning](https://arxiv.org/abs/1912.11370) by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby.
1. **[Blenderbot](model_doc/blenderbot)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BlenderbotSmall](model_doc/blenderbot-small)** (from Facebook) released with the paper [Recipes for building an open-domain chatbot](https://arxiv.org/abs/2004.13637) by Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Kurt Shuster, Eric M. Smith, Y-Lan Boureau, Jason Weston.
1. **[BLOOM](model_doc/bloom)** (from BigScience workshop) released by the [BigScience Workshop](https://bigscience.huggingface.co/).
1. **[BORT](model_doc/bort)** (from Alexa) released with the paper [Optimal Subarchitecture Extraction For BERT](https://arxiv.org/abs/2010.10499) by Adrian de Wynter and Daniel J. Perry.
1. **[ByT5](model_doc/byt5)** (from Google Research) released with the paper [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626) by Linting Xue, Aditya Barua, Noah Constant, Rami Al-Rfou, Sharan Narang, Mihir Kale, Adam Roberts, Colin Raffel.
1. **[CamemBERT](model_doc/camembert)** (from Inria/Facebook/Sorbonne) released with the paper [CamemBERT: a Tasty French Language Model](https://arxiv.org/abs/1911.03894) by Louis Martin*, Benjamin Muller*, Pedro Javier Ortiz Suárez*, Yoann Dupont, Laurent Romary, Éric Villemonte de la Clergerie, Djamé Seddah and Benoît Sagot.
1. **[CANINE](model_doc/canine)** (from Google Research) released with the paper [CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation](https://arxiv.org/abs/2103.06874) by Jonathan H. Clark, Dan Garrette, Iulia Turc, John Wieting.
1. **[Chinese-CLIP](model_doc/chinese_clip)** (from OFA-Sys) released with the paper [Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese](https://arxiv.org/abs/2211.01335) by An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou.
1. **[CLIP](model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
1. **[CLIPSeg](model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
1. **[CodeGen](model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
1. **[Conditional DETR](model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
1. **[ConvBERT](model_doc/convbert)** (from YituTech) released with the paper [ConvBERT: Improving BERT with Span-based Dynamic Convolution](https://arxiv.org/abs/2008.02496) by Zihang Jiang, Weihao Yu, Daquan Zhou, Yunpeng Chen, Jiashi Feng, Shuicheng Yan.
1. **[ConvNeXT](model_doc/convnext)** (from Facebook AI) released with the paper [A ConvNet for the 2020s](https://arxiv.org/abs/2201.03545) by Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie.
1. **[CPM](model_doc/cpm)** (from Tsinghua University) released with the paper [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://arxiv.org/abs/2012.00413) by Zhengyan Zhang, Xu Han, Hao Zhou, Pei Ke, Yuxian Gu, Deming Ye, Yujia Qin, Yusheng Su, Haozhe Ji, Jian Guan, Fanchao Qi, Xiaozhi Wang, Yanan Zheng, Guoyang Zeng, Huanqi Cao, Shengqi Chen, Daixuan Li, Zhenbo Sun, Zhiyuan Liu, Minlie Huang, Wentao Han, Jie Tang, Juanzi Li, Xiaoyan Zhu, Maosong Sun.
1. **[CTRL](model_doc/ctrl)** (from Salesforce) released with the paper [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://arxiv.org/abs/1909.05858) by Nitish Shirish Keskar*, Bryan McCann*, Lav R. Varshney, Caiming Xiong and Richard Socher.
1. **[CvT](model_doc/cvt)** (from Microsoft) released with the paper [CvT: Introducing Convolutions to Vision Transformers](https://arxiv.org/abs/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan, Lei Zhang.
1. **[Data2Vec](model_doc/data2vec)** (from Facebook) released with the paper [Data2Vec: A General Framework for Self-supervised Learning in Speech, Vision and Language](https://arxiv.org/abs/2202.03555) by Alexei Baevski, Wei-Ning Hsu, Qiantong Xu, Arun Babu, Jiatao Gu, Michael Auli.
1. **[DeBERTa](model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[DeBERTa-v2](model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
1. **[Decision Transformer](model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
1. **[Deformable DETR](model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
1. **[DeiT](model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
1. **[DETR](model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.
1. **[DialoGPT](model_doc/dialogpt)** (from Microsoft Research) released with the paper [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://arxiv.org/abs/1911.00536) by Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
1. **[DiNAT](model_doc/dinat)** (from SHI Labs) released with the paper [Dilated Neighborhood Attention Transformer](https://arxiv.org/abs/2209.15001) by Ali Hassani and Humphrey Shi.
1. **[DistilBERT](model_doc/distilbert)** (from HuggingFace), released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), RoBERTa into [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation), Multilingual BERT into [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/research_projects/distillation) and a German version of DistilBERT.
1. **[DiT](model_doc/dit)** (from Microsoft Research) released with the paper [DiT: Self-supervised Pre-training for Document Image Transformer](https://arxiv.org/abs/2203.02378) by Junlong Li, Yiheng Xu, Tengchao Lv, Lei Cui, Cha Zhang, Furu Wei.
1. **[Donut](model_doc/donut)** (from NAVER), released together with the paper [OCR-free Document Understanding Transformer](https://arxiv.org/abs/2111.15664) by Geewook Kim, Teakgyu Hong, Moonbin Yim, Jeongyeon Nam, Jinyoung Park, Jinyeong Yim, Wonseok Hwang, Sangdoo Yun, Dongyoon Han, Seunghyun Park.
1. **[DPR](model_doc/dpr)** (from Facebook) released with the paper [Dense Passage Retrieval for Open-Domain Question Answering](https://arxiv.org/abs/2004.04906) by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
1. **[DPT](master/model_doc/dpt)** (from Intel Labs) released with the paper [Vision Transformers for Dense Prediction](https://arxiv.org/abs/2103.13413) by René Ranftl, Alexey Bochkovskiy, Vladlen Koltun.
1. **[ELECTRA](model_doc/electra)** (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
1. **[EncoderDecoder](model_doc/encoder-decoder)** (from Google Research) released with the paper [Leveraging Pre-trained Checkpoints for Sequence Generation Tasks](https://arxiv.org/abs/1907.12461) by Sascha Rothe, Shashi Narayan, Aliaksei Severyn.
1. **[ERNIE](model_doc/ernie)** (from Baidu) released with the paper [ERNIE: Enhanced Representation through Knowledge Integration](https://arxiv.org/abs/1904.09223) by Yu Sun, Shuohuan Wang, Yukun Li, Shikun Feng, Xuyi Chen, Han Zhang, Xin Tian, Danxiang Zhu, Hao Tian, Hua Wu.
1. **[ESM](model_doc/esm)** (from Meta AI) are transformer protein language models. **ESM-1b** was released with the paper [Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences](https://www.pnas.org/content/118/15/e2016239118) by Alexander Rives, Joshua Meier, Tom Sercu, Siddharth Goyal, Zeming Lin, Jason Liu, Demi Guo, Myle Ott, C. Lawrence Zitnick, Jerry Ma, and Rob Fergus. **ESM-1v** was released with the paper [Language models enable zero-shot prediction of the effects of mutations on protein function](https://doi.org/10.1101/2021.07.09.450648) by Joshua Meier, Roshan Rao, Robert Verkuil, Jason Liu, Tom Sercu and Alexander Rives. **ESM-2 and ESMFold** were released with the paper [Language models of protein sequences at the scale of evolution enable accurate structure prediction](https://doi.org/10.1101/2022.07.20.500902) by Zeming Lin, Halil Akin, Roshan Rao, Brian Hie, Zhongkai Zhu, Wenting Lu, Allan dos Santos Costa, Maryam Fazel-Zarandi, Tom Sercu, Sal Candido, Alexander Rives.
1. **[FLAN-T5](model_doc/flan-t5)** (from Google AI) released in the repository [google-research/t5x](https://github.com/google-research/t5x/blob/main/docs/models.md#flan-t5-checkpoints) by Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping Huang, Andrew Dai, Hongkun Yu, Slav Petrov, Ed H. Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V. Le, and Jason Wei
1. **[FlauBERT](model_doc/flaubert)** (from CNRS) released with the paper [FlauBERT: Unsupervised Language Model Pre-training for French](https://arxiv.org/abs/1912.05372) by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
1. **[FLAVA](model_doc/flava)** (from Facebook AI) released with the paper [FLAVA: A Foundational Language And Vision Alignment Model](https://arxiv.org/abs/2112.04482) by Amanpreet Singh, Ronghang Hu, Vedanuj Goswami, Guillaume Couairon, Wojciech Galuba, Marcus Rohrbach, and Douwe Kiela.
1. **[FNet](model_doc/fnet)** (from Google Research) released with the paper [FNet: Mixing Tokens with Fourier Transforms](https://arxiv.org/abs/2105.03824) by James Lee-Thorp, Joshua Ainslie, Ilya Eckstein, Santiago Ontanon.
1. **[Funnel Transformer](model_doc/funnel)** (from CMU/Google Brain) released with the paper [Funnel-Transformer: Filtering out Sequential Redundancy for Efficient Language Processing](https://arxiv.org/abs/2006.03236) by Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
1. **[GLPN](model_doc/glpn)** (from KAIST) released with the paper [Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth](https://arxiv.org/abs/2201.07436) by Doyeon Kim, Woonghyun Ga, Pyungwhan Ahn, Donggyu Joo, Sehwan Chun, Junmo Kim.
1. **[GPT](model_doc/openai-gpt)** (from OpenAI) released with the paper [Improving Language Understanding by Generative Pre-Training](https://blog.openai.com/language-unsupervised/) by Alec Radford, Karthik Narasimhan, Tim Salimans and Ilya Sutskever.
1. **[GPT Neo](model_doc/gpt_neo)** (from EleutherAI) released in the repository [EleutherAI/gpt-neo](https://github.com/EleutherAI/gpt-neo) by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
1. **[GPT NeoX](model_doc/gpt_neox)** (from EleutherAI) released with the paper [GPT-NeoX-20B: An Open-Source Autoregressive Language Model](https://arxiv.org/abs/2204.06745) by Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, Michael Pieler, USVSN Sai Prashanth, Shivanshu Purohit, Laria Reynolds, Jonathan Tow, Ben Wang, Samuel Weinbach
1. **[GPT NeoX Japanese](model_doc/gpt_neox_japanese)** (from ABEJA) released by Shinya Otani, Takayoshi Makabe, Anuj Arora, and Kyo Hattori.
1. **[GPT-2](model_doc/gpt2)** (from OpenAI) released with the paper [Language Models are Unsupervised Multitask Learners](https://blog.openai.com/better-language-models/) by Alec Radford*, Jeffrey Wu*, Rewon Child, David Luan, Dario Amodei** and Ilya Sutskever**.
1. **[GPT-J](model_doc/gptj)** (from EleutherAI) released in the repository [kingoflolz/mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/) by Ben Wang and Aran Komatsuzaki.
1. **[GPT-Sw3](model_doc/gpt-sw3)** (from AI-Sweden) released with the paper [Lessons Learned from GPT-SW3: Building the First Large-Scale Generative Language Model for Swedish](http://www.lrec-conf.org/proceedings/lrec2022/pdf/2022.lrec-1.376.pdf) by Ariel Ekgren, Amaru Cuba Gyllensten, Evangelia Gogoulou, Alice Heiman, Severine Verlinden, Joey Öhman, Fredrik Carlsson, Magnus Sahlgren.
1. **[GroupViT](model_doc/groupvit)** (from UCSD, NVIDIA) released with the paper [GroupViT: Semantic Segmentation Emerges from Text Supervision](https://arxiv.org/abs/2202.11094) by Jiarui Xu, Shalini De Mello, Sifei Liu, Wonmin Byeon, Thomas Breuel, Jan Kautz, Xiaolong Wang.
1. **[Hubert](model_doc/hubert)** (from Facebook) released with the paper [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://arxiv.org/abs/2106.07447) by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
1. **[I-BERT](model_doc/ibert)** (from Berkeley) released with the paper [I-BERT: Integer-only BERT Quantization](https://arxiv.org/abs/2101.01321) by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer.
1. **[ImageGPT](model_doc/imagegpt)** (from OpenAI) released with the paper [Generative Pretraining from Pixels](https://openai.com/blog/image-gpt/) by Mark Chen, Alec Radford, Rewon Child, Jeffrey Wu, Heewoo Jun, David Luan, Ilya Sutskever.
1. **[Jukebox](model_doc/jukebox)** (from OpenAI) released with the paper [Jukebox: A Generative Model for Music](https://arxiv.org/pdf/2005.00341.pdf) by Prafulla Dhariwal, Heewoo Jun, Christine Payne, Jong Wook Kim, Alec Radford, Ilya Sutskever.
1. **[LayoutLM](model_doc/layoutlm)** (from Microsoft Research Asia) released with the paper [LayoutLM: Pre-training of Text and Layout for Document Image Understanding](https://arxiv.org/abs/1912.13318) by Yiheng Xu, Minghao Li, Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
1. **[LayoutLMv2](model_doc/layoutlmv2)** (from Microsoft Research Asia) released with the paper [LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding](https://arxiv.org/abs/2012.14740) by Yang Xu, Yiheng Xu, Tengchao Lv, Lei Cui, Furu Wei, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Wanxiang Che, Min Zhang, Lidong Zhou.
1. **[LayoutLMv3](model_doc/layoutlmv3)** (from Microsoft Research Asia) released with the paper [LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking](https://arxiv.org/abs/2204.08387) by Yupan Huang, Tengchao Lv, Lei Cui, Yutong Lu, Furu Wei.
1. **[LayoutXLM](model_doc/layoutxlm)** (from Microsoft Research Asia) released with the paper [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/abs/2104.08836) by Yiheng Xu, Tengchao Lv, Lei Cui, Guoxin Wang, Yijuan Lu, Dinei Florencio, Cha Zhang, Furu Wei.
1. **[LED](model_doc/led)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LeViT](model_doc/levit)** (from Meta AI) released with the paper [LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136) by Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze.
1. **[LiLT](model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
1. **[Longformer](model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
1. **[LongT5](model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
1. **[LUKE](model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
1. **[LXMERT](model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
1. **[M-CTC-T](model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
1. **[M2M100](model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
1. **[MarianMT](model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
1. **[MarkupLM](model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
1. **[MaskFormer](model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
1. **[mBART](model_doc/mbart)** (from Facebook) released with the paper [Multilingual Denoising Pre-training for Neural Machine Translation](https://arxiv.org/abs/2001.08210) by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
1. **[mBART-50](model_doc/mbart)** (from Facebook) released with the paper [Multilingual Translation with Extensible Multilingual Pretraining and Finetuning](https://arxiv.org/abs/2008.00401) by Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
1. **[Megatron-BERT](model_doc/megatron-bert)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[Megatron-GPT2](model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
1. **[mLUKE](model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
1. **[MobileBERT](model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[MobileNetV1](model_doc/mobilenet_v1)** (from Google Inc.) released with the paper [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam.
1. **[MobileNetV2](model_doc/mobilenet_v2)** (from Google Inc.) released with the paper [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381) by Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, Liang-Chieh Chen.
1. **[MobileViT](model_doc/mobilevit)** (from Apple) released with the paper [MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer](https://arxiv.org/abs/2110.02178) by Sachin Mehta and Mohammad Rastegari.
1. **[MPNet](model_doc/mpnet)** (from Microsoft Research) released with the paper [MPNet: Masked and Permuted Pre-training for Language Understanding](https://arxiv.org/abs/2004.09297) by Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, Tie-Yan Liu.
1. **[MT5](model_doc/mt5)** (from Google AI) released with the paper [mT5: A massively multilingual pre-trained text-to-text transformer](https://arxiv.org/abs/2010.11934) by Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
1. **[MVP](model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[NAT](model_doc/nat)** (from SHI Labs) released with the paper [Neighborhood Attention Transformer](https://arxiv.org/abs/2204.07143) by Ali Hassani, Steven Walton, Jiachen Li, Shen Li, and Humphrey Shi.
1. **[Nezha](model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[NLLB](model_doc/nllb)** (from Meta) released with the paper [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by the NLLB team.
1. **[Nyströmformer](model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
1. **[OWL-ViT](model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
1. **[Pegasus](model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
1. **[PEGASUS-X](model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
1. **[Perceiver IO](model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
1. **[PhoBERT](model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
1. **[PLBart](model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
1. **[PoolFormer](model_doc/poolformer)** (from Sea AI Labs) released with the paper [MetaFormer is Actually What You Need for Vision](https://arxiv.org/abs/2111.11418) by Yu, Weihao and Luo, Mi and Zhou, Pan and Si, Chenyang and Zhou, Yichen and Wang, Xinchao and Feng, Jiashi and Yan, Shuicheng.
1. **[ProphetNet](model_doc/prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[QDQBert](model_doc/qdqbert)** (from NVIDIA) released with the paper [Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation](https://arxiv.org/abs/2004.09602) by Hao Wu, Patrick Judd, Xiaojie Zhang, Mikhail Isaev and Paulius Micikevicius.
1. **[RAG](model_doc/rag)** (from Facebook) released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandara Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela.
1. **[REALM](model_doc/realm.html)** (from Google Research) released with the paper [REALM: Retrieval-Augmented Language Model Pre-Training](https://arxiv.org/abs/2002.08909) by Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat and Ming-Wei Chang.
1. **[Reformer](model_doc/reformer)** (from Google Research) released with the paper [Reformer: The Efficient Transformer](https://arxiv.org/abs/2001.04451) by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
1. **[RegNet](model_doc/regnet)** (from META Platforms) released with the paper [Designing Network Design Space](https://arxiv.org/abs/2003.13678) by Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár.
1. **[RemBERT](model_doc/rembert)** (from Google Research) released with the paper [Rethinking embedding coupling in pre-trained language models](https://arxiv.org/abs/2010.12821) by Hyung Won Chung, Thibault Févry, Henry Tsai, M. Johnson, Sebastian Ruder.
1. **[ResNet](model_doc/resnet)** (from Microsoft Research) released with the paper [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) by Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun.
1. **[RoBERTa](model_doc/roberta)** (from Facebook), released together with the paper [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://arxiv.org/abs/1907.11692) by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
1. **[RoCBert](model_doc/roc_bert)** (from WeChatAI) released with the paper [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf) by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
1. **[RoFormer](model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
1. **[SegFormer](model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
1. **[SEW](model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SEW-D](model_doc/sew_d)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
1. **[SpeechToTextTransformer](model_doc/speech_to_text)** (from Facebook), released together with the paper [fairseq S2T: Fast Speech-to-Text Modeling with fairseq](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
1. **[SpeechToTextTransformer2](model_doc/speech_to_text_2)** (from Facebook), released together with the paper [Large-Scale Self- and Semi-Supervised Learning for Speech Translation](https://arxiv.org/abs/2104.06678) by Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
1. **[Splinter](model_doc/splinter)** (from Tel Aviv University), released together with the paper [Few-Shot Question Answering by Pretraining Span Selection](https://arxiv.org/abs/2101.00438) by Ori Ram, Yuval Kirstain, Jonathan Berant, Amir Globerson, Omer Levy.
1. **[SqueezeBERT](model_doc/squeezebert)** (from Berkeley) released with the paper [SqueezeBERT: What can computer vision teach NLP about efficient neural networks?](https://arxiv.org/abs/2006.11316) by Forrest N. Iandola, Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
1. **[Swin Transformer](model_doc/swin)** (from Microsoft) released with the paper [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030) by Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo.
1. **[Swin Transformer V2](model_doc/swinv2)** (from Microsoft) released with the paper [Swin Transformer V2: Scaling Up Capacity and Resolution](https://arxiv.org/abs/2111.09883) by Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, Furu Wei, Baining Guo.
1. **[SwitchTransformers](model_doc/switch_transformers)** (from Google) released with the paper [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://arxiv.org/abs/2101.03961) by William Fedus, Barret Zoph, Noam Shazeer.
1. **[T5](model_doc/t5)** (from Google AI) released with the paper [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://arxiv.org/abs/1910.10683) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[T5v1.1](model_doc/t5v1.1)** (from Google AI) released in the repository [google-research/text-to-text-transfer-transformer](https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511) by Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
1. **[Table Transformer](model_doc/table-transformer)** (from Microsoft Research) released with the paper [PubTables-1M: Towards Comprehensive Table Extraction From Unstructured Documents](https://arxiv.org/abs/2110.00061) by Brandon Smock, Rohith Pesala, Robin Abraham.
1. **[TAPAS](model_doc/tapas)** (from Google AI) released with the paper [TAPAS: Weakly Supervised Table Parsing via Pre-training](https://arxiv.org/abs/2004.02349) by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno and Julian Martin Eisenschlos.
1. **[TAPEX](model_doc/tapex)** (from Microsoft Research) released with the paper [TAPEX: Table Pre-training via Learning a Neural SQL Executor](https://arxiv.org/abs/2107.07653) by Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, Jian-Guang Lou.
1. **[Time Series Transformer](model_doc/time_series_transformer)** (from HuggingFace).
1. **[TimeSformer](model_doc/timesformer)** (from Facebook) released with the paper [Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095) by Gedas Bertasius, Heng Wang, Lorenzo Torresani.
1. **[Trajectory Transformer](model_doc/trajectory_transformers)** (from the University of California at Berkeley) released with the paper [Offline Reinforcement Learning as One Big Sequence Modeling Problem](https://arxiv.org/abs/2106.02039) by Michael Janner, Qiyang Li, Sergey Levine
1. **[Transformer-XL](model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
1. **[TrOCR](model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
1. **[UL2](model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
1. **[UniSpeech](model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
1. **[UniSpeechSat](model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
1. **[VAN](model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
1. **[VideoMAE](model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.
1. **[ViLT](model_doc/vilt)** (from NAVER AI Lab/Kakao Enterprise/Kakao Brain) released with the paper [ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision](https://arxiv.org/abs/2102.03334) by Wonjae Kim, Bokyung Son, Ildoo Kim.
1. **[Vision Transformer (ViT)](model_doc/vit)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[VisualBERT](model_doc/visual_bert)** (from UCLA NLP) released with the paper [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://arxiv.org/pdf/1908.03557) by Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
1. **[ViT Hybrid](model_doc/vit_hybrid)** (from Google AI) released with the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
1. **[ViTMAE](model_doc/vit_mae)** (from Meta AI) released with the paper [Masked Autoencoders Are Scalable Vision Learners](https://arxiv.org/abs/2111.06377) by Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick.
1. **[ViTMSN](model_doc/vit_msn)** (from Meta AI) released with the paper [Masked Siamese Networks for Label-Efficient Learning](https://arxiv.org/abs/2204.07141) by Mahmoud Assran, Mathilde Caron, Ishan Misra, Piotr Bojanowski, Florian Bordes, Pascal Vincent, Armand Joulin, Michael Rabbat, Nicolas Ballas.
1. **[Wav2Vec2](model_doc/wav2vec2)** (from Facebook AI) released with the paper [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://arxiv.org/abs/2006.11477) by Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli.
1. **[Wav2Vec2-Conformer](model_doc/wav2vec2-conformer)** (from Facebook AI) released with the paper [FAIRSEQ S2T: Fast Speech-to-Text Modeling with FAIRSEQ](https://arxiv.org/abs/2010.05171) by Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino.
1. **[Wav2Vec2Phoneme](model_doc/wav2vec2_phoneme)** (from Facebook AI) released with the paper [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680) by Qiantong Xu, Alexei Baevski, Michael Auli.
1. **[WavLM](model_doc/wavlm)** (from Microsoft Research) released with the paper [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900) by Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Furu Wei.
1. **[Whisper](model_doc/whisper)** (from OpenAI) released with the paper [Robust Speech Recognition via Large-Scale Weak Supervision](https://cdn.openai.com/papers/whisper.pdf) by Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever.
1. **[X-CLIP](model_doc/xclip)** (from Microsoft Research) released with the paper [Expanding Language-Image Pretrained Models for General Video Recognition](https://arxiv.org/abs/2208.02816) by Bolin Ni, Houwen Peng, Minghao Chen, Songyang Zhang, Gaofeng Meng, Jianlong Fu, Shiming Xiang, Haibin Ling.
1. **[XGLM](model_doc/xglm)** (From Facebook AI) released with the paper [Few-shot Learning with Multilingual Language Models](https://arxiv.org/abs/2112.10668) by Xi Victoria Lin, Todor Mihaylov, Mikel Artetxe, Tianlu Wang, Shuohui Chen, Daniel Simig, Myle Ott, Naman Goyal, Shruti Bhosale, Jingfei Du, Ramakanth Pasunuru, Sam Shleifer, Punit Singh Koura, Vishrav Chaudhary, Brian O'Horo, Jeff Wang, Luke Zettlemoyer, Zornitsa Kozareva, Mona Diab, Veselin Stoyanov, Xian Li.
1. **[XLM](model_doc/xlm)** (from Facebook) released together with the paper [Cross-lingual Language Model Pretraining](https://arxiv.org/abs/1901.07291) by Guillaume Lample and Alexis Conneau.
1. **[XLM-ProphetNet](model_doc/xlm-prophetnet)** (from Microsoft Research) released with the paper [ProphetNet: Predicting Future N-gram for Sequence-to-Sequence Pre-training](https://arxiv.org/abs/2001.04063) by Yu Yan, Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
1. **[XLM-RoBERTa](model_doc/xlm-roberta)** (from Facebook AI), released together with the paper [Unsupervised Cross-lingual Representation Learning at Scale](https://arxiv.org/abs/1911.02116) by Alexis Conneau*, Kartikay Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer and Veselin Stoyanov.
1. **[XLM-RoBERTa-XL](model_doc/xlm-roberta-xl)** (from Facebook AI), released together with the paper [Larger-Scale Transformers for Multilingual Masked Language Modeling](https://arxiv.org/abs/2105.00572) by Naman Goyal, Jingfei Du, Myle Ott, Giri Anantharaman, Alexis Conneau.
1. **[XLNet](model_doc/xlnet)** (from Google/CMU) released with the paper [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://arxiv.org/abs/1906.08237) by Zhilin Yang*, Zihang Dai*, Yiming Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
1. **[XLS-R](model_doc/xls_r)** (from Facebook AI) released with the paper [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://arxiv.org/abs/2111.09296) by Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, Alexei Baevski, Alexis Conneau, Michael Auli.
1. **[XLSR-Wav2Vec2](model_doc/xlsr_wav2vec2)** (from Facebook AI) released with the paper [Unsupervised Cross-Lingual Representation Learning For Speech Recognition](https://arxiv.org/abs/2006.13979) by Alexis Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
1. **[YOLOS](model_doc/yolos)** (from Huazhong University of Science & Technology) released with the paper [You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection](https://arxiv.org/abs/2106.00666) by Yuxin Fang, Bencheng Liao, Xinggang Wang, Jiemin Fang, Jiyang Qi, Rui Wu, Jianwei Niu, Wenyu Liu.
1. **[YOSO](model_doc/yoso)** (from the University of Wisconsin - Madison) released with the paper [You Only Sample (Almost) Once: Linear Cost Self-Attention Via Bernoulli Sampling](https://arxiv.org/abs/2111.09714) by Zhanpeng Zeng, Yunyang Xiong, Sathya N. Ravi, Shailesh Acharya, Glenn Fung, Vikas Singh.
### Supported frameworks
The table below represents the current support in the library for each of those models, whether they have a Python
tokenizer (called "slow"). A "fast" tokenizer backed by the 🤗 Tokenizers library, whether they have support in Jax (via
Flax), PyTorch, and/or TensorFlow.
<!--This table is updated automatically from the auto modules with _make fix-copies_. Do not update manually!-->
| Model | Tokenizer slow | Tokenizer fast | PyTorch support | TensorFlow support | Flax Support |
|:-----------------------------:|:--------------:|:--------------:|:---------------:|:------------------:|:------------:|
| ALBERT | ✅ | ✅ | ✅ | ✅ | ✅ |
| Audio Spectrogram Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| BART | ✅ | ✅ | ✅ | ✅ | ✅ |
| BEiT | ❌ | ❌ | ✅ | ❌ | ✅ |
| BERT | ✅ | ✅ | ✅ | ✅ | ✅ |
| Bert Generation | ✅ | ❌ | ✅ | ❌ | ❌ |
| BigBird | ✅ | ✅ | ✅ | ❌ | ✅ |
| BigBird-Pegasus | ❌ | ❌ | ✅ | ❌ | ❌ |
| BioGpt | ✅ | ❌ | ✅ | ❌ | ❌ |
| BiT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Blenderbot | ✅ | ✅ | ✅ | ✅ | ✅ |
| BlenderbotSmall | ✅ | ✅ | ✅ | ✅ | ✅ |
| BLOOM | ❌ | ✅ | ✅ | ❌ | ❌ |
| CamemBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| CANINE | ✅ | ❌ | ✅ | ❌ | ❌ |
| Chinese-CLIP | ❌ | ❌ | ✅ | ❌ | ❌ |
| CLIP | ✅ | ✅ | ✅ | ✅ | ✅ |
| CLIPSeg | ❌ | ❌ | ✅ | ❌ | ❌ |
| CodeGen | ✅ | ✅ | ✅ | ❌ | ❌ |
| Conditional DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
| ConvBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| ConvNeXT | ❌ | ❌ | ✅ | ✅ | ❌ |
| CTRL | ✅ | ❌ | ✅ | ✅ | ❌ |
| CvT | ❌ | ❌ | ✅ | ✅ | ❌ |
| Data2VecAudio | ❌ | ❌ | ✅ | ❌ | ❌ |
| Data2VecText | ❌ | ❌ | ✅ | ❌ | ❌ |
| Data2VecVision | ❌ | ❌ | ✅ | ✅ | ❌ |
| DeBERTa | ✅ | ✅ | ✅ | ✅ | ❌ |
| DeBERTa-v2 | ✅ | ✅ | ✅ | ✅ | ❌ |
| Decision Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| Deformable DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
| DeiT | ❌ | ❌ | ✅ | ✅ | ❌ |
| DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
| DiNAT | ❌ | ❌ | ✅ | ❌ | ❌ |
| DistilBERT | ✅ | ✅ | ✅ | ✅ | ✅ |
| DonutSwin | ❌ | ❌ | ✅ | ❌ | ❌ |
| DPR | ✅ | ✅ | ✅ | ✅ | ❌ |
| DPT | ❌ | ❌ | ✅ | ❌ | ❌ |
| ELECTRA | ✅ | ✅ | ✅ | ✅ | ✅ |
| Encoder decoder | ❌ | ❌ | ✅ | ✅ | ✅ |
| ERNIE | ❌ | ❌ | ✅ | ❌ | ❌ |
| ESM | ✅ | ❌ | ✅ | ✅ | ❌ |
| FairSeq Machine-Translation | ✅ | ❌ | ✅ | ❌ | ❌ |
| FlauBERT | ✅ | ❌ | ✅ | ✅ | ❌ |
| FLAVA | ❌ | ❌ | ✅ | ❌ | ❌ |
| FNet | ✅ | ✅ | ✅ | ❌ | ❌ |
| Funnel Transformer | ✅ | ✅ | ✅ | ✅ | ❌ |
| GLPN | ❌ | ❌ | ✅ | ❌ | ❌ |
| GPT Neo | ❌ | ❌ | ✅ | ❌ | ✅ |
| GPT NeoX | ❌ | ✅ | ✅ | ❌ | ❌ |
| GPT NeoX Japanese | ✅ | ❌ | ✅ | ❌ | ❌ |
| GPT-J | ❌ | ❌ | ✅ | ✅ | ✅ |
| GPT-Sw3 | ✅ | ✅ | ✅ | ✅ | ✅ |
| GroupViT | ❌ | ❌ | ✅ | ✅ | ❌ |
| Hubert | ❌ | ❌ | ✅ | ✅ | ❌ |
| I-BERT | ❌ | ❌ | ✅ | ❌ | ❌ |
| ImageGPT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Jukebox | ✅ | ❌ | ✅ | ❌ | ❌ |
| LayoutLM | ✅ | ✅ | ✅ | ✅ | ❌ |
| LayoutLMv2 | ✅ | ✅ | ✅ | ❌ | ❌ |
| LayoutLMv3 | ✅ | ✅ | ✅ | ✅ | ❌ |
| LED | ✅ | ✅ | ✅ | ✅ | ❌ |
| LeViT | ❌ | ❌ | ✅ | ❌ | ❌ |
| LiLT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Longformer | ✅ | ✅ | ✅ | ✅ | ❌ |
| LongT5 | ❌ | ❌ | ✅ | ❌ | ✅ |
| LUKE | ✅ | ❌ | ✅ | ❌ | ❌ |
| LXMERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| M-CTC-T | ❌ | ❌ | ✅ | ❌ | ❌ |
| M2M100 | ✅ | ❌ | ✅ | ❌ | ❌ |
| Marian | ✅ | ❌ | ✅ | ✅ | ✅ |
| MarkupLM | ✅ | ✅ | ✅ | ❌ | ❌ |
| MaskFormer | ❌ | ❌ | ✅ | ❌ | ❌ |
| MaskFormerSwin | ❌ | ❌ | ❌ | ❌ | ❌ |
| mBART | ✅ | ✅ | ✅ | ✅ | ✅ |
| Megatron-BERT | ❌ | ❌ | ✅ | ❌ | ❌ |
| MobileBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| MobileNetV1 | ❌ | ❌ | ✅ | ❌ | ❌ |
| MobileNetV2 | ❌ | ❌ | ✅ | ❌ | ❌ |
| MobileViT | ❌ | ❌ | ✅ | ✅ | ❌ |
| MPNet | ✅ | ✅ | ✅ | ✅ | ❌ |
| MT5 | ✅ | ✅ | ✅ | ✅ | ✅ |
| MVP | ✅ | ✅ | ✅ | ❌ | ❌ |
| NAT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Nezha | ❌ | ❌ | ✅ | ❌ | ❌ |
| Nyströmformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| OpenAI GPT | ✅ | ✅ | ✅ | ✅ | ❌ |
| OpenAI GPT-2 | ✅ | ✅ | ✅ | ✅ | ✅ |
| OPT | ❌ | ❌ | ✅ | ✅ | ✅ |
| OWL-ViT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Pegasus | ✅ | ✅ | ✅ | ✅ | ✅ |
| PEGASUS-X | ❌ | ❌ | ✅ | ❌ | ❌ |
| Perceiver | ✅ | ❌ | ✅ | ❌ | ❌ |
| PLBart | ✅ | ❌ | ✅ | ❌ | ❌ |
| PoolFormer | ❌ | ❌ | ✅ | ❌ | ❌ |
| ProphetNet | ✅ | ❌ | ✅ | ❌ | ❌ |
| QDQBert | ❌ | ❌ | ✅ | ❌ | ❌ |
| RAG | ✅ | ❌ | ✅ | ✅ | ❌ |
| REALM | ✅ | ✅ | ✅ | ❌ | ❌ |
| Reformer | ✅ | ✅ | ✅ | ❌ | ❌ |
| RegNet | ❌ | ❌ | ✅ | ✅ | ❌ |
| RemBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| ResNet | ❌ | ❌ | ✅ | ✅ | ❌ |
| RetriBERT | ✅ | ✅ | ✅ | ❌ | ❌ |
| RoBERTa | ✅ | ✅ | ✅ | ✅ | ✅ |
| RoCBert | ✅ | ❌ | ✅ | ❌ | ❌ |
| RoFormer | ✅ | ✅ | ✅ | ✅ | ✅ |
| SegFormer | ❌ | ❌ | ✅ | ✅ | ❌ |
| SEW | ❌ | ❌ | ✅ | ❌ | ❌ |
| SEW-D | ❌ | ❌ | ✅ | ❌ | ❌ |
| Speech Encoder decoder | ❌ | ❌ | ✅ | ❌ | ✅ |
| Speech2Text | ✅ | ❌ | ✅ | ✅ | ❌ |
| Speech2Text2 | ✅ | ❌ | ❌ | ❌ | ❌ |
| Splinter | ✅ | ✅ | ✅ | ❌ | ❌ |
| SqueezeBERT | ✅ | ✅ | ✅ | ❌ | ❌ |
| Swin Transformer | ❌ | ❌ | ✅ | ✅ | ❌ |
| Swin Transformer V2 | ❌ | ❌ | ✅ | ❌ | ❌ |
| SwitchTransformers | ❌ | ❌ | ✅ | ❌ | ❌ |
| T5 | ✅ | ✅ | ✅ | ✅ | ✅ |
| Table Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| TAPAS | ✅ | ❌ | ✅ | ✅ | ❌ |
| Time Series Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| TimeSformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| Trajectory Transformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| Transformer-XL | ✅ | ❌ | ✅ | ✅ | ❌ |
| TrOCR | ❌ | ❌ | ✅ | ❌ | ❌ |
| UniSpeech | ❌ | ❌ | ✅ | ❌ | ❌ |
| UniSpeechSat | ❌ | ❌ | ✅ | ❌ | ❌ |
| VAN | ❌ | ❌ | ✅ | ❌ | ❌ |
| VideoMAE | ❌ | ❌ | ✅ | ❌ | ❌ |
| ViLT | ❌ | ❌ | ✅ | ❌ | ❌ |
| Vision Encoder decoder | ❌ | ❌ | ✅ | ✅ | ✅ |
| VisionTextDualEncoder | ❌ | ❌ | ✅ | ❌ | ✅ |
| VisualBERT | ❌ | ❌ | ✅ | ❌ | ❌ |
| ViT | ❌ | ❌ | ✅ | ✅ | ✅ |
| ViT Hybrid | ❌ | ❌ | ✅ | ❌ | ❌ |
| ViTMAE | ❌ | ❌ | ✅ | ✅ | ❌ |
| ViTMSN | ❌ | ❌ | ✅ | ❌ | ❌ |
| Wav2Vec2 | ✅ | ❌ | ✅ | ✅ | ✅ |
| Wav2Vec2-Conformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| WavLM | ❌ | ❌ | ✅ | ❌ | ❌ |
| Whisper | ✅ | ❌ | ✅ | ✅ | ❌ |
| X-CLIP | ❌ | ❌ | ✅ | ❌ | ❌ |
| XGLM | ✅ | ✅ | ✅ | ✅ | ✅ |
| XLM | ✅ | ❌ | ✅ | ✅ | ❌ |
| XLM-ProphetNet | ✅ | ❌ | ✅ | ❌ | ❌ |
| XLM-RoBERTa | ✅ | ✅ | ✅ | ✅ | ✅ |
| XLM-RoBERTa-XL | ❌ | ❌ | ✅ | ❌ | ❌ |
| XLNet | ✅ | ✅ | ✅ | ✅ | ❌ |
| YOLOS | ❌ | ❌ | ✅ | ❌ | ❌ |
| YOSO | ❌ | ❌ | ✅ | ❌ | ❌ |
<!-- End table--> | 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./docs/source/en/serialization.mdx | <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Export to ONNX
If you need to deploy 🤗 Transformers models in production environments, we recommend
exporting them to a serialized format that can be loaded and executed on specialized
runtimes and hardware. In this guide, we'll show you how to export 🤗 Transformers
models to [ONNX (Open Neural Network eXchange)](http://onnx.ai).
ONNX is an open standard that defines a common set of operators and a common file format
to represent deep learning models in a wide variety of frameworks, including PyTorch and
TensorFlow. When a model is exported to the ONNX format, these operators are used to
construct a computational graph (often called an _intermediate representation_) which
represents the flow of data through the neural network.
By exposing a graph with standardized operators and data types, ONNX makes it easy to
switch between frameworks. For example, a model trained in PyTorch can be exported to
ONNX format and then imported in TensorFlow (and vice versa).
🤗 Transformers provides a [`transformers.onnx`](main_classes/onnx) package that enables
you to convert model checkpoints to an ONNX graph by leveraging configuration objects.
These configuration objects come ready made for a number of model architectures, and are
designed to be easily extendable to other architectures.
<Tip>
You can also export 🤗 Transformers models with the [`optimum.exporters.onnx` package](https://huggingface.co/docs/optimum/exporters/onnx/usage_guides/export_a_model)
from 🤗 Optimum.
Once exported, a model can be:
- Optimized for inference via techniques such as quantization and graph optimization.
- Run with ONNX Runtime via [`ORTModelForXXX` classes](https://huggingface.co/docs/optimum/onnxruntime/package_reference/modeling_ort),
which follow the same `AutoModel` API as the one you are used to in 🤗 Transformers.
- Run with [optimized inference pipelines](https://huggingface.co/docs/optimum/main/en/onnxruntime/usage_guides/pipelines),
which has the same API as the [`pipeline`] function in 🤗 Transformers.
To explore all these features, check out the [🤗 Optimum library](https://github.com/huggingface/optimum).
</Tip>
Ready-made configurations include the following architectures:
<!--This table is automatically generated by `make fix-copies`, do not fill manually!-->
- ALBERT
- BART
- BEiT
- BERT
- BigBird
- BigBird-Pegasus
- Blenderbot
- BlenderbotSmall
- BLOOM
- CamemBERT
- Chinese-CLIP
- CLIP
- CodeGen
- Conditional DETR
- ConvBERT
- ConvNeXT
- Data2VecText
- Data2VecVision
- DeBERTa
- DeBERTa-v2
- DeiT
- DETR
- DistilBERT
- ELECTRA
- ERNIE
- FlauBERT
- GPT Neo
- GPT-J
- GroupViT
- I-BERT
- ImageGPT
- LayoutLM
- LayoutLMv3
- LeViT
- Longformer
- LongT5
- M2M100
- Marian
- mBART
- MobileBERT
- MobileNetV1
- MobileNetV2
- MobileViT
- MT5
- OpenAI GPT-2
- OWL-ViT
- Perceiver
- PLBart
- RemBERT
- ResNet
- RoBERTa
- RoFormer
- SegFormer
- SqueezeBERT
- Swin Transformer
- T5
- Table Transformer
- Vision Encoder decoder
- ViT
- Whisper
- XLM
- XLM-RoBERTa
- XLM-RoBERTa-XL
- YOLOS
In the next two sections, we'll show you how to:
* Export a supported model using the `transformers.onnx` package.
* Export a custom model for an unsupported architecture.
## Exporting a model to ONNX
<Tip>
The recommended way of exporting a model is now to use
[`optimum.exporters.onnx`](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/export_a_model#exporting-a-model-to-onnx-using-the-cli),
do not worry it is very similar to `transformers.onnx`!
</Tip>
To export a 🤗 Transformers model to ONNX, you'll first need to install some extra
dependencies:
```bash
pip install transformers[onnx]
```
The `transformers.onnx` package can then be used as a Python module:
```bash
python -m transformers.onnx --help
usage: Hugging Face Transformers ONNX exporter [-h] -m MODEL [--feature {causal-lm, ...}] [--opset OPSET] [--atol ATOL] output
positional arguments:
output Path indicating where to store generated ONNX model.
optional arguments:
-h, --help show this help message and exit
-m MODEL, --model MODEL
Model ID on huggingface.co or path on disk to load model from.
--feature {causal-lm, ...}
The type of features to export the model with.
--opset OPSET ONNX opset version to export the model with.
--atol ATOL Absolute difference tolerance when validating the model.
```
Exporting a checkpoint using a ready-made configuration can be done as follows:
```bash
python -m transformers.onnx --model=distilbert-base-uncased onnx/
```
You should see the following logs:
```bash
Validating ONNX model...
-[✓] ONNX model output names match reference model ({'last_hidden_state'})
- Validating ONNX Model output "last_hidden_state":
-[✓] (2, 8, 768) matches (2, 8, 768)
-[✓] all values close (atol: 1e-05)
All good, model saved at: onnx/model.onnx
```
This exports an ONNX graph of the checkpoint defined by the `--model` argument. In this
example, it is `distilbert-base-uncased`, but it can be any checkpoint on the Hugging
Face Hub or one that's stored locally.
The resulting `model.onnx` file can then be run on one of the [many
accelerators](https://onnx.ai/supported-tools.html#deployModel) that support the ONNX
standard. For example, we can load and run the model with [ONNX
Runtime](https://onnxruntime.ai/) as follows:
```python
>>> from transformers import AutoTokenizer
>>> from onnxruntime import InferenceSession
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
>>> session = InferenceSession("onnx/model.onnx")
>>> # ONNX Runtime expects NumPy arrays as input
>>> inputs = tokenizer("Using DistilBERT with ONNX Runtime!", return_tensors="np")
>>> outputs = session.run(output_names=["last_hidden_state"], input_feed=dict(inputs))
```
The required output names (like `["last_hidden_state"]`) can be obtained by taking a
look at the ONNX configuration of each model. For example, for DistilBERT we have:
```python
>>> from transformers.models.distilbert import DistilBertConfig, DistilBertOnnxConfig
>>> config = DistilBertConfig()
>>> onnx_config = DistilBertOnnxConfig(config)
>>> print(list(onnx_config.outputs.keys()))
["last_hidden_state"]
```
The process is identical for TensorFlow checkpoints on the Hub. For example, we can
export a pure TensorFlow checkpoint from the [Keras
organization](https://huggingface.co/keras-io) as follows:
```bash
python -m transformers.onnx --model=keras-io/transformers-qa onnx/
```
To export a model that's stored locally, you'll need to have the model's weights and
tokenizer files stored in a directory. For example, we can load and save a checkpoint as
follows:
<frameworkcontent> <pt>
```python
>>> from transformers import AutoTokenizer, AutoModelForSequenceClassification
>>> # Load tokenizer and PyTorch weights form the Hub
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
>>> pt_model = AutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased")
>>> # Save to disk
>>> tokenizer.save_pretrained("local-pt-checkpoint")
>>> pt_model.save_pretrained("local-pt-checkpoint")
```
Once the checkpoint is saved, we can export it to ONNX by pointing the `--model`
argument of the `transformers.onnx` package to the desired directory:
```bash
python -m transformers.onnx --model=local-pt-checkpoint onnx/
```
</pt> <tf>
```python
>>> from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
>>> # Load tokenizer and TensorFlow weights from the Hub
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
>>> tf_model = TFAutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased")
>>> # Save to disk
>>> tokenizer.save_pretrained("local-tf-checkpoint")
>>> tf_model.save_pretrained("local-tf-checkpoint")
```
Once the checkpoint is saved, we can export it to ONNX by pointing the `--model`
argument of the `transformers.onnx` package to the desired directory:
```bash
python -m transformers.onnx --model=local-tf-checkpoint onnx/
```
</tf> </frameworkcontent>
## Selecting features for different model tasks
<Tip>
The recommended way of exporting a model is now to use `optimum.exporters.onnx`.
You can check the [🤗 Optimum documentation](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/export_a_model#selecting-a-task)
to learn how to select a task.
</Tip>
Each ready-made configuration comes with a set of _features_ that enable you to export
models for different types of tasks. As shown in the table below, each feature is
associated with a different `AutoClass`:
| Feature | Auto Class |
| ------------------------------------ | ------------------------------------ |
| `causal-lm`, `causal-lm-with-past` | `AutoModelForCausalLM` |
| `default`, `default-with-past` | `AutoModel` |
| `masked-lm` | `AutoModelForMaskedLM` |
| `question-answering` | `AutoModelForQuestionAnswering` |
| `seq2seq-lm`, `seq2seq-lm-with-past` | `AutoModelForSeq2SeqLM` |
| `sequence-classification` | `AutoModelForSequenceClassification` |
| `token-classification` | `AutoModelForTokenClassification` |
For each configuration, you can find the list of supported features via the
[`~transformers.onnx.FeaturesManager`]. For example, for DistilBERT we have:
```python
>>> from transformers.onnx.features import FeaturesManager
>>> distilbert_features = list(FeaturesManager.get_supported_features_for_model_type("distilbert").keys())
>>> print(distilbert_features)
["default", "masked-lm", "causal-lm", "sequence-classification", "token-classification", "question-answering"]
```
You can then pass one of these features to the `--feature` argument in the
`transformers.onnx` package. For example, to export a text-classification model we can
pick a fine-tuned model from the Hub and run:
```bash
python -m transformers.onnx --model=distilbert-base-uncased-finetuned-sst-2-english \
--feature=sequence-classification onnx/
```
This displays the following logs:
```bash
Validating ONNX model...
-[✓] ONNX model output names match reference model ({'logits'})
- Validating ONNX Model output "logits":
-[✓] (2, 2) matches (2, 2)
-[✓] all values close (atol: 1e-05)
All good, model saved at: onnx/model.onnx
```
Notice that in this case, the output names from the fine-tuned model are `logits`
instead of the `last_hidden_state` we saw with the `distilbert-base-uncased` checkpoint
earlier. This is expected since the fine-tuned model has a sequence classification head.
<Tip>
The features that have a `with-past` suffix (like `causal-lm-with-past`) correspond to
model classes with precomputed hidden states (key and values in the attention blocks)
that can be used for fast autoregressive decoding.
</Tip>
<Tip>
For `VisionEncoderDecoder` type models, the encoder and decoder parts are
exported separately as two ONNX files named `encoder_model.onnx` and `decoder_model.onnx` respectively.
</Tip>
## Exporting a model for an unsupported architecture
<Tip>
If you wish to contribute by adding support for a model that cannot be currently exported, you should first check if it is
supported in [`optimum.exporters.onnx`](https://huggingface.co/docs/optimum/main/en/exporters/onnx/package_reference/configuration#supported-architectures),
and if it is not, [contribute to 🤗 Optimum](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/contribute)
directly.
</Tip>
If you wish to export a model whose architecture is not natively supported by the
library, there are three main steps to follow:
1. Implement a custom ONNX configuration.
2. Export the model to ONNX.
3. Validate the outputs of the PyTorch and exported models.
In this section, we'll look at how DistilBERT was implemented to show what's involved
with each step.
### Implementing a custom ONNX configuration
Let's start with the ONNX configuration object. We provide three abstract classes that
you should inherit from, depending on the type of model architecture you wish to export:
* Encoder-based models inherit from [`~onnx.config.OnnxConfig`]
* Decoder-based models inherit from [`~onnx.config.OnnxConfigWithPast`]
* Encoder-decoder models inherit from [`~onnx.config.OnnxSeq2SeqConfigWithPast`]
<Tip>
A good way to implement a custom ONNX configuration is to look at the existing
implementation in the `configuration_<model_name>.py` file of a similar architecture.
</Tip>
Since DistilBERT is an encoder-based model, its configuration inherits from
`OnnxConfig`:
```python
>>> from typing import Mapping, OrderedDict
>>> from transformers.onnx import OnnxConfig
>>> class DistilBertOnnxConfig(OnnxConfig):
... @property
... def inputs(self) -> Mapping[str, Mapping[int, str]]:
... return OrderedDict(
... [
... ("input_ids", {0: "batch", 1: "sequence"}),
... ("attention_mask", {0: "batch", 1: "sequence"}),
... ]
... )
```
Every configuration object must implement the `inputs` property and return a mapping,
where each key corresponds to an expected input, and each value indicates the axis of
that input. For DistilBERT, we can see that two inputs are required: `input_ids` and
`attention_mask`. These inputs have the same shape of `(batch_size, sequence_length)`
which is why we see the same axes used in the configuration.
<Tip>
Notice that `inputs` property for `DistilBertOnnxConfig` returns an `OrderedDict`. This
ensures that the inputs are matched with their relative position within the
`PreTrainedModel.forward()` method when tracing the graph. We recommend using an
`OrderedDict` for the `inputs` and `outputs` properties when implementing custom ONNX
configurations.
</Tip>
Once you have implemented an ONNX configuration, you can instantiate it by providing the
base model's configuration as follows:
```python
>>> from transformers import AutoConfig
>>> config = AutoConfig.from_pretrained("distilbert-base-uncased")
>>> onnx_config = DistilBertOnnxConfig(config)
```
The resulting object has several useful properties. For example, you can view the ONNX
operator set that will be used during the export:
```python
>>> print(onnx_config.default_onnx_opset)
11
```
You can also view the outputs associated with the model as follows:
```python
>>> print(onnx_config.outputs)
OrderedDict([("last_hidden_state", {0: "batch", 1: "sequence"})])
```
Notice that the outputs property follows the same structure as the inputs; it returns an
`OrderedDict` of named outputs and their shapes. The output structure is linked to the
choice of feature that the configuration is initialised with. By default, the ONNX
configuration is initialized with the `default` feature that corresponds to exporting a
model loaded with the `AutoModel` class. If you want to export a model for another task,
just provide a different feature to the `task` argument when you initialize the ONNX
configuration. For example, if we wished to export DistilBERT with a sequence
classification head, we could use:
```python
>>> from transformers import AutoConfig
>>> config = AutoConfig.from_pretrained("distilbert-base-uncased")
>>> onnx_config_for_seq_clf = DistilBertOnnxConfig(config, task="sequence-classification")
>>> print(onnx_config_for_seq_clf.outputs)
OrderedDict([('logits', {0: 'batch'})])
```
<Tip>
All of the base properties and methods associated with [`~onnx.config.OnnxConfig`] and
the other configuration classes can be overridden if needed. Check out [`BartOnnxConfig`]
for an advanced example.
</Tip>
### Exporting the model
Once you have implemented the ONNX configuration, the next step is to export the model.
Here we can use the `export()` function provided by the `transformers.onnx` package.
This function expects the ONNX configuration, along with the base model and tokenizer,
and the path to save the exported file:
```python
>>> from pathlib import Path
>>> from transformers.onnx import export
>>> from transformers import AutoTokenizer, AutoModel
>>> onnx_path = Path("model.onnx")
>>> model_ckpt = "distilbert-base-uncased"
>>> base_model = AutoModel.from_pretrained(model_ckpt)
>>> tokenizer = AutoTokenizer.from_pretrained(model_ckpt)
>>> onnx_inputs, onnx_outputs = export(tokenizer, base_model, onnx_config, onnx_config.default_onnx_opset, onnx_path)
```
The `onnx_inputs` and `onnx_outputs` returned by the `export()` function are lists of
the keys defined in the `inputs` and `outputs` properties of the configuration. Once the
model is exported, you can test that the model is well formed as follows:
```python
>>> import onnx
>>> onnx_model = onnx.load("model.onnx")
>>> onnx.checker.check_model(onnx_model)
```
<Tip>
If your model is larger than 2GB, you will see that many additional files are created
during the export. This is _expected_ because ONNX uses [Protocol
Buffers](https://developers.google.com/protocol-buffers/) to store the model and these
have a size limit of 2GB. See the [ONNX
documentation](https://github.com/onnx/onnx/blob/master/docs/ExternalData.md) for
instructions on how to load models with external data.
</Tip>
### Validating the model outputs
The final step is to validate that the outputs from the base and exported model agree
within some absolute tolerance. Here we can use the `validate_model_outputs()` function
provided by the `transformers.onnx` package as follows:
```python
>>> from transformers.onnx import validate_model_outputs
>>> validate_model_outputs(
... onnx_config, tokenizer, base_model, onnx_path, onnx_outputs, onnx_config.atol_for_validation
... )
```
This function uses the [`~transformers.onnx.OnnxConfig.generate_dummy_inputs`] method to
generate inputs for the base and exported model, and the absolute tolerance can be
defined in the configuration. We generally find numerical agreement in the 1e-6 to 1e-4
range, although anything smaller than 1e-3 is likely to be OK.
## Contributing a new configuration to 🤗 Transformers
We are looking to expand the set of ready-made configurations and welcome contributions
from the community! If you would like to contribute your addition to the library, you
will need to:
* Implement the ONNX configuration in the corresponding `configuration_<model_name>.py`
file
* Include the model architecture and corresponding features in
[`~onnx.features.FeatureManager`]
* Add your model architecture to the tests in `test_onnx_v2.py`
Check out how the configuration for [IBERT was
contributed](https://github.com/huggingface/transformers/pull/14868/files) to get an
idea of what's involved.
| <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# Export to ONNX
If you need to deploy 🤗 Transformers models in production environments, we recommend
exporting them to a serialized format that can be loaded and executed on specialized
runtimes and hardware. In this guide, we'll show you how to export 🤗 Transformers
models to [ONNX (Open Neural Network eXchange)](http://onnx.ai).
ONNX is an open standard that defines a common set of operators and a common file format
to represent deep learning models in a wide variety of frameworks, including PyTorch and
TensorFlow. When a model is exported to the ONNX format, these operators are used to
construct a computational graph (often called an _intermediate representation_) which
represents the flow of data through the neural network.
By exposing a graph with standardized operators and data types, ONNX makes it easy to
switch between frameworks. For example, a model trained in PyTorch can be exported to
ONNX format and then imported in TensorFlow (and vice versa).
🤗 Transformers provides a [`transformers.onnx`](main_classes/onnx) package that enables
you to convert model checkpoints to an ONNX graph by leveraging configuration objects.
These configuration objects come ready made for a number of model architectures, and are
designed to be easily extendable to other architectures.
<Tip>
You can also export 🤗 Transformers models with the [`optimum.exporters.onnx` package](https://huggingface.co/docs/optimum/exporters/onnx/usage_guides/export_a_model)
from 🤗 Optimum.
Once exported, a model can be:
- Optimized for inference via techniques such as quantization and graph optimization.
- Run with ONNX Runtime via [`ORTModelForXXX` classes](https://huggingface.co/docs/optimum/onnxruntime/package_reference/modeling_ort),
which follow the same `AutoModel` API as the one you are used to in 🤗 Transformers.
- Run with [optimized inference pipelines](https://huggingface.co/docs/optimum/main/en/onnxruntime/usage_guides/pipelines),
which has the same API as the [`pipeline`] function in 🤗 Transformers.
To explore all these features, check out the [🤗 Optimum library](https://github.com/huggingface/optimum).
</Tip>
Ready-made configurations include the following architectures:
<!--This table is automatically generated by `make fix-copies`, do not fill manually!-->
- ALBERT
- BART
- BEiT
- BERT
- BigBird
- BigBird-Pegasus
- Blenderbot
- BlenderbotSmall
- BLOOM
- CamemBERT
- Chinese-CLIP
- CLIP
- CodeGen
- Conditional DETR
- ConvBERT
- ConvNeXT
- Data2VecText
- Data2VecVision
- DeBERTa
- DeBERTa-v2
- DeiT
- DETR
- DistilBERT
- ELECTRA
- ERNIE
- FlauBERT
- GPT Neo
- GPT-J
- GPT-Sw3
- GroupViT
- I-BERT
- ImageGPT
- LayoutLM
- LayoutLMv3
- LeViT
- Longformer
- LongT5
- M2M100
- Marian
- mBART
- MobileBERT
- MobileNetV1
- MobileNetV2
- MobileViT
- MT5
- OpenAI GPT-2
- OWL-ViT
- Perceiver
- PLBart
- RemBERT
- ResNet
- RoBERTa
- RoFormer
- SegFormer
- SqueezeBERT
- Swin Transformer
- T5
- Table Transformer
- Vision Encoder decoder
- ViT
- Whisper
- XLM
- XLM-RoBERTa
- XLM-RoBERTa-XL
- YOLOS
In the next two sections, we'll show you how to:
* Export a supported model using the `transformers.onnx` package.
* Export a custom model for an unsupported architecture.
## Exporting a model to ONNX
<Tip>
The recommended way of exporting a model is now to use
[`optimum.exporters.onnx`](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/export_a_model#exporting-a-model-to-onnx-using-the-cli),
do not worry it is very similar to `transformers.onnx`!
</Tip>
To export a 🤗 Transformers model to ONNX, you'll first need to install some extra
dependencies:
```bash
pip install transformers[onnx]
```
The `transformers.onnx` package can then be used as a Python module:
```bash
python -m transformers.onnx --help
usage: Hugging Face Transformers ONNX exporter [-h] -m MODEL [--feature {causal-lm, ...}] [--opset OPSET] [--atol ATOL] output
positional arguments:
output Path indicating where to store generated ONNX model.
optional arguments:
-h, --help show this help message and exit
-m MODEL, --model MODEL
Model ID on huggingface.co or path on disk to load model from.
--feature {causal-lm, ...}
The type of features to export the model with.
--opset OPSET ONNX opset version to export the model with.
--atol ATOL Absolute difference tolerance when validating the model.
```
Exporting a checkpoint using a ready-made configuration can be done as follows:
```bash
python -m transformers.onnx --model=distilbert-base-uncased onnx/
```
You should see the following logs:
```bash
Validating ONNX model...
-[✓] ONNX model output names match reference model ({'last_hidden_state'})
- Validating ONNX Model output "last_hidden_state":
-[✓] (2, 8, 768) matches (2, 8, 768)
-[✓] all values close (atol: 1e-05)
All good, model saved at: onnx/model.onnx
```
This exports an ONNX graph of the checkpoint defined by the `--model` argument. In this
example, it is `distilbert-base-uncased`, but it can be any checkpoint on the Hugging
Face Hub or one that's stored locally.
The resulting `model.onnx` file can then be run on one of the [many
accelerators](https://onnx.ai/supported-tools.html#deployModel) that support the ONNX
standard. For example, we can load and run the model with [ONNX
Runtime](https://onnxruntime.ai/) as follows:
```python
>>> from transformers import AutoTokenizer
>>> from onnxruntime import InferenceSession
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
>>> session = InferenceSession("onnx/model.onnx")
>>> # ONNX Runtime expects NumPy arrays as input
>>> inputs = tokenizer("Using DistilBERT with ONNX Runtime!", return_tensors="np")
>>> outputs = session.run(output_names=["last_hidden_state"], input_feed=dict(inputs))
```
The required output names (like `["last_hidden_state"]`) can be obtained by taking a
look at the ONNX configuration of each model. For example, for DistilBERT we have:
```python
>>> from transformers.models.distilbert import DistilBertConfig, DistilBertOnnxConfig
>>> config = DistilBertConfig()
>>> onnx_config = DistilBertOnnxConfig(config)
>>> print(list(onnx_config.outputs.keys()))
["last_hidden_state"]
```
The process is identical for TensorFlow checkpoints on the Hub. For example, we can
export a pure TensorFlow checkpoint from the [Keras
organization](https://huggingface.co/keras-io) as follows:
```bash
python -m transformers.onnx --model=keras-io/transformers-qa onnx/
```
To export a model that's stored locally, you'll need to have the model's weights and
tokenizer files stored in a directory. For example, we can load and save a checkpoint as
follows:
<frameworkcontent> <pt>
```python
>>> from transformers import AutoTokenizer, AutoModelForSequenceClassification
>>> # Load tokenizer and PyTorch weights form the Hub
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
>>> pt_model = AutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased")
>>> # Save to disk
>>> tokenizer.save_pretrained("local-pt-checkpoint")
>>> pt_model.save_pretrained("local-pt-checkpoint")
```
Once the checkpoint is saved, we can export it to ONNX by pointing the `--model`
argument of the `transformers.onnx` package to the desired directory:
```bash
python -m transformers.onnx --model=local-pt-checkpoint onnx/
```
</pt> <tf>
```python
>>> from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
>>> # Load tokenizer and TensorFlow weights from the Hub
>>> tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
>>> tf_model = TFAutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased")
>>> # Save to disk
>>> tokenizer.save_pretrained("local-tf-checkpoint")
>>> tf_model.save_pretrained("local-tf-checkpoint")
```
Once the checkpoint is saved, we can export it to ONNX by pointing the `--model`
argument of the `transformers.onnx` package to the desired directory:
```bash
python -m transformers.onnx --model=local-tf-checkpoint onnx/
```
</tf> </frameworkcontent>
## Selecting features for different model tasks
<Tip>
The recommended way of exporting a model is now to use `optimum.exporters.onnx`.
You can check the [🤗 Optimum documentation](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/export_a_model#selecting-a-task)
to learn how to select a task.
</Tip>
Each ready-made configuration comes with a set of _features_ that enable you to export
models for different types of tasks. As shown in the table below, each feature is
associated with a different `AutoClass`:
| Feature | Auto Class |
| ------------------------------------ | ------------------------------------ |
| `causal-lm`, `causal-lm-with-past` | `AutoModelForCausalLM` |
| `default`, `default-with-past` | `AutoModel` |
| `masked-lm` | `AutoModelForMaskedLM` |
| `question-answering` | `AutoModelForQuestionAnswering` |
| `seq2seq-lm`, `seq2seq-lm-with-past` | `AutoModelForSeq2SeqLM` |
| `sequence-classification` | `AutoModelForSequenceClassification` |
| `token-classification` | `AutoModelForTokenClassification` |
For each configuration, you can find the list of supported features via the
[`~transformers.onnx.FeaturesManager`]. For example, for DistilBERT we have:
```python
>>> from transformers.onnx.features import FeaturesManager
>>> distilbert_features = list(FeaturesManager.get_supported_features_for_model_type("distilbert").keys())
>>> print(distilbert_features)
["default", "masked-lm", "causal-lm", "sequence-classification", "token-classification", "question-answering"]
```
You can then pass one of these features to the `--feature` argument in the
`transformers.onnx` package. For example, to export a text-classification model we can
pick a fine-tuned model from the Hub and run:
```bash
python -m transformers.onnx --model=distilbert-base-uncased-finetuned-sst-2-english \
--feature=sequence-classification onnx/
```
This displays the following logs:
```bash
Validating ONNX model...
-[✓] ONNX model output names match reference model ({'logits'})
- Validating ONNX Model output "logits":
-[✓] (2, 2) matches (2, 2)
-[✓] all values close (atol: 1e-05)
All good, model saved at: onnx/model.onnx
```
Notice that in this case, the output names from the fine-tuned model are `logits`
instead of the `last_hidden_state` we saw with the `distilbert-base-uncased` checkpoint
earlier. This is expected since the fine-tuned model has a sequence classification head.
<Tip>
The features that have a `with-past` suffix (like `causal-lm-with-past`) correspond to
model classes with precomputed hidden states (key and values in the attention blocks)
that can be used for fast autoregressive decoding.
</Tip>
<Tip>
For `VisionEncoderDecoder` type models, the encoder and decoder parts are
exported separately as two ONNX files named `encoder_model.onnx` and `decoder_model.onnx` respectively.
</Tip>
## Exporting a model for an unsupported architecture
<Tip>
If you wish to contribute by adding support for a model that cannot be currently exported, you should first check if it is
supported in [`optimum.exporters.onnx`](https://huggingface.co/docs/optimum/main/en/exporters/onnx/package_reference/configuration#supported-architectures),
and if it is not, [contribute to 🤗 Optimum](https://huggingface.co/docs/optimum/main/en/exporters/onnx/usage_guides/contribute)
directly.
</Tip>
If you wish to export a model whose architecture is not natively supported by the
library, there are three main steps to follow:
1. Implement a custom ONNX configuration.
2. Export the model to ONNX.
3. Validate the outputs of the PyTorch and exported models.
In this section, we'll look at how DistilBERT was implemented to show what's involved
with each step.
### Implementing a custom ONNX configuration
Let's start with the ONNX configuration object. We provide three abstract classes that
you should inherit from, depending on the type of model architecture you wish to export:
* Encoder-based models inherit from [`~onnx.config.OnnxConfig`]
* Decoder-based models inherit from [`~onnx.config.OnnxConfigWithPast`]
* Encoder-decoder models inherit from [`~onnx.config.OnnxSeq2SeqConfigWithPast`]
<Tip>
A good way to implement a custom ONNX configuration is to look at the existing
implementation in the `configuration_<model_name>.py` file of a similar architecture.
</Tip>
Since DistilBERT is an encoder-based model, its configuration inherits from
`OnnxConfig`:
```python
>>> from typing import Mapping, OrderedDict
>>> from transformers.onnx import OnnxConfig
>>> class DistilBertOnnxConfig(OnnxConfig):
... @property
... def inputs(self) -> Mapping[str, Mapping[int, str]]:
... return OrderedDict(
... [
... ("input_ids", {0: "batch", 1: "sequence"}),
... ("attention_mask", {0: "batch", 1: "sequence"}),
... ]
... )
```
Every configuration object must implement the `inputs` property and return a mapping,
where each key corresponds to an expected input, and each value indicates the axis of
that input. For DistilBERT, we can see that two inputs are required: `input_ids` and
`attention_mask`. These inputs have the same shape of `(batch_size, sequence_length)`
which is why we see the same axes used in the configuration.
<Tip>
Notice that `inputs` property for `DistilBertOnnxConfig` returns an `OrderedDict`. This
ensures that the inputs are matched with their relative position within the
`PreTrainedModel.forward()` method when tracing the graph. We recommend using an
`OrderedDict` for the `inputs` and `outputs` properties when implementing custom ONNX
configurations.
</Tip>
Once you have implemented an ONNX configuration, you can instantiate it by providing the
base model's configuration as follows:
```python
>>> from transformers import AutoConfig
>>> config = AutoConfig.from_pretrained("distilbert-base-uncased")
>>> onnx_config = DistilBertOnnxConfig(config)
```
The resulting object has several useful properties. For example, you can view the ONNX
operator set that will be used during the export:
```python
>>> print(onnx_config.default_onnx_opset)
11
```
You can also view the outputs associated with the model as follows:
```python
>>> print(onnx_config.outputs)
OrderedDict([("last_hidden_state", {0: "batch", 1: "sequence"})])
```
Notice that the outputs property follows the same structure as the inputs; it returns an
`OrderedDict` of named outputs and their shapes. The output structure is linked to the
choice of feature that the configuration is initialised with. By default, the ONNX
configuration is initialized with the `default` feature that corresponds to exporting a
model loaded with the `AutoModel` class. If you want to export a model for another task,
just provide a different feature to the `task` argument when you initialize the ONNX
configuration. For example, if we wished to export DistilBERT with a sequence
classification head, we could use:
```python
>>> from transformers import AutoConfig
>>> config = AutoConfig.from_pretrained("distilbert-base-uncased")
>>> onnx_config_for_seq_clf = DistilBertOnnxConfig(config, task="sequence-classification")
>>> print(onnx_config_for_seq_clf.outputs)
OrderedDict([('logits', {0: 'batch'})])
```
<Tip>
All of the base properties and methods associated with [`~onnx.config.OnnxConfig`] and
the other configuration classes can be overridden if needed. Check out [`BartOnnxConfig`]
for an advanced example.
</Tip>
### Exporting the model
Once you have implemented the ONNX configuration, the next step is to export the model.
Here we can use the `export()` function provided by the `transformers.onnx` package.
This function expects the ONNX configuration, along with the base model and tokenizer,
and the path to save the exported file:
```python
>>> from pathlib import Path
>>> from transformers.onnx import export
>>> from transformers import AutoTokenizer, AutoModel
>>> onnx_path = Path("model.onnx")
>>> model_ckpt = "distilbert-base-uncased"
>>> base_model = AutoModel.from_pretrained(model_ckpt)
>>> tokenizer = AutoTokenizer.from_pretrained(model_ckpt)
>>> onnx_inputs, onnx_outputs = export(tokenizer, base_model, onnx_config, onnx_config.default_onnx_opset, onnx_path)
```
The `onnx_inputs` and `onnx_outputs` returned by the `export()` function are lists of
the keys defined in the `inputs` and `outputs` properties of the configuration. Once the
model is exported, you can test that the model is well formed as follows:
```python
>>> import onnx
>>> onnx_model = onnx.load("model.onnx")
>>> onnx.checker.check_model(onnx_model)
```
<Tip>
If your model is larger than 2GB, you will see that many additional files are created
during the export. This is _expected_ because ONNX uses [Protocol
Buffers](https://developers.google.com/protocol-buffers/) to store the model and these
have a size limit of 2GB. See the [ONNX
documentation](https://github.com/onnx/onnx/blob/master/docs/ExternalData.md) for
instructions on how to load models with external data.
</Tip>
### Validating the model outputs
The final step is to validate that the outputs from the base and exported model agree
within some absolute tolerance. Here we can use the `validate_model_outputs()` function
provided by the `transformers.onnx` package as follows:
```python
>>> from transformers.onnx import validate_model_outputs
>>> validate_model_outputs(
... onnx_config, tokenizer, base_model, onnx_path, onnx_outputs, onnx_config.atol_for_validation
... )
```
This function uses the [`~transformers.onnx.OnnxConfig.generate_dummy_inputs`] method to
generate inputs for the base and exported model, and the absolute tolerance can be
defined in the configuration. We generally find numerical agreement in the 1e-6 to 1e-4
range, although anything smaller than 1e-3 is likely to be OK.
## Contributing a new configuration to 🤗 Transformers
We are looking to expand the set of ready-made configurations and welcome contributions
from the community! If you would like to contribute your addition to the library, you
will need to:
* Implement the ONNX configuration in the corresponding `configuration_<model_name>.py`
file
* Include the model architecture and corresponding features in
[`~onnx.features.FeatureManager`]
* Add your model architecture to the tests in `test_onnx_v2.py`
Check out how the configuration for [IBERT was
contributed](https://github.com/huggingface/transformers/pull/14868/files) to get an
idea of what's involved.
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/__init__.py | # flake8: noqa
# There's no way to ignore "F401 '...' imported but unused" warnings in this
# module, but to preserve other warnings. So, don't check this module at all.
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# When adding a new object to this init, remember to add it twice: once inside the `_import_structure` dictionary and
# once inside the `if TYPE_CHECKING` branch. The `TYPE_CHECKING` should have import statements as usual, but they are
# only there for type checking. The `_import_structure` is a dictionary submodule to list of object names, and is used
# to defer the actual importing for when the objects are requested. This way `import transformers` provides the names
# in the namespace without actually importing anything (and especially none of the backends).
__version__ = "4.26.0.dev0"
from typing import TYPE_CHECKING
# Check the dependencies satisfy the minimal versions required.
from . import dependency_versions_check
from .utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_keras_nlp_available,
is_sentencepiece_available,
is_speech_available,
is_tensorflow_text_available,
is_tf_available,
is_timm_available,
is_tokenizers_available,
is_torch_available,
is_vision_available,
logging,
)
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
# Base objects, independent of any specific backend
_import_structure = {
"benchmark": [],
"commands": [],
"configuration_utils": ["PretrainedConfig"],
"convert_graph_to_onnx": [],
"convert_slow_tokenizers_checkpoints_to_fast": [],
"convert_tf_hub_seq_to_seq_bert_to_pytorch": [],
"data": [
"DataProcessor",
"InputExample",
"InputFeatures",
"SingleSentenceClassificationProcessor",
"SquadExample",
"SquadFeatures",
"SquadV1Processor",
"SquadV2Processor",
"glue_compute_metrics",
"glue_convert_examples_to_features",
"glue_output_modes",
"glue_processors",
"glue_tasks_num_labels",
"squad_convert_examples_to_features",
"xnli_compute_metrics",
"xnli_output_modes",
"xnli_processors",
"xnli_tasks_num_labels",
],
"data.data_collator": [
"DataCollator",
"DataCollatorForLanguageModeling",
"DataCollatorForPermutationLanguageModeling",
"DataCollatorForSeq2Seq",
"DataCollatorForSOP",
"DataCollatorForTokenClassification",
"DataCollatorForWholeWordMask",
"DataCollatorWithPadding",
"DefaultDataCollator",
"default_data_collator",
],
"data.metrics": [],
"data.processors": [],
"debug_utils": [],
"dependency_versions_check": [],
"dependency_versions_table": [],
"dynamic_module_utils": [],
"feature_extraction_sequence_utils": ["SequenceFeatureExtractor"],
"feature_extraction_utils": ["BatchFeature", "FeatureExtractionMixin"],
"file_utils": [],
"generation": ["GenerationConfig"],
"hf_argparser": ["HfArgumentParser"],
"integrations": [
"is_clearml_available",
"is_comet_available",
"is_neptune_available",
"is_optuna_available",
"is_ray_available",
"is_ray_tune_available",
"is_sigopt_available",
"is_tensorboard_available",
"is_wandb_available",
],
"modelcard": ["ModelCard"],
"modeling_tf_pytorch_utils": [
"convert_tf_weight_name_to_pt_weight_name",
"load_pytorch_checkpoint_in_tf2_model",
"load_pytorch_model_in_tf2_model",
"load_pytorch_weights_in_tf2_model",
"load_tf2_checkpoint_in_pytorch_model",
"load_tf2_model_in_pytorch_model",
"load_tf2_weights_in_pytorch_model",
],
"models": [],
# Models
"models.albert": ["ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "AlbertConfig"],
"models.audio_spectrogram_transformer": [
"AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ASTConfig",
],
"models.auto": [
"ALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"CONFIG_MAPPING",
"FEATURE_EXTRACTOR_MAPPING",
"IMAGE_PROCESSOR_MAPPING",
"MODEL_NAMES_MAPPING",
"PROCESSOR_MAPPING",
"TOKENIZER_MAPPING",
"AutoConfig",
"AutoFeatureExtractor",
"AutoImageProcessor",
"AutoProcessor",
"AutoTokenizer",
],
"models.bart": ["BartConfig", "BartTokenizer"],
"models.barthez": [],
"models.bartpho": [],
"models.beit": ["BEIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BeitConfig"],
"models.bert": [
"BERT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BasicTokenizer",
"BertConfig",
"BertTokenizer",
"WordpieceTokenizer",
],
"models.bert_generation": ["BertGenerationConfig"],
"models.bert_japanese": ["BertJapaneseTokenizer", "CharacterTokenizer", "MecabTokenizer"],
"models.bertweet": ["BertweetTokenizer"],
"models.big_bird": ["BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP", "BigBirdConfig"],
"models.bigbird_pegasus": [
"BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BigBirdPegasusConfig",
],
"models.biogpt": ["BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BioGptConfig", "BioGptTokenizer"],
"models.bit": ["BIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BitConfig"],
"models.blenderbot": ["BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BlenderbotConfig", "BlenderbotTokenizer"],
"models.blenderbot_small": [
"BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BlenderbotSmallConfig",
"BlenderbotSmallTokenizer",
],
"models.bloom": ["BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP", "BloomConfig"],
"models.bort": [],
"models.byt5": ["ByT5Tokenizer"],
"models.camembert": ["CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "CamembertConfig"],
"models.canine": ["CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP", "CanineConfig", "CanineTokenizer"],
"models.chinese_clip": [
"CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ChineseCLIPConfig",
"ChineseCLIPProcessor",
"ChineseCLIPTextConfig",
"ChineseCLIPVisionConfig",
],
"models.clip": [
"CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
"CLIPConfig",
"CLIPProcessor",
"CLIPTextConfig",
"CLIPTokenizer",
"CLIPVisionConfig",
],
"models.clipseg": [
"CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP",
"CLIPSegConfig",
"CLIPSegProcessor",
"CLIPSegTextConfig",
"CLIPSegVisionConfig",
],
"models.codegen": ["CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP", "CodeGenConfig", "CodeGenTokenizer"],
"models.conditional_detr": ["CONDITIONAL_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP", "ConditionalDetrConfig"],
"models.convbert": ["CONVBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ConvBertConfig", "ConvBertTokenizer"],
"models.convnext": ["CONVNEXT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ConvNextConfig"],
"models.cpm": [],
"models.ctrl": ["CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP", "CTRLConfig", "CTRLTokenizer"],
"models.cvt": ["CVT_PRETRAINED_CONFIG_ARCHIVE_MAP", "CvtConfig"],
"models.data2vec": [
"DATA2VEC_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"DATA2VEC_VISION_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Data2VecAudioConfig",
"Data2VecTextConfig",
"Data2VecVisionConfig",
],
"models.deberta": ["DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP", "DebertaConfig", "DebertaTokenizer"],
"models.deberta_v2": ["DEBERTA_V2_PRETRAINED_CONFIG_ARCHIVE_MAP", "DebertaV2Config"],
"models.decision_transformer": ["DECISION_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "DecisionTransformerConfig"],
"models.deformable_detr": ["DEFORMABLE_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP", "DeformableDetrConfig"],
"models.deit": ["DEIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DeiTConfig"],
"models.detr": ["DETR_PRETRAINED_CONFIG_ARCHIVE_MAP", "DetrConfig"],
"models.dialogpt": [],
"models.dinat": ["DINAT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DinatConfig"],
"models.distilbert": ["DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DistilBertConfig", "DistilBertTokenizer"],
"models.dit": [],
"models.donut": ["DONUT_SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP", "DonutProcessor", "DonutSwinConfig"],
"models.dpr": [
"DPR_PRETRAINED_CONFIG_ARCHIVE_MAP",
"DPRConfig",
"DPRContextEncoderTokenizer",
"DPRQuestionEncoderTokenizer",
"DPRReaderOutput",
"DPRReaderTokenizer",
],
"models.dpt": ["DPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DPTConfig"],
"models.electra": ["ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP", "ElectraConfig", "ElectraTokenizer"],
"models.encoder_decoder": ["EncoderDecoderConfig"],
"models.ernie": [
"ERNIE_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ErnieConfig",
],
"models.esm": ["ESM_PRETRAINED_CONFIG_ARCHIVE_MAP", "EsmConfig", "EsmTokenizer"],
"models.flaubert": ["FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "FlaubertConfig", "FlaubertTokenizer"],
"models.flava": [
"FLAVA_PRETRAINED_CONFIG_ARCHIVE_MAP",
"FlavaConfig",
"FlavaImageCodebookConfig",
"FlavaImageConfig",
"FlavaMultimodalConfig",
"FlavaTextConfig",
],
"models.fnet": ["FNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "FNetConfig"],
"models.fsmt": ["FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP", "FSMTConfig", "FSMTTokenizer"],
"models.funnel": ["FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP", "FunnelConfig", "FunnelTokenizer"],
"models.glpn": ["GLPN_PRETRAINED_CONFIG_ARCHIVE_MAP", "GLPNConfig"],
"models.gpt2": ["GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPT2Config", "GPT2Tokenizer"],
"models.gpt_neo": ["GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoConfig"],
"models.gpt_neox": ["GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXConfig"],
"models.gpt_neox_japanese": ["GPT_NEOX_JAPANESE_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXJapaneseConfig"],
"models.gptj": ["GPTJ_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTJConfig"],
"models.groupvit": [
"GROUPVIT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"GroupViTConfig",
"GroupViTTextConfig",
"GroupViTVisionConfig",
],
"models.herbert": ["HerbertTokenizer"],
"models.hubert": ["HUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "HubertConfig"],
"models.ibert": ["IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "IBertConfig"],
"models.imagegpt": ["IMAGEGPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ImageGPTConfig"],
"models.jukebox": [
"JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP",
"JukeboxConfig",
"JukeboxPriorConfig",
"JukeboxTokenizer",
"JukeboxVQVAEConfig",
],
"models.layoutlm": ["LAYOUTLM_PRETRAINED_CONFIG_ARCHIVE_MAP", "LayoutLMConfig", "LayoutLMTokenizer"],
"models.layoutlmv2": [
"LAYOUTLMV2_PRETRAINED_CONFIG_ARCHIVE_MAP",
"LayoutLMv2Config",
"LayoutLMv2FeatureExtractor",
"LayoutLMv2ImageProcessor",
"LayoutLMv2Processor",
"LayoutLMv2Tokenizer",
],
"models.layoutlmv3": [
"LAYOUTLMV3_PRETRAINED_CONFIG_ARCHIVE_MAP",
"LayoutLMv3Config",
"LayoutLMv3FeatureExtractor",
"LayoutLMv3ImageProcessor",
"LayoutLMv3Processor",
"LayoutLMv3Tokenizer",
],
"models.layoutxlm": ["LayoutXLMProcessor"],
"models.led": ["LED_PRETRAINED_CONFIG_ARCHIVE_MAP", "LEDConfig", "LEDTokenizer"],
"models.levit": ["LEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "LevitConfig"],
"models.lilt": ["LILT_PRETRAINED_CONFIG_ARCHIVE_MAP", "LiltConfig"],
"models.longformer": ["LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "LongformerConfig", "LongformerTokenizer"],
"models.longt5": ["LONGT5_PRETRAINED_CONFIG_ARCHIVE_MAP", "LongT5Config"],
"models.luke": ["LUKE_PRETRAINED_CONFIG_ARCHIVE_MAP", "LukeConfig", "LukeTokenizer"],
"models.lxmert": ["LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "LxmertConfig", "LxmertTokenizer"],
"models.m2m_100": ["M2M_100_PRETRAINED_CONFIG_ARCHIVE_MAP", "M2M100Config"],
"models.marian": ["MarianConfig"],
"models.markuplm": [
"MARKUPLM_PRETRAINED_CONFIG_ARCHIVE_MAP",
"MarkupLMConfig",
"MarkupLMFeatureExtractor",
"MarkupLMProcessor",
"MarkupLMTokenizer",
],
"models.maskformer": ["MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "MaskFormerConfig", "MaskFormerSwinConfig"],
"models.mbart": ["MBartConfig"],
"models.mbart50": [],
"models.mctct": ["MCTCT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MCTCTConfig", "MCTCTProcessor"],
"models.megatron_bert": ["MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MegatronBertConfig"],
"models.megatron_gpt2": [],
"models.mluke": [],
"models.mmbt": ["MMBTConfig"],
"models.mobilebert": ["MOBILEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileBertConfig", "MobileBertTokenizer"],
"models.mobilenet_v1": ["MOBILENET_V1_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileNetV1Config"],
"models.mobilenet_v2": ["MOBILENET_V2_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileNetV2Config"],
"models.mobilevit": ["MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileViTConfig"],
"models.mpnet": ["MPNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "MPNetConfig", "MPNetTokenizer"],
"models.mt5": ["MT5Config"],
"models.mvp": ["MvpConfig", "MvpTokenizer"],
"models.nat": ["NAT_PRETRAINED_CONFIG_ARCHIVE_MAP", "NatConfig"],
"models.nezha": ["NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP", "NezhaConfig"],
"models.nllb": [],
"models.nystromformer": [
"NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"NystromformerConfig",
],
"models.openai": ["OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "OpenAIGPTConfig", "OpenAIGPTTokenizer"],
"models.opt": ["OPTConfig"],
"models.owlvit": [
"OWLVIT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"OwlViTConfig",
"OwlViTProcessor",
"OwlViTTextConfig",
"OwlViTVisionConfig",
],
"models.pegasus": ["PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP", "PegasusConfig", "PegasusTokenizer"],
"models.pegasus_x": ["PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP", "PegasusXConfig"],
"models.perceiver": ["PERCEIVER_PRETRAINED_CONFIG_ARCHIVE_MAP", "PerceiverConfig", "PerceiverTokenizer"],
"models.phobert": ["PhobertTokenizer"],
"models.plbart": ["PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP", "PLBartConfig"],
"models.poolformer": ["POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "PoolFormerConfig"],
"models.prophetnet": ["PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "ProphetNetConfig", "ProphetNetTokenizer"],
"models.qdqbert": ["QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "QDQBertConfig"],
"models.rag": ["RagConfig", "RagRetriever", "RagTokenizer"],
"models.realm": ["REALM_PRETRAINED_CONFIG_ARCHIVE_MAP", "RealmConfig", "RealmTokenizer"],
"models.reformer": ["REFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "ReformerConfig"],
"models.regnet": ["REGNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "RegNetConfig"],
"models.rembert": ["REMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "RemBertConfig"],
"models.resnet": ["RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "ResNetConfig"],
"models.retribert": ["RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "RetriBertConfig", "RetriBertTokenizer"],
"models.roberta": ["ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP", "RobertaConfig", "RobertaTokenizer"],
"models.roc_bert": ["ROC_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "RoCBertConfig", "RoCBertTokenizer"],
"models.roformer": ["ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "RoFormerConfig", "RoFormerTokenizer"],
"models.segformer": ["SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "SegformerConfig"],
"models.sew": ["SEW_PRETRAINED_CONFIG_ARCHIVE_MAP", "SEWConfig"],
"models.sew_d": ["SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP", "SEWDConfig"],
"models.speech_encoder_decoder": ["SpeechEncoderDecoderConfig"],
"models.speech_to_text": [
"SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Speech2TextConfig",
],
"models.speech_to_text_2": [
"SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Speech2Text2Config",
"Speech2Text2Processor",
"Speech2Text2Tokenizer",
],
"models.splinter": ["SPLINTER_PRETRAINED_CONFIG_ARCHIVE_MAP", "SplinterConfig", "SplinterTokenizer"],
"models.squeezebert": ["SQUEEZEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "SqueezeBertConfig", "SqueezeBertTokenizer"],
"models.swin": ["SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP", "SwinConfig"],
"models.swinv2": ["SWINV2_PRETRAINED_CONFIG_ARCHIVE_MAP", "Swinv2Config"],
"models.switch_transformers": ["SWITCH_TRANSFORMERS_PRETRAINED_CONFIG_ARCHIVE_MAP", "SwitchTransformersConfig"],
"models.t5": ["T5_PRETRAINED_CONFIG_ARCHIVE_MAP", "T5Config"],
"models.table_transformer": ["TABLE_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "TableTransformerConfig"],
"models.tapas": ["TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP", "TapasConfig", "TapasTokenizer"],
"models.tapex": ["TapexTokenizer"],
"models.time_series_transformer": [
"TIME_SERIES_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TimeSeriesTransformerConfig",
],
"models.timesformer": ["TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "TimesformerConfig"],
"models.trajectory_transformer": [
"TRAJECTORY_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TrajectoryTransformerConfig",
],
"models.transfo_xl": [
"TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TransfoXLConfig",
"TransfoXLCorpus",
"TransfoXLTokenizer",
],
"models.trocr": [
"TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TrOCRConfig",
"TrOCRProcessor",
],
"models.unispeech": [
"UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP",
"UniSpeechConfig",
],
"models.unispeech_sat": [
"UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"UniSpeechSatConfig",
],
"models.van": ["VAN_PRETRAINED_CONFIG_ARCHIVE_MAP", "VanConfig"],
"models.videomae": ["VIDEOMAE_PRETRAINED_CONFIG_ARCHIVE_MAP", "VideoMAEConfig"],
"models.vilt": [
"VILT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ViltConfig",
"ViltFeatureExtractor",
"ViltImageProcessor",
"ViltProcessor",
],
"models.vision_encoder_decoder": ["VisionEncoderDecoderConfig"],
"models.vision_text_dual_encoder": ["VisionTextDualEncoderConfig", "VisionTextDualEncoderProcessor"],
"models.visual_bert": ["VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "VisualBertConfig"],
"models.vit": ["VIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTConfig"],
"models.vit_hybrid": ["VIT_HYBRID_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTHybridConfig"],
"models.vit_mae": ["VIT_MAE_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTMAEConfig"],
"models.vit_msn": ["VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTMSNConfig"],
"models.wav2vec2": [
"WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Wav2Vec2Config",
"Wav2Vec2CTCTokenizer",
"Wav2Vec2FeatureExtractor",
"Wav2Vec2Processor",
"Wav2Vec2Tokenizer",
],
"models.wav2vec2_conformer": [
"WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Wav2Vec2ConformerConfig",
],
"models.wav2vec2_phoneme": ["Wav2Vec2PhonemeCTCTokenizer"],
"models.wav2vec2_with_lm": ["Wav2Vec2ProcessorWithLM"],
"models.wavlm": [
"WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP",
"WavLMConfig",
],
"models.whisper": [
"WHISPER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"WhisperConfig",
"WhisperFeatureExtractor",
"WhisperProcessor",
"WhisperTokenizer",
],
"models.x_clip": [
"XCLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
"XCLIPConfig",
"XCLIPProcessor",
"XCLIPTextConfig",
"XCLIPVisionConfig",
],
"models.xglm": ["XGLM_PRETRAINED_CONFIG_ARCHIVE_MAP", "XGLMConfig"],
"models.xlm": ["XLM_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMConfig", "XLMTokenizer"],
"models.xlm_prophetnet": ["XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMProphetNetConfig"],
"models.xlm_roberta": ["XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMRobertaConfig"],
"models.xlm_roberta_xl": ["XLM_ROBERTA_XL_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMRobertaXLConfig"],
"models.xlnet": ["XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLNetConfig"],
"models.yolos": ["YOLOS_PRETRAINED_CONFIG_ARCHIVE_MAP", "YolosConfig"],
"models.yoso": ["YOSO_PRETRAINED_CONFIG_ARCHIVE_MAP", "YosoConfig"],
"onnx": [],
"pipelines": [
"AudioClassificationPipeline",
"AutomaticSpeechRecognitionPipeline",
"Conversation",
"ConversationalPipeline",
"CsvPipelineDataFormat",
"DepthEstimationPipeline",
"DocumentQuestionAnsweringPipeline",
"FeatureExtractionPipeline",
"FillMaskPipeline",
"ImageClassificationPipeline",
"ImageSegmentationPipeline",
"ImageToTextPipeline",
"JsonPipelineDataFormat",
"NerPipeline",
"ObjectDetectionPipeline",
"PipedPipelineDataFormat",
"Pipeline",
"PipelineDataFormat",
"QuestionAnsweringPipeline",
"SummarizationPipeline",
"TableQuestionAnsweringPipeline",
"Text2TextGenerationPipeline",
"TextClassificationPipeline",
"TextGenerationPipeline",
"TokenClassificationPipeline",
"TranslationPipeline",
"VideoClassificationPipeline",
"VisualQuestionAnsweringPipeline",
"ZeroShotClassificationPipeline",
"ZeroShotImageClassificationPipeline",
"ZeroShotObjectDetectionPipeline",
"pipeline",
],
"processing_utils": ["ProcessorMixin"],
"testing_utils": [],
"tokenization_utils": ["PreTrainedTokenizer"],
"tokenization_utils_base": [
"AddedToken",
"BatchEncoding",
"CharSpan",
"PreTrainedTokenizerBase",
"SpecialTokensMixin",
"TokenSpan",
],
"trainer_callback": [
"DefaultFlowCallback",
"EarlyStoppingCallback",
"PrinterCallback",
"ProgressCallback",
"TrainerCallback",
"TrainerControl",
"TrainerState",
],
"trainer_utils": ["EvalPrediction", "IntervalStrategy", "SchedulerType", "enable_full_determinism", "set_seed"],
"training_args": ["TrainingArguments"],
"training_args_seq2seq": ["Seq2SeqTrainingArguments"],
"training_args_tf": ["TFTrainingArguments"],
"utils": [
"CONFIG_NAME",
"MODEL_CARD_NAME",
"PYTORCH_PRETRAINED_BERT_CACHE",
"PYTORCH_TRANSFORMERS_CACHE",
"SPIECE_UNDERLINE",
"TF2_WEIGHTS_NAME",
"TF_WEIGHTS_NAME",
"TRANSFORMERS_CACHE",
"WEIGHTS_NAME",
"TensorType",
"add_end_docstrings",
"add_start_docstrings",
"is_apex_available",
"is_datasets_available",
"is_decord_available",
"is_faiss_available",
"is_flax_available",
"is_keras_nlp_available",
"is_phonemizer_available",
"is_psutil_available",
"is_py3nvml_available",
"is_pyctcdecode_available",
"is_safetensors_available",
"is_scipy_available",
"is_sentencepiece_available",
"is_sklearn_available",
"is_speech_available",
"is_tensorflow_text_available",
"is_tf_available",
"is_timm_available",
"is_tokenizers_available",
"is_torch_available",
"is_torch_tpu_available",
"is_vision_available",
"logging",
],
"utils.bitsandbytes": [],
}
# sentencepiece-backed objects
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_sentencepiece_objects
_import_structure["utils.dummy_sentencepiece_objects"] = [
name for name in dir(dummy_sentencepiece_objects) if not name.startswith("_")
]
else:
_import_structure["models.albert"].append("AlbertTokenizer")
_import_structure["models.barthez"].append("BarthezTokenizer")
_import_structure["models.bartpho"].append("BartphoTokenizer")
_import_structure["models.bert_generation"].append("BertGenerationTokenizer")
_import_structure["models.big_bird"].append("BigBirdTokenizer")
_import_structure["models.camembert"].append("CamembertTokenizer")
_import_structure["models.cpm"].append("CpmTokenizer")
_import_structure["models.deberta_v2"].append("DebertaV2Tokenizer")
_import_structure["models.fnet"].append("FNetTokenizer")
_import_structure["models.layoutxlm"].append("LayoutXLMTokenizer")
_import_structure["models.m2m_100"].append("M2M100Tokenizer")
_import_structure["models.marian"].append("MarianTokenizer")
_import_structure["models.mbart"].append("MBartTokenizer")
_import_structure["models.mbart50"].append("MBart50Tokenizer")
_import_structure["models.mluke"].append("MLukeTokenizer")
_import_structure["models.mt5"].append("MT5Tokenizer")
_import_structure["models.nllb"].append("NllbTokenizer")
_import_structure["models.pegasus"].append("PegasusTokenizer")
_import_structure["models.plbart"].append("PLBartTokenizer")
_import_structure["models.reformer"].append("ReformerTokenizer")
_import_structure["models.rembert"].append("RemBertTokenizer")
_import_structure["models.speech_to_text"].append("Speech2TextTokenizer")
_import_structure["models.t5"].append("T5Tokenizer")
_import_structure["models.xglm"].append("XGLMTokenizer")
_import_structure["models.xlm_prophetnet"].append("XLMProphetNetTokenizer")
_import_structure["models.xlm_roberta"].append("XLMRobertaTokenizer")
_import_structure["models.xlnet"].append("XLNetTokenizer")
# tokenizers-backed objects
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_tokenizers_objects
_import_structure["utils.dummy_tokenizers_objects"] = [
name for name in dir(dummy_tokenizers_objects) if not name.startswith("_")
]
else:
# Fast tokenizers structure
_import_structure["models.albert"].append("AlbertTokenizerFast")
_import_structure["models.bart"].append("BartTokenizerFast")
_import_structure["models.barthez"].append("BarthezTokenizerFast")
_import_structure["models.bert"].append("BertTokenizerFast")
_import_structure["models.big_bird"].append("BigBirdTokenizerFast")
_import_structure["models.blenderbot"].append("BlenderbotTokenizerFast")
_import_structure["models.blenderbot_small"].append("BlenderbotSmallTokenizerFast")
_import_structure["models.bloom"].append("BloomTokenizerFast")
_import_structure["models.camembert"].append("CamembertTokenizerFast")
_import_structure["models.clip"].append("CLIPTokenizerFast")
_import_structure["models.codegen"].append("CodeGenTokenizerFast")
_import_structure["models.convbert"].append("ConvBertTokenizerFast")
_import_structure["models.cpm"].append("CpmTokenizerFast")
_import_structure["models.deberta"].append("DebertaTokenizerFast")
_import_structure["models.deberta_v2"].append("DebertaV2TokenizerFast")
_import_structure["models.distilbert"].append("DistilBertTokenizerFast")
_import_structure["models.dpr"].extend(
["DPRContextEncoderTokenizerFast", "DPRQuestionEncoderTokenizerFast", "DPRReaderTokenizerFast"]
)
_import_structure["models.electra"].append("ElectraTokenizerFast")
_import_structure["models.fnet"].append("FNetTokenizerFast")
_import_structure["models.funnel"].append("FunnelTokenizerFast")
_import_structure["models.gpt2"].append("GPT2TokenizerFast")
_import_structure["models.gpt_neox"].append("GPTNeoXTokenizerFast")
_import_structure["models.gpt_neox_japanese"].append("GPTNeoXJapaneseTokenizer")
_import_structure["models.herbert"].append("HerbertTokenizerFast")
_import_structure["models.layoutlm"].append("LayoutLMTokenizerFast")
_import_structure["models.layoutlmv2"].append("LayoutLMv2TokenizerFast")
_import_structure["models.layoutlmv3"].append("LayoutLMv3TokenizerFast")
_import_structure["models.layoutxlm"].append("LayoutXLMTokenizerFast")
_import_structure["models.led"].append("LEDTokenizerFast")
_import_structure["models.longformer"].append("LongformerTokenizerFast")
_import_structure["models.lxmert"].append("LxmertTokenizerFast")
_import_structure["models.markuplm"].append("MarkupLMTokenizerFast")
_import_structure["models.mbart"].append("MBartTokenizerFast")
_import_structure["models.mbart50"].append("MBart50TokenizerFast")
_import_structure["models.mobilebert"].append("MobileBertTokenizerFast")
_import_structure["models.mpnet"].append("MPNetTokenizerFast")
_import_structure["models.mt5"].append("MT5TokenizerFast")
_import_structure["models.mvp"].append("MvpTokenizerFast")
_import_structure["models.nllb"].append("NllbTokenizerFast")
_import_structure["models.openai"].append("OpenAIGPTTokenizerFast")
_import_structure["models.pegasus"].append("PegasusTokenizerFast")
_import_structure["models.realm"].append("RealmTokenizerFast")
_import_structure["models.reformer"].append("ReformerTokenizerFast")
_import_structure["models.rembert"].append("RemBertTokenizerFast")
_import_structure["models.retribert"].append("RetriBertTokenizerFast")
_import_structure["models.roberta"].append("RobertaTokenizerFast")
_import_structure["models.roformer"].append("RoFormerTokenizerFast")
_import_structure["models.splinter"].append("SplinterTokenizerFast")
_import_structure["models.squeezebert"].append("SqueezeBertTokenizerFast")
_import_structure["models.t5"].append("T5TokenizerFast")
_import_structure["models.xglm"].append("XGLMTokenizerFast")
_import_structure["models.xlm_roberta"].append("XLMRobertaTokenizerFast")
_import_structure["models.xlnet"].append("XLNetTokenizerFast")
_import_structure["tokenization_utils_fast"] = ["PreTrainedTokenizerFast"]
try:
if not (is_sentencepiece_available() and is_tokenizers_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_sentencepiece_and_tokenizers_objects
_import_structure["utils.dummy_sentencepiece_and_tokenizers_objects"] = [
name for name in dir(dummy_sentencepiece_and_tokenizers_objects) if not name.startswith("_")
]
else:
_import_structure["convert_slow_tokenizer"] = ["SLOW_TO_FAST_CONVERTERS", "convert_slow_tokenizer"]
# Speech-specific objects
try:
if not is_speech_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_speech_objects
_import_structure["utils.dummy_speech_objects"] = [
name for name in dir(dummy_speech_objects) if not name.startswith("_")
]
else:
_import_structure["models.audio_spectrogram_transformer"].append("ASTFeatureExtractor")
_import_structure["models.mctct"].append("MCTCTFeatureExtractor")
_import_structure["models.speech_to_text"].append("Speech2TextFeatureExtractor")
# Tensorflow-text-specific objects
try:
if not is_tensorflow_text_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_tensorflow_text_objects
_import_structure["utils.dummy_tensorflow_text_objects"] = [
name for name in dir(dummy_tensorflow_text_objects) if not name.startswith("_")
]
else:
_import_structure["models.bert"].append("TFBertTokenizer")
# keras-nlp-specific objects
try:
if not is_keras_nlp_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_keras_nlp_objects
_import_structure["utils.dummy_keras_nlp_objects"] = [
name for name in dir(dummy_keras_nlp_objects) if not name.startswith("_")
]
else:
_import_structure["models.gpt2"].append("TFGPT2Tokenizer")
try:
if not (is_sentencepiece_available() and is_speech_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_sentencepiece_and_speech_objects
_import_structure["utils.dummy_sentencepiece_and_speech_objects"] = [
name for name in dir(dummy_sentencepiece_and_speech_objects) if not name.startswith("_")
]
else:
_import_structure["models.speech_to_text"].append("Speech2TextProcessor")
# Vision-specific objects
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_vision_objects
_import_structure["utils.dummy_vision_objects"] = [
name for name in dir(dummy_vision_objects) if not name.startswith("_")
]
else:
_import_structure["image_processing_utils"] = ["ImageProcessingMixin"]
_import_structure["image_transforms"] = ["rescale", "resize", "to_pil_image"]
_import_structure["image_utils"] = ["ImageFeatureExtractionMixin"]
_import_structure["models.beit"].extend(["BeitFeatureExtractor", "BeitImageProcessor"])
_import_structure["models.bit"].extend(["BitImageProcessor"])
_import_structure["models.chinese_clip"].extend(["ChineseCLIPFeatureExtractor", "ChineseCLIPImageProcessor"])
_import_structure["models.clip"].extend(["CLIPFeatureExtractor", "CLIPImageProcessor"])
_import_structure["models.conditional_detr"].extend(
["ConditionalDetrFeatureExtractor", "ConditionalDetrImageProcessor"]
)
_import_structure["models.convnext"].extend(["ConvNextFeatureExtractor", "ConvNextImageProcessor"])
_import_structure["models.deformable_detr"].extend(
["DeformableDetrFeatureExtractor", "DeformableDetrImageProcessor"]
)
_import_structure["models.deit"].extend(["DeiTFeatureExtractor", "DeiTImageProcessor"])
_import_structure["models.detr"].extend(["DetrFeatureExtractor", "DetrImageProcessor"])
_import_structure["models.donut"].extend(["DonutFeatureExtractor", "DonutImageProcessor"])
_import_structure["models.dpt"].extend(["DPTFeatureExtractor", "DPTImageProcessor"])
_import_structure["models.flava"].extend(["FlavaFeatureExtractor", "FlavaImageProcessor", "FlavaProcessor"])
_import_structure["models.glpn"].extend(["GLPNFeatureExtractor", "GLPNImageProcessor"])
_import_structure["models.imagegpt"].extend(["ImageGPTFeatureExtractor", "ImageGPTImageProcessor"])
_import_structure["models.layoutlmv2"].extend(["LayoutLMv2FeatureExtractor", "LayoutLMv2ImageProcessor"])
_import_structure["models.layoutlmv3"].extend(["LayoutLMv3FeatureExtractor", "LayoutLMv3ImageProcessor"])
_import_structure["models.levit"].extend(["LevitFeatureExtractor", "LevitImageProcessor"])
_import_structure["models.maskformer"].extend(["MaskFormerFeatureExtractor", "MaskFormerImageProcessor"])
_import_structure["models.mobilenet_v1"].extend(["MobileNetV1FeatureExtractor", "MobileNetV1ImageProcessor"])
_import_structure["models.mobilenet_v2"].extend(["MobileNetV2FeatureExtractor", "MobileNetV2ImageProcessor"])
_import_structure["models.mobilevit"].extend(["MobileViTFeatureExtractor", "MobileViTImageProcessor"])
_import_structure["models.owlvit"].extend(["OwlViTFeatureExtractor", "OwlViTImageProcessor"])
_import_structure["models.perceiver"].extend(["PerceiverFeatureExtractor", "PerceiverImageProcessor"])
_import_structure["models.poolformer"].extend(["PoolFormerFeatureExtractor", "PoolFormerImageProcessor"])
_import_structure["models.segformer"].extend(["SegformerFeatureExtractor", "SegformerImageProcessor"])
_import_structure["models.videomae"].extend(["VideoMAEFeatureExtractor", "VideoMAEImageProcessor"])
_import_structure["models.vilt"].extend(["ViltFeatureExtractor", "ViltImageProcessor", "ViltProcessor"])
_import_structure["models.vit"].extend(["ViTFeatureExtractor", "ViTImageProcessor"])
_import_structure["models.vit_hybrid"].extend(["ViTHybridImageProcessor"])
_import_structure["models.yolos"].extend(["YolosFeatureExtractor", "YolosImageProcessor"])
# Timm-backed objects
try:
if not (is_timm_available() and is_vision_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_timm_and_vision_objects
_import_structure["utils.dummy_timm_and_vision_objects"] = [
name for name in dir(dummy_timm_and_vision_objects) if not name.startswith("_")
]
else:
_import_structure["models.deformable_detr"].extend(
[
"DEFORMABLE_DETR_PRETRAINED_MODEL_ARCHIVE_LIST",
"DeformableDetrForObjectDetection",
"DeformableDetrModel",
"DeformableDetrPreTrainedModel",
]
)
_import_structure["models.detr"].extend(
[
"DETR_PRETRAINED_MODEL_ARCHIVE_LIST",
"DetrForObjectDetection",
"DetrForSegmentation",
"DetrModel",
"DetrPreTrainedModel",
]
)
_import_structure["models.table_transformer"].extend(
[
"TABLE_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TableTransformerForObjectDetection",
"TableTransformerModel",
"TableTransformerPreTrainedModel",
]
)
_import_structure["models.conditional_detr"].extend(
[
"CONDITIONAL_DETR_PRETRAINED_MODEL_ARCHIVE_LIST",
"ConditionalDetrForObjectDetection",
"ConditionalDetrForSegmentation",
"ConditionalDetrModel",
"ConditionalDetrPreTrainedModel",
]
)
# PyTorch-backed objects
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_pt_objects
_import_structure["utils.dummy_pt_objects"] = [name for name in dir(dummy_pt_objects) if not name.startswith("_")]
else:
_import_structure["activations"] = []
_import_structure["benchmark.benchmark"] = ["PyTorchBenchmark"]
_import_structure["benchmark.benchmark_args"] = ["PyTorchBenchmarkArguments"]
_import_structure["data.datasets"] = [
"GlueDataset",
"GlueDataTrainingArguments",
"LineByLineTextDataset",
"LineByLineWithRefDataset",
"LineByLineWithSOPTextDataset",
"SquadDataset",
"SquadDataTrainingArguments",
"TextDataset",
"TextDatasetForNextSentencePrediction",
]
_import_structure["deepspeed"] = []
_import_structure["generation"].extend(
[
"BeamScorer",
"BeamSearchScorer",
"ConstrainedBeamSearchScorer",
"Constraint",
"ConstraintListState",
"DisjunctiveConstraint",
"ForcedBOSTokenLogitsProcessor",
"ForcedEOSTokenLogitsProcessor",
"GenerationMixin",
"HammingDiversityLogitsProcessor",
"InfNanRemoveLogitsProcessor",
"LogitsProcessor",
"LogitsProcessorList",
"LogitsWarper",
"MaxLengthCriteria",
"MaxTimeCriteria",
"MinLengthLogitsProcessor",
"NoBadWordsLogitsProcessor",
"NoRepeatNGramLogitsProcessor",
"PhrasalConstraint",
"PrefixConstrainedLogitsProcessor",
"RepetitionPenaltyLogitsProcessor",
"StoppingCriteria",
"StoppingCriteriaList",
"TemperatureLogitsWarper",
"TopKLogitsWarper",
"TopPLogitsWarper",
"TypicalLogitsWarper",
"top_k_top_p_filtering",
]
)
_import_structure["generation_utils"] = []
_import_structure["modeling_outputs"] = []
_import_structure["modeling_utils"] = ["PreTrainedModel"]
# PyTorch models structure
_import_structure["models.albert"].extend(
[
"ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"AlbertForMaskedLM",
"AlbertForMultipleChoice",
"AlbertForPreTraining",
"AlbertForQuestionAnswering",
"AlbertForSequenceClassification",
"AlbertForTokenClassification",
"AlbertModel",
"AlbertPreTrainedModel",
"load_tf_weights_in_albert",
]
)
_import_structure["models.audio_spectrogram_transformer"].extend(
[
"AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"ASTForAudioClassification",
"ASTModel",
"ASTPreTrainedModel",
]
)
_import_structure["models.auto"].extend(
[
"MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING",
"MODEL_FOR_AUDIO_XVECTOR_MAPPING",
"MODEL_FOR_BACKBONE_MAPPING",
"MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING",
"MODEL_FOR_CAUSAL_LM_MAPPING",
"MODEL_FOR_CTC_MAPPING",
"MODEL_FOR_DEPTH_ESTIMATION_MAPPING",
"MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING",
"MODEL_FOR_IMAGE_SEGMENTATION_MAPPING",
"MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING",
"MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING",
"MODEL_FOR_MASKED_LM_MAPPING",
"MODEL_FOR_MULTIPLE_CHOICE_MAPPING",
"MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING",
"MODEL_FOR_OBJECT_DETECTION_MAPPING",
"MODEL_FOR_PRETRAINING_MAPPING",
"MODEL_FOR_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING",
"MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING",
"MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING",
"MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING",
"MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING",
"MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING",
"MODEL_FOR_VISION_2_SEQ_MAPPING",
"MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING",
"MODEL_MAPPING",
"MODEL_WITH_LM_HEAD_MAPPING",
"AutoBackbone",
"AutoModel",
"AutoModelForAudioClassification",
"AutoModelForAudioFrameClassification",
"AutoModelForAudioXVector",
"AutoModelForCausalLM",
"AutoModelForCTC",
"AutoModelForDepthEstimation",
"AutoModelForDocumentQuestionAnswering",
"AutoModelForImageClassification",
"AutoModelForImageSegmentation",
"AutoModelForInstanceSegmentation",
"AutoModelForMaskedImageModeling",
"AutoModelForMaskedLM",
"AutoModelForMultipleChoice",
"AutoModelForNextSentencePrediction",
"AutoModelForObjectDetection",
"AutoModelForPreTraining",
"AutoModelForQuestionAnswering",
"AutoModelForSemanticSegmentation",
"AutoModelForSeq2SeqLM",
"AutoModelForSequenceClassification",
"AutoModelForSpeechSeq2Seq",
"AutoModelForTableQuestionAnswering",
"AutoModelForTokenClassification",
"AutoModelForVideoClassification",
"AutoModelForVision2Seq",
"AutoModelForVisualQuestionAnswering",
"AutoModelForZeroShotObjectDetection",
"AutoModelWithLMHead",
]
)
_import_structure["models.bart"].extend(
[
"BART_PRETRAINED_MODEL_ARCHIVE_LIST",
"BartForCausalLM",
"BartForConditionalGeneration",
"BartForQuestionAnswering",
"BartForSequenceClassification",
"BartModel",
"BartPretrainedModel",
"PretrainedBartModel",
]
)
_import_structure["models.beit"].extend(
[
"BEIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BeitForImageClassification",
"BeitForMaskedImageModeling",
"BeitForSemanticSegmentation",
"BeitModel",
"BeitPreTrainedModel",
]
)
_import_structure["models.bert"].extend(
[
"BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BertForMaskedLM",
"BertForMultipleChoice",
"BertForNextSentencePrediction",
"BertForPreTraining",
"BertForQuestionAnswering",
"BertForSequenceClassification",
"BertForTokenClassification",
"BertLayer",
"BertLMHeadModel",
"BertModel",
"BertPreTrainedModel",
"load_tf_weights_in_bert",
]
)
_import_structure["models.bert_generation"].extend(
[
"BertGenerationDecoder",
"BertGenerationEncoder",
"BertGenerationPreTrainedModel",
"load_tf_weights_in_bert_generation",
]
)
_import_structure["models.big_bird"].extend(
[
"BIG_BIRD_PRETRAINED_MODEL_ARCHIVE_LIST",
"BigBirdForCausalLM",
"BigBirdForMaskedLM",
"BigBirdForMultipleChoice",
"BigBirdForPreTraining",
"BigBirdForQuestionAnswering",
"BigBirdForSequenceClassification",
"BigBirdForTokenClassification",
"BigBirdLayer",
"BigBirdModel",
"BigBirdPreTrainedModel",
"load_tf_weights_in_big_bird",
]
)
_import_structure["models.bigbird_pegasus"].extend(
[
"BIGBIRD_PEGASUS_PRETRAINED_MODEL_ARCHIVE_LIST",
"BigBirdPegasusForCausalLM",
"BigBirdPegasusForConditionalGeneration",
"BigBirdPegasusForQuestionAnswering",
"BigBirdPegasusForSequenceClassification",
"BigBirdPegasusModel",
"BigBirdPegasusPreTrainedModel",
]
)
_import_structure["models.biogpt"].extend(
[
"BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BioGptForCausalLM",
"BioGptModel",
"BioGptPreTrainedModel",
]
)
_import_structure["models.bit"].extend(
[
"BIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BitBackbone",
"BitForImageClassification",
"BitModel",
"BitPreTrainedModel",
]
)
_import_structure["models.blenderbot"].extend(
[
"BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BlenderbotForCausalLM",
"BlenderbotForConditionalGeneration",
"BlenderbotModel",
"BlenderbotPreTrainedModel",
]
)
_import_structure["models.blenderbot_small"].extend(
[
"BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST",
"BlenderbotSmallForCausalLM",
"BlenderbotSmallForConditionalGeneration",
"BlenderbotSmallModel",
"BlenderbotSmallPreTrainedModel",
]
)
_import_structure["models.bloom"].extend(
[
"BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST",
"BloomForCausalLM",
"BloomForQuestionAnswering",
"BloomForSequenceClassification",
"BloomForTokenClassification",
"BloomModel",
"BloomPreTrainedModel",
]
)
_import_structure["models.camembert"].extend(
[
"CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"CamembertForCausalLM",
"CamembertForMaskedLM",
"CamembertForMultipleChoice",
"CamembertForQuestionAnswering",
"CamembertForSequenceClassification",
"CamembertForTokenClassification",
"CamembertModel",
"CamembertPreTrainedModel",
]
)
_import_structure["models.canine"].extend(
[
"CANINE_PRETRAINED_MODEL_ARCHIVE_LIST",
"CanineForMultipleChoice",
"CanineForQuestionAnswering",
"CanineForSequenceClassification",
"CanineForTokenClassification",
"CanineLayer",
"CanineModel",
"CaninePreTrainedModel",
"load_tf_weights_in_canine",
]
)
_import_structure["models.chinese_clip"].extend(
[
"CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"ChineseCLIPModel",
"ChineseCLIPPreTrainedModel",
"ChineseCLIPTextModel",
"ChineseCLIPVisionModel",
]
)
_import_structure["models.clip"].extend(
[
"CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"CLIPModel",
"CLIPPreTrainedModel",
"CLIPTextModel",
"CLIPTextModelWithProjection",
"CLIPVisionModel",
"CLIPVisionModelWithProjection",
]
)
_import_structure["models.clipseg"].extend(
[
"CLIPSEG_PRETRAINED_MODEL_ARCHIVE_LIST",
"CLIPSegForImageSegmentation",
"CLIPSegModel",
"CLIPSegPreTrainedModel",
"CLIPSegTextModel",
"CLIPSegVisionModel",
]
)
_import_structure["models.codegen"].extend(
[
"CODEGEN_PRETRAINED_MODEL_ARCHIVE_LIST",
"CodeGenForCausalLM",
"CodeGenModel",
"CodeGenPreTrainedModel",
]
)
_import_structure["models.convbert"].extend(
[
"CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ConvBertForMaskedLM",
"ConvBertForMultipleChoice",
"ConvBertForQuestionAnswering",
"ConvBertForSequenceClassification",
"ConvBertForTokenClassification",
"ConvBertLayer",
"ConvBertModel",
"ConvBertPreTrainedModel",
"load_tf_weights_in_convbert",
]
)
_import_structure["models.convnext"].extend(
[
"CONVNEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ConvNextForImageClassification",
"ConvNextModel",
"ConvNextPreTrainedModel",
]
)
_import_structure["models.ctrl"].extend(
[
"CTRL_PRETRAINED_MODEL_ARCHIVE_LIST",
"CTRLForSequenceClassification",
"CTRLLMHeadModel",
"CTRLModel",
"CTRLPreTrainedModel",
]
)
_import_structure["models.cvt"].extend(
[
"CVT_PRETRAINED_MODEL_ARCHIVE_LIST",
"CvtForImageClassification",
"CvtModel",
"CvtPreTrainedModel",
]
)
_import_structure["models.data2vec"].extend(
[
"DATA2VEC_AUDIO_PRETRAINED_MODEL_ARCHIVE_LIST",
"DATA2VEC_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DATA2VEC_VISION_PRETRAINED_MODEL_ARCHIVE_LIST",
"Data2VecAudioForAudioFrameClassification",
"Data2VecAudioForCTC",
"Data2VecAudioForSequenceClassification",
"Data2VecAudioForXVector",
"Data2VecAudioModel",
"Data2VecAudioPreTrainedModel",
"Data2VecTextForCausalLM",
"Data2VecTextForMaskedLM",
"Data2VecTextForMultipleChoice",
"Data2VecTextForQuestionAnswering",
"Data2VecTextForSequenceClassification",
"Data2VecTextForTokenClassification",
"Data2VecTextModel",
"Data2VecTextPreTrainedModel",
"Data2VecVisionForImageClassification",
"Data2VecVisionForSemanticSegmentation",
"Data2VecVisionModel",
"Data2VecVisionPreTrainedModel",
]
)
_import_structure["models.deberta"].extend(
[
"DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"DebertaForMaskedLM",
"DebertaForQuestionAnswering",
"DebertaForSequenceClassification",
"DebertaForTokenClassification",
"DebertaModel",
"DebertaPreTrainedModel",
]
)
_import_structure["models.deberta_v2"].extend(
[
"DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST",
"DebertaV2ForMaskedLM",
"DebertaV2ForMultipleChoice",
"DebertaV2ForQuestionAnswering",
"DebertaV2ForSequenceClassification",
"DebertaV2ForTokenClassification",
"DebertaV2Model",
"DebertaV2PreTrainedModel",
]
)
_import_structure["models.decision_transformer"].extend(
[
"DECISION_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DecisionTransformerGPT2Model",
"DecisionTransformerGPT2PreTrainedModel",
"DecisionTransformerModel",
"DecisionTransformerPreTrainedModel",
]
)
_import_structure["models.deit"].extend(
[
"DEIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DeiTForImageClassification",
"DeiTForImageClassificationWithTeacher",
"DeiTForMaskedImageModeling",
"DeiTModel",
"DeiTPreTrainedModel",
]
)
_import_structure["models.dinat"].extend(
[
"DINAT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DinatForImageClassification",
"DinatModel",
"DinatPreTrainedModel",
]
)
_import_structure["models.distilbert"].extend(
[
"DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DistilBertForMaskedLM",
"DistilBertForMultipleChoice",
"DistilBertForQuestionAnswering",
"DistilBertForSequenceClassification",
"DistilBertForTokenClassification",
"DistilBertModel",
"DistilBertPreTrainedModel",
]
)
_import_structure["models.donut"].extend(
[
"DONUT_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST",
"DonutSwinModel",
"DonutSwinPreTrainedModel",
]
)
_import_structure["models.dpr"].extend(
[
"DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPRContextEncoder",
"DPRPretrainedContextEncoder",
"DPRPreTrainedModel",
"DPRPretrainedQuestionEncoder",
"DPRPretrainedReader",
"DPRQuestionEncoder",
"DPRReader",
]
)
_import_structure["models.dpt"].extend(
[
"DPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPTForDepthEstimation",
"DPTForSemanticSegmentation",
"DPTModel",
"DPTPreTrainedModel",
]
)
_import_structure["models.electra"].extend(
[
"ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"ElectraForCausalLM",
"ElectraForMaskedLM",
"ElectraForMultipleChoice",
"ElectraForPreTraining",
"ElectraForQuestionAnswering",
"ElectraForSequenceClassification",
"ElectraForTokenClassification",
"ElectraModel",
"ElectraPreTrainedModel",
"load_tf_weights_in_electra",
]
)
_import_structure["models.encoder_decoder"].append("EncoderDecoderModel")
_import_structure["models.ernie"].extend(
[
"ERNIE_PRETRAINED_MODEL_ARCHIVE_LIST",
"ErnieForCausalLM",
"ErnieForMaskedLM",
"ErnieForMultipleChoice",
"ErnieForNextSentencePrediction",
"ErnieForPreTraining",
"ErnieForQuestionAnswering",
"ErnieForSequenceClassification",
"ErnieForTokenClassification",
"ErnieModel",
"ErniePreTrainedModel",
]
)
_import_structure["models.esm"].extend(
[
"ESM_PRETRAINED_MODEL_ARCHIVE_LIST",
"EsmFoldPreTrainedModel",
"EsmForMaskedLM",
"EsmForProteinFolding",
"EsmForSequenceClassification",
"EsmForTokenClassification",
"EsmModel",
"EsmPreTrainedModel",
]
)
_import_structure["models.flaubert"].extend(
[
"FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"FlaubertForMultipleChoice",
"FlaubertForQuestionAnswering",
"FlaubertForQuestionAnsweringSimple",
"FlaubertForSequenceClassification",
"FlaubertForTokenClassification",
"FlaubertModel",
"FlaubertPreTrainedModel",
"FlaubertWithLMHeadModel",
]
)
_import_structure["models.flava"].extend(
[
"FLAVA_PRETRAINED_MODEL_ARCHIVE_LIST",
"FlavaForPreTraining",
"FlavaImageCodebook",
"FlavaImageModel",
"FlavaModel",
"FlavaMultimodalModel",
"FlavaPreTrainedModel",
"FlavaTextModel",
]
)
_import_structure["models.fnet"].extend(
[
"FNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"FNetForMaskedLM",
"FNetForMultipleChoice",
"FNetForNextSentencePrediction",
"FNetForPreTraining",
"FNetForQuestionAnswering",
"FNetForSequenceClassification",
"FNetForTokenClassification",
"FNetLayer",
"FNetModel",
"FNetPreTrainedModel",
]
)
_import_structure["models.fsmt"].extend(["FSMTForConditionalGeneration", "FSMTModel", "PretrainedFSMTModel"])
_import_structure["models.funnel"].extend(
[
"FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST",
"FunnelBaseModel",
"FunnelForMaskedLM",
"FunnelForMultipleChoice",
"FunnelForPreTraining",
"FunnelForQuestionAnswering",
"FunnelForSequenceClassification",
"FunnelForTokenClassification",
"FunnelModel",
"FunnelPreTrainedModel",
"load_tf_weights_in_funnel",
]
)
_import_structure["models.glpn"].extend(
[
"GLPN_PRETRAINED_MODEL_ARCHIVE_LIST",
"GLPNForDepthEstimation",
"GLPNModel",
"GLPNPreTrainedModel",
]
)
_import_structure["models.gpt2"].extend(
[
"GPT2_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPT2DoubleHeadsModel",
"GPT2ForSequenceClassification",
"GPT2ForTokenClassification",
"GPT2LMHeadModel",
"GPT2Model",
"GPT2PreTrainedModel",
"load_tf_weights_in_gpt2",
]
)
_import_structure["models.gpt_neo"].extend(
[
"GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoForCausalLM",
"GPTNeoForSequenceClassification",
"GPTNeoModel",
"GPTNeoPreTrainedModel",
"load_tf_weights_in_gpt_neo",
]
)
_import_structure["models.gpt_neox"].extend(
[
"GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoXForCausalLM",
"GPTNeoXLayer",
"GPTNeoXModel",
"GPTNeoXPreTrainedModel",
]
)
_import_structure["models.gpt_neox_japanese"].extend(
[
"GPT_NEOX_JAPANESE_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoXJapaneseForCausalLM",
"GPTNeoXJapaneseLayer",
"GPTNeoXJapaneseModel",
"GPTNeoXJapanesePreTrainedModel",
]
)
_import_structure["models.gptj"].extend(
[
"GPTJ_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTJForCausalLM",
"GPTJForQuestionAnswering",
"GPTJForSequenceClassification",
"GPTJModel",
"GPTJPreTrainedModel",
]
)
_import_structure["models.groupvit"].extend(
[
"GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"GroupViTModel",
"GroupViTPreTrainedModel",
"GroupViTTextModel",
"GroupViTVisionModel",
]
)
_import_structure["models.hubert"].extend(
[
"HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"HubertForCTC",
"HubertForSequenceClassification",
"HubertModel",
"HubertPreTrainedModel",
]
)
_import_structure["models.ibert"].extend(
[
"IBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"IBertForMaskedLM",
"IBertForMultipleChoice",
"IBertForQuestionAnswering",
"IBertForSequenceClassification",
"IBertForTokenClassification",
"IBertModel",
"IBertPreTrainedModel",
]
)
_import_structure["models.imagegpt"].extend(
[
"IMAGEGPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ImageGPTForCausalImageModeling",
"ImageGPTForImageClassification",
"ImageGPTModel",
"ImageGPTPreTrainedModel",
"load_tf_weights_in_imagegpt",
]
)
_import_structure["models.jukebox"].extend(
[
"JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST",
"JukeboxModel",
"JukeboxPreTrainedModel",
"JukeboxPrior",
"JukeboxVQVAE",
]
)
_import_structure["models.layoutlm"].extend(
[
"LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"LayoutLMForMaskedLM",
"LayoutLMForQuestionAnswering",
"LayoutLMForSequenceClassification",
"LayoutLMForTokenClassification",
"LayoutLMModel",
"LayoutLMPreTrainedModel",
]
)
_import_structure["models.layoutlmv2"].extend(
[
"LAYOUTLMV2_PRETRAINED_MODEL_ARCHIVE_LIST",
"LayoutLMv2ForQuestionAnswering",
"LayoutLMv2ForSequenceClassification",
"LayoutLMv2ForTokenClassification",
"LayoutLMv2Model",
"LayoutLMv2PreTrainedModel",
]
)
_import_structure["models.layoutlmv3"].extend(
[
"LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST",
"LayoutLMv3ForQuestionAnswering",
"LayoutLMv3ForSequenceClassification",
"LayoutLMv3ForTokenClassification",
"LayoutLMv3Model",
"LayoutLMv3PreTrainedModel",
]
)
_import_structure["models.led"].extend(
[
"LED_PRETRAINED_MODEL_ARCHIVE_LIST",
"LEDForConditionalGeneration",
"LEDForQuestionAnswering",
"LEDForSequenceClassification",
"LEDModel",
"LEDPreTrainedModel",
]
)
_import_structure["models.levit"].extend(
[
"LEVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"LevitForImageClassification",
"LevitForImageClassificationWithTeacher",
"LevitModel",
"LevitPreTrainedModel",
]
)
_import_structure["models.lilt"].extend(
[
"LILT_PRETRAINED_MODEL_ARCHIVE_LIST",
"LiltForQuestionAnswering",
"LiltForSequenceClassification",
"LiltForTokenClassification",
"LiltModel",
"LiltPreTrainedModel",
]
)
_import_structure["models.longformer"].extend(
[
"LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"LongformerForMaskedLM",
"LongformerForMultipleChoice",
"LongformerForQuestionAnswering",
"LongformerForSequenceClassification",
"LongformerForTokenClassification",
"LongformerModel",
"LongformerPreTrainedModel",
"LongformerSelfAttention",
]
)
_import_structure["models.longt5"].extend(
[
"LONGT5_PRETRAINED_MODEL_ARCHIVE_LIST",
"LongT5EncoderModel",
"LongT5ForConditionalGeneration",
"LongT5Model",
"LongT5PreTrainedModel",
]
)
_import_structure["models.luke"].extend(
[
"LUKE_PRETRAINED_MODEL_ARCHIVE_LIST",
"LukeForEntityClassification",
"LukeForEntityPairClassification",
"LukeForEntitySpanClassification",
"LukeForMaskedLM",
"LukeForMultipleChoice",
"LukeForQuestionAnswering",
"LukeForSequenceClassification",
"LukeForTokenClassification",
"LukeModel",
"LukePreTrainedModel",
]
)
_import_structure["models.lxmert"].extend(
[
"LxmertEncoder",
"LxmertForPreTraining",
"LxmertForQuestionAnswering",
"LxmertModel",
"LxmertPreTrainedModel",
"LxmertVisualFeatureEncoder",
"LxmertXLayer",
]
)
_import_structure["models.m2m_100"].extend(
[
"M2M_100_PRETRAINED_MODEL_ARCHIVE_LIST",
"M2M100ForConditionalGeneration",
"M2M100Model",
"M2M100PreTrainedModel",
]
)
_import_structure["models.marian"].extend(["MarianForCausalLM", "MarianModel", "MarianMTModel"])
_import_structure["models.markuplm"].extend(
[
"MARKUPLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"MarkupLMForQuestionAnswering",
"MarkupLMForSequenceClassification",
"MarkupLMForTokenClassification",
"MarkupLMModel",
"MarkupLMPreTrainedModel",
]
)
_import_structure["models.maskformer"].extend(
[
"MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"MaskFormerForInstanceSegmentation",
"MaskFormerModel",
"MaskFormerPreTrainedModel",
"MaskFormerSwinBackbone",
]
)
_import_structure["models.mbart"].extend(
[
"MBartForCausalLM",
"MBartForConditionalGeneration",
"MBartForQuestionAnswering",
"MBartForSequenceClassification",
"MBartModel",
"MBartPreTrainedModel",
]
)
_import_structure["models.mctct"].extend(
[
"MCTCT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MCTCTForCTC",
"MCTCTModel",
"MCTCTPreTrainedModel",
]
)
_import_structure["models.megatron_bert"].extend(
[
"MEGATRON_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MegatronBertForCausalLM",
"MegatronBertForMaskedLM",
"MegatronBertForMultipleChoice",
"MegatronBertForNextSentencePrediction",
"MegatronBertForPreTraining",
"MegatronBertForQuestionAnswering",
"MegatronBertForSequenceClassification",
"MegatronBertForTokenClassification",
"MegatronBertModel",
"MegatronBertPreTrainedModel",
]
)
_import_structure["models.mmbt"].extend(["MMBTForClassification", "MMBTModel", "ModalEmbeddings"])
_import_structure["models.mobilebert"].extend(
[
"MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileBertForMaskedLM",
"MobileBertForMultipleChoice",
"MobileBertForNextSentencePrediction",
"MobileBertForPreTraining",
"MobileBertForQuestionAnswering",
"MobileBertForSequenceClassification",
"MobileBertForTokenClassification",
"MobileBertLayer",
"MobileBertModel",
"MobileBertPreTrainedModel",
"load_tf_weights_in_mobilebert",
]
)
_import_structure["models.mobilenet_v1"].extend(
[
"MOBILENET_V1_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileNetV1ForImageClassification",
"MobileNetV1Model",
"MobileNetV1PreTrainedModel",
"load_tf_weights_in_mobilenet_v1",
]
)
_import_structure["models.mobilenet_v2"].extend(
[
"MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileNetV2ForImageClassification",
"MobileNetV2ForSemanticSegmentation",
"MobileNetV2Model",
"MobileNetV2PreTrainedModel",
"load_tf_weights_in_mobilenet_v2",
]
)
_import_structure["models.mobilevit"].extend(
[
"MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileViTForImageClassification",
"MobileViTForSemanticSegmentation",
"MobileViTModel",
"MobileViTPreTrainedModel",
]
)
_import_structure["models.mpnet"].extend(
[
"MPNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"MPNetForMaskedLM",
"MPNetForMultipleChoice",
"MPNetForQuestionAnswering",
"MPNetForSequenceClassification",
"MPNetForTokenClassification",
"MPNetLayer",
"MPNetModel",
"MPNetPreTrainedModel",
]
)
_import_structure["models.mt5"].extend(["MT5EncoderModel", "MT5ForConditionalGeneration", "MT5Model"])
_import_structure["models.mvp"].extend(
[
"MVP_PRETRAINED_MODEL_ARCHIVE_LIST",
"MvpForCausalLM",
"MvpForConditionalGeneration",
"MvpForQuestionAnswering",
"MvpForSequenceClassification",
"MvpModel",
"MvpPreTrainedModel",
]
)
_import_structure["models.nat"].extend(
[
"NAT_PRETRAINED_MODEL_ARCHIVE_LIST",
"NatForImageClassification",
"NatModel",
"NatPreTrainedModel",
]
)
_import_structure["models.nezha"].extend(
[
"NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST",
"NezhaForMaskedLM",
"NezhaForMultipleChoice",
"NezhaForNextSentencePrediction",
"NezhaForPreTraining",
"NezhaForQuestionAnswering",
"NezhaForSequenceClassification",
"NezhaForTokenClassification",
"NezhaModel",
"NezhaPreTrainedModel",
]
)
_import_structure["models.nystromformer"].extend(
[
"NYSTROMFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"NystromformerForMaskedLM",
"NystromformerForMultipleChoice",
"NystromformerForQuestionAnswering",
"NystromformerForSequenceClassification",
"NystromformerForTokenClassification",
"NystromformerLayer",
"NystromformerModel",
"NystromformerPreTrainedModel",
]
)
_import_structure["models.openai"].extend(
[
"OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"OpenAIGPTDoubleHeadsModel",
"OpenAIGPTForSequenceClassification",
"OpenAIGPTLMHeadModel",
"OpenAIGPTModel",
"OpenAIGPTPreTrainedModel",
"load_tf_weights_in_openai_gpt",
]
)
_import_structure["models.opt"].extend(
[
"OPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"OPTForCausalLM",
"OPTForQuestionAnswering",
"OPTForSequenceClassification",
"OPTModel",
"OPTPreTrainedModel",
]
)
_import_structure["models.owlvit"].extend(
[
"OWLVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"OwlViTForObjectDetection",
"OwlViTModel",
"OwlViTPreTrainedModel",
"OwlViTTextModel",
"OwlViTVisionModel",
]
)
_import_structure["models.pegasus"].extend(
["PegasusForCausalLM", "PegasusForConditionalGeneration", "PegasusModel", "PegasusPreTrainedModel"]
)
_import_structure["models.pegasus_x"].extend(
[
"PEGASUS_X_PRETRAINED_MODEL_ARCHIVE_LIST",
"PegasusXForConditionalGeneration",
"PegasusXModel",
"PegasusXPreTrainedModel",
]
)
_import_structure["models.perceiver"].extend(
[
"PERCEIVER_PRETRAINED_MODEL_ARCHIVE_LIST",
"PerceiverForImageClassificationConvProcessing",
"PerceiverForImageClassificationFourier",
"PerceiverForImageClassificationLearned",
"PerceiverForMaskedLM",
"PerceiverForMultimodalAutoencoding",
"PerceiverForOpticalFlow",
"PerceiverForSequenceClassification",
"PerceiverLayer",
"PerceiverModel",
"PerceiverPreTrainedModel",
]
)
_import_structure["models.plbart"].extend(
[
"PLBART_PRETRAINED_MODEL_ARCHIVE_LIST",
"PLBartForCausalLM",
"PLBartForConditionalGeneration",
"PLBartForSequenceClassification",
"PLBartModel",
"PLBartPreTrainedModel",
]
)
_import_structure["models.poolformer"].extend(
[
"POOLFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"PoolFormerForImageClassification",
"PoolFormerModel",
"PoolFormerPreTrainedModel",
]
)
_import_structure["models.prophetnet"].extend(
[
"PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"ProphetNetDecoder",
"ProphetNetEncoder",
"ProphetNetForCausalLM",
"ProphetNetForConditionalGeneration",
"ProphetNetModel",
"ProphetNetPreTrainedModel",
]
)
_import_structure["models.qdqbert"].extend(
[
"QDQBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"QDQBertForMaskedLM",
"QDQBertForMultipleChoice",
"QDQBertForNextSentencePrediction",
"QDQBertForQuestionAnswering",
"QDQBertForSequenceClassification",
"QDQBertForTokenClassification",
"QDQBertLayer",
"QDQBertLMHeadModel",
"QDQBertModel",
"QDQBertPreTrainedModel",
"load_tf_weights_in_qdqbert",
]
)
_import_structure["models.rag"].extend(
["RagModel", "RagPreTrainedModel", "RagSequenceForGeneration", "RagTokenForGeneration"]
)
_import_structure["models.realm"].extend(
[
"REALM_PRETRAINED_MODEL_ARCHIVE_LIST",
"RealmEmbedder",
"RealmForOpenQA",
"RealmKnowledgeAugEncoder",
"RealmPreTrainedModel",
"RealmReader",
"RealmRetriever",
"RealmScorer",
"load_tf_weights_in_realm",
]
)
_import_structure["models.reformer"].extend(
[
"REFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"ReformerAttention",
"ReformerForMaskedLM",
"ReformerForQuestionAnswering",
"ReformerForSequenceClassification",
"ReformerLayer",
"ReformerModel",
"ReformerModelWithLMHead",
"ReformerPreTrainedModel",
]
)
_import_structure["models.regnet"].extend(
[
"REGNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"RegNetForImageClassification",
"RegNetModel",
"RegNetPreTrainedModel",
]
)
_import_structure["models.rembert"].extend(
[
"REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"RemBertForCausalLM",
"RemBertForMaskedLM",
"RemBertForMultipleChoice",
"RemBertForQuestionAnswering",
"RemBertForSequenceClassification",
"RemBertForTokenClassification",
"RemBertLayer",
"RemBertModel",
"RemBertPreTrainedModel",
"load_tf_weights_in_rembert",
]
)
_import_structure["models.resnet"].extend(
[
"RESNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"ResNetBackbone",
"ResNetForImageClassification",
"ResNetModel",
"ResNetPreTrainedModel",
]
)
_import_structure["models.retribert"].extend(
["RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST", "RetriBertModel", "RetriBertPreTrainedModel"]
)
_import_structure["models.roberta"].extend(
[
"ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"RobertaForCausalLM",
"RobertaForMaskedLM",
"RobertaForMultipleChoice",
"RobertaForQuestionAnswering",
"RobertaForSequenceClassification",
"RobertaForTokenClassification",
"RobertaModel",
"RobertaPreTrainedModel",
]
)
_import_structure["models.roc_bert"].extend(
[
"ROC_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"RoCBertForCausalLM",
"RoCBertForMaskedLM",
"RoCBertForMultipleChoice",
"RoCBertForPreTraining",
"RoCBertForQuestionAnswering",
"RoCBertForSequenceClassification",
"RoCBertForTokenClassification",
"RoCBertLayer",
"RoCBertModel",
"RoCBertPreTrainedModel",
"load_tf_weights_in_roc_bert",
]
)
_import_structure["models.roformer"].extend(
[
"ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"RoFormerForCausalLM",
"RoFormerForMaskedLM",
"RoFormerForMultipleChoice",
"RoFormerForQuestionAnswering",
"RoFormerForSequenceClassification",
"RoFormerForTokenClassification",
"RoFormerLayer",
"RoFormerModel",
"RoFormerPreTrainedModel",
"load_tf_weights_in_roformer",
]
)
_import_structure["models.segformer"].extend(
[
"SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"SegformerDecodeHead",
"SegformerForImageClassification",
"SegformerForSemanticSegmentation",
"SegformerLayer",
"SegformerModel",
"SegformerPreTrainedModel",
]
)
_import_structure["models.sew"].extend(
[
"SEW_PRETRAINED_MODEL_ARCHIVE_LIST",
"SEWForCTC",
"SEWForSequenceClassification",
"SEWModel",
"SEWPreTrainedModel",
]
)
_import_structure["models.sew_d"].extend(
[
"SEW_D_PRETRAINED_MODEL_ARCHIVE_LIST",
"SEWDForCTC",
"SEWDForSequenceClassification",
"SEWDModel",
"SEWDPreTrainedModel",
]
)
_import_structure["models.speech_encoder_decoder"].extend(["SpeechEncoderDecoderModel"])
_import_structure["models.speech_to_text"].extend(
[
"SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"Speech2TextForConditionalGeneration",
"Speech2TextModel",
"Speech2TextPreTrainedModel",
]
)
_import_structure["models.speech_to_text_2"].extend(["Speech2Text2ForCausalLM", "Speech2Text2PreTrainedModel"])
_import_structure["models.splinter"].extend(
[
"SPLINTER_PRETRAINED_MODEL_ARCHIVE_LIST",
"SplinterForPreTraining",
"SplinterForQuestionAnswering",
"SplinterLayer",
"SplinterModel",
"SplinterPreTrainedModel",
]
)
_import_structure["models.squeezebert"].extend(
[
"SQUEEZEBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"SqueezeBertForMaskedLM",
"SqueezeBertForMultipleChoice",
"SqueezeBertForQuestionAnswering",
"SqueezeBertForSequenceClassification",
"SqueezeBertForTokenClassification",
"SqueezeBertModel",
"SqueezeBertModule",
"SqueezeBertPreTrainedModel",
]
)
_import_structure["models.swin"].extend(
[
"SWIN_PRETRAINED_MODEL_ARCHIVE_LIST",
"SwinForImageClassification",
"SwinForMaskedImageModeling",
"SwinModel",
"SwinPreTrainedModel",
]
)
_import_structure["models.swinv2"].extend(
[
"SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST",
"Swinv2ForImageClassification",
"Swinv2ForMaskedImageModeling",
"Swinv2Model",
"Swinv2PreTrainedModel",
]
)
_import_structure["models.switch_transformers"].extend(
[
"SWITCH_TRANSFORMERS_PRETRAINED_MODEL_ARCHIVE_LIST",
"SwitchTransformersEncoderModel",
"SwitchTransformersForConditionalGeneration",
"SwitchTransformersModel",
"SwitchTransformersPreTrainedModel",
"SwitchTransformersSparseMLP",
"SwitchTransformersTop1Router",
]
)
_import_structure["models.t5"].extend(
[
"T5_PRETRAINED_MODEL_ARCHIVE_LIST",
"T5EncoderModel",
"T5ForConditionalGeneration",
"T5Model",
"T5PreTrainedModel",
"load_tf_weights_in_t5",
]
)
_import_structure["models.tapas"].extend(
[
"TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TapasForMaskedLM",
"TapasForQuestionAnswering",
"TapasForSequenceClassification",
"TapasModel",
"TapasPreTrainedModel",
"load_tf_weights_in_tapas",
]
)
_import_structure["models.time_series_transformer"].extend(
[
"TIME_SERIES_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TimeSeriesTransformerForPrediction",
"TimeSeriesTransformerModel",
"TimeSeriesTransformerPreTrainedModel",
]
)
_import_structure["models.timesformer"].extend(
[
"TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TimesformerForVideoClassification",
"TimesformerModel",
"TimesformerPreTrainedModel",
]
)
_import_structure["models.trajectory_transformer"].extend(
[
"TRAJECTORY_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TrajectoryTransformerModel",
"TrajectoryTransformerPreTrainedModel",
]
)
_import_structure["models.transfo_xl"].extend(
[
"TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST",
"AdaptiveEmbedding",
"TransfoXLForSequenceClassification",
"TransfoXLLMHeadModel",
"TransfoXLModel",
"TransfoXLPreTrainedModel",
"load_tf_weights_in_transfo_xl",
]
)
_import_structure["models.trocr"].extend(
["TROCR_PRETRAINED_MODEL_ARCHIVE_LIST", "TrOCRForCausalLM", "TrOCRPreTrainedModel"]
)
_import_structure["models.unispeech"].extend(
[
"UNISPEECH_PRETRAINED_MODEL_ARCHIVE_LIST",
"UniSpeechForCTC",
"UniSpeechForPreTraining",
"UniSpeechForSequenceClassification",
"UniSpeechModel",
"UniSpeechPreTrainedModel",
]
)
_import_structure["models.unispeech_sat"].extend(
[
"UNISPEECH_SAT_PRETRAINED_MODEL_ARCHIVE_LIST",
"UniSpeechSatForAudioFrameClassification",
"UniSpeechSatForCTC",
"UniSpeechSatForPreTraining",
"UniSpeechSatForSequenceClassification",
"UniSpeechSatForXVector",
"UniSpeechSatModel",
"UniSpeechSatPreTrainedModel",
]
)
_import_structure["models.van"].extend(
[
"VAN_PRETRAINED_MODEL_ARCHIVE_LIST",
"VanForImageClassification",
"VanModel",
"VanPreTrainedModel",
]
)
_import_structure["models.videomae"].extend(
[
"VIDEOMAE_PRETRAINED_MODEL_ARCHIVE_LIST",
"VideoMAEForPreTraining",
"VideoMAEForVideoClassification",
"VideoMAEModel",
"VideoMAEPreTrainedModel",
]
)
_import_structure["models.vilt"].extend(
[
"VILT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViltForImageAndTextRetrieval",
"ViltForImagesAndTextClassification",
"ViltForMaskedLM",
"ViltForQuestionAnswering",
"ViltForTokenClassification",
"ViltLayer",
"ViltModel",
"ViltPreTrainedModel",
]
)
_import_structure["models.vision_encoder_decoder"].extend(["VisionEncoderDecoderModel"])
_import_structure["models.vision_text_dual_encoder"].extend(["VisionTextDualEncoderModel"])
_import_structure["models.visual_bert"].extend(
[
"VISUAL_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"VisualBertForMultipleChoice",
"VisualBertForPreTraining",
"VisualBertForQuestionAnswering",
"VisualBertForRegionToPhraseAlignment",
"VisualBertForVisualReasoning",
"VisualBertLayer",
"VisualBertModel",
"VisualBertPreTrainedModel",
]
)
_import_structure["models.vit"].extend(
[
"VIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTForImageClassification",
"ViTForMaskedImageModeling",
"ViTModel",
"ViTPreTrainedModel",
]
)
_import_structure["models.vit_hybrid"].extend(
[
"VIT_HYBRID_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTHybridForImageClassification",
"ViTHybridModel",
"ViTHybridPreTrainedModel",
]
)
_import_structure["models.vit_mae"].extend(
[
"VIT_MAE_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTMAEForPreTraining",
"ViTMAELayer",
"ViTMAEModel",
"ViTMAEPreTrainedModel",
]
)
_import_structure["models.vit_msn"].extend(
[
"VIT_MSN_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTMSNForImageClassification",
"ViTMSNModel",
"ViTMSNPreTrainedModel",
]
)
_import_structure["models.wav2vec2"].extend(
[
"WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST",
"Wav2Vec2ForAudioFrameClassification",
"Wav2Vec2ForCTC",
"Wav2Vec2ForMaskedLM",
"Wav2Vec2ForPreTraining",
"Wav2Vec2ForSequenceClassification",
"Wav2Vec2ForXVector",
"Wav2Vec2Model",
"Wav2Vec2PreTrainedModel",
]
)
_import_structure["models.wav2vec2_conformer"].extend(
[
"WAV2VEC2_CONFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"Wav2Vec2ConformerForAudioFrameClassification",
"Wav2Vec2ConformerForCTC",
"Wav2Vec2ConformerForPreTraining",
"Wav2Vec2ConformerForSequenceClassification",
"Wav2Vec2ConformerForXVector",
"Wav2Vec2ConformerModel",
"Wav2Vec2ConformerPreTrainedModel",
]
)
_import_structure["models.wavlm"].extend(
[
"WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"WavLMForAudioFrameClassification",
"WavLMForCTC",
"WavLMForSequenceClassification",
"WavLMForXVector",
"WavLMModel",
"WavLMPreTrainedModel",
]
)
_import_structure["models.whisper"].extend(
[
"WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST",
"WhisperForConditionalGeneration",
"WhisperModel",
"WhisperPreTrainedModel",
]
)
_import_structure["models.x_clip"].extend(
[
"XCLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"XCLIPModel",
"XCLIPPreTrainedModel",
"XCLIPTextModel",
"XCLIPVisionModel",
]
)
_import_structure["models.xglm"].extend(
[
"XGLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"XGLMForCausalLM",
"XGLMModel",
"XGLMPreTrainedModel",
]
)
_import_structure["models.xlm"].extend(
[
"XLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMForMultipleChoice",
"XLMForQuestionAnswering",
"XLMForQuestionAnsweringSimple",
"XLMForSequenceClassification",
"XLMForTokenClassification",
"XLMModel",
"XLMPreTrainedModel",
"XLMWithLMHeadModel",
]
)
_import_structure["models.xlm_prophetnet"].extend(
[
"XLM_PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMProphetNetDecoder",
"XLMProphetNetEncoder",
"XLMProphetNetForCausalLM",
"XLMProphetNetForConditionalGeneration",
"XLMProphetNetModel",
"XLMProphetNetPreTrainedModel",
]
)
_import_structure["models.xlm_roberta"].extend(
[
"XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMRobertaForCausalLM",
"XLMRobertaForMaskedLM",
"XLMRobertaForMultipleChoice",
"XLMRobertaForQuestionAnswering",
"XLMRobertaForSequenceClassification",
"XLMRobertaForTokenClassification",
"XLMRobertaModel",
"XLMRobertaPreTrainedModel",
]
)
_import_structure["models.xlm_roberta_xl"].extend(
[
"XLM_ROBERTA_XL_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMRobertaXLForCausalLM",
"XLMRobertaXLForMaskedLM",
"XLMRobertaXLForMultipleChoice",
"XLMRobertaXLForQuestionAnswering",
"XLMRobertaXLForSequenceClassification",
"XLMRobertaXLForTokenClassification",
"XLMRobertaXLModel",
"XLMRobertaXLPreTrainedModel",
]
)
_import_structure["models.xlnet"].extend(
[
"XLNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLNetForMultipleChoice",
"XLNetForQuestionAnswering",
"XLNetForQuestionAnsweringSimple",
"XLNetForSequenceClassification",
"XLNetForTokenClassification",
"XLNetLMHeadModel",
"XLNetModel",
"XLNetPreTrainedModel",
"load_tf_weights_in_xlnet",
]
)
_import_structure["models.yolos"].extend(
[
"YOLOS_PRETRAINED_MODEL_ARCHIVE_LIST",
"YolosForObjectDetection",
"YolosModel",
"YolosPreTrainedModel",
]
)
_import_structure["models.yoso"].extend(
[
"YOSO_PRETRAINED_MODEL_ARCHIVE_LIST",
"YosoForMaskedLM",
"YosoForMultipleChoice",
"YosoForQuestionAnswering",
"YosoForSequenceClassification",
"YosoForTokenClassification",
"YosoLayer",
"YosoModel",
"YosoPreTrainedModel",
]
)
_import_structure["optimization"] = [
"Adafactor",
"AdamW",
"get_constant_schedule",
"get_constant_schedule_with_warmup",
"get_cosine_schedule_with_warmup",
"get_cosine_with_hard_restarts_schedule_with_warmup",
"get_linear_schedule_with_warmup",
"get_polynomial_decay_schedule_with_warmup",
"get_scheduler",
]
_import_structure["pytorch_utils"] = ["Conv1D", "apply_chunking_to_forward", "prune_layer"]
_import_structure["sagemaker"] = []
_import_structure["trainer"] = ["Trainer"]
_import_structure["trainer_pt_utils"] = ["torch_distributed_zero_first"]
_import_structure["trainer_seq2seq"] = ["Seq2SeqTrainer"]
# TensorFlow-backed objects
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_tf_objects
_import_structure["utils.dummy_tf_objects"] = [name for name in dir(dummy_tf_objects) if not name.startswith("_")]
else:
_import_structure["activations_tf"] = []
_import_structure["benchmark.benchmark_args_tf"] = ["TensorFlowBenchmarkArguments"]
_import_structure["benchmark.benchmark_tf"] = ["TensorFlowBenchmark"]
_import_structure["generation"].extend(
[
"TFForcedBOSTokenLogitsProcessor",
"TFForcedEOSTokenLogitsProcessor",
"TFGenerationMixin",
"TFLogitsProcessor",
"TFLogitsProcessorList",
"TFLogitsWarper",
"TFMinLengthLogitsProcessor",
"TFNoBadWordsLogitsProcessor",
"TFNoRepeatNGramLogitsProcessor",
"TFRepetitionPenaltyLogitsProcessor",
"TFTemperatureLogitsWarper",
"TFTopKLogitsWarper",
"TFTopPLogitsWarper",
"tf_top_k_top_p_filtering",
]
)
_import_structure["generation_tf_utils"] = []
_import_structure["keras_callbacks"] = ["KerasMetricCallback", "PushToHubCallback"]
_import_structure["modeling_tf_outputs"] = []
_import_structure["modeling_tf_utils"] = [
"TFPreTrainedModel",
"TFSequenceSummary",
"TFSharedEmbeddings",
"shape_list",
]
# TensorFlow models structure
_import_structure["models.albert"].extend(
[
"TF_ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFAlbertForMaskedLM",
"TFAlbertForMultipleChoice",
"TFAlbertForPreTraining",
"TFAlbertForQuestionAnswering",
"TFAlbertForSequenceClassification",
"TFAlbertForTokenClassification",
"TFAlbertMainLayer",
"TFAlbertModel",
"TFAlbertPreTrainedModel",
]
)
_import_structure["models.auto"].extend(
[
"TF_MODEL_FOR_CAUSAL_LM_MAPPING",
"TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING",
"TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING",
"TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING",
"TF_MODEL_FOR_MASKED_LM_MAPPING",
"TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING",
"TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING",
"TF_MODEL_FOR_PRETRAINING_MAPPING",
"TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING",
"TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING",
"TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING",
"TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING",
"TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING",
"TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING",
"TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING",
"TF_MODEL_FOR_VISION_2_SEQ_MAPPING",
"TF_MODEL_MAPPING",
"TF_MODEL_WITH_LM_HEAD_MAPPING",
"TFAutoModel",
"TFAutoModelForCausalLM",
"TFAutoModelForDocumentQuestionAnswering",
"TFAutoModelForImageClassification",
"TFAutoModelForMaskedLM",
"TFAutoModelForMultipleChoice",
"TFAutoModelForNextSentencePrediction",
"TFAutoModelForPreTraining",
"TFAutoModelForQuestionAnswering",
"TFAutoModelForSemanticSegmentation",
"TFAutoModelForSeq2SeqLM",
"TFAutoModelForSequenceClassification",
"TFAutoModelForSpeechSeq2Seq",
"TFAutoModelForTableQuestionAnswering",
"TFAutoModelForTokenClassification",
"TFAutoModelForVision2Seq",
"TFAutoModelWithLMHead",
]
)
_import_structure["models.bart"].extend(
["TFBartForConditionalGeneration", "TFBartForSequenceClassification", "TFBartModel", "TFBartPretrainedModel"]
)
_import_structure["models.bert"].extend(
[
"TF_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFBertEmbeddings",
"TFBertForMaskedLM",
"TFBertForMultipleChoice",
"TFBertForNextSentencePrediction",
"TFBertForPreTraining",
"TFBertForQuestionAnswering",
"TFBertForSequenceClassification",
"TFBertForTokenClassification",
"TFBertLMHeadModel",
"TFBertMainLayer",
"TFBertModel",
"TFBertPreTrainedModel",
]
)
_import_structure["models.blenderbot"].extend(
["TFBlenderbotForConditionalGeneration", "TFBlenderbotModel", "TFBlenderbotPreTrainedModel"]
)
_import_structure["models.blenderbot_small"].extend(
["TFBlenderbotSmallForConditionalGeneration", "TFBlenderbotSmallModel", "TFBlenderbotSmallPreTrainedModel"]
)
_import_structure["models.camembert"].extend(
[
"TF_CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCamembertForCausalLM",
"TFCamembertForMaskedLM",
"TFCamembertForMultipleChoice",
"TFCamembertForQuestionAnswering",
"TFCamembertForSequenceClassification",
"TFCamembertForTokenClassification",
"TFCamembertModel",
"TFCamembertPreTrainedModel",
]
)
_import_structure["models.clip"].extend(
[
"TF_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCLIPModel",
"TFCLIPPreTrainedModel",
"TFCLIPTextModel",
"TFCLIPVisionModel",
]
)
_import_structure["models.convbert"].extend(
[
"TF_CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFConvBertForMaskedLM",
"TFConvBertForMultipleChoice",
"TFConvBertForQuestionAnswering",
"TFConvBertForSequenceClassification",
"TFConvBertForTokenClassification",
"TFConvBertLayer",
"TFConvBertModel",
"TFConvBertPreTrainedModel",
]
)
_import_structure["models.convnext"].extend(
[
"TFConvNextForImageClassification",
"TFConvNextModel",
"TFConvNextPreTrainedModel",
]
)
_import_structure["models.ctrl"].extend(
[
"TF_CTRL_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCTRLForSequenceClassification",
"TFCTRLLMHeadModel",
"TFCTRLModel",
"TFCTRLPreTrainedModel",
]
)
_import_structure["models.cvt"].extend(
[
"TF_CVT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCvtForImageClassification",
"TFCvtModel",
"TFCvtPreTrainedModel",
]
)
_import_structure["models.data2vec"].extend(
[
"TFData2VecVisionForImageClassification",
"TFData2VecVisionForSemanticSegmentation",
"TFData2VecVisionModel",
"TFData2VecVisionPreTrainedModel",
]
)
_import_structure["models.deberta"].extend(
[
"TF_DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDebertaForMaskedLM",
"TFDebertaForQuestionAnswering",
"TFDebertaForSequenceClassification",
"TFDebertaForTokenClassification",
"TFDebertaModel",
"TFDebertaPreTrainedModel",
]
)
_import_structure["models.deberta_v2"].extend(
[
"TF_DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDebertaV2ForMaskedLM",
"TFDebertaV2ForQuestionAnswering",
"TFDebertaV2ForSequenceClassification",
"TFDebertaV2ForTokenClassification",
"TFDebertaV2Model",
"TFDebertaV2PreTrainedModel",
]
)
_import_structure["models.deit"].extend(
[
"TF_DEIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDeiTForImageClassification",
"TFDeiTForImageClassificationWithTeacher",
"TFDeiTForMaskedImageModeling",
"TFDeiTModel",
"TFDeiTPreTrainedModel",
]
)
_import_structure["models.distilbert"].extend(
[
"TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDistilBertForMaskedLM",
"TFDistilBertForMultipleChoice",
"TFDistilBertForQuestionAnswering",
"TFDistilBertForSequenceClassification",
"TFDistilBertForTokenClassification",
"TFDistilBertMainLayer",
"TFDistilBertModel",
"TFDistilBertPreTrainedModel",
]
)
_import_structure["models.dpr"].extend(
[
"TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TF_DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TF_DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDPRContextEncoder",
"TFDPRPretrainedContextEncoder",
"TFDPRPretrainedQuestionEncoder",
"TFDPRPretrainedReader",
"TFDPRQuestionEncoder",
"TFDPRReader",
]
)
_import_structure["models.electra"].extend(
[
"TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFElectraForMaskedLM",
"TFElectraForMultipleChoice",
"TFElectraForPreTraining",
"TFElectraForQuestionAnswering",
"TFElectraForSequenceClassification",
"TFElectraForTokenClassification",
"TFElectraModel",
"TFElectraPreTrainedModel",
]
)
_import_structure["models.encoder_decoder"].append("TFEncoderDecoderModel")
_import_structure["models.esm"].extend(
[
"ESM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFEsmForMaskedLM",
"TFEsmForSequenceClassification",
"TFEsmForTokenClassification",
"TFEsmModel",
"TFEsmPreTrainedModel",
]
)
_import_structure["models.flaubert"].extend(
[
"TF_FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFFlaubertForMultipleChoice",
"TFFlaubertForQuestionAnsweringSimple",
"TFFlaubertForSequenceClassification",
"TFFlaubertForTokenClassification",
"TFFlaubertModel",
"TFFlaubertPreTrainedModel",
"TFFlaubertWithLMHeadModel",
]
)
_import_structure["models.funnel"].extend(
[
"TF_FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFFunnelBaseModel",
"TFFunnelForMaskedLM",
"TFFunnelForMultipleChoice",
"TFFunnelForPreTraining",
"TFFunnelForQuestionAnswering",
"TFFunnelForSequenceClassification",
"TFFunnelForTokenClassification",
"TFFunnelModel",
"TFFunnelPreTrainedModel",
]
)
_import_structure["models.gpt2"].extend(
[
"TF_GPT2_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFGPT2DoubleHeadsModel",
"TFGPT2ForSequenceClassification",
"TFGPT2LMHeadModel",
"TFGPT2MainLayer",
"TFGPT2Model",
"TFGPT2PreTrainedModel",
]
)
_import_structure["models.gptj"].extend(
[
"TFGPTJForCausalLM",
"TFGPTJForQuestionAnswering",
"TFGPTJForSequenceClassification",
"TFGPTJModel",
"TFGPTJPreTrainedModel",
]
)
_import_structure["models.groupvit"].extend(
[
"TF_GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFGroupViTModel",
"TFGroupViTPreTrainedModel",
"TFGroupViTTextModel",
"TFGroupViTVisionModel",
]
)
_import_structure["models.hubert"].extend(
[
"TF_HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFHubertForCTC",
"TFHubertModel",
"TFHubertPreTrainedModel",
]
)
_import_structure["models.layoutlm"].extend(
[
"TF_LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLayoutLMForMaskedLM",
"TFLayoutLMForQuestionAnswering",
"TFLayoutLMForSequenceClassification",
"TFLayoutLMForTokenClassification",
"TFLayoutLMMainLayer",
"TFLayoutLMModel",
"TFLayoutLMPreTrainedModel",
]
)
_import_structure["models.layoutlmv3"].extend(
[
"TF_LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLayoutLMv3ForQuestionAnswering",
"TFLayoutLMv3ForSequenceClassification",
"TFLayoutLMv3ForTokenClassification",
"TFLayoutLMv3Model",
"TFLayoutLMv3PreTrainedModel",
]
)
_import_structure["models.led"].extend(["TFLEDForConditionalGeneration", "TFLEDModel", "TFLEDPreTrainedModel"])
_import_structure["models.longformer"].extend(
[
"TF_LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLongformerForMaskedLM",
"TFLongformerForMultipleChoice",
"TFLongformerForQuestionAnswering",
"TFLongformerForSequenceClassification",
"TFLongformerForTokenClassification",
"TFLongformerModel",
"TFLongformerPreTrainedModel",
"TFLongformerSelfAttention",
]
)
_import_structure["models.lxmert"].extend(
[
"TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLxmertForPreTraining",
"TFLxmertMainLayer",
"TFLxmertModel",
"TFLxmertPreTrainedModel",
"TFLxmertVisualFeatureEncoder",
]
)
_import_structure["models.marian"].extend(["TFMarianModel", "TFMarianMTModel", "TFMarianPreTrainedModel"])
_import_structure["models.mbart"].extend(
["TFMBartForConditionalGeneration", "TFMBartModel", "TFMBartPreTrainedModel"]
)
_import_structure["models.mobilebert"].extend(
[
"TF_MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFMobileBertForMaskedLM",
"TFMobileBertForMultipleChoice",
"TFMobileBertForNextSentencePrediction",
"TFMobileBertForPreTraining",
"TFMobileBertForQuestionAnswering",
"TFMobileBertForSequenceClassification",
"TFMobileBertForTokenClassification",
"TFMobileBertMainLayer",
"TFMobileBertModel",
"TFMobileBertPreTrainedModel",
]
)
_import_structure["models.mobilevit"].extend(
[
"TF_MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFMobileViTForImageClassification",
"TFMobileViTForSemanticSegmentation",
"TFMobileViTModel",
"TFMobileViTPreTrainedModel",
]
)
_import_structure["models.mpnet"].extend(
[
"TF_MPNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFMPNetForMaskedLM",
"TFMPNetForMultipleChoice",
"TFMPNetForQuestionAnswering",
"TFMPNetForSequenceClassification",
"TFMPNetForTokenClassification",
"TFMPNetMainLayer",
"TFMPNetModel",
"TFMPNetPreTrainedModel",
]
)
_import_structure["models.mt5"].extend(["TFMT5EncoderModel", "TFMT5ForConditionalGeneration", "TFMT5Model"])
_import_structure["models.openai"].extend(
[
"TF_OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFOpenAIGPTDoubleHeadsModel",
"TFOpenAIGPTForSequenceClassification",
"TFOpenAIGPTLMHeadModel",
"TFOpenAIGPTMainLayer",
"TFOpenAIGPTModel",
"TFOpenAIGPTPreTrainedModel",
]
)
_import_structure["models.opt"].extend(
[
"TFOPTForCausalLM",
"TFOPTModel",
"TFOPTPreTrainedModel",
]
)
_import_structure["models.pegasus"].extend(
["TFPegasusForConditionalGeneration", "TFPegasusModel", "TFPegasusPreTrainedModel"]
)
_import_structure["models.rag"].extend(
[
"TFRagModel",
"TFRagPreTrainedModel",
"TFRagSequenceForGeneration",
"TFRagTokenForGeneration",
]
)
_import_structure["models.regnet"].extend(
[
"TF_REGNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRegNetForImageClassification",
"TFRegNetModel",
"TFRegNetPreTrainedModel",
]
)
_import_structure["models.rembert"].extend(
[
"TF_REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRemBertForCausalLM",
"TFRemBertForMaskedLM",
"TFRemBertForMultipleChoice",
"TFRemBertForQuestionAnswering",
"TFRemBertForSequenceClassification",
"TFRemBertForTokenClassification",
"TFRemBertLayer",
"TFRemBertModel",
"TFRemBertPreTrainedModel",
]
)
_import_structure["models.resnet"].extend(
[
"TF_RESNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFResNetForImageClassification",
"TFResNetModel",
"TFResNetPreTrainedModel",
]
)
_import_structure["models.roberta"].extend(
[
"TF_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRobertaForCausalLM",
"TFRobertaForMaskedLM",
"TFRobertaForMultipleChoice",
"TFRobertaForQuestionAnswering",
"TFRobertaForSequenceClassification",
"TFRobertaForTokenClassification",
"TFRobertaMainLayer",
"TFRobertaModel",
"TFRobertaPreTrainedModel",
]
)
_import_structure["models.roformer"].extend(
[
"TF_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRoFormerForCausalLM",
"TFRoFormerForMaskedLM",
"TFRoFormerForMultipleChoice",
"TFRoFormerForQuestionAnswering",
"TFRoFormerForSequenceClassification",
"TFRoFormerForTokenClassification",
"TFRoFormerLayer",
"TFRoFormerModel",
"TFRoFormerPreTrainedModel",
]
)
_import_structure["models.segformer"].extend(
[
"TF_SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFSegformerDecodeHead",
"TFSegformerForImageClassification",
"TFSegformerForSemanticSegmentation",
"TFSegformerModel",
"TFSegformerPreTrainedModel",
]
)
_import_structure["models.speech_to_text"].extend(
[
"TF_SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFSpeech2TextForConditionalGeneration",
"TFSpeech2TextModel",
"TFSpeech2TextPreTrainedModel",
]
)
_import_structure["models.swin"].extend(
[
"TF_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFSwinForImageClassification",
"TFSwinForMaskedImageModeling",
"TFSwinModel",
"TFSwinPreTrainedModel",
]
)
_import_structure["models.t5"].extend(
[
"TF_T5_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFT5EncoderModel",
"TFT5ForConditionalGeneration",
"TFT5Model",
"TFT5PreTrainedModel",
]
)
_import_structure["models.tapas"].extend(
[
"TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFTapasForMaskedLM",
"TFTapasForQuestionAnswering",
"TFTapasForSequenceClassification",
"TFTapasModel",
"TFTapasPreTrainedModel",
]
)
_import_structure["models.transfo_xl"].extend(
[
"TF_TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFAdaptiveEmbedding",
"TFTransfoXLForSequenceClassification",
"TFTransfoXLLMHeadModel",
"TFTransfoXLMainLayer",
"TFTransfoXLModel",
"TFTransfoXLPreTrainedModel",
]
)
_import_structure["models.vision_encoder_decoder"].extend(["TFVisionEncoderDecoderModel"])
_import_structure["models.vit"].extend(
[
"TFViTForImageClassification",
"TFViTModel",
"TFViTPreTrainedModel",
]
)
_import_structure["models.vit_mae"].extend(
[
"TFViTMAEForPreTraining",
"TFViTMAEModel",
"TFViTMAEPreTrainedModel",
]
)
_import_structure["models.wav2vec2"].extend(
[
"TF_WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFWav2Vec2ForCTC",
"TFWav2Vec2Model",
"TFWav2Vec2PreTrainedModel",
]
)
_import_structure["models.whisper"].extend(
[
"TF_WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFWhisperForConditionalGeneration",
"TFWhisperModel",
"TFWhisperPreTrainedModel",
]
)
_import_structure["models.xglm"].extend(
[
"TF_XGLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXGLMForCausalLM",
"TFXGLMModel",
"TFXGLMPreTrainedModel",
]
)
_import_structure["models.xlm"].extend(
[
"TF_XLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXLMForMultipleChoice",
"TFXLMForQuestionAnsweringSimple",
"TFXLMForSequenceClassification",
"TFXLMForTokenClassification",
"TFXLMMainLayer",
"TFXLMModel",
"TFXLMPreTrainedModel",
"TFXLMWithLMHeadModel",
]
)
_import_structure["models.xlm_roberta"].extend(
[
"TF_XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXLMRobertaForMaskedLM",
"TFXLMRobertaForMultipleChoice",
"TFXLMRobertaForQuestionAnswering",
"TFXLMRobertaForSequenceClassification",
"TFXLMRobertaForTokenClassification",
"TFXLMRobertaModel",
]
)
_import_structure["models.xlnet"].extend(
[
"TF_XLNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXLNetForMultipleChoice",
"TFXLNetForQuestionAnsweringSimple",
"TFXLNetForSequenceClassification",
"TFXLNetForTokenClassification",
"TFXLNetLMHeadModel",
"TFXLNetMainLayer",
"TFXLNetModel",
"TFXLNetPreTrainedModel",
]
)
_import_structure["optimization_tf"] = ["AdamWeightDecay", "GradientAccumulator", "WarmUp", "create_optimizer"]
_import_structure["tf_utils"] = []
_import_structure["trainer_tf"] = ["TFTrainer"]
# FLAX-backed objects
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_flax_objects
_import_structure["utils.dummy_flax_objects"] = [
name for name in dir(dummy_flax_objects) if not name.startswith("_")
]
else:
_import_structure["generation"].extend(
[
"FlaxForcedBOSTokenLogitsProcessor",
"FlaxForcedEOSTokenLogitsProcessor",
"FlaxGenerationMixin",
"FlaxLogitsProcessor",
"FlaxLogitsProcessorList",
"FlaxLogitsWarper",
"FlaxMinLengthLogitsProcessor",
"FlaxTemperatureLogitsWarper",
"FlaxTopKLogitsWarper",
"FlaxTopPLogitsWarper",
]
)
_import_structure["generation_flax_utils"] = []
_import_structure["modeling_flax_outputs"] = []
_import_structure["modeling_flax_utils"] = ["FlaxPreTrainedModel"]
_import_structure["models.albert"].extend(
[
"FlaxAlbertForMaskedLM",
"FlaxAlbertForMultipleChoice",
"FlaxAlbertForPreTraining",
"FlaxAlbertForQuestionAnswering",
"FlaxAlbertForSequenceClassification",
"FlaxAlbertForTokenClassification",
"FlaxAlbertModel",
"FlaxAlbertPreTrainedModel",
]
)
_import_structure["models.auto"].extend(
[
"FLAX_MODEL_FOR_CAUSAL_LM_MAPPING",
"FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING",
"FLAX_MODEL_FOR_MASKED_LM_MAPPING",
"FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING",
"FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING",
"FLAX_MODEL_FOR_PRETRAINING_MAPPING",
"FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING",
"FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING",
"FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING",
"FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING",
"FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING",
"FLAX_MODEL_MAPPING",
"FlaxAutoModel",
"FlaxAutoModelForCausalLM",
"FlaxAutoModelForImageClassification",
"FlaxAutoModelForMaskedLM",
"FlaxAutoModelForMultipleChoice",
"FlaxAutoModelForNextSentencePrediction",
"FlaxAutoModelForPreTraining",
"FlaxAutoModelForQuestionAnswering",
"FlaxAutoModelForSeq2SeqLM",
"FlaxAutoModelForSequenceClassification",
"FlaxAutoModelForTokenClassification",
"FlaxAutoModelForVision2Seq",
]
)
# Flax models structure
_import_structure["models.bart"].extend(
[
"FlaxBartDecoderPreTrainedModel",
"FlaxBartForCausalLM",
"FlaxBartForConditionalGeneration",
"FlaxBartForQuestionAnswering",
"FlaxBartForSequenceClassification",
"FlaxBartModel",
"FlaxBartPreTrainedModel",
]
)
_import_structure["models.beit"].extend(
[
"FlaxBeitForImageClassification",
"FlaxBeitForMaskedImageModeling",
"FlaxBeitModel",
"FlaxBeitPreTrainedModel",
]
)
_import_structure["models.bert"].extend(
[
"FlaxBertForCausalLM",
"FlaxBertForMaskedLM",
"FlaxBertForMultipleChoice",
"FlaxBertForNextSentencePrediction",
"FlaxBertForPreTraining",
"FlaxBertForQuestionAnswering",
"FlaxBertForSequenceClassification",
"FlaxBertForTokenClassification",
"FlaxBertModel",
"FlaxBertPreTrainedModel",
]
)
_import_structure["models.big_bird"].extend(
[
"FlaxBigBirdForCausalLM",
"FlaxBigBirdForMaskedLM",
"FlaxBigBirdForMultipleChoice",
"FlaxBigBirdForPreTraining",
"FlaxBigBirdForQuestionAnswering",
"FlaxBigBirdForSequenceClassification",
"FlaxBigBirdForTokenClassification",
"FlaxBigBirdModel",
"FlaxBigBirdPreTrainedModel",
]
)
_import_structure["models.blenderbot"].extend(
["FlaxBlenderbotForConditionalGeneration", "FlaxBlenderbotModel", "FlaxBlenderbotPreTrainedModel"]
)
_import_structure["models.blenderbot_small"].extend(
[
"FlaxBlenderbotSmallForConditionalGeneration",
"FlaxBlenderbotSmallModel",
"FlaxBlenderbotSmallPreTrainedModel",
]
)
_import_structure["models.clip"].extend(
[
"FlaxCLIPModel",
"FlaxCLIPPreTrainedModel",
"FlaxCLIPTextModel",
"FlaxCLIPTextPreTrainedModel",
"FlaxCLIPVisionModel",
"FlaxCLIPVisionPreTrainedModel",
]
)
_import_structure["models.distilbert"].extend(
[
"FlaxDistilBertForMaskedLM",
"FlaxDistilBertForMultipleChoice",
"FlaxDistilBertForQuestionAnswering",
"FlaxDistilBertForSequenceClassification",
"FlaxDistilBertForTokenClassification",
"FlaxDistilBertModel",
"FlaxDistilBertPreTrainedModel",
]
)
_import_structure["models.electra"].extend(
[
"FlaxElectraForCausalLM",
"FlaxElectraForMaskedLM",
"FlaxElectraForMultipleChoice",
"FlaxElectraForPreTraining",
"FlaxElectraForQuestionAnswering",
"FlaxElectraForSequenceClassification",
"FlaxElectraForTokenClassification",
"FlaxElectraModel",
"FlaxElectraPreTrainedModel",
]
)
_import_structure["models.encoder_decoder"].append("FlaxEncoderDecoderModel")
_import_structure["models.gpt2"].extend(["FlaxGPT2LMHeadModel", "FlaxGPT2Model", "FlaxGPT2PreTrainedModel"])
_import_structure["models.gpt_neo"].extend(
["FlaxGPTNeoForCausalLM", "FlaxGPTNeoModel", "FlaxGPTNeoPreTrainedModel"]
)
_import_structure["models.gptj"].extend(["FlaxGPTJForCausalLM", "FlaxGPTJModel", "FlaxGPTJPreTrainedModel"])
_import_structure["models.longt5"].extend(
["FlaxLongT5ForConditionalGeneration", "FlaxLongT5Model", "FlaxLongT5PreTrainedModel"]
)
_import_structure["models.marian"].extend(
[
"FlaxMarianModel",
"FlaxMarianMTModel",
"FlaxMarianPreTrainedModel",
]
)
_import_structure["models.mbart"].extend(
[
"FlaxMBartForConditionalGeneration",
"FlaxMBartForQuestionAnswering",
"FlaxMBartForSequenceClassification",
"FlaxMBartModel",
"FlaxMBartPreTrainedModel",
]
)
_import_structure["models.mt5"].extend(["FlaxMT5EncoderModel", "FlaxMT5ForConditionalGeneration", "FlaxMT5Model"])
_import_structure["models.opt"].extend(
[
"FlaxOPTForCausalLM",
"FlaxOPTModel",
"FlaxOPTPreTrainedModel",
]
)
_import_structure["models.pegasus"].extend(
[
"FlaxPegasusForConditionalGeneration",
"FlaxPegasusModel",
"FlaxPegasusPreTrainedModel",
]
)
_import_structure["models.roberta"].extend(
[
"FlaxRobertaForCausalLM",
"FlaxRobertaForMaskedLM",
"FlaxRobertaForMultipleChoice",
"FlaxRobertaForQuestionAnswering",
"FlaxRobertaForSequenceClassification",
"FlaxRobertaForTokenClassification",
"FlaxRobertaModel",
"FlaxRobertaPreTrainedModel",
]
)
_import_structure["models.roformer"].extend(
[
"FlaxRoFormerForMaskedLM",
"FlaxRoFormerForMultipleChoice",
"FlaxRoFormerForQuestionAnswering",
"FlaxRoFormerForSequenceClassification",
"FlaxRoFormerForTokenClassification",
"FlaxRoFormerModel",
"FlaxRoFormerPreTrainedModel",
]
)
_import_structure["models.speech_encoder_decoder"].append("FlaxSpeechEncoderDecoderModel")
_import_structure["models.t5"].extend(
["FlaxT5EncoderModel", "FlaxT5ForConditionalGeneration", "FlaxT5Model", "FlaxT5PreTrainedModel"]
)
_import_structure["models.vision_encoder_decoder"].append("FlaxVisionEncoderDecoderModel")
_import_structure["models.vision_text_dual_encoder"].extend(["FlaxVisionTextDualEncoderModel"])
_import_structure["models.vit"].extend(["FlaxViTForImageClassification", "FlaxViTModel", "FlaxViTPreTrainedModel"])
_import_structure["models.wav2vec2"].extend(
["FlaxWav2Vec2ForCTC", "FlaxWav2Vec2ForPreTraining", "FlaxWav2Vec2Model", "FlaxWav2Vec2PreTrainedModel"]
)
_import_structure["models.xglm"].extend(
[
"FlaxXGLMForCausalLM",
"FlaxXGLMModel",
"FlaxXGLMPreTrainedModel",
]
)
_import_structure["models.xlm_roberta"].extend(
[
"FlaxXLMRobertaForMaskedLM",
"FlaxXLMRobertaForMultipleChoice",
"FlaxXLMRobertaForQuestionAnswering",
"FlaxXLMRobertaForSequenceClassification",
"FlaxXLMRobertaForTokenClassification",
"FlaxXLMRobertaModel",
]
)
# Direct imports for type-checking
if TYPE_CHECKING:
# Configuration
from .configuration_utils import PretrainedConfig
# Data
from .data import (
DataProcessor,
InputExample,
InputFeatures,
SingleSentenceClassificationProcessor,
SquadExample,
SquadFeatures,
SquadV1Processor,
SquadV2Processor,
glue_compute_metrics,
glue_convert_examples_to_features,
glue_output_modes,
glue_processors,
glue_tasks_num_labels,
squad_convert_examples_to_features,
xnli_compute_metrics,
xnli_output_modes,
xnli_processors,
xnli_tasks_num_labels,
)
from .data.data_collator import (
DataCollator,
DataCollatorForLanguageModeling,
DataCollatorForPermutationLanguageModeling,
DataCollatorForSeq2Seq,
DataCollatorForSOP,
DataCollatorForTokenClassification,
DataCollatorForWholeWordMask,
DataCollatorWithPadding,
DefaultDataCollator,
default_data_collator,
)
from .feature_extraction_sequence_utils import SequenceFeatureExtractor
# Feature Extractor
from .feature_extraction_utils import BatchFeature, FeatureExtractionMixin
# Generation
from .generation import GenerationConfig
from .hf_argparser import HfArgumentParser
# Integrations
from .integrations import (
is_clearml_available,
is_comet_available,
is_neptune_available,
is_optuna_available,
is_ray_available,
is_ray_tune_available,
is_sigopt_available,
is_tensorboard_available,
is_wandb_available,
)
# Model Cards
from .modelcard import ModelCard
# TF 2.0 <=> PyTorch conversion utilities
from .modeling_tf_pytorch_utils import (
convert_tf_weight_name_to_pt_weight_name,
load_pytorch_checkpoint_in_tf2_model,
load_pytorch_model_in_tf2_model,
load_pytorch_weights_in_tf2_model,
load_tf2_checkpoint_in_pytorch_model,
load_tf2_model_in_pytorch_model,
load_tf2_weights_in_pytorch_model,
)
from .models.albert import ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, AlbertConfig
from .models.audio_spectrogram_transformer import (
AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
ASTConfig,
)
from .models.auto import (
ALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
CONFIG_MAPPING,
FEATURE_EXTRACTOR_MAPPING,
IMAGE_PROCESSOR_MAPPING,
MODEL_NAMES_MAPPING,
PROCESSOR_MAPPING,
TOKENIZER_MAPPING,
AutoConfig,
AutoFeatureExtractor,
AutoImageProcessor,
AutoProcessor,
AutoTokenizer,
)
from .models.bart import BartConfig, BartTokenizer
from .models.beit import BEIT_PRETRAINED_CONFIG_ARCHIVE_MAP, BeitConfig
from .models.bert import (
BERT_PRETRAINED_CONFIG_ARCHIVE_MAP,
BasicTokenizer,
BertConfig,
BertTokenizer,
WordpieceTokenizer,
)
from .models.bert_generation import BertGenerationConfig
from .models.bert_japanese import BertJapaneseTokenizer, CharacterTokenizer, MecabTokenizer
from .models.bertweet import BertweetTokenizer
from .models.big_bird import BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP, BigBirdConfig
from .models.bigbird_pegasus import BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP, BigBirdPegasusConfig
from .models.biogpt import BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, BioGptConfig, BioGptTokenizer
from .models.bit import BIT_PRETRAINED_CONFIG_ARCHIVE_MAP, BitConfig
from .models.blenderbot import BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP, BlenderbotConfig, BlenderbotTokenizer
from .models.blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
BlenderbotSmallConfig,
BlenderbotSmallTokenizer,
)
from .models.bloom import BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP, BloomConfig
from .models.byt5 import ByT5Tokenizer
from .models.camembert import CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, CamembertConfig
from .models.canine import CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP, CanineConfig, CanineTokenizer
from .models.chinese_clip import (
CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
ChineseCLIPConfig,
ChineseCLIPProcessor,
ChineseCLIPTextConfig,
ChineseCLIPVisionConfig,
)
from .models.clip import (
CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
CLIPConfig,
CLIPProcessor,
CLIPTextConfig,
CLIPTokenizer,
CLIPVisionConfig,
)
from .models.clipseg import (
CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP,
CLIPSegConfig,
CLIPSegProcessor,
CLIPSegTextConfig,
CLIPSegVisionConfig,
)
from .models.codegen import CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP, CodeGenConfig, CodeGenTokenizer
from .models.conditional_detr import CONDITIONAL_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP, ConditionalDetrConfig
from .models.convbert import CONVBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, ConvBertConfig, ConvBertTokenizer
from .models.convnext import CONVNEXT_PRETRAINED_CONFIG_ARCHIVE_MAP, ConvNextConfig
from .models.ctrl import CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP, CTRLConfig, CTRLTokenizer
from .models.cvt import CVT_PRETRAINED_CONFIG_ARCHIVE_MAP, CvtConfig
from .models.data2vec import (
DATA2VEC_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP,
DATA2VEC_VISION_PRETRAINED_CONFIG_ARCHIVE_MAP,
Data2VecAudioConfig,
Data2VecTextConfig,
Data2VecVisionConfig,
)
from .models.deberta import DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaConfig, DebertaTokenizer
from .models.deberta_v2 import DEBERTA_V2_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaV2Config
from .models.decision_transformer import (
DECISION_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
DecisionTransformerConfig,
)
from .models.deformable_detr import DEFORMABLE_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP, DeformableDetrConfig
from .models.deit import DEIT_PRETRAINED_CONFIG_ARCHIVE_MAP, DeiTConfig
from .models.detr import DETR_PRETRAINED_CONFIG_ARCHIVE_MAP, DetrConfig
from .models.dinat import DINAT_PRETRAINED_CONFIG_ARCHIVE_MAP, DinatConfig
from .models.distilbert import DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, DistilBertConfig, DistilBertTokenizer
from .models.donut import DONUT_SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP, DonutProcessor, DonutSwinConfig
from .models.dpr import (
DPR_PRETRAINED_CONFIG_ARCHIVE_MAP,
DPRConfig,
DPRContextEncoderTokenizer,
DPRQuestionEncoderTokenizer,
DPRReaderOutput,
DPRReaderTokenizer,
)
from .models.dpt import DPT_PRETRAINED_CONFIG_ARCHIVE_MAP, DPTConfig
from .models.electra import ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, ElectraConfig, ElectraTokenizer
from .models.encoder_decoder import EncoderDecoderConfig
from .models.ernie import ERNIE_PRETRAINED_CONFIG_ARCHIVE_MAP, ErnieConfig
from .models.esm import ESM_PRETRAINED_CONFIG_ARCHIVE_MAP, EsmConfig, EsmTokenizer
from .models.flaubert import FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, FlaubertConfig, FlaubertTokenizer
from .models.flava import (
FLAVA_PRETRAINED_CONFIG_ARCHIVE_MAP,
FlavaConfig,
FlavaImageCodebookConfig,
FlavaImageConfig,
FlavaMultimodalConfig,
FlavaTextConfig,
)
from .models.fnet import FNET_PRETRAINED_CONFIG_ARCHIVE_MAP, FNetConfig
from .models.fsmt import FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP, FSMTConfig, FSMTTokenizer
from .models.funnel import FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP, FunnelConfig, FunnelTokenizer
from .models.glpn import GLPN_PRETRAINED_CONFIG_ARCHIVE_MAP, GLPNConfig
from .models.gpt2 import GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP, GPT2Config, GPT2Tokenizer
from .models.gpt_neo import GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoConfig
from .models.gpt_neox import GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXConfig
from .models.gpt_neox_japanese import GPT_NEOX_JAPANESE_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXJapaneseConfig
from .models.gptj import GPTJ_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTJConfig
from .models.groupvit import (
GROUPVIT_PRETRAINED_CONFIG_ARCHIVE_MAP,
GroupViTConfig,
GroupViTTextConfig,
GroupViTVisionConfig,
)
from .models.herbert import HerbertTokenizer
from .models.hubert import HUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, HubertConfig
from .models.ibert import IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, IBertConfig
from .models.imagegpt import IMAGEGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, ImageGPTConfig
from .models.jukebox import (
JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP,
JukeboxConfig,
JukeboxPriorConfig,
JukeboxTokenizer,
JukeboxVQVAEConfig,
)
from .models.layoutlm import LAYOUTLM_PRETRAINED_CONFIG_ARCHIVE_MAP, LayoutLMConfig, LayoutLMTokenizer
from .models.layoutlmv2 import (
LAYOUTLMV2_PRETRAINED_CONFIG_ARCHIVE_MAP,
LayoutLMv2Config,
LayoutLMv2FeatureExtractor,
LayoutLMv2ImageProcessor,
LayoutLMv2Processor,
LayoutLMv2Tokenizer,
)
from .models.layoutlmv3 import (
LAYOUTLMV3_PRETRAINED_CONFIG_ARCHIVE_MAP,
LayoutLMv3Config,
LayoutLMv3FeatureExtractor,
LayoutLMv3ImageProcessor,
LayoutLMv3Processor,
LayoutLMv3Tokenizer,
)
from .models.layoutxlm import LayoutXLMProcessor
from .models.led import LED_PRETRAINED_CONFIG_ARCHIVE_MAP, LEDConfig, LEDTokenizer
from .models.levit import LEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP, LevitConfig
from .models.lilt import LILT_PRETRAINED_CONFIG_ARCHIVE_MAP, LiltConfig
from .models.longformer import LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, LongformerConfig, LongformerTokenizer
from .models.longt5 import LONGT5_PRETRAINED_CONFIG_ARCHIVE_MAP, LongT5Config
from .models.luke import LUKE_PRETRAINED_CONFIG_ARCHIVE_MAP, LukeConfig, LukeTokenizer
from .models.lxmert import LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP, LxmertConfig, LxmertTokenizer
from .models.m2m_100 import M2M_100_PRETRAINED_CONFIG_ARCHIVE_MAP, M2M100Config
from .models.marian import MarianConfig
from .models.markuplm import (
MARKUPLM_PRETRAINED_CONFIG_ARCHIVE_MAP,
MarkupLMConfig,
MarkupLMFeatureExtractor,
MarkupLMProcessor,
MarkupLMTokenizer,
)
from .models.maskformer import MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskFormerConfig, MaskFormerSwinConfig
from .models.mbart import MBartConfig
from .models.mctct import MCTCT_PRETRAINED_CONFIG_ARCHIVE_MAP, MCTCTConfig, MCTCTProcessor
from .models.megatron_bert import MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, MegatronBertConfig
from .models.mmbt import MMBTConfig
from .models.mobilebert import MOBILEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileBertConfig, MobileBertTokenizer
from .models.mobilenet_v1 import MOBILENET_V1_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileNetV1Config
from .models.mobilenet_v2 import MOBILENET_V2_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileNetV2Config
from .models.mobilevit import MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileViTConfig
from .models.mpnet import MPNET_PRETRAINED_CONFIG_ARCHIVE_MAP, MPNetConfig, MPNetTokenizer
from .models.mt5 import MT5Config
from .models.mvp import MvpConfig, MvpTokenizer
from .models.nat import NAT_PRETRAINED_CONFIG_ARCHIVE_MAP, NatConfig
from .models.nezha import NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP, NezhaConfig
from .models.nystromformer import NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, NystromformerConfig
from .models.openai import OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP, OpenAIGPTConfig, OpenAIGPTTokenizer
from .models.opt import OPTConfig
from .models.owlvit import (
OWLVIT_PRETRAINED_CONFIG_ARCHIVE_MAP,
OwlViTConfig,
OwlViTProcessor,
OwlViTTextConfig,
OwlViTVisionConfig,
)
from .models.pegasus import PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP, PegasusConfig, PegasusTokenizer
from .models.pegasus_x import PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP, PegasusXConfig
from .models.perceiver import PERCEIVER_PRETRAINED_CONFIG_ARCHIVE_MAP, PerceiverConfig, PerceiverTokenizer
from .models.phobert import PhobertTokenizer
from .models.plbart import PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP, PLBartConfig
from .models.poolformer import POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, PoolFormerConfig
from .models.prophetnet import PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ProphetNetConfig, ProphetNetTokenizer
from .models.qdqbert import QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, QDQBertConfig
from .models.rag import RagConfig, RagRetriever, RagTokenizer
from .models.realm import REALM_PRETRAINED_CONFIG_ARCHIVE_MAP, RealmConfig, RealmTokenizer
from .models.reformer import REFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, ReformerConfig
from .models.regnet import REGNET_PRETRAINED_CONFIG_ARCHIVE_MAP, RegNetConfig
from .models.rembert import REMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, RemBertConfig
from .models.resnet import RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ResNetConfig
from .models.retribert import RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, RetriBertConfig, RetriBertTokenizer
from .models.roberta import ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, RobertaConfig, RobertaTokenizer
from .models.roc_bert import ROC_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, RoCBertConfig, RoCBertTokenizer
from .models.roformer import ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, RoFormerConfig, RoFormerTokenizer
from .models.segformer import SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, SegformerConfig
from .models.sew import SEW_PRETRAINED_CONFIG_ARCHIVE_MAP, SEWConfig
from .models.sew_d import SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP, SEWDConfig
from .models.speech_encoder_decoder import SpeechEncoderDecoderConfig
from .models.speech_to_text import SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP, Speech2TextConfig
from .models.speech_to_text_2 import (
SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP,
Speech2Text2Config,
Speech2Text2Processor,
Speech2Text2Tokenizer,
)
from .models.splinter import SPLINTER_PRETRAINED_CONFIG_ARCHIVE_MAP, SplinterConfig, SplinterTokenizer
from .models.squeezebert import SQUEEZEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, SqueezeBertConfig, SqueezeBertTokenizer
from .models.swin import SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP, SwinConfig
from .models.swinv2 import SWINV2_PRETRAINED_CONFIG_ARCHIVE_MAP, Swinv2Config
from .models.switch_transformers import SWITCH_TRANSFORMERS_PRETRAINED_CONFIG_ARCHIVE_MAP, SwitchTransformersConfig
from .models.t5 import T5_PRETRAINED_CONFIG_ARCHIVE_MAP, T5Config
from .models.table_transformer import TABLE_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, TableTransformerConfig
from .models.tapas import TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP, TapasConfig, TapasTokenizer
from .models.tapex import TapexTokenizer
from .models.time_series_transformer import (
TIME_SERIES_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
TimeSeriesTransformerConfig,
)
from .models.timesformer import TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, TimesformerConfig
from .models.trajectory_transformer import (
TRAJECTORY_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
TrajectoryTransformerConfig,
)
from .models.transfo_xl import (
TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP,
TransfoXLConfig,
TransfoXLCorpus,
TransfoXLTokenizer,
)
from .models.trocr import TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP, TrOCRConfig, TrOCRProcessor
from .models.unispeech import UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP, UniSpeechConfig
from .models.unispeech_sat import UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP, UniSpeechSatConfig
from .models.van import VAN_PRETRAINED_CONFIG_ARCHIVE_MAP, VanConfig
from .models.videomae import VIDEOMAE_PRETRAINED_CONFIG_ARCHIVE_MAP, VideoMAEConfig
from .models.vilt import (
VILT_PRETRAINED_CONFIG_ARCHIVE_MAP,
ViltConfig,
ViltFeatureExtractor,
ViltImageProcessor,
ViltProcessor,
)
from .models.vision_encoder_decoder import VisionEncoderDecoderConfig
from .models.vision_text_dual_encoder import VisionTextDualEncoderConfig, VisionTextDualEncoderProcessor
from .models.visual_bert import VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, VisualBertConfig
from .models.vit import VIT_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTConfig
from .models.vit_hybrid import VIT_HYBRID_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTHybridConfig
from .models.vit_mae import VIT_MAE_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTMAEConfig
from .models.vit_msn import VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTMSNConfig
from .models.wav2vec2 import (
WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP,
Wav2Vec2Config,
Wav2Vec2CTCTokenizer,
Wav2Vec2FeatureExtractor,
Wav2Vec2Processor,
Wav2Vec2Tokenizer,
)
from .models.wav2vec2_conformer import WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, Wav2Vec2ConformerConfig
from .models.wav2vec2_phoneme import Wav2Vec2PhonemeCTCTokenizer
from .models.wav2vec2_with_lm import Wav2Vec2ProcessorWithLM
from .models.wavlm import WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP, WavLMConfig
from .models.whisper import (
WHISPER_PRETRAINED_CONFIG_ARCHIVE_MAP,
WhisperConfig,
WhisperFeatureExtractor,
WhisperProcessor,
WhisperTokenizer,
)
from .models.x_clip import (
XCLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
XCLIPConfig,
XCLIPProcessor,
XCLIPTextConfig,
XCLIPVisionConfig,
)
from .models.xglm import XGLM_PRETRAINED_CONFIG_ARCHIVE_MAP, XGLMConfig
from .models.xlm import XLM_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMConfig, XLMTokenizer
from .models.xlm_prophetnet import XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMProphetNetConfig
from .models.xlm_roberta import XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMRobertaConfig
from .models.xlm_roberta_xl import XLM_ROBERTA_XL_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMRobertaXLConfig
from .models.xlnet import XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP, XLNetConfig
from .models.yolos import YOLOS_PRETRAINED_CONFIG_ARCHIVE_MAP, YolosConfig
from .models.yoso import YOSO_PRETRAINED_CONFIG_ARCHIVE_MAP, YosoConfig
# Pipelines
from .pipelines import (
AudioClassificationPipeline,
AutomaticSpeechRecognitionPipeline,
Conversation,
ConversationalPipeline,
CsvPipelineDataFormat,
DepthEstimationPipeline,
DocumentQuestionAnsweringPipeline,
FeatureExtractionPipeline,
FillMaskPipeline,
ImageClassificationPipeline,
ImageSegmentationPipeline,
ImageToTextPipeline,
JsonPipelineDataFormat,
NerPipeline,
ObjectDetectionPipeline,
PipedPipelineDataFormat,
Pipeline,
PipelineDataFormat,
QuestionAnsweringPipeline,
SummarizationPipeline,
TableQuestionAnsweringPipeline,
Text2TextGenerationPipeline,
TextClassificationPipeline,
TextGenerationPipeline,
TokenClassificationPipeline,
TranslationPipeline,
VideoClassificationPipeline,
VisualQuestionAnsweringPipeline,
ZeroShotClassificationPipeline,
ZeroShotImageClassificationPipeline,
ZeroShotObjectDetectionPipeline,
pipeline,
)
from .processing_utils import ProcessorMixin
# Tokenization
from .tokenization_utils import PreTrainedTokenizer
from .tokenization_utils_base import (
AddedToken,
BatchEncoding,
CharSpan,
PreTrainedTokenizerBase,
SpecialTokensMixin,
TokenSpan,
)
# Trainer
from .trainer_callback import (
DefaultFlowCallback,
EarlyStoppingCallback,
PrinterCallback,
ProgressCallback,
TrainerCallback,
TrainerControl,
TrainerState,
)
from .trainer_utils import EvalPrediction, IntervalStrategy, SchedulerType, enable_full_determinism, set_seed
from .training_args import TrainingArguments
from .training_args_seq2seq import Seq2SeqTrainingArguments
from .training_args_tf import TFTrainingArguments
# Files and general utilities
from .utils import (
CONFIG_NAME,
MODEL_CARD_NAME,
PYTORCH_PRETRAINED_BERT_CACHE,
PYTORCH_TRANSFORMERS_CACHE,
SPIECE_UNDERLINE,
TF2_WEIGHTS_NAME,
TF_WEIGHTS_NAME,
TRANSFORMERS_CACHE,
WEIGHTS_NAME,
TensorType,
add_end_docstrings,
add_start_docstrings,
is_apex_available,
is_datasets_available,
is_decord_available,
is_faiss_available,
is_flax_available,
is_keras_nlp_available,
is_phonemizer_available,
is_psutil_available,
is_py3nvml_available,
is_pyctcdecode_available,
is_safetensors_available,
is_scipy_available,
is_sentencepiece_available,
is_sklearn_available,
is_speech_available,
is_tensorflow_text_available,
is_tf_available,
is_timm_available,
is_tokenizers_available,
is_torch_available,
is_torch_tpu_available,
is_vision_available,
logging,
)
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_sentencepiece_objects import *
else:
from .models.albert import AlbertTokenizer
from .models.barthez import BarthezTokenizer
from .models.bartpho import BartphoTokenizer
from .models.bert_generation import BertGenerationTokenizer
from .models.big_bird import BigBirdTokenizer
from .models.camembert import CamembertTokenizer
from .models.cpm import CpmTokenizer
from .models.deberta_v2 import DebertaV2Tokenizer
from .models.fnet import FNetTokenizer
from .models.layoutxlm import LayoutXLMTokenizer
from .models.m2m_100 import M2M100Tokenizer
from .models.marian import MarianTokenizer
from .models.mbart import MBart50Tokenizer, MBartTokenizer
from .models.mluke import MLukeTokenizer
from .models.mt5 import MT5Tokenizer
from .models.nllb import NllbTokenizer
from .models.pegasus import PegasusTokenizer
from .models.plbart import PLBartTokenizer
from .models.reformer import ReformerTokenizer
from .models.rembert import RemBertTokenizer
from .models.speech_to_text import Speech2TextTokenizer
from .models.t5 import T5Tokenizer
from .models.xglm import XGLMTokenizer
from .models.xlm_prophetnet import XLMProphetNetTokenizer
from .models.xlm_roberta import XLMRobertaTokenizer
from .models.xlnet import XLNetTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_tokenizers_objects import *
else:
# Fast tokenizers imports
from .models.albert import AlbertTokenizerFast
from .models.bart import BartTokenizerFast
from .models.barthez import BarthezTokenizerFast
from .models.bert import BertTokenizerFast
from .models.big_bird import BigBirdTokenizerFast
from .models.blenderbot import BlenderbotTokenizerFast
from .models.blenderbot_small import BlenderbotSmallTokenizerFast
from .models.bloom import BloomTokenizerFast
from .models.camembert import CamembertTokenizerFast
from .models.clip import CLIPTokenizerFast
from .models.codegen import CodeGenTokenizerFast
from .models.convbert import ConvBertTokenizerFast
from .models.cpm import CpmTokenizerFast
from .models.deberta import DebertaTokenizerFast
from .models.deberta_v2 import DebertaV2TokenizerFast
from .models.distilbert import DistilBertTokenizerFast
from .models.dpr import DPRContextEncoderTokenizerFast, DPRQuestionEncoderTokenizerFast, DPRReaderTokenizerFast
from .models.electra import ElectraTokenizerFast
from .models.fnet import FNetTokenizerFast
from .models.funnel import FunnelTokenizerFast
from .models.gpt2 import GPT2TokenizerFast
from .models.gpt_neox import GPTNeoXTokenizerFast
from .models.gpt_neox_japanese import GPTNeoXJapaneseTokenizer
from .models.herbert import HerbertTokenizerFast
from .models.layoutlm import LayoutLMTokenizerFast
from .models.layoutlmv2 import LayoutLMv2TokenizerFast
from .models.layoutlmv3 import LayoutLMv3TokenizerFast
from .models.layoutxlm import LayoutXLMTokenizerFast
from .models.led import LEDTokenizerFast
from .models.longformer import LongformerTokenizerFast
from .models.lxmert import LxmertTokenizerFast
from .models.markuplm import MarkupLMTokenizerFast
from .models.mbart import MBartTokenizerFast
from .models.mbart50 import MBart50TokenizerFast
from .models.mobilebert import MobileBertTokenizerFast
from .models.mpnet import MPNetTokenizerFast
from .models.mt5 import MT5TokenizerFast
from .models.mvp import MvpTokenizerFast
from .models.nllb import NllbTokenizerFast
from .models.openai import OpenAIGPTTokenizerFast
from .models.pegasus import PegasusTokenizerFast
from .models.realm import RealmTokenizerFast
from .models.reformer import ReformerTokenizerFast
from .models.rembert import RemBertTokenizerFast
from .models.retribert import RetriBertTokenizerFast
from .models.roberta import RobertaTokenizerFast
from .models.roformer import RoFormerTokenizerFast
from .models.splinter import SplinterTokenizerFast
from .models.squeezebert import SqueezeBertTokenizerFast
from .models.t5 import T5TokenizerFast
from .models.xglm import XGLMTokenizerFast
from .models.xlm_roberta import XLMRobertaTokenizerFast
from .models.xlnet import XLNetTokenizerFast
from .tokenization_utils_fast import PreTrainedTokenizerFast
try:
if not (is_sentencepiece_available() and is_tokenizers_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummies_sentencepiece_and_tokenizers_objects import *
else:
from .convert_slow_tokenizer import SLOW_TO_FAST_CONVERTERS, convert_slow_tokenizer
try:
if not is_speech_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_speech_objects import *
else:
from .models.audio_spectrogram_transformer import ASTFeatureExtractor
from .models.mctct import MCTCTFeatureExtractor
from .models.speech_to_text import Speech2TextFeatureExtractor
try:
if not is_tensorflow_text_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_tensorflow_text_objects import *
else:
from .models.bert import TFBertTokenizer
try:
if not is_keras_nlp_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_keras_nlp_objects import *
else:
from .models.gpt2 import TFGPT2Tokenizer
try:
if not (is_speech_available() and is_sentencepiece_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_sentencepiece_and_speech_objects import *
else:
from .models.speech_to_text import Speech2TextProcessor
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_vision_objects import *
else:
from .image_processing_utils import ImageProcessingMixin
from .image_transforms import rescale, resize, to_pil_image
from .image_utils import ImageFeatureExtractionMixin
from .models.beit import BeitFeatureExtractor, BeitImageProcessor
from .models.bit import BitImageProcessor
from .models.chinese_clip import ChineseCLIPFeatureExtractor, ChineseCLIPImageProcessor
from .models.clip import CLIPFeatureExtractor, CLIPImageProcessor
from .models.conditional_detr import ConditionalDetrFeatureExtractor, ConditionalDetrImageProcessor
from .models.convnext import ConvNextFeatureExtractor, ConvNextImageProcessor
from .models.deformable_detr import DeformableDetrFeatureExtractor, DeformableDetrImageProcessor
from .models.deit import DeiTFeatureExtractor, DeiTImageProcessor
from .models.detr import DetrFeatureExtractor, DetrImageProcessor
from .models.donut import DonutFeatureExtractor, DonutImageProcessor
from .models.dpt import DPTFeatureExtractor, DPTImageProcessor
from .models.flava import FlavaFeatureExtractor, FlavaImageProcessor, FlavaProcessor
from .models.glpn import GLPNFeatureExtractor, GLPNImageProcessor
from .models.imagegpt import ImageGPTFeatureExtractor, ImageGPTImageProcessor
from .models.layoutlmv2 import LayoutLMv2FeatureExtractor, LayoutLMv2ImageProcessor
from .models.layoutlmv3 import LayoutLMv3FeatureExtractor, LayoutLMv3ImageProcessor
from .models.levit import LevitFeatureExtractor, LevitImageProcessor
from .models.maskformer import MaskFormerFeatureExtractor, MaskFormerImageProcessor
from .models.mobilenet_v1 import MobileNetV1FeatureExtractor, MobileNetV1ImageProcessor
from .models.mobilenet_v2 import MobileNetV2FeatureExtractor, MobileNetV2ImageProcessor
from .models.mobilevit import MobileViTFeatureExtractor, MobileViTImageProcessor
from .models.owlvit import OwlViTFeatureExtractor, OwlViTImageProcessor
from .models.perceiver import PerceiverFeatureExtractor, PerceiverImageProcessor
from .models.poolformer import PoolFormerFeatureExtractor, PoolFormerImageProcessor
from .models.segformer import SegformerFeatureExtractor, SegformerImageProcessor
from .models.videomae import VideoMAEFeatureExtractor, VideoMAEImageProcessor
from .models.vilt import ViltFeatureExtractor, ViltImageProcessor, ViltProcessor
from .models.vit import ViTFeatureExtractor, ViTImageProcessor
from .models.vit_hybrid import ViTHybridImageProcessor
from .models.yolos import YolosFeatureExtractor, YolosImageProcessor
# Modeling
try:
if not (is_timm_available() and is_vision_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_timm_and_vision_objects import *
else:
from .models.conditional_detr import (
CONDITIONAL_DETR_PRETRAINED_MODEL_ARCHIVE_LIST,
ConditionalDetrForObjectDetection,
ConditionalDetrForSegmentation,
ConditionalDetrModel,
ConditionalDetrPreTrainedModel,
)
from .models.deformable_detr import (
DEFORMABLE_DETR_PRETRAINED_MODEL_ARCHIVE_LIST,
DeformableDetrForObjectDetection,
DeformableDetrModel,
DeformableDetrPreTrainedModel,
)
from .models.detr import (
DETR_PRETRAINED_MODEL_ARCHIVE_LIST,
DetrForObjectDetection,
DetrForSegmentation,
DetrModel,
DetrPreTrainedModel,
)
from .models.table_transformer import (
TABLE_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TableTransformerForObjectDetection,
TableTransformerModel,
TableTransformerPreTrainedModel,
)
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_pt_objects import *
else:
# Benchmarks
from .benchmark.benchmark import PyTorchBenchmark
from .benchmark.benchmark_args import PyTorchBenchmarkArguments
from .data.datasets import (
GlueDataset,
GlueDataTrainingArguments,
LineByLineTextDataset,
LineByLineWithRefDataset,
LineByLineWithSOPTextDataset,
SquadDataset,
SquadDataTrainingArguments,
TextDataset,
TextDatasetForNextSentencePrediction,
)
from .generation import (
BeamScorer,
BeamSearchScorer,
ConstrainedBeamSearchScorer,
Constraint,
ConstraintListState,
DisjunctiveConstraint,
ForcedBOSTokenLogitsProcessor,
ForcedEOSTokenLogitsProcessor,
GenerationMixin,
HammingDiversityLogitsProcessor,
InfNanRemoveLogitsProcessor,
LogitsProcessor,
LogitsProcessorList,
LogitsWarper,
MaxLengthCriteria,
MaxTimeCriteria,
MinLengthLogitsProcessor,
NoBadWordsLogitsProcessor,
NoRepeatNGramLogitsProcessor,
PhrasalConstraint,
PrefixConstrainedLogitsProcessor,
RepetitionPenaltyLogitsProcessor,
StoppingCriteria,
StoppingCriteriaList,
TemperatureLogitsWarper,
TopKLogitsWarper,
TopPLogitsWarper,
TypicalLogitsWarper,
top_k_top_p_filtering,
)
from .modeling_utils import PreTrainedModel
# PyTorch model imports
from .models.albert import (
ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
AlbertForMaskedLM,
AlbertForMultipleChoice,
AlbertForPreTraining,
AlbertForQuestionAnswering,
AlbertForSequenceClassification,
AlbertForTokenClassification,
AlbertModel,
AlbertPreTrainedModel,
load_tf_weights_in_albert,
)
from .models.audio_spectrogram_transformer import (
AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
ASTForAudioClassification,
ASTModel,
ASTPreTrainedModel,
)
from .models.auto import (
MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING,
MODEL_FOR_AUDIO_XVECTOR_MAPPING,
MODEL_FOR_BACKBONE_MAPPING,
MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING,
MODEL_FOR_CAUSAL_LM_MAPPING,
MODEL_FOR_CTC_MAPPING,
MODEL_FOR_DEPTH_ESTIMATION_MAPPING,
MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
MODEL_FOR_IMAGE_SEGMENTATION_MAPPING,
MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING,
MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
MODEL_FOR_MULTIPLE_CHOICE_MAPPING,
MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING,
MODEL_FOR_OBJECT_DETECTION_MAPPING,
MODEL_FOR_PRETRAINING_MAPPING,
MODEL_FOR_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING,
MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING,
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING,
MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
MODEL_FOR_VISION_2_SEQ_MAPPING,
MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING,
MODEL_MAPPING,
MODEL_WITH_LM_HEAD_MAPPING,
AutoBackbone,
AutoModel,
AutoModelForAudioClassification,
AutoModelForAudioFrameClassification,
AutoModelForAudioXVector,
AutoModelForCausalLM,
AutoModelForCTC,
AutoModelForDepthEstimation,
AutoModelForDocumentQuestionAnswering,
AutoModelForImageClassification,
AutoModelForImageSegmentation,
AutoModelForInstanceSegmentation,
AutoModelForMaskedImageModeling,
AutoModelForMaskedLM,
AutoModelForMultipleChoice,
AutoModelForNextSentencePrediction,
AutoModelForObjectDetection,
AutoModelForPreTraining,
AutoModelForQuestionAnswering,
AutoModelForSemanticSegmentation,
AutoModelForSeq2SeqLM,
AutoModelForSequenceClassification,
AutoModelForSpeechSeq2Seq,
AutoModelForTableQuestionAnswering,
AutoModelForTokenClassification,
AutoModelForVideoClassification,
AutoModelForVision2Seq,
AutoModelForVisualQuestionAnswering,
AutoModelForZeroShotObjectDetection,
AutoModelWithLMHead,
)
from .models.bart import (
BART_PRETRAINED_MODEL_ARCHIVE_LIST,
BartForCausalLM,
BartForConditionalGeneration,
BartForQuestionAnswering,
BartForSequenceClassification,
BartModel,
BartPretrainedModel,
PretrainedBartModel,
)
from .models.beit import (
BEIT_PRETRAINED_MODEL_ARCHIVE_LIST,
BeitForImageClassification,
BeitForMaskedImageModeling,
BeitForSemanticSegmentation,
BeitModel,
BeitPreTrainedModel,
)
from .models.bert import (
BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
BertForMaskedLM,
BertForMultipleChoice,
BertForNextSentencePrediction,
BertForPreTraining,
BertForQuestionAnswering,
BertForSequenceClassification,
BertForTokenClassification,
BertLayer,
BertLMHeadModel,
BertModel,
BertPreTrainedModel,
load_tf_weights_in_bert,
)
from .models.bert_generation import (
BertGenerationDecoder,
BertGenerationEncoder,
BertGenerationPreTrainedModel,
load_tf_weights_in_bert_generation,
)
from .models.big_bird import (
BIG_BIRD_PRETRAINED_MODEL_ARCHIVE_LIST,
BigBirdForCausalLM,
BigBirdForMaskedLM,
BigBirdForMultipleChoice,
BigBirdForPreTraining,
BigBirdForQuestionAnswering,
BigBirdForSequenceClassification,
BigBirdForTokenClassification,
BigBirdLayer,
BigBirdModel,
BigBirdPreTrainedModel,
load_tf_weights_in_big_bird,
)
from .models.bigbird_pegasus import (
BIGBIRD_PEGASUS_PRETRAINED_MODEL_ARCHIVE_LIST,
BigBirdPegasusForCausalLM,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForQuestionAnswering,
BigBirdPegasusForSequenceClassification,
BigBirdPegasusModel,
BigBirdPegasusPreTrainedModel,
)
from .models.biogpt import (
BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST,
BioGptForCausalLM,
BioGptModel,
BioGptPreTrainedModel,
)
from .models.bit import (
BIT_PRETRAINED_MODEL_ARCHIVE_LIST,
BitBackbone,
BitForImageClassification,
BitModel,
BitPreTrainedModel,
)
from .models.blenderbot import (
BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotForCausalLM,
BlenderbotForConditionalGeneration,
BlenderbotModel,
BlenderbotPreTrainedModel,
)
from .models.blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotSmallForCausalLM,
BlenderbotSmallForConditionalGeneration,
BlenderbotSmallModel,
BlenderbotSmallPreTrainedModel,
)
from .models.bloom import (
BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST,
BloomForCausalLM,
BloomForQuestionAnswering,
BloomForSequenceClassification,
BloomForTokenClassification,
BloomModel,
BloomPreTrainedModel,
)
from .models.camembert import (
CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
CamembertForCausalLM,
CamembertForMaskedLM,
CamembertForMultipleChoice,
CamembertForQuestionAnswering,
CamembertForSequenceClassification,
CamembertForTokenClassification,
CamembertModel,
CamembertPreTrainedModel,
)
from .models.canine import (
CANINE_PRETRAINED_MODEL_ARCHIVE_LIST,
CanineForMultipleChoice,
CanineForQuestionAnswering,
CanineForSequenceClassification,
CanineForTokenClassification,
CanineLayer,
CanineModel,
CaninePreTrainedModel,
load_tf_weights_in_canine,
)
from .models.chinese_clip import (
CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
ChineseCLIPModel,
ChineseCLIPPreTrainedModel,
ChineseCLIPTextModel,
ChineseCLIPVisionModel,
)
from .models.clip import (
CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
CLIPModel,
CLIPPreTrainedModel,
CLIPTextModel,
CLIPTextModelWithProjection,
CLIPVisionModel,
CLIPVisionModelWithProjection,
)
from .models.clipseg import (
CLIPSEG_PRETRAINED_MODEL_ARCHIVE_LIST,
CLIPSegForImageSegmentation,
CLIPSegModel,
CLIPSegPreTrainedModel,
CLIPSegTextModel,
CLIPSegVisionModel,
)
from .models.codegen import (
CODEGEN_PRETRAINED_MODEL_ARCHIVE_LIST,
CodeGenForCausalLM,
CodeGenModel,
CodeGenPreTrainedModel,
)
from .models.convbert import (
CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
ConvBertForMaskedLM,
ConvBertForMultipleChoice,
ConvBertForQuestionAnswering,
ConvBertForSequenceClassification,
ConvBertForTokenClassification,
ConvBertLayer,
ConvBertModel,
ConvBertPreTrainedModel,
load_tf_weights_in_convbert,
)
from .models.convnext import (
CONVNEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
ConvNextForImageClassification,
ConvNextModel,
ConvNextPreTrainedModel,
)
from .models.ctrl import (
CTRL_PRETRAINED_MODEL_ARCHIVE_LIST,
CTRLForSequenceClassification,
CTRLLMHeadModel,
CTRLModel,
CTRLPreTrainedModel,
)
from .models.cvt import (
CVT_PRETRAINED_MODEL_ARCHIVE_LIST,
CvtForImageClassification,
CvtModel,
CvtPreTrainedModel,
)
from .models.data2vec import (
DATA2VEC_AUDIO_PRETRAINED_MODEL_ARCHIVE_LIST,
DATA2VEC_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
DATA2VEC_VISION_PRETRAINED_MODEL_ARCHIVE_LIST,
Data2VecAudioForAudioFrameClassification,
Data2VecAudioForCTC,
Data2VecAudioForSequenceClassification,
Data2VecAudioForXVector,
Data2VecAudioModel,
Data2VecAudioPreTrainedModel,
Data2VecTextForCausalLM,
Data2VecTextForMaskedLM,
Data2VecTextForMultipleChoice,
Data2VecTextForQuestionAnswering,
Data2VecTextForSequenceClassification,
Data2VecTextForTokenClassification,
Data2VecTextModel,
Data2VecTextPreTrainedModel,
Data2VecVisionForImageClassification,
Data2VecVisionForSemanticSegmentation,
Data2VecVisionModel,
Data2VecVisionPreTrainedModel,
)
from .models.deberta import (
DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
DebertaForMaskedLM,
DebertaForQuestionAnswering,
DebertaForSequenceClassification,
DebertaForTokenClassification,
DebertaModel,
DebertaPreTrainedModel,
)
from .models.deberta_v2 import (
DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST,
DebertaV2ForMaskedLM,
DebertaV2ForMultipleChoice,
DebertaV2ForQuestionAnswering,
DebertaV2ForSequenceClassification,
DebertaV2ForTokenClassification,
DebertaV2Model,
DebertaV2PreTrainedModel,
)
from .models.decision_transformer import (
DECISION_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
DecisionTransformerGPT2Model,
DecisionTransformerGPT2PreTrainedModel,
DecisionTransformerModel,
DecisionTransformerPreTrainedModel,
)
from .models.deit import (
DEIT_PRETRAINED_MODEL_ARCHIVE_LIST,
DeiTForImageClassification,
DeiTForImageClassificationWithTeacher,
DeiTForMaskedImageModeling,
DeiTModel,
DeiTPreTrainedModel,
)
from .models.dinat import (
DINAT_PRETRAINED_MODEL_ARCHIVE_LIST,
DinatForImageClassification,
DinatModel,
DinatPreTrainedModel,
)
from .models.distilbert import (
DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
DistilBertForMaskedLM,
DistilBertForMultipleChoice,
DistilBertForQuestionAnswering,
DistilBertForSequenceClassification,
DistilBertForTokenClassification,
DistilBertModel,
DistilBertPreTrainedModel,
)
from .models.donut import DONUT_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST, DonutSwinModel, DonutSwinPreTrainedModel
from .models.dpr import (
DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST,
DPRContextEncoder,
DPRPretrainedContextEncoder,
DPRPreTrainedModel,
DPRPretrainedQuestionEncoder,
DPRPretrainedReader,
DPRQuestionEncoder,
DPRReader,
)
from .models.dpt import (
DPT_PRETRAINED_MODEL_ARCHIVE_LIST,
DPTForDepthEstimation,
DPTForSemanticSegmentation,
DPTModel,
DPTPreTrainedModel,
)
from .models.electra import (
ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
ElectraForCausalLM,
ElectraForMaskedLM,
ElectraForMultipleChoice,
ElectraForPreTraining,
ElectraForQuestionAnswering,
ElectraForSequenceClassification,
ElectraForTokenClassification,
ElectraModel,
ElectraPreTrainedModel,
load_tf_weights_in_electra,
)
from .models.encoder_decoder import EncoderDecoderModel
from .models.ernie import (
ERNIE_PRETRAINED_MODEL_ARCHIVE_LIST,
ErnieForCausalLM,
ErnieForMaskedLM,
ErnieForMultipleChoice,
ErnieForNextSentencePrediction,
ErnieForPreTraining,
ErnieForQuestionAnswering,
ErnieForSequenceClassification,
ErnieForTokenClassification,
ErnieModel,
ErniePreTrainedModel,
)
from .models.esm import (
ESM_PRETRAINED_MODEL_ARCHIVE_LIST,
EsmFoldPreTrainedModel,
EsmForMaskedLM,
EsmForProteinFolding,
EsmForSequenceClassification,
EsmForTokenClassification,
EsmModel,
EsmPreTrainedModel,
)
from .models.flaubert import (
FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
FlaubertForMultipleChoice,
FlaubertForQuestionAnswering,
FlaubertForQuestionAnsweringSimple,
FlaubertForSequenceClassification,
FlaubertForTokenClassification,
FlaubertModel,
FlaubertPreTrainedModel,
FlaubertWithLMHeadModel,
)
from .models.flava import (
FLAVA_PRETRAINED_MODEL_ARCHIVE_LIST,
FlavaForPreTraining,
FlavaImageCodebook,
FlavaImageModel,
FlavaModel,
FlavaMultimodalModel,
FlavaPreTrainedModel,
FlavaTextModel,
)
from .models.fnet import (
FNET_PRETRAINED_MODEL_ARCHIVE_LIST,
FNetForMaskedLM,
FNetForMultipleChoice,
FNetForNextSentencePrediction,
FNetForPreTraining,
FNetForQuestionAnswering,
FNetForSequenceClassification,
FNetForTokenClassification,
FNetLayer,
FNetModel,
FNetPreTrainedModel,
)
from .models.fsmt import FSMTForConditionalGeneration, FSMTModel, PretrainedFSMTModel
from .models.funnel import (
FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST,
FunnelBaseModel,
FunnelForMaskedLM,
FunnelForMultipleChoice,
FunnelForPreTraining,
FunnelForQuestionAnswering,
FunnelForSequenceClassification,
FunnelForTokenClassification,
FunnelModel,
FunnelPreTrainedModel,
load_tf_weights_in_funnel,
)
from .models.glpn import (
GLPN_PRETRAINED_MODEL_ARCHIVE_LIST,
GLPNForDepthEstimation,
GLPNModel,
GLPNPreTrainedModel,
)
from .models.gpt2 import (
GPT2_PRETRAINED_MODEL_ARCHIVE_LIST,
GPT2DoubleHeadsModel,
GPT2ForSequenceClassification,
GPT2ForTokenClassification,
GPT2LMHeadModel,
GPT2Model,
GPT2PreTrainedModel,
load_tf_weights_in_gpt2,
)
from .models.gpt_neo import (
GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoForCausalLM,
GPTNeoForSequenceClassification,
GPTNeoModel,
GPTNeoPreTrainedModel,
load_tf_weights_in_gpt_neo,
)
from .models.gpt_neox import (
GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoXForCausalLM,
GPTNeoXLayer,
GPTNeoXModel,
GPTNeoXPreTrainedModel,
)
from .models.gpt_neox_japanese import (
GPT_NEOX_JAPANESE_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoXJapaneseForCausalLM,
GPTNeoXJapaneseLayer,
GPTNeoXJapaneseModel,
GPTNeoXJapanesePreTrainedModel,
)
from .models.gptj import (
GPTJ_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTJForCausalLM,
GPTJForQuestionAnswering,
GPTJForSequenceClassification,
GPTJModel,
GPTJPreTrainedModel,
)
from .models.groupvit import (
GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
GroupViTModel,
GroupViTPreTrainedModel,
GroupViTTextModel,
GroupViTVisionModel,
)
from .models.hubert import (
HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
HubertForCTC,
HubertForSequenceClassification,
HubertModel,
HubertPreTrainedModel,
)
from .models.ibert import (
IBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
IBertForMaskedLM,
IBertForMultipleChoice,
IBertForQuestionAnswering,
IBertForSequenceClassification,
IBertForTokenClassification,
IBertModel,
IBertPreTrainedModel,
)
from .models.imagegpt import (
IMAGEGPT_PRETRAINED_MODEL_ARCHIVE_LIST,
ImageGPTForCausalImageModeling,
ImageGPTForImageClassification,
ImageGPTModel,
ImageGPTPreTrainedModel,
load_tf_weights_in_imagegpt,
)
from .models.jukebox import (
JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST,
JukeboxModel,
JukeboxPreTrainedModel,
JukeboxPrior,
JukeboxVQVAE,
)
from .models.layoutlm import (
LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST,
LayoutLMForMaskedLM,
LayoutLMForQuestionAnswering,
LayoutLMForSequenceClassification,
LayoutLMForTokenClassification,
LayoutLMModel,
LayoutLMPreTrainedModel,
)
from .models.layoutlmv2 import (
LAYOUTLMV2_PRETRAINED_MODEL_ARCHIVE_LIST,
LayoutLMv2ForQuestionAnswering,
LayoutLMv2ForSequenceClassification,
LayoutLMv2ForTokenClassification,
LayoutLMv2Model,
LayoutLMv2PreTrainedModel,
)
from .models.layoutlmv3 import (
LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST,
LayoutLMv3ForQuestionAnswering,
LayoutLMv3ForSequenceClassification,
LayoutLMv3ForTokenClassification,
LayoutLMv3Model,
LayoutLMv3PreTrainedModel,
)
from .models.led import (
LED_PRETRAINED_MODEL_ARCHIVE_LIST,
LEDForConditionalGeneration,
LEDForQuestionAnswering,
LEDForSequenceClassification,
LEDModel,
LEDPreTrainedModel,
)
from .models.levit import (
LEVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
LevitForImageClassification,
LevitForImageClassificationWithTeacher,
LevitModel,
LevitPreTrainedModel,
)
from .models.lilt import (
LILT_PRETRAINED_MODEL_ARCHIVE_LIST,
LiltForQuestionAnswering,
LiltForSequenceClassification,
LiltForTokenClassification,
LiltModel,
LiltPreTrainedModel,
)
from .models.longformer import (
LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
LongformerForMaskedLM,
LongformerForMultipleChoice,
LongformerForQuestionAnswering,
LongformerForSequenceClassification,
LongformerForTokenClassification,
LongformerModel,
LongformerPreTrainedModel,
LongformerSelfAttention,
)
from .models.longt5 import (
LONGT5_PRETRAINED_MODEL_ARCHIVE_LIST,
LongT5EncoderModel,
LongT5ForConditionalGeneration,
LongT5Model,
LongT5PreTrainedModel,
)
from .models.luke import (
LUKE_PRETRAINED_MODEL_ARCHIVE_LIST,
LukeForEntityClassification,
LukeForEntityPairClassification,
LukeForEntitySpanClassification,
LukeForMaskedLM,
LukeForMultipleChoice,
LukeForQuestionAnswering,
LukeForSequenceClassification,
LukeForTokenClassification,
LukeModel,
LukePreTrainedModel,
)
from .models.lxmert import (
LxmertEncoder,
LxmertForPreTraining,
LxmertForQuestionAnswering,
LxmertModel,
LxmertPreTrainedModel,
LxmertVisualFeatureEncoder,
LxmertXLayer,
)
from .models.m2m_100 import (
M2M_100_PRETRAINED_MODEL_ARCHIVE_LIST,
M2M100ForConditionalGeneration,
M2M100Model,
M2M100PreTrainedModel,
)
from .models.marian import MarianForCausalLM, MarianModel, MarianMTModel
from .models.markuplm import (
MARKUPLM_PRETRAINED_MODEL_ARCHIVE_LIST,
MarkupLMForQuestionAnswering,
MarkupLMForSequenceClassification,
MarkupLMForTokenClassification,
MarkupLMModel,
MarkupLMPreTrainedModel,
)
from .models.maskformer import (
MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
MaskFormerForInstanceSegmentation,
MaskFormerModel,
MaskFormerPreTrainedModel,
MaskFormerSwinBackbone,
)
from .models.mbart import (
MBartForCausalLM,
MBartForConditionalGeneration,
MBartForQuestionAnswering,
MBartForSequenceClassification,
MBartModel,
MBartPreTrainedModel,
)
from .models.mctct import MCTCT_PRETRAINED_MODEL_ARCHIVE_LIST, MCTCTForCTC, MCTCTModel, MCTCTPreTrainedModel
from .models.megatron_bert import (
MEGATRON_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
MegatronBertForCausalLM,
MegatronBertForMaskedLM,
MegatronBertForMultipleChoice,
MegatronBertForNextSentencePrediction,
MegatronBertForPreTraining,
MegatronBertForQuestionAnswering,
MegatronBertForSequenceClassification,
MegatronBertForTokenClassification,
MegatronBertModel,
MegatronBertPreTrainedModel,
)
from .models.mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
from .models.mobilebert import (
MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
MobileBertLayer,
MobileBertModel,
MobileBertPreTrainedModel,
load_tf_weights_in_mobilebert,
)
from .models.mobilenet_v1 import (
MOBILENET_V1_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileNetV1ForImageClassification,
MobileNetV1Model,
MobileNetV1PreTrainedModel,
load_tf_weights_in_mobilenet_v1,
)
from .models.mobilenet_v2 import (
MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileNetV2ForImageClassification,
MobileNetV2ForSemanticSegmentation,
MobileNetV2Model,
MobileNetV2PreTrainedModel,
load_tf_weights_in_mobilenet_v2,
)
from .models.mobilevit import (
MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileViTForImageClassification,
MobileViTForSemanticSegmentation,
MobileViTModel,
MobileViTPreTrainedModel,
)
from .models.mpnet import (
MPNET_PRETRAINED_MODEL_ARCHIVE_LIST,
MPNetForMaskedLM,
MPNetForMultipleChoice,
MPNetForQuestionAnswering,
MPNetForSequenceClassification,
MPNetForTokenClassification,
MPNetLayer,
MPNetModel,
MPNetPreTrainedModel,
)
from .models.mt5 import MT5EncoderModel, MT5ForConditionalGeneration, MT5Model
from .models.mvp import (
MVP_PRETRAINED_MODEL_ARCHIVE_LIST,
MvpForCausalLM,
MvpForConditionalGeneration,
MvpForQuestionAnswering,
MvpForSequenceClassification,
MvpModel,
MvpPreTrainedModel,
)
from .models.nat import (
NAT_PRETRAINED_MODEL_ARCHIVE_LIST,
NatForImageClassification,
NatModel,
NatPreTrainedModel,
)
from .models.nezha import (
NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST,
NezhaForMaskedLM,
NezhaForMultipleChoice,
NezhaForNextSentencePrediction,
NezhaForPreTraining,
NezhaForQuestionAnswering,
NezhaForSequenceClassification,
NezhaForTokenClassification,
NezhaModel,
NezhaPreTrainedModel,
)
from .models.nystromformer import (
NYSTROMFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
NystromformerForMaskedLM,
NystromformerForMultipleChoice,
NystromformerForQuestionAnswering,
NystromformerForSequenceClassification,
NystromformerForTokenClassification,
NystromformerLayer,
NystromformerModel,
NystromformerPreTrainedModel,
)
from .models.openai import (
OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
OpenAIGPTDoubleHeadsModel,
OpenAIGPTForSequenceClassification,
OpenAIGPTLMHeadModel,
OpenAIGPTModel,
OpenAIGPTPreTrainedModel,
load_tf_weights_in_openai_gpt,
)
from .models.opt import (
OPT_PRETRAINED_MODEL_ARCHIVE_LIST,
OPTForCausalLM,
OPTForQuestionAnswering,
OPTForSequenceClassification,
OPTModel,
OPTPreTrainedModel,
)
from .models.owlvit import (
OWLVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
OwlViTForObjectDetection,
OwlViTModel,
OwlViTPreTrainedModel,
OwlViTTextModel,
OwlViTVisionModel,
)
from .models.pegasus import (
PegasusForCausalLM,
PegasusForConditionalGeneration,
PegasusModel,
PegasusPreTrainedModel,
)
from .models.pegasus_x import (
PEGASUS_X_PRETRAINED_MODEL_ARCHIVE_LIST,
PegasusXForConditionalGeneration,
PegasusXModel,
PegasusXPreTrainedModel,
)
from .models.perceiver import (
PERCEIVER_PRETRAINED_MODEL_ARCHIVE_LIST,
PerceiverForImageClassificationConvProcessing,
PerceiverForImageClassificationFourier,
PerceiverForImageClassificationLearned,
PerceiverForMaskedLM,
PerceiverForMultimodalAutoencoding,
PerceiverForOpticalFlow,
PerceiverForSequenceClassification,
PerceiverLayer,
PerceiverModel,
PerceiverPreTrainedModel,
)
from .models.plbart import (
PLBART_PRETRAINED_MODEL_ARCHIVE_LIST,
PLBartForCausalLM,
PLBartForConditionalGeneration,
PLBartForSequenceClassification,
PLBartModel,
PLBartPreTrainedModel,
)
from .models.poolformer import (
POOLFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
PoolFormerForImageClassification,
PoolFormerModel,
PoolFormerPreTrainedModel,
)
from .models.prophetnet import (
PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST,
ProphetNetDecoder,
ProphetNetEncoder,
ProphetNetForCausalLM,
ProphetNetForConditionalGeneration,
ProphetNetModel,
ProphetNetPreTrainedModel,
)
from .models.qdqbert import (
QDQBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
QDQBertForMaskedLM,
QDQBertForMultipleChoice,
QDQBertForNextSentencePrediction,
QDQBertForQuestionAnswering,
QDQBertForSequenceClassification,
QDQBertForTokenClassification,
QDQBertLayer,
QDQBertLMHeadModel,
QDQBertModel,
QDQBertPreTrainedModel,
load_tf_weights_in_qdqbert,
)
from .models.rag import RagModel, RagPreTrainedModel, RagSequenceForGeneration, RagTokenForGeneration
from .models.realm import (
REALM_PRETRAINED_MODEL_ARCHIVE_LIST,
RealmEmbedder,
RealmForOpenQA,
RealmKnowledgeAugEncoder,
RealmPreTrainedModel,
RealmReader,
RealmRetriever,
RealmScorer,
load_tf_weights_in_realm,
)
from .models.reformer import (
REFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
ReformerAttention,
ReformerForMaskedLM,
ReformerForQuestionAnswering,
ReformerForSequenceClassification,
ReformerLayer,
ReformerModel,
ReformerModelWithLMHead,
ReformerPreTrainedModel,
)
from .models.regnet import (
REGNET_PRETRAINED_MODEL_ARCHIVE_LIST,
RegNetForImageClassification,
RegNetModel,
RegNetPreTrainedModel,
)
from .models.rembert import (
REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
RemBertForCausalLM,
RemBertForMaskedLM,
RemBertForMultipleChoice,
RemBertForQuestionAnswering,
RemBertForSequenceClassification,
RemBertForTokenClassification,
RemBertLayer,
RemBertModel,
RemBertPreTrainedModel,
load_tf_weights_in_rembert,
)
from .models.resnet import (
RESNET_PRETRAINED_MODEL_ARCHIVE_LIST,
ResNetBackbone,
ResNetForImageClassification,
ResNetModel,
ResNetPreTrainedModel,
)
from .models.retribert import RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST, RetriBertModel, RetriBertPreTrainedModel
from .models.roberta import (
ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
RobertaForCausalLM,
RobertaForMaskedLM,
RobertaForMultipleChoice,
RobertaForQuestionAnswering,
RobertaForSequenceClassification,
RobertaForTokenClassification,
RobertaModel,
RobertaPreTrainedModel,
)
from .models.roc_bert import (
ROC_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
RoCBertForCausalLM,
RoCBertForMaskedLM,
RoCBertForMultipleChoice,
RoCBertForPreTraining,
RoCBertForQuestionAnswering,
RoCBertForSequenceClassification,
RoCBertForTokenClassification,
RoCBertLayer,
RoCBertModel,
RoCBertPreTrainedModel,
load_tf_weights_in_roc_bert,
)
from .models.roformer import (
ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
RoFormerForCausalLM,
RoFormerForMaskedLM,
RoFormerForMultipleChoice,
RoFormerForQuestionAnswering,
RoFormerForSequenceClassification,
RoFormerForTokenClassification,
RoFormerLayer,
RoFormerModel,
RoFormerPreTrainedModel,
load_tf_weights_in_roformer,
)
from .models.segformer import (
SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
SegformerDecodeHead,
SegformerForImageClassification,
SegformerForSemanticSegmentation,
SegformerLayer,
SegformerModel,
SegformerPreTrainedModel,
)
from .models.sew import (
SEW_PRETRAINED_MODEL_ARCHIVE_LIST,
SEWForCTC,
SEWForSequenceClassification,
SEWModel,
SEWPreTrainedModel,
)
from .models.sew_d import (
SEW_D_PRETRAINED_MODEL_ARCHIVE_LIST,
SEWDForCTC,
SEWDForSequenceClassification,
SEWDModel,
SEWDPreTrainedModel,
)
from .models.speech_encoder_decoder import SpeechEncoderDecoderModel
from .models.speech_to_text import (
SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
Speech2TextForConditionalGeneration,
Speech2TextModel,
Speech2TextPreTrainedModel,
)
from .models.speech_to_text_2 import Speech2Text2ForCausalLM, Speech2Text2PreTrainedModel
from .models.splinter import (
SPLINTER_PRETRAINED_MODEL_ARCHIVE_LIST,
SplinterForPreTraining,
SplinterForQuestionAnswering,
SplinterLayer,
SplinterModel,
SplinterPreTrainedModel,
)
from .models.squeezebert import (
SQUEEZEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
SqueezeBertForMaskedLM,
SqueezeBertForMultipleChoice,
SqueezeBertForQuestionAnswering,
SqueezeBertForSequenceClassification,
SqueezeBertForTokenClassification,
SqueezeBertModel,
SqueezeBertModule,
SqueezeBertPreTrainedModel,
)
from .models.swin import (
SWIN_PRETRAINED_MODEL_ARCHIVE_LIST,
SwinForImageClassification,
SwinForMaskedImageModeling,
SwinModel,
SwinPreTrainedModel,
)
from .models.swinv2 import (
SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST,
Swinv2ForImageClassification,
Swinv2ForMaskedImageModeling,
Swinv2Model,
Swinv2PreTrainedModel,
)
from .models.switch_transformers import (
SWITCH_TRANSFORMERS_PRETRAINED_MODEL_ARCHIVE_LIST,
SwitchTransformersEncoderModel,
SwitchTransformersForConditionalGeneration,
SwitchTransformersModel,
SwitchTransformersPreTrainedModel,
SwitchTransformersSparseMLP,
SwitchTransformersTop1Router,
)
from .models.t5 import (
T5_PRETRAINED_MODEL_ARCHIVE_LIST,
T5EncoderModel,
T5ForConditionalGeneration,
T5Model,
T5PreTrainedModel,
load_tf_weights_in_t5,
)
from .models.tapas import (
TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TapasForMaskedLM,
TapasForQuestionAnswering,
TapasForSequenceClassification,
TapasModel,
TapasPreTrainedModel,
load_tf_weights_in_tapas,
)
from .models.time_series_transformer import (
TIME_SERIES_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TimeSeriesTransformerForPrediction,
TimeSeriesTransformerModel,
TimeSeriesTransformerPreTrainedModel,
)
from .models.timesformer import (
TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TimesformerForVideoClassification,
TimesformerModel,
TimesformerPreTrainedModel,
)
from .models.trajectory_transformer import (
TRAJECTORY_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TrajectoryTransformerModel,
TrajectoryTransformerPreTrainedModel,
)
from .models.transfo_xl import (
TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST,
AdaptiveEmbedding,
TransfoXLForSequenceClassification,
TransfoXLLMHeadModel,
TransfoXLModel,
TransfoXLPreTrainedModel,
load_tf_weights_in_transfo_xl,
)
from .models.trocr import TROCR_PRETRAINED_MODEL_ARCHIVE_LIST, TrOCRForCausalLM, TrOCRPreTrainedModel
from .models.unispeech import (
UNISPEECH_PRETRAINED_MODEL_ARCHIVE_LIST,
UniSpeechForCTC,
UniSpeechForPreTraining,
UniSpeechForSequenceClassification,
UniSpeechModel,
UniSpeechPreTrainedModel,
)
from .models.unispeech_sat import (
UNISPEECH_SAT_PRETRAINED_MODEL_ARCHIVE_LIST,
UniSpeechSatForAudioFrameClassification,
UniSpeechSatForCTC,
UniSpeechSatForPreTraining,
UniSpeechSatForSequenceClassification,
UniSpeechSatForXVector,
UniSpeechSatModel,
UniSpeechSatPreTrainedModel,
)
from .models.van import (
VAN_PRETRAINED_MODEL_ARCHIVE_LIST,
VanForImageClassification,
VanModel,
VanPreTrainedModel,
)
from .models.videomae import (
VIDEOMAE_PRETRAINED_MODEL_ARCHIVE_LIST,
VideoMAEForPreTraining,
VideoMAEForVideoClassification,
VideoMAEModel,
VideoMAEPreTrainedModel,
)
from .models.vilt import (
VILT_PRETRAINED_MODEL_ARCHIVE_LIST,
ViltForImageAndTextRetrieval,
ViltForImagesAndTextClassification,
ViltForMaskedLM,
ViltForQuestionAnswering,
ViltForTokenClassification,
ViltLayer,
ViltModel,
ViltPreTrainedModel,
)
from .models.vision_encoder_decoder import VisionEncoderDecoderModel
from .models.vision_text_dual_encoder import VisionTextDualEncoderModel
from .models.visual_bert import (
VISUAL_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
VisualBertForMultipleChoice,
VisualBertForPreTraining,
VisualBertForQuestionAnswering,
VisualBertForRegionToPhraseAlignment,
VisualBertForVisualReasoning,
VisualBertLayer,
VisualBertModel,
VisualBertPreTrainedModel,
)
from .models.vit import (
VIT_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTForImageClassification,
ViTForMaskedImageModeling,
ViTModel,
ViTPreTrainedModel,
)
from .models.vit_hybrid import (
VIT_HYBRID_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTHybridForImageClassification,
ViTHybridModel,
ViTHybridPreTrainedModel,
)
from .models.vit_mae import (
VIT_MAE_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTMAEForPreTraining,
ViTMAELayer,
ViTMAEModel,
ViTMAEPreTrainedModel,
)
from .models.vit_msn import (
VIT_MSN_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTMSNForImageClassification,
ViTMSNModel,
ViTMSNPreTrainedModel,
)
from .models.wav2vec2 import (
WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST,
Wav2Vec2ForAudioFrameClassification,
Wav2Vec2ForCTC,
Wav2Vec2ForMaskedLM,
Wav2Vec2ForPreTraining,
Wav2Vec2ForSequenceClassification,
Wav2Vec2ForXVector,
Wav2Vec2Model,
Wav2Vec2PreTrainedModel,
)
from .models.wav2vec2_conformer import (
WAV2VEC2_CONFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
Wav2Vec2ConformerForAudioFrameClassification,
Wav2Vec2ConformerForCTC,
Wav2Vec2ConformerForPreTraining,
Wav2Vec2ConformerForSequenceClassification,
Wav2Vec2ConformerForXVector,
Wav2Vec2ConformerModel,
Wav2Vec2ConformerPreTrainedModel,
)
from .models.wavlm import (
WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST,
WavLMForAudioFrameClassification,
WavLMForCTC,
WavLMForSequenceClassification,
WavLMForXVector,
WavLMModel,
WavLMPreTrainedModel,
)
from .models.whisper import (
WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST,
WhisperForConditionalGeneration,
WhisperModel,
WhisperPreTrainedModel,
)
from .models.x_clip import (
XCLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
XCLIPModel,
XCLIPPreTrainedModel,
XCLIPTextModel,
XCLIPVisionModel,
)
from .models.xglm import XGLM_PRETRAINED_MODEL_ARCHIVE_LIST, XGLMForCausalLM, XGLMModel, XGLMPreTrainedModel
from .models.xlm import (
XLM_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMForMultipleChoice,
XLMForQuestionAnswering,
XLMForQuestionAnsweringSimple,
XLMForSequenceClassification,
XLMForTokenClassification,
XLMModel,
XLMPreTrainedModel,
XLMWithLMHeadModel,
)
from .models.xlm_prophetnet import (
XLM_PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMProphetNetDecoder,
XLMProphetNetEncoder,
XLMProphetNetForCausalLM,
XLMProphetNetForConditionalGeneration,
XLMProphetNetModel,
XLMProphetNetPreTrainedModel,
)
from .models.xlm_roberta import (
XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMRobertaForCausalLM,
XLMRobertaForMaskedLM,
XLMRobertaForMultipleChoice,
XLMRobertaForQuestionAnswering,
XLMRobertaForSequenceClassification,
XLMRobertaForTokenClassification,
XLMRobertaModel,
XLMRobertaPreTrainedModel,
)
from .models.xlm_roberta_xl import (
XLM_ROBERTA_XL_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMRobertaXLForCausalLM,
XLMRobertaXLForMaskedLM,
XLMRobertaXLForMultipleChoice,
XLMRobertaXLForQuestionAnswering,
XLMRobertaXLForSequenceClassification,
XLMRobertaXLForTokenClassification,
XLMRobertaXLModel,
XLMRobertaXLPreTrainedModel,
)
from .models.xlnet import (
XLNET_PRETRAINED_MODEL_ARCHIVE_LIST,
XLNetForMultipleChoice,
XLNetForQuestionAnswering,
XLNetForQuestionAnsweringSimple,
XLNetForSequenceClassification,
XLNetForTokenClassification,
XLNetLMHeadModel,
XLNetModel,
XLNetPreTrainedModel,
load_tf_weights_in_xlnet,
)
from .models.yolos import (
YOLOS_PRETRAINED_MODEL_ARCHIVE_LIST,
YolosForObjectDetection,
YolosModel,
YolosPreTrainedModel,
)
from .models.yoso import (
YOSO_PRETRAINED_MODEL_ARCHIVE_LIST,
YosoForMaskedLM,
YosoForMultipleChoice,
YosoForQuestionAnswering,
YosoForSequenceClassification,
YosoForTokenClassification,
YosoLayer,
YosoModel,
YosoPreTrainedModel,
)
# Optimization
from .optimization import (
Adafactor,
AdamW,
get_constant_schedule,
get_constant_schedule_with_warmup,
get_cosine_schedule_with_warmup,
get_cosine_with_hard_restarts_schedule_with_warmup,
get_linear_schedule_with_warmup,
get_polynomial_decay_schedule_with_warmup,
get_scheduler,
)
from .pytorch_utils import Conv1D, apply_chunking_to_forward, prune_layer
# Trainer
from .trainer import Trainer
from .trainer_pt_utils import torch_distributed_zero_first
from .trainer_seq2seq import Seq2SeqTrainer
# TensorFlow
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
# Import the same objects as dummies to get them in the namespace.
# They will raise an import error if the user tries to instantiate / use them.
from .utils.dummy_tf_objects import *
else:
from .benchmark.benchmark_args_tf import TensorFlowBenchmarkArguments
# Benchmarks
from .benchmark.benchmark_tf import TensorFlowBenchmark
from .generation import (
TFForcedBOSTokenLogitsProcessor,
TFForcedEOSTokenLogitsProcessor,
TFGenerationMixin,
TFLogitsProcessor,
TFLogitsProcessorList,
TFLogitsWarper,
TFMinLengthLogitsProcessor,
TFNoBadWordsLogitsProcessor,
TFNoRepeatNGramLogitsProcessor,
TFRepetitionPenaltyLogitsProcessor,
TFTemperatureLogitsWarper,
TFTopKLogitsWarper,
TFTopPLogitsWarper,
tf_top_k_top_p_filtering,
)
from .keras_callbacks import KerasMetricCallback, PushToHubCallback
from .modeling_tf_layoutlm import (
TF_LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLayoutLMForMaskedLM,
TFLayoutLMForQuestionAnswering,
TFLayoutLMForSequenceClassification,
TFLayoutLMForTokenClassification,
TFLayoutLMMainLayer,
TFLayoutLMModel,
TFLayoutLMPreTrainedModel,
)
from .modeling_tf_utils import TFPreTrainedModel, TFSequenceSummary, TFSharedEmbeddings, shape_list
# TensorFlow model imports
from .models.albert import (
TF_ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFAlbertForMaskedLM,
TFAlbertForMultipleChoice,
TFAlbertForPreTraining,
TFAlbertForQuestionAnswering,
TFAlbertForSequenceClassification,
TFAlbertForTokenClassification,
TFAlbertMainLayer,
TFAlbertModel,
TFAlbertPreTrainedModel,
)
from .models.auto import (
TF_MODEL_FOR_CAUSAL_LM_MAPPING,
TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING,
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING,
TF_MODEL_FOR_MASKED_LM_MAPPING,
TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING,
TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING,
TF_MODEL_FOR_PRETRAINING_MAPPING,
TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING,
TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING,
TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING,
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING,
TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING,
TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_VISION_2_SEQ_MAPPING,
TF_MODEL_MAPPING,
TF_MODEL_WITH_LM_HEAD_MAPPING,
TFAutoModel,
TFAutoModelForCausalLM,
TFAutoModelForDocumentQuestionAnswering,
TFAutoModelForImageClassification,
TFAutoModelForMaskedLM,
TFAutoModelForMultipleChoice,
TFAutoModelForNextSentencePrediction,
TFAutoModelForPreTraining,
TFAutoModelForQuestionAnswering,
TFAutoModelForSemanticSegmentation,
TFAutoModelForSeq2SeqLM,
TFAutoModelForSequenceClassification,
TFAutoModelForSpeechSeq2Seq,
TFAutoModelForTableQuestionAnswering,
TFAutoModelForTokenClassification,
TFAutoModelForVision2Seq,
TFAutoModelWithLMHead,
)
from .models.bart import (
TFBartForConditionalGeneration,
TFBartForSequenceClassification,
TFBartModel,
TFBartPretrainedModel,
)
from .models.bert import (
TF_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFBertEmbeddings,
TFBertForMaskedLM,
TFBertForMultipleChoice,
TFBertForNextSentencePrediction,
TFBertForPreTraining,
TFBertForQuestionAnswering,
TFBertForSequenceClassification,
TFBertForTokenClassification,
TFBertLMHeadModel,
TFBertMainLayer,
TFBertModel,
TFBertPreTrainedModel,
)
from .models.blenderbot import (
TFBlenderbotForConditionalGeneration,
TFBlenderbotModel,
TFBlenderbotPreTrainedModel,
)
from .models.blenderbot_small import (
TFBlenderbotSmallForConditionalGeneration,
TFBlenderbotSmallModel,
TFBlenderbotSmallPreTrainedModel,
)
from .models.camembert import (
TF_CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCamembertForCausalLM,
TFCamembertForMaskedLM,
TFCamembertForMultipleChoice,
TFCamembertForQuestionAnswering,
TFCamembertForSequenceClassification,
TFCamembertForTokenClassification,
TFCamembertModel,
TFCamembertPreTrainedModel,
)
from .models.clip import (
TF_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCLIPModel,
TFCLIPPreTrainedModel,
TFCLIPTextModel,
TFCLIPVisionModel,
)
from .models.convbert import (
TF_CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFConvBertForMaskedLM,
TFConvBertForMultipleChoice,
TFConvBertForQuestionAnswering,
TFConvBertForSequenceClassification,
TFConvBertForTokenClassification,
TFConvBertLayer,
TFConvBertModel,
TFConvBertPreTrainedModel,
)
from .models.convnext import TFConvNextForImageClassification, TFConvNextModel, TFConvNextPreTrainedModel
from .models.ctrl import (
TF_CTRL_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCTRLForSequenceClassification,
TFCTRLLMHeadModel,
TFCTRLModel,
TFCTRLPreTrainedModel,
)
from .models.cvt import (
TF_CVT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCvtForImageClassification,
TFCvtModel,
TFCvtPreTrainedModel,
)
from .models.data2vec import (
TFData2VecVisionForImageClassification,
TFData2VecVisionForSemanticSegmentation,
TFData2VecVisionModel,
TFData2VecVisionPreTrainedModel,
)
from .models.deberta import (
TF_DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDebertaForMaskedLM,
TFDebertaForQuestionAnswering,
TFDebertaForSequenceClassification,
TFDebertaForTokenClassification,
TFDebertaModel,
TFDebertaPreTrainedModel,
)
from .models.deberta_v2 import (
TF_DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDebertaV2ForMaskedLM,
TFDebertaV2ForQuestionAnswering,
TFDebertaV2ForSequenceClassification,
TFDebertaV2ForTokenClassification,
TFDebertaV2Model,
TFDebertaV2PreTrainedModel,
)
from .models.deit import (
TF_DEIT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDeiTForImageClassification,
TFDeiTForImageClassificationWithTeacher,
TFDeiTForMaskedImageModeling,
TFDeiTModel,
TFDeiTPreTrainedModel,
)
from .models.distilbert import (
TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDistilBertForMaskedLM,
TFDistilBertForMultipleChoice,
TFDistilBertForQuestionAnswering,
TFDistilBertForSequenceClassification,
TFDistilBertForTokenClassification,
TFDistilBertMainLayer,
TFDistilBertModel,
TFDistilBertPreTrainedModel,
)
from .models.dpr import (
TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDPRContextEncoder,
TFDPRPretrainedContextEncoder,
TFDPRPretrainedQuestionEncoder,
TFDPRPretrainedReader,
TFDPRQuestionEncoder,
TFDPRReader,
)
from .models.electra import (
TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFElectraForMaskedLM,
TFElectraForMultipleChoice,
TFElectraForPreTraining,
TFElectraForQuestionAnswering,
TFElectraForSequenceClassification,
TFElectraForTokenClassification,
TFElectraModel,
TFElectraPreTrainedModel,
)
from .models.encoder_decoder import TFEncoderDecoderModel
from .models.esm import (
ESM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFEsmForMaskedLM,
TFEsmForSequenceClassification,
TFEsmForTokenClassification,
TFEsmModel,
TFEsmPreTrainedModel,
)
from .models.flaubert import (
TF_FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFFlaubertForMultipleChoice,
TFFlaubertForQuestionAnsweringSimple,
TFFlaubertForSequenceClassification,
TFFlaubertForTokenClassification,
TFFlaubertModel,
TFFlaubertPreTrainedModel,
TFFlaubertWithLMHeadModel,
)
from .models.funnel import (
TF_FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST,
TFFunnelBaseModel,
TFFunnelForMaskedLM,
TFFunnelForMultipleChoice,
TFFunnelForPreTraining,
TFFunnelForQuestionAnswering,
TFFunnelForSequenceClassification,
TFFunnelForTokenClassification,
TFFunnelModel,
TFFunnelPreTrainedModel,
)
from .models.gpt2 import (
TF_GPT2_PRETRAINED_MODEL_ARCHIVE_LIST,
TFGPT2DoubleHeadsModel,
TFGPT2ForSequenceClassification,
TFGPT2LMHeadModel,
TFGPT2MainLayer,
TFGPT2Model,
TFGPT2PreTrainedModel,
)
from .models.gptj import (
TFGPTJForCausalLM,
TFGPTJForQuestionAnswering,
TFGPTJForSequenceClassification,
TFGPTJModel,
TFGPTJPreTrainedModel,
)
from .models.groupvit import (
TF_GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFGroupViTModel,
TFGroupViTPreTrainedModel,
TFGroupViTTextModel,
TFGroupViTVisionModel,
)
from .models.hubert import (
TF_HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFHubertForCTC,
TFHubertModel,
TFHubertPreTrainedModel,
)
from .models.layoutlmv3 import (
TF_LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLayoutLMv3ForQuestionAnswering,
TFLayoutLMv3ForSequenceClassification,
TFLayoutLMv3ForTokenClassification,
TFLayoutLMv3Model,
TFLayoutLMv3PreTrainedModel,
)
from .models.led import TFLEDForConditionalGeneration, TFLEDModel, TFLEDPreTrainedModel
from .models.longformer import (
TF_LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLongformerForMaskedLM,
TFLongformerForMultipleChoice,
TFLongformerForQuestionAnswering,
TFLongformerForSequenceClassification,
TFLongformerForTokenClassification,
TFLongformerModel,
TFLongformerPreTrainedModel,
TFLongformerSelfAttention,
)
from .models.lxmert import (
TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLxmertForPreTraining,
TFLxmertMainLayer,
TFLxmertModel,
TFLxmertPreTrainedModel,
TFLxmertVisualFeatureEncoder,
)
from .models.marian import TFMarianModel, TFMarianMTModel, TFMarianPreTrainedModel
from .models.mbart import TFMBartForConditionalGeneration, TFMBartModel, TFMBartPreTrainedModel
from .models.mobilebert import (
TF_MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFMobileBertForMaskedLM,
TFMobileBertForMultipleChoice,
TFMobileBertForNextSentencePrediction,
TFMobileBertForPreTraining,
TFMobileBertForQuestionAnswering,
TFMobileBertForSequenceClassification,
TFMobileBertForTokenClassification,
TFMobileBertMainLayer,
TFMobileBertModel,
TFMobileBertPreTrainedModel,
TFMobileViTForImageClassification,
TFMobileViTForSemanticSegmentation,
TFMobileViTModel,
TFMobileViTPreTrainedModel,
)
from .models.mpnet import (
TF_MPNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFMPNetForMaskedLM,
TFMPNetForMultipleChoice,
TFMPNetForQuestionAnswering,
TFMPNetForSequenceClassification,
TFMPNetForTokenClassification,
TFMPNetMainLayer,
TFMPNetModel,
TFMPNetPreTrainedModel,
)
from .models.mt5 import TFMT5EncoderModel, TFMT5ForConditionalGeneration, TFMT5Model
from .models.openai import (
TF_OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFOpenAIGPTDoubleHeadsModel,
TFOpenAIGPTForSequenceClassification,
TFOpenAIGPTLMHeadModel,
TFOpenAIGPTMainLayer,
TFOpenAIGPTModel,
TFOpenAIGPTPreTrainedModel,
)
from .models.opt import TFOPTForCausalLM, TFOPTModel, TFOPTPreTrainedModel
from .models.pegasus import TFPegasusForConditionalGeneration, TFPegasusModel, TFPegasusPreTrainedModel
from .models.rag import TFRagModel, TFRagPreTrainedModel, TFRagSequenceForGeneration, TFRagTokenForGeneration
from .models.regnet import (
TF_REGNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRegNetForImageClassification,
TFRegNetModel,
TFRegNetPreTrainedModel,
)
from .models.rembert import (
TF_REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRemBertForCausalLM,
TFRemBertForMaskedLM,
TFRemBertForMultipleChoice,
TFRemBertForQuestionAnswering,
TFRemBertForSequenceClassification,
TFRemBertForTokenClassification,
TFRemBertLayer,
TFRemBertModel,
TFRemBertPreTrainedModel,
)
from .models.resnet import (
TF_RESNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFResNetForImageClassification,
TFResNetModel,
TFResNetPreTrainedModel,
)
from .models.roberta import (
TF_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRobertaForCausalLM,
TFRobertaForMaskedLM,
TFRobertaForMultipleChoice,
TFRobertaForQuestionAnswering,
TFRobertaForSequenceClassification,
TFRobertaForTokenClassification,
TFRobertaMainLayer,
TFRobertaModel,
TFRobertaPreTrainedModel,
)
from .models.roformer import (
TF_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRoFormerForCausalLM,
TFRoFormerForMaskedLM,
TFRoFormerForMultipleChoice,
TFRoFormerForQuestionAnswering,
TFRoFormerForSequenceClassification,
TFRoFormerForTokenClassification,
TFRoFormerLayer,
TFRoFormerModel,
TFRoFormerPreTrainedModel,
)
from .models.segformer import (
TF_SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFSegformerDecodeHead,
TFSegformerForImageClassification,
TFSegformerForSemanticSegmentation,
TFSegformerModel,
TFSegformerPreTrainedModel,
)
from .models.speech_to_text import (
TF_SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFSpeech2TextForConditionalGeneration,
TFSpeech2TextModel,
TFSpeech2TextPreTrainedModel,
)
from .models.swin import (
TF_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST,
TFSwinForImageClassification,
TFSwinForMaskedImageModeling,
TFSwinModel,
TFSwinPreTrainedModel,
)
from .models.t5 import (
TF_T5_PRETRAINED_MODEL_ARCHIVE_LIST,
TFT5EncoderModel,
TFT5ForConditionalGeneration,
TFT5Model,
TFT5PreTrainedModel,
)
from .models.tapas import (
TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TFTapasForMaskedLM,
TFTapasForQuestionAnswering,
TFTapasForSequenceClassification,
TFTapasModel,
TFTapasPreTrainedModel,
)
from .models.transfo_xl import (
TF_TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST,
TFAdaptiveEmbedding,
TFTransfoXLForSequenceClassification,
TFTransfoXLLMHeadModel,
TFTransfoXLMainLayer,
TFTransfoXLModel,
TFTransfoXLPreTrainedModel,
)
from .models.vision_encoder_decoder import TFVisionEncoderDecoderModel
from .models.vit import TFViTForImageClassification, TFViTModel, TFViTPreTrainedModel
from .models.vit_mae import TFViTMAEForPreTraining, TFViTMAEModel, TFViTMAEPreTrainedModel
from .models.wav2vec2 import (
TF_WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST,
TFWav2Vec2ForCTC,
TFWav2Vec2Model,
TFWav2Vec2PreTrainedModel,
)
from .models.whisper import (
TF_WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFWhisperForConditionalGeneration,
TFWhisperModel,
TFWhisperPreTrainedModel,
)
from .models.xglm import (
TF_XGLM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXGLMForCausalLM,
TFXGLMModel,
TFXGLMPreTrainedModel,
)
from .models.xlm import (
TF_XLM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXLMForMultipleChoice,
TFXLMForQuestionAnsweringSimple,
TFXLMForSequenceClassification,
TFXLMForTokenClassification,
TFXLMMainLayer,
TFXLMModel,
TFXLMPreTrainedModel,
TFXLMWithLMHeadModel,
)
from .models.xlm_roberta import (
TF_XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXLMRobertaForMaskedLM,
TFXLMRobertaForMultipleChoice,
TFXLMRobertaForQuestionAnswering,
TFXLMRobertaForSequenceClassification,
TFXLMRobertaForTokenClassification,
TFXLMRobertaModel,
)
from .models.xlnet import (
TF_XLNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXLNetForMultipleChoice,
TFXLNetForQuestionAnsweringSimple,
TFXLNetForSequenceClassification,
TFXLNetForTokenClassification,
TFXLNetLMHeadModel,
TFXLNetMainLayer,
TFXLNetModel,
TFXLNetPreTrainedModel,
)
# Optimization
from .optimization_tf import AdamWeightDecay, GradientAccumulator, WarmUp, create_optimizer
# Trainer
from .trainer_tf import TFTrainer
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
# Import the same objects as dummies to get them in the namespace.
# They will raise an import error if the user tries to instantiate / use them.
from .utils.dummy_flax_objects import *
else:
from .generation import (
FlaxForcedBOSTokenLogitsProcessor,
FlaxForcedEOSTokenLogitsProcessor,
FlaxGenerationMixin,
FlaxLogitsProcessor,
FlaxLogitsProcessorList,
FlaxLogitsWarper,
FlaxMinLengthLogitsProcessor,
FlaxTemperatureLogitsWarper,
FlaxTopKLogitsWarper,
FlaxTopPLogitsWarper,
)
from .modeling_flax_utils import FlaxPreTrainedModel
# Flax model imports
from .models.albert import (
FlaxAlbertForMaskedLM,
FlaxAlbertForMultipleChoice,
FlaxAlbertForPreTraining,
FlaxAlbertForQuestionAnswering,
FlaxAlbertForSequenceClassification,
FlaxAlbertForTokenClassification,
FlaxAlbertModel,
FlaxAlbertPreTrainedModel,
)
from .models.auto import (
FLAX_MODEL_FOR_CAUSAL_LM_MAPPING,
FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
FLAX_MODEL_FOR_MASKED_LM_MAPPING,
FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING,
FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING,
FLAX_MODEL_FOR_PRETRAINING_MAPPING,
FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING,
FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING,
FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING,
FLAX_MODEL_MAPPING,
FlaxAutoModel,
FlaxAutoModelForCausalLM,
FlaxAutoModelForImageClassification,
FlaxAutoModelForMaskedLM,
FlaxAutoModelForMultipleChoice,
FlaxAutoModelForNextSentencePrediction,
FlaxAutoModelForPreTraining,
FlaxAutoModelForQuestionAnswering,
FlaxAutoModelForSeq2SeqLM,
FlaxAutoModelForSequenceClassification,
FlaxAutoModelForTokenClassification,
FlaxAutoModelForVision2Seq,
)
from .models.bart import (
FlaxBartDecoderPreTrainedModel,
FlaxBartForCausalLM,
FlaxBartForConditionalGeneration,
FlaxBartForQuestionAnswering,
FlaxBartForSequenceClassification,
FlaxBartModel,
FlaxBartPreTrainedModel,
)
from .models.beit import (
FlaxBeitForImageClassification,
FlaxBeitForMaskedImageModeling,
FlaxBeitModel,
FlaxBeitPreTrainedModel,
)
from .models.bert import (
FlaxBertForCausalLM,
FlaxBertForMaskedLM,
FlaxBertForMultipleChoice,
FlaxBertForNextSentencePrediction,
FlaxBertForPreTraining,
FlaxBertForQuestionAnswering,
FlaxBertForSequenceClassification,
FlaxBertForTokenClassification,
FlaxBertModel,
FlaxBertPreTrainedModel,
)
from .models.big_bird import (
FlaxBigBirdForCausalLM,
FlaxBigBirdForMaskedLM,
FlaxBigBirdForMultipleChoice,
FlaxBigBirdForPreTraining,
FlaxBigBirdForQuestionAnswering,
FlaxBigBirdForSequenceClassification,
FlaxBigBirdForTokenClassification,
FlaxBigBirdModel,
FlaxBigBirdPreTrainedModel,
)
from .models.blenderbot import (
FlaxBlenderbotForConditionalGeneration,
FlaxBlenderbotModel,
FlaxBlenderbotPreTrainedModel,
)
from .models.blenderbot_small import (
FlaxBlenderbotSmallForConditionalGeneration,
FlaxBlenderbotSmallModel,
FlaxBlenderbotSmallPreTrainedModel,
)
from .models.clip import (
FlaxCLIPModel,
FlaxCLIPPreTrainedModel,
FlaxCLIPTextModel,
FlaxCLIPTextPreTrainedModel,
FlaxCLIPVisionModel,
FlaxCLIPVisionPreTrainedModel,
)
from .models.distilbert import (
FlaxDistilBertForMaskedLM,
FlaxDistilBertForMultipleChoice,
FlaxDistilBertForQuestionAnswering,
FlaxDistilBertForSequenceClassification,
FlaxDistilBertForTokenClassification,
FlaxDistilBertModel,
FlaxDistilBertPreTrainedModel,
)
from .models.electra import (
FlaxElectraForCausalLM,
FlaxElectraForMaskedLM,
FlaxElectraForMultipleChoice,
FlaxElectraForPreTraining,
FlaxElectraForQuestionAnswering,
FlaxElectraForSequenceClassification,
FlaxElectraForTokenClassification,
FlaxElectraModel,
FlaxElectraPreTrainedModel,
)
from .models.encoder_decoder import FlaxEncoderDecoderModel
from .models.gpt2 import FlaxGPT2LMHeadModel, FlaxGPT2Model, FlaxGPT2PreTrainedModel
from .models.gpt_neo import FlaxGPTNeoForCausalLM, FlaxGPTNeoModel, FlaxGPTNeoPreTrainedModel
from .models.gptj import FlaxGPTJForCausalLM, FlaxGPTJModel, FlaxGPTJPreTrainedModel
from .models.longt5 import FlaxLongT5ForConditionalGeneration, FlaxLongT5Model, FlaxLongT5PreTrainedModel
from .models.marian import FlaxMarianModel, FlaxMarianMTModel, FlaxMarianPreTrainedModel
from .models.mbart import (
FlaxMBartForConditionalGeneration,
FlaxMBartForQuestionAnswering,
FlaxMBartForSequenceClassification,
FlaxMBartModel,
FlaxMBartPreTrainedModel,
)
from .models.mt5 import FlaxMT5EncoderModel, FlaxMT5ForConditionalGeneration, FlaxMT5Model
from .models.opt import FlaxOPTForCausalLM, FlaxOPTModel, FlaxOPTPreTrainedModel
from .models.pegasus import FlaxPegasusForConditionalGeneration, FlaxPegasusModel, FlaxPegasusPreTrainedModel
from .models.roberta import (
FlaxRobertaForCausalLM,
FlaxRobertaForMaskedLM,
FlaxRobertaForMultipleChoice,
FlaxRobertaForQuestionAnswering,
FlaxRobertaForSequenceClassification,
FlaxRobertaForTokenClassification,
FlaxRobertaModel,
FlaxRobertaPreTrainedModel,
)
from .models.roformer import (
FlaxRoFormerForMaskedLM,
FlaxRoFormerForMultipleChoice,
FlaxRoFormerForQuestionAnswering,
FlaxRoFormerForSequenceClassification,
FlaxRoFormerForTokenClassification,
FlaxRoFormerModel,
FlaxRoFormerPreTrainedModel,
)
from .models.speech_encoder_decoder import FlaxSpeechEncoderDecoderModel
from .models.t5 import FlaxT5EncoderModel, FlaxT5ForConditionalGeneration, FlaxT5Model, FlaxT5PreTrainedModel
from .models.vision_encoder_decoder import FlaxVisionEncoderDecoderModel
from .models.vision_text_dual_encoder import FlaxVisionTextDualEncoderModel
from .models.vit import FlaxViTForImageClassification, FlaxViTModel, FlaxViTPreTrainedModel
from .models.wav2vec2 import (
FlaxWav2Vec2ForCTC,
FlaxWav2Vec2ForPreTraining,
FlaxWav2Vec2Model,
FlaxWav2Vec2PreTrainedModel,
)
from .models.xglm import FlaxXGLMForCausalLM, FlaxXGLMModel, FlaxXGLMPreTrainedModel
from .models.xlm_roberta import (
FlaxXLMRobertaForMaskedLM,
FlaxXLMRobertaForMultipleChoice,
FlaxXLMRobertaForQuestionAnswering,
FlaxXLMRobertaForSequenceClassification,
FlaxXLMRobertaForTokenClassification,
FlaxXLMRobertaModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(
__name__,
globals()["__file__"],
_import_structure,
module_spec=__spec__,
extra_objects={"__version__": __version__},
)
if not is_tf_available() and not is_torch_available() and not is_flax_available():
logger.warning(
"None of PyTorch, TensorFlow >= 2.0, or Flax have been found. "
"Models won't be available and only tokenizers, configuration "
"and file/data utilities can be used."
)
| # flake8: noqa
# There's no way to ignore "F401 '...' imported but unused" warnings in this
# module, but to preserve other warnings. So, don't check this module at all.
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# When adding a new object to this init, remember to add it twice: once inside the `_import_structure` dictionary and
# once inside the `if TYPE_CHECKING` branch. The `TYPE_CHECKING` should have import statements as usual, but they are
# only there for type checking. The `_import_structure` is a dictionary submodule to list of object names, and is used
# to defer the actual importing for when the objects are requested. This way `import transformers` provides the names
# in the namespace without actually importing anything (and especially none of the backends).
__version__ = "4.26.0.dev0"
from typing import TYPE_CHECKING
# Check the dependencies satisfy the minimal versions required.
from . import dependency_versions_check
from .utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_keras_nlp_available,
is_sentencepiece_available,
is_speech_available,
is_tensorflow_text_available,
is_tf_available,
is_timm_available,
is_tokenizers_available,
is_torch_available,
is_vision_available,
logging,
)
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
# Base objects, independent of any specific backend
_import_structure = {
"benchmark": [],
"commands": [],
"configuration_utils": ["PretrainedConfig"],
"convert_graph_to_onnx": [],
"convert_slow_tokenizers_checkpoints_to_fast": [],
"convert_tf_hub_seq_to_seq_bert_to_pytorch": [],
"data": [
"DataProcessor",
"InputExample",
"InputFeatures",
"SingleSentenceClassificationProcessor",
"SquadExample",
"SquadFeatures",
"SquadV1Processor",
"SquadV2Processor",
"glue_compute_metrics",
"glue_convert_examples_to_features",
"glue_output_modes",
"glue_processors",
"glue_tasks_num_labels",
"squad_convert_examples_to_features",
"xnli_compute_metrics",
"xnli_output_modes",
"xnli_processors",
"xnli_tasks_num_labels",
],
"data.data_collator": [
"DataCollator",
"DataCollatorForLanguageModeling",
"DataCollatorForPermutationLanguageModeling",
"DataCollatorForSeq2Seq",
"DataCollatorForSOP",
"DataCollatorForTokenClassification",
"DataCollatorForWholeWordMask",
"DataCollatorWithPadding",
"DefaultDataCollator",
"default_data_collator",
],
"data.metrics": [],
"data.processors": [],
"debug_utils": [],
"dependency_versions_check": [],
"dependency_versions_table": [],
"dynamic_module_utils": [],
"feature_extraction_sequence_utils": ["SequenceFeatureExtractor"],
"feature_extraction_utils": ["BatchFeature", "FeatureExtractionMixin"],
"file_utils": [],
"generation": ["GenerationConfig"],
"hf_argparser": ["HfArgumentParser"],
"integrations": [
"is_clearml_available",
"is_comet_available",
"is_neptune_available",
"is_optuna_available",
"is_ray_available",
"is_ray_tune_available",
"is_sigopt_available",
"is_tensorboard_available",
"is_wandb_available",
],
"modelcard": ["ModelCard"],
"modeling_tf_pytorch_utils": [
"convert_tf_weight_name_to_pt_weight_name",
"load_pytorch_checkpoint_in_tf2_model",
"load_pytorch_model_in_tf2_model",
"load_pytorch_weights_in_tf2_model",
"load_tf2_checkpoint_in_pytorch_model",
"load_tf2_model_in_pytorch_model",
"load_tf2_weights_in_pytorch_model",
],
"models": [],
# Models
"models.albert": ["ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "AlbertConfig"],
"models.audio_spectrogram_transformer": [
"AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ASTConfig",
],
"models.auto": [
"ALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"CONFIG_MAPPING",
"FEATURE_EXTRACTOR_MAPPING",
"IMAGE_PROCESSOR_MAPPING",
"MODEL_NAMES_MAPPING",
"PROCESSOR_MAPPING",
"TOKENIZER_MAPPING",
"AutoConfig",
"AutoFeatureExtractor",
"AutoImageProcessor",
"AutoProcessor",
"AutoTokenizer",
],
"models.bart": ["BartConfig", "BartTokenizer"],
"models.barthez": [],
"models.bartpho": [],
"models.beit": ["BEIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BeitConfig"],
"models.bert": [
"BERT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BasicTokenizer",
"BertConfig",
"BertTokenizer",
"WordpieceTokenizer",
],
"models.bert_generation": ["BertGenerationConfig"],
"models.bert_japanese": ["BertJapaneseTokenizer", "CharacterTokenizer", "MecabTokenizer"],
"models.bertweet": ["BertweetTokenizer"],
"models.big_bird": ["BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP", "BigBirdConfig"],
"models.bigbird_pegasus": [
"BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BigBirdPegasusConfig",
],
"models.biogpt": ["BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BioGptConfig", "BioGptTokenizer"],
"models.bit": ["BIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BitConfig"],
"models.blenderbot": ["BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP", "BlenderbotConfig", "BlenderbotTokenizer"],
"models.blenderbot_small": [
"BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"BlenderbotSmallConfig",
"BlenderbotSmallTokenizer",
],
"models.bloom": ["BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP", "BloomConfig"],
"models.bort": [],
"models.byt5": ["ByT5Tokenizer"],
"models.camembert": ["CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "CamembertConfig"],
"models.canine": ["CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP", "CanineConfig", "CanineTokenizer"],
"models.chinese_clip": [
"CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ChineseCLIPConfig",
"ChineseCLIPProcessor",
"ChineseCLIPTextConfig",
"ChineseCLIPVisionConfig",
],
"models.clip": [
"CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
"CLIPConfig",
"CLIPProcessor",
"CLIPTextConfig",
"CLIPTokenizer",
"CLIPVisionConfig",
],
"models.clipseg": [
"CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP",
"CLIPSegConfig",
"CLIPSegProcessor",
"CLIPSegTextConfig",
"CLIPSegVisionConfig",
],
"models.codegen": ["CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP", "CodeGenConfig", "CodeGenTokenizer"],
"models.conditional_detr": ["CONDITIONAL_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP", "ConditionalDetrConfig"],
"models.convbert": ["CONVBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ConvBertConfig", "ConvBertTokenizer"],
"models.convnext": ["CONVNEXT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ConvNextConfig"],
"models.cpm": [],
"models.ctrl": ["CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP", "CTRLConfig", "CTRLTokenizer"],
"models.cvt": ["CVT_PRETRAINED_CONFIG_ARCHIVE_MAP", "CvtConfig"],
"models.data2vec": [
"DATA2VEC_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"DATA2VEC_VISION_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Data2VecAudioConfig",
"Data2VecTextConfig",
"Data2VecVisionConfig",
],
"models.deberta": ["DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP", "DebertaConfig", "DebertaTokenizer"],
"models.deberta_v2": ["DEBERTA_V2_PRETRAINED_CONFIG_ARCHIVE_MAP", "DebertaV2Config"],
"models.decision_transformer": ["DECISION_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "DecisionTransformerConfig"],
"models.deformable_detr": ["DEFORMABLE_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP", "DeformableDetrConfig"],
"models.deit": ["DEIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DeiTConfig"],
"models.detr": ["DETR_PRETRAINED_CONFIG_ARCHIVE_MAP", "DetrConfig"],
"models.dialogpt": [],
"models.dinat": ["DINAT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DinatConfig"],
"models.distilbert": ["DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DistilBertConfig", "DistilBertTokenizer"],
"models.dit": [],
"models.donut": ["DONUT_SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP", "DonutProcessor", "DonutSwinConfig"],
"models.dpr": [
"DPR_PRETRAINED_CONFIG_ARCHIVE_MAP",
"DPRConfig",
"DPRContextEncoderTokenizer",
"DPRQuestionEncoderTokenizer",
"DPRReaderOutput",
"DPRReaderTokenizer",
],
"models.dpt": ["DPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "DPTConfig"],
"models.electra": ["ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP", "ElectraConfig", "ElectraTokenizer"],
"models.encoder_decoder": ["EncoderDecoderConfig"],
"models.ernie": [
"ERNIE_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ErnieConfig",
],
"models.esm": ["ESM_PRETRAINED_CONFIG_ARCHIVE_MAP", "EsmConfig", "EsmTokenizer"],
"models.flaubert": ["FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "FlaubertConfig", "FlaubertTokenizer"],
"models.flava": [
"FLAVA_PRETRAINED_CONFIG_ARCHIVE_MAP",
"FlavaConfig",
"FlavaImageCodebookConfig",
"FlavaImageConfig",
"FlavaMultimodalConfig",
"FlavaTextConfig",
],
"models.fnet": ["FNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "FNetConfig"],
"models.fsmt": ["FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP", "FSMTConfig", "FSMTTokenizer"],
"models.funnel": ["FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP", "FunnelConfig", "FunnelTokenizer"],
"models.glpn": ["GLPN_PRETRAINED_CONFIG_ARCHIVE_MAP", "GLPNConfig"],
"models.gpt2": ["GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPT2Config", "GPT2Tokenizer"],
"models.gpt_neo": ["GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoConfig"],
"models.gpt_neox": ["GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXConfig"],
"models.gpt_neox_japanese": ["GPT_NEOX_JAPANESE_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXJapaneseConfig"],
"models.gpt_sw3": [],
"models.gptj": ["GPTJ_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTJConfig"],
"models.groupvit": [
"GROUPVIT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"GroupViTConfig",
"GroupViTTextConfig",
"GroupViTVisionConfig",
],
"models.herbert": ["HerbertTokenizer"],
"models.hubert": ["HUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "HubertConfig"],
"models.ibert": ["IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "IBertConfig"],
"models.imagegpt": ["IMAGEGPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ImageGPTConfig"],
"models.jukebox": [
"JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP",
"JukeboxConfig",
"JukeboxPriorConfig",
"JukeboxTokenizer",
"JukeboxVQVAEConfig",
],
"models.layoutlm": ["LAYOUTLM_PRETRAINED_CONFIG_ARCHIVE_MAP", "LayoutLMConfig", "LayoutLMTokenizer"],
"models.layoutlmv2": [
"LAYOUTLMV2_PRETRAINED_CONFIG_ARCHIVE_MAP",
"LayoutLMv2Config",
"LayoutLMv2FeatureExtractor",
"LayoutLMv2ImageProcessor",
"LayoutLMv2Processor",
"LayoutLMv2Tokenizer",
],
"models.layoutlmv3": [
"LAYOUTLMV3_PRETRAINED_CONFIG_ARCHIVE_MAP",
"LayoutLMv3Config",
"LayoutLMv3FeatureExtractor",
"LayoutLMv3ImageProcessor",
"LayoutLMv3Processor",
"LayoutLMv3Tokenizer",
],
"models.layoutxlm": ["LayoutXLMProcessor"],
"models.led": ["LED_PRETRAINED_CONFIG_ARCHIVE_MAP", "LEDConfig", "LEDTokenizer"],
"models.levit": ["LEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "LevitConfig"],
"models.lilt": ["LILT_PRETRAINED_CONFIG_ARCHIVE_MAP", "LiltConfig"],
"models.longformer": ["LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "LongformerConfig", "LongformerTokenizer"],
"models.longt5": ["LONGT5_PRETRAINED_CONFIG_ARCHIVE_MAP", "LongT5Config"],
"models.luke": ["LUKE_PRETRAINED_CONFIG_ARCHIVE_MAP", "LukeConfig", "LukeTokenizer"],
"models.lxmert": ["LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "LxmertConfig", "LxmertTokenizer"],
"models.m2m_100": ["M2M_100_PRETRAINED_CONFIG_ARCHIVE_MAP", "M2M100Config"],
"models.marian": ["MarianConfig"],
"models.markuplm": [
"MARKUPLM_PRETRAINED_CONFIG_ARCHIVE_MAP",
"MarkupLMConfig",
"MarkupLMFeatureExtractor",
"MarkupLMProcessor",
"MarkupLMTokenizer",
],
"models.maskformer": ["MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "MaskFormerConfig", "MaskFormerSwinConfig"],
"models.mbart": ["MBartConfig"],
"models.mbart50": [],
"models.mctct": ["MCTCT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MCTCTConfig", "MCTCTProcessor"],
"models.megatron_bert": ["MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MegatronBertConfig"],
"models.megatron_gpt2": [],
"models.mluke": [],
"models.mmbt": ["MMBTConfig"],
"models.mobilebert": ["MOBILEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileBertConfig", "MobileBertTokenizer"],
"models.mobilenet_v1": ["MOBILENET_V1_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileNetV1Config"],
"models.mobilenet_v2": ["MOBILENET_V2_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileNetV2Config"],
"models.mobilevit": ["MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileViTConfig"],
"models.mpnet": ["MPNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "MPNetConfig", "MPNetTokenizer"],
"models.mt5": ["MT5Config"],
"models.mvp": ["MvpConfig", "MvpTokenizer"],
"models.nat": ["NAT_PRETRAINED_CONFIG_ARCHIVE_MAP", "NatConfig"],
"models.nezha": ["NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP", "NezhaConfig"],
"models.nllb": [],
"models.nystromformer": [
"NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"NystromformerConfig",
],
"models.openai": ["OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "OpenAIGPTConfig", "OpenAIGPTTokenizer"],
"models.opt": ["OPTConfig"],
"models.owlvit": [
"OWLVIT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"OwlViTConfig",
"OwlViTProcessor",
"OwlViTTextConfig",
"OwlViTVisionConfig",
],
"models.pegasus": ["PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP", "PegasusConfig", "PegasusTokenizer"],
"models.pegasus_x": ["PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP", "PegasusXConfig"],
"models.perceiver": ["PERCEIVER_PRETRAINED_CONFIG_ARCHIVE_MAP", "PerceiverConfig", "PerceiverTokenizer"],
"models.phobert": ["PhobertTokenizer"],
"models.plbart": ["PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP", "PLBartConfig"],
"models.poolformer": ["POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "PoolFormerConfig"],
"models.prophetnet": ["PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "ProphetNetConfig", "ProphetNetTokenizer"],
"models.qdqbert": ["QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "QDQBertConfig"],
"models.rag": ["RagConfig", "RagRetriever", "RagTokenizer"],
"models.realm": ["REALM_PRETRAINED_CONFIG_ARCHIVE_MAP", "RealmConfig", "RealmTokenizer"],
"models.reformer": ["REFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "ReformerConfig"],
"models.regnet": ["REGNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "RegNetConfig"],
"models.rembert": ["REMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "RemBertConfig"],
"models.resnet": ["RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "ResNetConfig"],
"models.retribert": ["RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "RetriBertConfig", "RetriBertTokenizer"],
"models.roberta": ["ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP", "RobertaConfig", "RobertaTokenizer"],
"models.roc_bert": ["ROC_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "RoCBertConfig", "RoCBertTokenizer"],
"models.roformer": ["ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "RoFormerConfig", "RoFormerTokenizer"],
"models.segformer": ["SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "SegformerConfig"],
"models.sew": ["SEW_PRETRAINED_CONFIG_ARCHIVE_MAP", "SEWConfig"],
"models.sew_d": ["SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP", "SEWDConfig"],
"models.speech_encoder_decoder": ["SpeechEncoderDecoderConfig"],
"models.speech_to_text": [
"SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Speech2TextConfig",
],
"models.speech_to_text_2": [
"SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Speech2Text2Config",
"Speech2Text2Processor",
"Speech2Text2Tokenizer",
],
"models.splinter": ["SPLINTER_PRETRAINED_CONFIG_ARCHIVE_MAP", "SplinterConfig", "SplinterTokenizer"],
"models.squeezebert": ["SQUEEZEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "SqueezeBertConfig", "SqueezeBertTokenizer"],
"models.swin": ["SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP", "SwinConfig"],
"models.swinv2": ["SWINV2_PRETRAINED_CONFIG_ARCHIVE_MAP", "Swinv2Config"],
"models.switch_transformers": ["SWITCH_TRANSFORMERS_PRETRAINED_CONFIG_ARCHIVE_MAP", "SwitchTransformersConfig"],
"models.t5": ["T5_PRETRAINED_CONFIG_ARCHIVE_MAP", "T5Config"],
"models.table_transformer": ["TABLE_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "TableTransformerConfig"],
"models.tapas": ["TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP", "TapasConfig", "TapasTokenizer"],
"models.tapex": ["TapexTokenizer"],
"models.time_series_transformer": [
"TIME_SERIES_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TimeSeriesTransformerConfig",
],
"models.timesformer": ["TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "TimesformerConfig"],
"models.trajectory_transformer": [
"TRAJECTORY_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TrajectoryTransformerConfig",
],
"models.transfo_xl": [
"TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TransfoXLConfig",
"TransfoXLCorpus",
"TransfoXLTokenizer",
],
"models.trocr": [
"TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP",
"TrOCRConfig",
"TrOCRProcessor",
],
"models.unispeech": [
"UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP",
"UniSpeechConfig",
],
"models.unispeech_sat": [
"UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"UniSpeechSatConfig",
],
"models.van": ["VAN_PRETRAINED_CONFIG_ARCHIVE_MAP", "VanConfig"],
"models.videomae": ["VIDEOMAE_PRETRAINED_CONFIG_ARCHIVE_MAP", "VideoMAEConfig"],
"models.vilt": [
"VILT_PRETRAINED_CONFIG_ARCHIVE_MAP",
"ViltConfig",
"ViltFeatureExtractor",
"ViltImageProcessor",
"ViltProcessor",
],
"models.vision_encoder_decoder": ["VisionEncoderDecoderConfig"],
"models.vision_text_dual_encoder": ["VisionTextDualEncoderConfig", "VisionTextDualEncoderProcessor"],
"models.visual_bert": ["VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "VisualBertConfig"],
"models.vit": ["VIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTConfig"],
"models.vit_hybrid": ["VIT_HYBRID_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTHybridConfig"],
"models.vit_mae": ["VIT_MAE_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTMAEConfig"],
"models.vit_msn": ["VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP", "ViTMSNConfig"],
"models.wav2vec2": [
"WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Wav2Vec2Config",
"Wav2Vec2CTCTokenizer",
"Wav2Vec2FeatureExtractor",
"Wav2Vec2Processor",
"Wav2Vec2Tokenizer",
],
"models.wav2vec2_conformer": [
"WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"Wav2Vec2ConformerConfig",
],
"models.wav2vec2_phoneme": ["Wav2Vec2PhonemeCTCTokenizer"],
"models.wav2vec2_with_lm": ["Wav2Vec2ProcessorWithLM"],
"models.wavlm": [
"WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP",
"WavLMConfig",
],
"models.whisper": [
"WHISPER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"WhisperConfig",
"WhisperFeatureExtractor",
"WhisperProcessor",
"WhisperTokenizer",
],
"models.x_clip": [
"XCLIP_PRETRAINED_CONFIG_ARCHIVE_MAP",
"XCLIPConfig",
"XCLIPProcessor",
"XCLIPTextConfig",
"XCLIPVisionConfig",
],
"models.xglm": ["XGLM_PRETRAINED_CONFIG_ARCHIVE_MAP", "XGLMConfig"],
"models.xlm": ["XLM_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMConfig", "XLMTokenizer"],
"models.xlm_prophetnet": ["XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMProphetNetConfig"],
"models.xlm_roberta": ["XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMRobertaConfig"],
"models.xlm_roberta_xl": ["XLM_ROBERTA_XL_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLMRobertaXLConfig"],
"models.xlnet": ["XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "XLNetConfig"],
"models.yolos": ["YOLOS_PRETRAINED_CONFIG_ARCHIVE_MAP", "YolosConfig"],
"models.yoso": ["YOSO_PRETRAINED_CONFIG_ARCHIVE_MAP", "YosoConfig"],
"onnx": [],
"pipelines": [
"AudioClassificationPipeline",
"AutomaticSpeechRecognitionPipeline",
"Conversation",
"ConversationalPipeline",
"CsvPipelineDataFormat",
"DepthEstimationPipeline",
"DocumentQuestionAnsweringPipeline",
"FeatureExtractionPipeline",
"FillMaskPipeline",
"ImageClassificationPipeline",
"ImageSegmentationPipeline",
"ImageToTextPipeline",
"JsonPipelineDataFormat",
"NerPipeline",
"ObjectDetectionPipeline",
"PipedPipelineDataFormat",
"Pipeline",
"PipelineDataFormat",
"QuestionAnsweringPipeline",
"SummarizationPipeline",
"TableQuestionAnsweringPipeline",
"Text2TextGenerationPipeline",
"TextClassificationPipeline",
"TextGenerationPipeline",
"TokenClassificationPipeline",
"TranslationPipeline",
"VideoClassificationPipeline",
"VisualQuestionAnsweringPipeline",
"ZeroShotClassificationPipeline",
"ZeroShotImageClassificationPipeline",
"ZeroShotObjectDetectionPipeline",
"pipeline",
],
"processing_utils": ["ProcessorMixin"],
"testing_utils": [],
"tokenization_utils": ["PreTrainedTokenizer"],
"tokenization_utils_base": [
"AddedToken",
"BatchEncoding",
"CharSpan",
"PreTrainedTokenizerBase",
"SpecialTokensMixin",
"TokenSpan",
],
"trainer_callback": [
"DefaultFlowCallback",
"EarlyStoppingCallback",
"PrinterCallback",
"ProgressCallback",
"TrainerCallback",
"TrainerControl",
"TrainerState",
],
"trainer_utils": ["EvalPrediction", "IntervalStrategy", "SchedulerType", "enable_full_determinism", "set_seed"],
"training_args": ["TrainingArguments"],
"training_args_seq2seq": ["Seq2SeqTrainingArguments"],
"training_args_tf": ["TFTrainingArguments"],
"utils": [
"CONFIG_NAME",
"MODEL_CARD_NAME",
"PYTORCH_PRETRAINED_BERT_CACHE",
"PYTORCH_TRANSFORMERS_CACHE",
"SPIECE_UNDERLINE",
"TF2_WEIGHTS_NAME",
"TF_WEIGHTS_NAME",
"TRANSFORMERS_CACHE",
"WEIGHTS_NAME",
"TensorType",
"add_end_docstrings",
"add_start_docstrings",
"is_apex_available",
"is_datasets_available",
"is_decord_available",
"is_faiss_available",
"is_flax_available",
"is_keras_nlp_available",
"is_phonemizer_available",
"is_psutil_available",
"is_py3nvml_available",
"is_pyctcdecode_available",
"is_safetensors_available",
"is_scipy_available",
"is_sentencepiece_available",
"is_sklearn_available",
"is_speech_available",
"is_tensorflow_text_available",
"is_tf_available",
"is_timm_available",
"is_tokenizers_available",
"is_torch_available",
"is_torch_tpu_available",
"is_vision_available",
"logging",
],
"utils.bitsandbytes": [],
}
# sentencepiece-backed objects
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_sentencepiece_objects
_import_structure["utils.dummy_sentencepiece_objects"] = [
name for name in dir(dummy_sentencepiece_objects) if not name.startswith("_")
]
else:
_import_structure["models.albert"].append("AlbertTokenizer")
_import_structure["models.barthez"].append("BarthezTokenizer")
_import_structure["models.bartpho"].append("BartphoTokenizer")
_import_structure["models.bert_generation"].append("BertGenerationTokenizer")
_import_structure["models.big_bird"].append("BigBirdTokenizer")
_import_structure["models.camembert"].append("CamembertTokenizer")
_import_structure["models.cpm"].append("CpmTokenizer")
_import_structure["models.deberta_v2"].append("DebertaV2Tokenizer")
_import_structure["models.fnet"].append("FNetTokenizer")
_import_structure["models.gpt_sw3"].append("GPTSw3Tokenizer")
_import_structure["models.layoutxlm"].append("LayoutXLMTokenizer")
_import_structure["models.m2m_100"].append("M2M100Tokenizer")
_import_structure["models.marian"].append("MarianTokenizer")
_import_structure["models.mbart"].append("MBartTokenizer")
_import_structure["models.mbart50"].append("MBart50Tokenizer")
_import_structure["models.mluke"].append("MLukeTokenizer")
_import_structure["models.mt5"].append("MT5Tokenizer")
_import_structure["models.nllb"].append("NllbTokenizer")
_import_structure["models.pegasus"].append("PegasusTokenizer")
_import_structure["models.plbart"].append("PLBartTokenizer")
_import_structure["models.reformer"].append("ReformerTokenizer")
_import_structure["models.rembert"].append("RemBertTokenizer")
_import_structure["models.speech_to_text"].append("Speech2TextTokenizer")
_import_structure["models.t5"].append("T5Tokenizer")
_import_structure["models.xglm"].append("XGLMTokenizer")
_import_structure["models.xlm_prophetnet"].append("XLMProphetNetTokenizer")
_import_structure["models.xlm_roberta"].append("XLMRobertaTokenizer")
_import_structure["models.xlnet"].append("XLNetTokenizer")
# tokenizers-backed objects
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_tokenizers_objects
_import_structure["utils.dummy_tokenizers_objects"] = [
name for name in dir(dummy_tokenizers_objects) if not name.startswith("_")
]
else:
# Fast tokenizers structure
_import_structure["models.albert"].append("AlbertTokenizerFast")
_import_structure["models.bart"].append("BartTokenizerFast")
_import_structure["models.barthez"].append("BarthezTokenizerFast")
_import_structure["models.bert"].append("BertTokenizerFast")
_import_structure["models.big_bird"].append("BigBirdTokenizerFast")
_import_structure["models.blenderbot"].append("BlenderbotTokenizerFast")
_import_structure["models.blenderbot_small"].append("BlenderbotSmallTokenizerFast")
_import_structure["models.bloom"].append("BloomTokenizerFast")
_import_structure["models.camembert"].append("CamembertTokenizerFast")
_import_structure["models.clip"].append("CLIPTokenizerFast")
_import_structure["models.codegen"].append("CodeGenTokenizerFast")
_import_structure["models.convbert"].append("ConvBertTokenizerFast")
_import_structure["models.cpm"].append("CpmTokenizerFast")
_import_structure["models.deberta"].append("DebertaTokenizerFast")
_import_structure["models.deberta_v2"].append("DebertaV2TokenizerFast")
_import_structure["models.distilbert"].append("DistilBertTokenizerFast")
_import_structure["models.dpr"].extend(
["DPRContextEncoderTokenizerFast", "DPRQuestionEncoderTokenizerFast", "DPRReaderTokenizerFast"]
)
_import_structure["models.electra"].append("ElectraTokenizerFast")
_import_structure["models.fnet"].append("FNetTokenizerFast")
_import_structure["models.funnel"].append("FunnelTokenizerFast")
_import_structure["models.gpt2"].append("GPT2TokenizerFast")
_import_structure["models.gpt_neox"].append("GPTNeoXTokenizerFast")
_import_structure["models.gpt_neox_japanese"].append("GPTNeoXJapaneseTokenizer")
_import_structure["models.herbert"].append("HerbertTokenizerFast")
_import_structure["models.layoutlm"].append("LayoutLMTokenizerFast")
_import_structure["models.layoutlmv2"].append("LayoutLMv2TokenizerFast")
_import_structure["models.layoutlmv3"].append("LayoutLMv3TokenizerFast")
_import_structure["models.layoutxlm"].append("LayoutXLMTokenizerFast")
_import_structure["models.led"].append("LEDTokenizerFast")
_import_structure["models.longformer"].append("LongformerTokenizerFast")
_import_structure["models.lxmert"].append("LxmertTokenizerFast")
_import_structure["models.markuplm"].append("MarkupLMTokenizerFast")
_import_structure["models.mbart"].append("MBartTokenizerFast")
_import_structure["models.mbart50"].append("MBart50TokenizerFast")
_import_structure["models.mobilebert"].append("MobileBertTokenizerFast")
_import_structure["models.mpnet"].append("MPNetTokenizerFast")
_import_structure["models.mt5"].append("MT5TokenizerFast")
_import_structure["models.mvp"].append("MvpTokenizerFast")
_import_structure["models.nllb"].append("NllbTokenizerFast")
_import_structure["models.openai"].append("OpenAIGPTTokenizerFast")
_import_structure["models.pegasus"].append("PegasusTokenizerFast")
_import_structure["models.realm"].append("RealmTokenizerFast")
_import_structure["models.reformer"].append("ReformerTokenizerFast")
_import_structure["models.rembert"].append("RemBertTokenizerFast")
_import_structure["models.retribert"].append("RetriBertTokenizerFast")
_import_structure["models.roberta"].append("RobertaTokenizerFast")
_import_structure["models.roformer"].append("RoFormerTokenizerFast")
_import_structure["models.splinter"].append("SplinterTokenizerFast")
_import_structure["models.squeezebert"].append("SqueezeBertTokenizerFast")
_import_structure["models.t5"].append("T5TokenizerFast")
_import_structure["models.xglm"].append("XGLMTokenizerFast")
_import_structure["models.xlm_roberta"].append("XLMRobertaTokenizerFast")
_import_structure["models.xlnet"].append("XLNetTokenizerFast")
_import_structure["tokenization_utils_fast"] = ["PreTrainedTokenizerFast"]
try:
if not (is_sentencepiece_available() and is_tokenizers_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_sentencepiece_and_tokenizers_objects
_import_structure["utils.dummy_sentencepiece_and_tokenizers_objects"] = [
name for name in dir(dummy_sentencepiece_and_tokenizers_objects) if not name.startswith("_")
]
else:
_import_structure["convert_slow_tokenizer"] = ["SLOW_TO_FAST_CONVERTERS", "convert_slow_tokenizer"]
# Speech-specific objects
try:
if not is_speech_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_speech_objects
_import_structure["utils.dummy_speech_objects"] = [
name for name in dir(dummy_speech_objects) if not name.startswith("_")
]
else:
_import_structure["models.audio_spectrogram_transformer"].append("ASTFeatureExtractor")
_import_structure["models.mctct"].append("MCTCTFeatureExtractor")
_import_structure["models.speech_to_text"].append("Speech2TextFeatureExtractor")
# Tensorflow-text-specific objects
try:
if not is_tensorflow_text_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_tensorflow_text_objects
_import_structure["utils.dummy_tensorflow_text_objects"] = [
name for name in dir(dummy_tensorflow_text_objects) if not name.startswith("_")
]
else:
_import_structure["models.bert"].append("TFBertTokenizer")
# keras-nlp-specific objects
try:
if not is_keras_nlp_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_keras_nlp_objects
_import_structure["utils.dummy_keras_nlp_objects"] = [
name for name in dir(dummy_keras_nlp_objects) if not name.startswith("_")
]
else:
_import_structure["models.gpt2"].append("TFGPT2Tokenizer")
try:
if not (is_sentencepiece_available() and is_speech_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_sentencepiece_and_speech_objects
_import_structure["utils.dummy_sentencepiece_and_speech_objects"] = [
name for name in dir(dummy_sentencepiece_and_speech_objects) if not name.startswith("_")
]
else:
_import_structure["models.speech_to_text"].append("Speech2TextProcessor")
# Vision-specific objects
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_vision_objects
_import_structure["utils.dummy_vision_objects"] = [
name for name in dir(dummy_vision_objects) if not name.startswith("_")
]
else:
_import_structure["image_processing_utils"] = ["ImageProcessingMixin"]
_import_structure["image_transforms"] = ["rescale", "resize", "to_pil_image"]
_import_structure["image_utils"] = ["ImageFeatureExtractionMixin"]
_import_structure["models.beit"].extend(["BeitFeatureExtractor", "BeitImageProcessor"])
_import_structure["models.bit"].extend(["BitImageProcessor"])
_import_structure["models.chinese_clip"].extend(["ChineseCLIPFeatureExtractor", "ChineseCLIPImageProcessor"])
_import_structure["models.clip"].extend(["CLIPFeatureExtractor", "CLIPImageProcessor"])
_import_structure["models.conditional_detr"].extend(
["ConditionalDetrFeatureExtractor", "ConditionalDetrImageProcessor"]
)
_import_structure["models.convnext"].extend(["ConvNextFeatureExtractor", "ConvNextImageProcessor"])
_import_structure["models.deformable_detr"].extend(
["DeformableDetrFeatureExtractor", "DeformableDetrImageProcessor"]
)
_import_structure["models.deit"].extend(["DeiTFeatureExtractor", "DeiTImageProcessor"])
_import_structure["models.detr"].extend(["DetrFeatureExtractor", "DetrImageProcessor"])
_import_structure["models.donut"].extend(["DonutFeatureExtractor", "DonutImageProcessor"])
_import_structure["models.dpt"].extend(["DPTFeatureExtractor", "DPTImageProcessor"])
_import_structure["models.flava"].extend(["FlavaFeatureExtractor", "FlavaImageProcessor", "FlavaProcessor"])
_import_structure["models.glpn"].extend(["GLPNFeatureExtractor", "GLPNImageProcessor"])
_import_structure["models.imagegpt"].extend(["ImageGPTFeatureExtractor", "ImageGPTImageProcessor"])
_import_structure["models.layoutlmv2"].extend(["LayoutLMv2FeatureExtractor", "LayoutLMv2ImageProcessor"])
_import_structure["models.layoutlmv3"].extend(["LayoutLMv3FeatureExtractor", "LayoutLMv3ImageProcessor"])
_import_structure["models.levit"].extend(["LevitFeatureExtractor", "LevitImageProcessor"])
_import_structure["models.maskformer"].extend(["MaskFormerFeatureExtractor", "MaskFormerImageProcessor"])
_import_structure["models.mobilenet_v1"].extend(["MobileNetV1FeatureExtractor", "MobileNetV1ImageProcessor"])
_import_structure["models.mobilenet_v2"].extend(["MobileNetV2FeatureExtractor", "MobileNetV2ImageProcessor"])
_import_structure["models.mobilevit"].extend(["MobileViTFeatureExtractor", "MobileViTImageProcessor"])
_import_structure["models.owlvit"].extend(["OwlViTFeatureExtractor", "OwlViTImageProcessor"])
_import_structure["models.perceiver"].extend(["PerceiverFeatureExtractor", "PerceiverImageProcessor"])
_import_structure["models.poolformer"].extend(["PoolFormerFeatureExtractor", "PoolFormerImageProcessor"])
_import_structure["models.segformer"].extend(["SegformerFeatureExtractor", "SegformerImageProcessor"])
_import_structure["models.videomae"].extend(["VideoMAEFeatureExtractor", "VideoMAEImageProcessor"])
_import_structure["models.vilt"].extend(["ViltFeatureExtractor", "ViltImageProcessor", "ViltProcessor"])
_import_structure["models.vit"].extend(["ViTFeatureExtractor", "ViTImageProcessor"])
_import_structure["models.vit_hybrid"].extend(["ViTHybridImageProcessor"])
_import_structure["models.yolos"].extend(["YolosFeatureExtractor", "YolosImageProcessor"])
# Timm-backed objects
try:
if not (is_timm_available() and is_vision_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_timm_and_vision_objects
_import_structure["utils.dummy_timm_and_vision_objects"] = [
name for name in dir(dummy_timm_and_vision_objects) if not name.startswith("_")
]
else:
_import_structure["models.deformable_detr"].extend(
[
"DEFORMABLE_DETR_PRETRAINED_MODEL_ARCHIVE_LIST",
"DeformableDetrForObjectDetection",
"DeformableDetrModel",
"DeformableDetrPreTrainedModel",
]
)
_import_structure["models.detr"].extend(
[
"DETR_PRETRAINED_MODEL_ARCHIVE_LIST",
"DetrForObjectDetection",
"DetrForSegmentation",
"DetrModel",
"DetrPreTrainedModel",
]
)
_import_structure["models.table_transformer"].extend(
[
"TABLE_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TableTransformerForObjectDetection",
"TableTransformerModel",
"TableTransformerPreTrainedModel",
]
)
_import_structure["models.conditional_detr"].extend(
[
"CONDITIONAL_DETR_PRETRAINED_MODEL_ARCHIVE_LIST",
"ConditionalDetrForObjectDetection",
"ConditionalDetrForSegmentation",
"ConditionalDetrModel",
"ConditionalDetrPreTrainedModel",
]
)
# PyTorch-backed objects
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_pt_objects
_import_structure["utils.dummy_pt_objects"] = [name for name in dir(dummy_pt_objects) if not name.startswith("_")]
else:
_import_structure["activations"] = []
_import_structure["benchmark.benchmark"] = ["PyTorchBenchmark"]
_import_structure["benchmark.benchmark_args"] = ["PyTorchBenchmarkArguments"]
_import_structure["data.datasets"] = [
"GlueDataset",
"GlueDataTrainingArguments",
"LineByLineTextDataset",
"LineByLineWithRefDataset",
"LineByLineWithSOPTextDataset",
"SquadDataset",
"SquadDataTrainingArguments",
"TextDataset",
"TextDatasetForNextSentencePrediction",
]
_import_structure["deepspeed"] = []
_import_structure["generation"].extend(
[
"BeamScorer",
"BeamSearchScorer",
"ConstrainedBeamSearchScorer",
"Constraint",
"ConstraintListState",
"DisjunctiveConstraint",
"ForcedBOSTokenLogitsProcessor",
"ForcedEOSTokenLogitsProcessor",
"GenerationMixin",
"HammingDiversityLogitsProcessor",
"InfNanRemoveLogitsProcessor",
"LogitsProcessor",
"LogitsProcessorList",
"LogitsWarper",
"MaxLengthCriteria",
"MaxTimeCriteria",
"MinLengthLogitsProcessor",
"NoBadWordsLogitsProcessor",
"NoRepeatNGramLogitsProcessor",
"PhrasalConstraint",
"PrefixConstrainedLogitsProcessor",
"RepetitionPenaltyLogitsProcessor",
"StoppingCriteria",
"StoppingCriteriaList",
"TemperatureLogitsWarper",
"TopKLogitsWarper",
"TopPLogitsWarper",
"TypicalLogitsWarper",
"top_k_top_p_filtering",
]
)
_import_structure["generation_utils"] = []
_import_structure["modeling_outputs"] = []
_import_structure["modeling_utils"] = ["PreTrainedModel"]
# PyTorch models structure
_import_structure["models.albert"].extend(
[
"ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"AlbertForMaskedLM",
"AlbertForMultipleChoice",
"AlbertForPreTraining",
"AlbertForQuestionAnswering",
"AlbertForSequenceClassification",
"AlbertForTokenClassification",
"AlbertModel",
"AlbertPreTrainedModel",
"load_tf_weights_in_albert",
]
)
_import_structure["models.audio_spectrogram_transformer"].extend(
[
"AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"ASTForAudioClassification",
"ASTModel",
"ASTPreTrainedModel",
]
)
_import_structure["models.auto"].extend(
[
"MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING",
"MODEL_FOR_AUDIO_XVECTOR_MAPPING",
"MODEL_FOR_BACKBONE_MAPPING",
"MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING",
"MODEL_FOR_CAUSAL_LM_MAPPING",
"MODEL_FOR_CTC_MAPPING",
"MODEL_FOR_DEPTH_ESTIMATION_MAPPING",
"MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING",
"MODEL_FOR_IMAGE_SEGMENTATION_MAPPING",
"MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING",
"MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING",
"MODEL_FOR_MASKED_LM_MAPPING",
"MODEL_FOR_MULTIPLE_CHOICE_MAPPING",
"MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING",
"MODEL_FOR_OBJECT_DETECTION_MAPPING",
"MODEL_FOR_PRETRAINING_MAPPING",
"MODEL_FOR_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING",
"MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING",
"MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING",
"MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING",
"MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING",
"MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING",
"MODEL_FOR_VISION_2_SEQ_MAPPING",
"MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING",
"MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING",
"MODEL_MAPPING",
"MODEL_WITH_LM_HEAD_MAPPING",
"AutoBackbone",
"AutoModel",
"AutoModelForAudioClassification",
"AutoModelForAudioFrameClassification",
"AutoModelForAudioXVector",
"AutoModelForCausalLM",
"AutoModelForCTC",
"AutoModelForDepthEstimation",
"AutoModelForDocumentQuestionAnswering",
"AutoModelForImageClassification",
"AutoModelForImageSegmentation",
"AutoModelForInstanceSegmentation",
"AutoModelForMaskedImageModeling",
"AutoModelForMaskedLM",
"AutoModelForMultipleChoice",
"AutoModelForNextSentencePrediction",
"AutoModelForObjectDetection",
"AutoModelForPreTraining",
"AutoModelForQuestionAnswering",
"AutoModelForSemanticSegmentation",
"AutoModelForSeq2SeqLM",
"AutoModelForSequenceClassification",
"AutoModelForSpeechSeq2Seq",
"AutoModelForTableQuestionAnswering",
"AutoModelForTokenClassification",
"AutoModelForVideoClassification",
"AutoModelForVision2Seq",
"AutoModelForVisualQuestionAnswering",
"AutoModelForZeroShotObjectDetection",
"AutoModelWithLMHead",
]
)
_import_structure["models.bart"].extend(
[
"BART_PRETRAINED_MODEL_ARCHIVE_LIST",
"BartForCausalLM",
"BartForConditionalGeneration",
"BartForQuestionAnswering",
"BartForSequenceClassification",
"BartModel",
"BartPretrainedModel",
"PretrainedBartModel",
]
)
_import_structure["models.beit"].extend(
[
"BEIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BeitForImageClassification",
"BeitForMaskedImageModeling",
"BeitForSemanticSegmentation",
"BeitModel",
"BeitPreTrainedModel",
]
)
_import_structure["models.bert"].extend(
[
"BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BertForMaskedLM",
"BertForMultipleChoice",
"BertForNextSentencePrediction",
"BertForPreTraining",
"BertForQuestionAnswering",
"BertForSequenceClassification",
"BertForTokenClassification",
"BertLayer",
"BertLMHeadModel",
"BertModel",
"BertPreTrainedModel",
"load_tf_weights_in_bert",
]
)
_import_structure["models.bert_generation"].extend(
[
"BertGenerationDecoder",
"BertGenerationEncoder",
"BertGenerationPreTrainedModel",
"load_tf_weights_in_bert_generation",
]
)
_import_structure["models.big_bird"].extend(
[
"BIG_BIRD_PRETRAINED_MODEL_ARCHIVE_LIST",
"BigBirdForCausalLM",
"BigBirdForMaskedLM",
"BigBirdForMultipleChoice",
"BigBirdForPreTraining",
"BigBirdForQuestionAnswering",
"BigBirdForSequenceClassification",
"BigBirdForTokenClassification",
"BigBirdLayer",
"BigBirdModel",
"BigBirdPreTrainedModel",
"load_tf_weights_in_big_bird",
]
)
_import_structure["models.bigbird_pegasus"].extend(
[
"BIGBIRD_PEGASUS_PRETRAINED_MODEL_ARCHIVE_LIST",
"BigBirdPegasusForCausalLM",
"BigBirdPegasusForConditionalGeneration",
"BigBirdPegasusForQuestionAnswering",
"BigBirdPegasusForSequenceClassification",
"BigBirdPegasusModel",
"BigBirdPegasusPreTrainedModel",
]
)
_import_structure["models.biogpt"].extend(
[
"BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BioGptForCausalLM",
"BioGptModel",
"BioGptPreTrainedModel",
]
)
_import_structure["models.bit"].extend(
[
"BIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BitBackbone",
"BitForImageClassification",
"BitModel",
"BitPreTrainedModel",
]
)
_import_structure["models.blenderbot"].extend(
[
"BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST",
"BlenderbotForCausalLM",
"BlenderbotForConditionalGeneration",
"BlenderbotModel",
"BlenderbotPreTrainedModel",
]
)
_import_structure["models.blenderbot_small"].extend(
[
"BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST",
"BlenderbotSmallForCausalLM",
"BlenderbotSmallForConditionalGeneration",
"BlenderbotSmallModel",
"BlenderbotSmallPreTrainedModel",
]
)
_import_structure["models.bloom"].extend(
[
"BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST",
"BloomForCausalLM",
"BloomForQuestionAnswering",
"BloomForSequenceClassification",
"BloomForTokenClassification",
"BloomModel",
"BloomPreTrainedModel",
]
)
_import_structure["models.camembert"].extend(
[
"CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"CamembertForCausalLM",
"CamembertForMaskedLM",
"CamembertForMultipleChoice",
"CamembertForQuestionAnswering",
"CamembertForSequenceClassification",
"CamembertForTokenClassification",
"CamembertModel",
"CamembertPreTrainedModel",
]
)
_import_structure["models.canine"].extend(
[
"CANINE_PRETRAINED_MODEL_ARCHIVE_LIST",
"CanineForMultipleChoice",
"CanineForQuestionAnswering",
"CanineForSequenceClassification",
"CanineForTokenClassification",
"CanineLayer",
"CanineModel",
"CaninePreTrainedModel",
"load_tf_weights_in_canine",
]
)
_import_structure["models.chinese_clip"].extend(
[
"CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"ChineseCLIPModel",
"ChineseCLIPPreTrainedModel",
"ChineseCLIPTextModel",
"ChineseCLIPVisionModel",
]
)
_import_structure["models.clip"].extend(
[
"CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"CLIPModel",
"CLIPPreTrainedModel",
"CLIPTextModel",
"CLIPTextModelWithProjection",
"CLIPVisionModel",
"CLIPVisionModelWithProjection",
]
)
_import_structure["models.clipseg"].extend(
[
"CLIPSEG_PRETRAINED_MODEL_ARCHIVE_LIST",
"CLIPSegForImageSegmentation",
"CLIPSegModel",
"CLIPSegPreTrainedModel",
"CLIPSegTextModel",
"CLIPSegVisionModel",
]
)
_import_structure["models.codegen"].extend(
[
"CODEGEN_PRETRAINED_MODEL_ARCHIVE_LIST",
"CodeGenForCausalLM",
"CodeGenModel",
"CodeGenPreTrainedModel",
]
)
_import_structure["models.convbert"].extend(
[
"CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ConvBertForMaskedLM",
"ConvBertForMultipleChoice",
"ConvBertForQuestionAnswering",
"ConvBertForSequenceClassification",
"ConvBertForTokenClassification",
"ConvBertLayer",
"ConvBertModel",
"ConvBertPreTrainedModel",
"load_tf_weights_in_convbert",
]
)
_import_structure["models.convnext"].extend(
[
"CONVNEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ConvNextForImageClassification",
"ConvNextModel",
"ConvNextPreTrainedModel",
]
)
_import_structure["models.ctrl"].extend(
[
"CTRL_PRETRAINED_MODEL_ARCHIVE_LIST",
"CTRLForSequenceClassification",
"CTRLLMHeadModel",
"CTRLModel",
"CTRLPreTrainedModel",
]
)
_import_structure["models.cvt"].extend(
[
"CVT_PRETRAINED_MODEL_ARCHIVE_LIST",
"CvtForImageClassification",
"CvtModel",
"CvtPreTrainedModel",
]
)
_import_structure["models.data2vec"].extend(
[
"DATA2VEC_AUDIO_PRETRAINED_MODEL_ARCHIVE_LIST",
"DATA2VEC_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DATA2VEC_VISION_PRETRAINED_MODEL_ARCHIVE_LIST",
"Data2VecAudioForAudioFrameClassification",
"Data2VecAudioForCTC",
"Data2VecAudioForSequenceClassification",
"Data2VecAudioForXVector",
"Data2VecAudioModel",
"Data2VecAudioPreTrainedModel",
"Data2VecTextForCausalLM",
"Data2VecTextForMaskedLM",
"Data2VecTextForMultipleChoice",
"Data2VecTextForQuestionAnswering",
"Data2VecTextForSequenceClassification",
"Data2VecTextForTokenClassification",
"Data2VecTextModel",
"Data2VecTextPreTrainedModel",
"Data2VecVisionForImageClassification",
"Data2VecVisionForSemanticSegmentation",
"Data2VecVisionModel",
"Data2VecVisionPreTrainedModel",
]
)
_import_structure["models.deberta"].extend(
[
"DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"DebertaForMaskedLM",
"DebertaForQuestionAnswering",
"DebertaForSequenceClassification",
"DebertaForTokenClassification",
"DebertaModel",
"DebertaPreTrainedModel",
]
)
_import_structure["models.deberta_v2"].extend(
[
"DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST",
"DebertaV2ForMaskedLM",
"DebertaV2ForMultipleChoice",
"DebertaV2ForQuestionAnswering",
"DebertaV2ForSequenceClassification",
"DebertaV2ForTokenClassification",
"DebertaV2Model",
"DebertaV2PreTrainedModel",
]
)
_import_structure["models.decision_transformer"].extend(
[
"DECISION_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DecisionTransformerGPT2Model",
"DecisionTransformerGPT2PreTrainedModel",
"DecisionTransformerModel",
"DecisionTransformerPreTrainedModel",
]
)
_import_structure["models.deit"].extend(
[
"DEIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DeiTForImageClassification",
"DeiTForImageClassificationWithTeacher",
"DeiTForMaskedImageModeling",
"DeiTModel",
"DeiTPreTrainedModel",
]
)
_import_structure["models.dinat"].extend(
[
"DINAT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DinatForImageClassification",
"DinatModel",
"DinatPreTrainedModel",
]
)
_import_structure["models.distilbert"].extend(
[
"DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DistilBertForMaskedLM",
"DistilBertForMultipleChoice",
"DistilBertForQuestionAnswering",
"DistilBertForSequenceClassification",
"DistilBertForTokenClassification",
"DistilBertModel",
"DistilBertPreTrainedModel",
]
)
_import_structure["models.donut"].extend(
[
"DONUT_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST",
"DonutSwinModel",
"DonutSwinPreTrainedModel",
]
)
_import_structure["models.dpr"].extend(
[
"DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPRContextEncoder",
"DPRPretrainedContextEncoder",
"DPRPreTrainedModel",
"DPRPretrainedQuestionEncoder",
"DPRPretrainedReader",
"DPRQuestionEncoder",
"DPRReader",
]
)
_import_structure["models.dpt"].extend(
[
"DPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"DPTForDepthEstimation",
"DPTForSemanticSegmentation",
"DPTModel",
"DPTPreTrainedModel",
]
)
_import_structure["models.electra"].extend(
[
"ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"ElectraForCausalLM",
"ElectraForMaskedLM",
"ElectraForMultipleChoice",
"ElectraForPreTraining",
"ElectraForQuestionAnswering",
"ElectraForSequenceClassification",
"ElectraForTokenClassification",
"ElectraModel",
"ElectraPreTrainedModel",
"load_tf_weights_in_electra",
]
)
_import_structure["models.encoder_decoder"].append("EncoderDecoderModel")
_import_structure["models.ernie"].extend(
[
"ERNIE_PRETRAINED_MODEL_ARCHIVE_LIST",
"ErnieForCausalLM",
"ErnieForMaskedLM",
"ErnieForMultipleChoice",
"ErnieForNextSentencePrediction",
"ErnieForPreTraining",
"ErnieForQuestionAnswering",
"ErnieForSequenceClassification",
"ErnieForTokenClassification",
"ErnieModel",
"ErniePreTrainedModel",
]
)
_import_structure["models.esm"].extend(
[
"ESM_PRETRAINED_MODEL_ARCHIVE_LIST",
"EsmFoldPreTrainedModel",
"EsmForMaskedLM",
"EsmForProteinFolding",
"EsmForSequenceClassification",
"EsmForTokenClassification",
"EsmModel",
"EsmPreTrainedModel",
]
)
_import_structure["models.flaubert"].extend(
[
"FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"FlaubertForMultipleChoice",
"FlaubertForQuestionAnswering",
"FlaubertForQuestionAnsweringSimple",
"FlaubertForSequenceClassification",
"FlaubertForTokenClassification",
"FlaubertModel",
"FlaubertPreTrainedModel",
"FlaubertWithLMHeadModel",
]
)
_import_structure["models.flava"].extend(
[
"FLAVA_PRETRAINED_MODEL_ARCHIVE_LIST",
"FlavaForPreTraining",
"FlavaImageCodebook",
"FlavaImageModel",
"FlavaModel",
"FlavaMultimodalModel",
"FlavaPreTrainedModel",
"FlavaTextModel",
]
)
_import_structure["models.fnet"].extend(
[
"FNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"FNetForMaskedLM",
"FNetForMultipleChoice",
"FNetForNextSentencePrediction",
"FNetForPreTraining",
"FNetForQuestionAnswering",
"FNetForSequenceClassification",
"FNetForTokenClassification",
"FNetLayer",
"FNetModel",
"FNetPreTrainedModel",
]
)
_import_structure["models.fsmt"].extend(["FSMTForConditionalGeneration", "FSMTModel", "PretrainedFSMTModel"])
_import_structure["models.funnel"].extend(
[
"FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST",
"FunnelBaseModel",
"FunnelForMaskedLM",
"FunnelForMultipleChoice",
"FunnelForPreTraining",
"FunnelForQuestionAnswering",
"FunnelForSequenceClassification",
"FunnelForTokenClassification",
"FunnelModel",
"FunnelPreTrainedModel",
"load_tf_weights_in_funnel",
]
)
_import_structure["models.glpn"].extend(
[
"GLPN_PRETRAINED_MODEL_ARCHIVE_LIST",
"GLPNForDepthEstimation",
"GLPNModel",
"GLPNPreTrainedModel",
]
)
_import_structure["models.gpt2"].extend(
[
"GPT2_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPT2DoubleHeadsModel",
"GPT2ForSequenceClassification",
"GPT2ForTokenClassification",
"GPT2LMHeadModel",
"GPT2Model",
"GPT2PreTrainedModel",
"load_tf_weights_in_gpt2",
]
)
_import_structure["models.gpt_neo"].extend(
[
"GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoForCausalLM",
"GPTNeoForSequenceClassification",
"GPTNeoModel",
"GPTNeoPreTrainedModel",
"load_tf_weights_in_gpt_neo",
]
)
_import_structure["models.gpt_neox"].extend(
[
"GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoXForCausalLM",
"GPTNeoXLayer",
"GPTNeoXModel",
"GPTNeoXPreTrainedModel",
]
)
_import_structure["models.gpt_neox_japanese"].extend(
[
"GPT_NEOX_JAPANESE_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoXJapaneseForCausalLM",
"GPTNeoXJapaneseLayer",
"GPTNeoXJapaneseModel",
"GPTNeoXJapanesePreTrainedModel",
]
)
_import_structure["models.gptj"].extend(
[
"GPTJ_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTJForCausalLM",
"GPTJForQuestionAnswering",
"GPTJForSequenceClassification",
"GPTJModel",
"GPTJPreTrainedModel",
]
)
_import_structure["models.groupvit"].extend(
[
"GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"GroupViTModel",
"GroupViTPreTrainedModel",
"GroupViTTextModel",
"GroupViTVisionModel",
]
)
_import_structure["models.hubert"].extend(
[
"HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"HubertForCTC",
"HubertForSequenceClassification",
"HubertModel",
"HubertPreTrainedModel",
]
)
_import_structure["models.ibert"].extend(
[
"IBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"IBertForMaskedLM",
"IBertForMultipleChoice",
"IBertForQuestionAnswering",
"IBertForSequenceClassification",
"IBertForTokenClassification",
"IBertModel",
"IBertPreTrainedModel",
]
)
_import_structure["models.imagegpt"].extend(
[
"IMAGEGPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ImageGPTForCausalImageModeling",
"ImageGPTForImageClassification",
"ImageGPTModel",
"ImageGPTPreTrainedModel",
"load_tf_weights_in_imagegpt",
]
)
_import_structure["models.jukebox"].extend(
[
"JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST",
"JukeboxModel",
"JukeboxPreTrainedModel",
"JukeboxPrior",
"JukeboxVQVAE",
]
)
_import_structure["models.layoutlm"].extend(
[
"LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"LayoutLMForMaskedLM",
"LayoutLMForQuestionAnswering",
"LayoutLMForSequenceClassification",
"LayoutLMForTokenClassification",
"LayoutLMModel",
"LayoutLMPreTrainedModel",
]
)
_import_structure["models.layoutlmv2"].extend(
[
"LAYOUTLMV2_PRETRAINED_MODEL_ARCHIVE_LIST",
"LayoutLMv2ForQuestionAnswering",
"LayoutLMv2ForSequenceClassification",
"LayoutLMv2ForTokenClassification",
"LayoutLMv2Model",
"LayoutLMv2PreTrainedModel",
]
)
_import_structure["models.layoutlmv3"].extend(
[
"LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST",
"LayoutLMv3ForQuestionAnswering",
"LayoutLMv3ForSequenceClassification",
"LayoutLMv3ForTokenClassification",
"LayoutLMv3Model",
"LayoutLMv3PreTrainedModel",
]
)
_import_structure["models.led"].extend(
[
"LED_PRETRAINED_MODEL_ARCHIVE_LIST",
"LEDForConditionalGeneration",
"LEDForQuestionAnswering",
"LEDForSequenceClassification",
"LEDModel",
"LEDPreTrainedModel",
]
)
_import_structure["models.levit"].extend(
[
"LEVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"LevitForImageClassification",
"LevitForImageClassificationWithTeacher",
"LevitModel",
"LevitPreTrainedModel",
]
)
_import_structure["models.lilt"].extend(
[
"LILT_PRETRAINED_MODEL_ARCHIVE_LIST",
"LiltForQuestionAnswering",
"LiltForSequenceClassification",
"LiltForTokenClassification",
"LiltModel",
"LiltPreTrainedModel",
]
)
_import_structure["models.longformer"].extend(
[
"LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"LongformerForMaskedLM",
"LongformerForMultipleChoice",
"LongformerForQuestionAnswering",
"LongformerForSequenceClassification",
"LongformerForTokenClassification",
"LongformerModel",
"LongformerPreTrainedModel",
"LongformerSelfAttention",
]
)
_import_structure["models.longt5"].extend(
[
"LONGT5_PRETRAINED_MODEL_ARCHIVE_LIST",
"LongT5EncoderModel",
"LongT5ForConditionalGeneration",
"LongT5Model",
"LongT5PreTrainedModel",
]
)
_import_structure["models.luke"].extend(
[
"LUKE_PRETRAINED_MODEL_ARCHIVE_LIST",
"LukeForEntityClassification",
"LukeForEntityPairClassification",
"LukeForEntitySpanClassification",
"LukeForMaskedLM",
"LukeForMultipleChoice",
"LukeForQuestionAnswering",
"LukeForSequenceClassification",
"LukeForTokenClassification",
"LukeModel",
"LukePreTrainedModel",
]
)
_import_structure["models.lxmert"].extend(
[
"LxmertEncoder",
"LxmertForPreTraining",
"LxmertForQuestionAnswering",
"LxmertModel",
"LxmertPreTrainedModel",
"LxmertVisualFeatureEncoder",
"LxmertXLayer",
]
)
_import_structure["models.m2m_100"].extend(
[
"M2M_100_PRETRAINED_MODEL_ARCHIVE_LIST",
"M2M100ForConditionalGeneration",
"M2M100Model",
"M2M100PreTrainedModel",
]
)
_import_structure["models.marian"].extend(["MarianForCausalLM", "MarianModel", "MarianMTModel"])
_import_structure["models.markuplm"].extend(
[
"MARKUPLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"MarkupLMForQuestionAnswering",
"MarkupLMForSequenceClassification",
"MarkupLMForTokenClassification",
"MarkupLMModel",
"MarkupLMPreTrainedModel",
]
)
_import_structure["models.maskformer"].extend(
[
"MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"MaskFormerForInstanceSegmentation",
"MaskFormerModel",
"MaskFormerPreTrainedModel",
"MaskFormerSwinBackbone",
]
)
_import_structure["models.mbart"].extend(
[
"MBartForCausalLM",
"MBartForConditionalGeneration",
"MBartForQuestionAnswering",
"MBartForSequenceClassification",
"MBartModel",
"MBartPreTrainedModel",
]
)
_import_structure["models.mctct"].extend(
[
"MCTCT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MCTCTForCTC",
"MCTCTModel",
"MCTCTPreTrainedModel",
]
)
_import_structure["models.megatron_bert"].extend(
[
"MEGATRON_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MegatronBertForCausalLM",
"MegatronBertForMaskedLM",
"MegatronBertForMultipleChoice",
"MegatronBertForNextSentencePrediction",
"MegatronBertForPreTraining",
"MegatronBertForQuestionAnswering",
"MegatronBertForSequenceClassification",
"MegatronBertForTokenClassification",
"MegatronBertModel",
"MegatronBertPreTrainedModel",
]
)
_import_structure["models.mmbt"].extend(["MMBTForClassification", "MMBTModel", "ModalEmbeddings"])
_import_structure["models.mobilebert"].extend(
[
"MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileBertForMaskedLM",
"MobileBertForMultipleChoice",
"MobileBertForNextSentencePrediction",
"MobileBertForPreTraining",
"MobileBertForQuestionAnswering",
"MobileBertForSequenceClassification",
"MobileBertForTokenClassification",
"MobileBertLayer",
"MobileBertModel",
"MobileBertPreTrainedModel",
"load_tf_weights_in_mobilebert",
]
)
_import_structure["models.mobilenet_v1"].extend(
[
"MOBILENET_V1_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileNetV1ForImageClassification",
"MobileNetV1Model",
"MobileNetV1PreTrainedModel",
"load_tf_weights_in_mobilenet_v1",
]
)
_import_structure["models.mobilenet_v2"].extend(
[
"MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileNetV2ForImageClassification",
"MobileNetV2ForSemanticSegmentation",
"MobileNetV2Model",
"MobileNetV2PreTrainedModel",
"load_tf_weights_in_mobilenet_v2",
]
)
_import_structure["models.mobilevit"].extend(
[
"MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"MobileViTForImageClassification",
"MobileViTForSemanticSegmentation",
"MobileViTModel",
"MobileViTPreTrainedModel",
]
)
_import_structure["models.mpnet"].extend(
[
"MPNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"MPNetForMaskedLM",
"MPNetForMultipleChoice",
"MPNetForQuestionAnswering",
"MPNetForSequenceClassification",
"MPNetForTokenClassification",
"MPNetLayer",
"MPNetModel",
"MPNetPreTrainedModel",
]
)
_import_structure["models.mt5"].extend(["MT5EncoderModel", "MT5ForConditionalGeneration", "MT5Model"])
_import_structure["models.mvp"].extend(
[
"MVP_PRETRAINED_MODEL_ARCHIVE_LIST",
"MvpForCausalLM",
"MvpForConditionalGeneration",
"MvpForQuestionAnswering",
"MvpForSequenceClassification",
"MvpModel",
"MvpPreTrainedModel",
]
)
_import_structure["models.nat"].extend(
[
"NAT_PRETRAINED_MODEL_ARCHIVE_LIST",
"NatForImageClassification",
"NatModel",
"NatPreTrainedModel",
]
)
_import_structure["models.nezha"].extend(
[
"NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST",
"NezhaForMaskedLM",
"NezhaForMultipleChoice",
"NezhaForNextSentencePrediction",
"NezhaForPreTraining",
"NezhaForQuestionAnswering",
"NezhaForSequenceClassification",
"NezhaForTokenClassification",
"NezhaModel",
"NezhaPreTrainedModel",
]
)
_import_structure["models.nystromformer"].extend(
[
"NYSTROMFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"NystromformerForMaskedLM",
"NystromformerForMultipleChoice",
"NystromformerForQuestionAnswering",
"NystromformerForSequenceClassification",
"NystromformerForTokenClassification",
"NystromformerLayer",
"NystromformerModel",
"NystromformerPreTrainedModel",
]
)
_import_structure["models.openai"].extend(
[
"OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"OpenAIGPTDoubleHeadsModel",
"OpenAIGPTForSequenceClassification",
"OpenAIGPTLMHeadModel",
"OpenAIGPTModel",
"OpenAIGPTPreTrainedModel",
"load_tf_weights_in_openai_gpt",
]
)
_import_structure["models.opt"].extend(
[
"OPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"OPTForCausalLM",
"OPTForQuestionAnswering",
"OPTForSequenceClassification",
"OPTModel",
"OPTPreTrainedModel",
]
)
_import_structure["models.owlvit"].extend(
[
"OWLVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"OwlViTForObjectDetection",
"OwlViTModel",
"OwlViTPreTrainedModel",
"OwlViTTextModel",
"OwlViTVisionModel",
]
)
_import_structure["models.pegasus"].extend(
["PegasusForCausalLM", "PegasusForConditionalGeneration", "PegasusModel", "PegasusPreTrainedModel"]
)
_import_structure["models.pegasus_x"].extend(
[
"PEGASUS_X_PRETRAINED_MODEL_ARCHIVE_LIST",
"PegasusXForConditionalGeneration",
"PegasusXModel",
"PegasusXPreTrainedModel",
]
)
_import_structure["models.perceiver"].extend(
[
"PERCEIVER_PRETRAINED_MODEL_ARCHIVE_LIST",
"PerceiverForImageClassificationConvProcessing",
"PerceiverForImageClassificationFourier",
"PerceiverForImageClassificationLearned",
"PerceiverForMaskedLM",
"PerceiverForMultimodalAutoencoding",
"PerceiverForOpticalFlow",
"PerceiverForSequenceClassification",
"PerceiverLayer",
"PerceiverModel",
"PerceiverPreTrainedModel",
]
)
_import_structure["models.plbart"].extend(
[
"PLBART_PRETRAINED_MODEL_ARCHIVE_LIST",
"PLBartForCausalLM",
"PLBartForConditionalGeneration",
"PLBartForSequenceClassification",
"PLBartModel",
"PLBartPreTrainedModel",
]
)
_import_structure["models.poolformer"].extend(
[
"POOLFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"PoolFormerForImageClassification",
"PoolFormerModel",
"PoolFormerPreTrainedModel",
]
)
_import_structure["models.prophetnet"].extend(
[
"PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"ProphetNetDecoder",
"ProphetNetEncoder",
"ProphetNetForCausalLM",
"ProphetNetForConditionalGeneration",
"ProphetNetModel",
"ProphetNetPreTrainedModel",
]
)
_import_structure["models.qdqbert"].extend(
[
"QDQBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"QDQBertForMaskedLM",
"QDQBertForMultipleChoice",
"QDQBertForNextSentencePrediction",
"QDQBertForQuestionAnswering",
"QDQBertForSequenceClassification",
"QDQBertForTokenClassification",
"QDQBertLayer",
"QDQBertLMHeadModel",
"QDQBertModel",
"QDQBertPreTrainedModel",
"load_tf_weights_in_qdqbert",
]
)
_import_structure["models.rag"].extend(
["RagModel", "RagPreTrainedModel", "RagSequenceForGeneration", "RagTokenForGeneration"]
)
_import_structure["models.realm"].extend(
[
"REALM_PRETRAINED_MODEL_ARCHIVE_LIST",
"RealmEmbedder",
"RealmForOpenQA",
"RealmKnowledgeAugEncoder",
"RealmPreTrainedModel",
"RealmReader",
"RealmRetriever",
"RealmScorer",
"load_tf_weights_in_realm",
]
)
_import_structure["models.reformer"].extend(
[
"REFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"ReformerAttention",
"ReformerForMaskedLM",
"ReformerForQuestionAnswering",
"ReformerForSequenceClassification",
"ReformerLayer",
"ReformerModel",
"ReformerModelWithLMHead",
"ReformerPreTrainedModel",
]
)
_import_structure["models.regnet"].extend(
[
"REGNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"RegNetForImageClassification",
"RegNetModel",
"RegNetPreTrainedModel",
]
)
_import_structure["models.rembert"].extend(
[
"REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"RemBertForCausalLM",
"RemBertForMaskedLM",
"RemBertForMultipleChoice",
"RemBertForQuestionAnswering",
"RemBertForSequenceClassification",
"RemBertForTokenClassification",
"RemBertLayer",
"RemBertModel",
"RemBertPreTrainedModel",
"load_tf_weights_in_rembert",
]
)
_import_structure["models.resnet"].extend(
[
"RESNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"ResNetBackbone",
"ResNetForImageClassification",
"ResNetModel",
"ResNetPreTrainedModel",
]
)
_import_structure["models.retribert"].extend(
["RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST", "RetriBertModel", "RetriBertPreTrainedModel"]
)
_import_structure["models.roberta"].extend(
[
"ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"RobertaForCausalLM",
"RobertaForMaskedLM",
"RobertaForMultipleChoice",
"RobertaForQuestionAnswering",
"RobertaForSequenceClassification",
"RobertaForTokenClassification",
"RobertaModel",
"RobertaPreTrainedModel",
]
)
_import_structure["models.roc_bert"].extend(
[
"ROC_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"RoCBertForCausalLM",
"RoCBertForMaskedLM",
"RoCBertForMultipleChoice",
"RoCBertForPreTraining",
"RoCBertForQuestionAnswering",
"RoCBertForSequenceClassification",
"RoCBertForTokenClassification",
"RoCBertLayer",
"RoCBertModel",
"RoCBertPreTrainedModel",
"load_tf_weights_in_roc_bert",
]
)
_import_structure["models.roformer"].extend(
[
"ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"RoFormerForCausalLM",
"RoFormerForMaskedLM",
"RoFormerForMultipleChoice",
"RoFormerForQuestionAnswering",
"RoFormerForSequenceClassification",
"RoFormerForTokenClassification",
"RoFormerLayer",
"RoFormerModel",
"RoFormerPreTrainedModel",
"load_tf_weights_in_roformer",
]
)
_import_structure["models.segformer"].extend(
[
"SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"SegformerDecodeHead",
"SegformerForImageClassification",
"SegformerForSemanticSegmentation",
"SegformerLayer",
"SegformerModel",
"SegformerPreTrainedModel",
]
)
_import_structure["models.sew"].extend(
[
"SEW_PRETRAINED_MODEL_ARCHIVE_LIST",
"SEWForCTC",
"SEWForSequenceClassification",
"SEWModel",
"SEWPreTrainedModel",
]
)
_import_structure["models.sew_d"].extend(
[
"SEW_D_PRETRAINED_MODEL_ARCHIVE_LIST",
"SEWDForCTC",
"SEWDForSequenceClassification",
"SEWDModel",
"SEWDPreTrainedModel",
]
)
_import_structure["models.speech_encoder_decoder"].extend(["SpeechEncoderDecoderModel"])
_import_structure["models.speech_to_text"].extend(
[
"SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"Speech2TextForConditionalGeneration",
"Speech2TextModel",
"Speech2TextPreTrainedModel",
]
)
_import_structure["models.speech_to_text_2"].extend(["Speech2Text2ForCausalLM", "Speech2Text2PreTrainedModel"])
_import_structure["models.splinter"].extend(
[
"SPLINTER_PRETRAINED_MODEL_ARCHIVE_LIST",
"SplinterForPreTraining",
"SplinterForQuestionAnswering",
"SplinterLayer",
"SplinterModel",
"SplinterPreTrainedModel",
]
)
_import_structure["models.squeezebert"].extend(
[
"SQUEEZEBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"SqueezeBertForMaskedLM",
"SqueezeBertForMultipleChoice",
"SqueezeBertForQuestionAnswering",
"SqueezeBertForSequenceClassification",
"SqueezeBertForTokenClassification",
"SqueezeBertModel",
"SqueezeBertModule",
"SqueezeBertPreTrainedModel",
]
)
_import_structure["models.swin"].extend(
[
"SWIN_PRETRAINED_MODEL_ARCHIVE_LIST",
"SwinForImageClassification",
"SwinForMaskedImageModeling",
"SwinModel",
"SwinPreTrainedModel",
]
)
_import_structure["models.swinv2"].extend(
[
"SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST",
"Swinv2ForImageClassification",
"Swinv2ForMaskedImageModeling",
"Swinv2Model",
"Swinv2PreTrainedModel",
]
)
_import_structure["models.switch_transformers"].extend(
[
"SWITCH_TRANSFORMERS_PRETRAINED_MODEL_ARCHIVE_LIST",
"SwitchTransformersEncoderModel",
"SwitchTransformersForConditionalGeneration",
"SwitchTransformersModel",
"SwitchTransformersPreTrainedModel",
"SwitchTransformersSparseMLP",
"SwitchTransformersTop1Router",
]
)
_import_structure["models.t5"].extend(
[
"T5_PRETRAINED_MODEL_ARCHIVE_LIST",
"T5EncoderModel",
"T5ForConditionalGeneration",
"T5Model",
"T5PreTrainedModel",
"load_tf_weights_in_t5",
]
)
_import_structure["models.tapas"].extend(
[
"TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TapasForMaskedLM",
"TapasForQuestionAnswering",
"TapasForSequenceClassification",
"TapasModel",
"TapasPreTrainedModel",
"load_tf_weights_in_tapas",
]
)
_import_structure["models.time_series_transformer"].extend(
[
"TIME_SERIES_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TimeSeriesTransformerForPrediction",
"TimeSeriesTransformerModel",
"TimeSeriesTransformerPreTrainedModel",
]
)
_import_structure["models.timesformer"].extend(
[
"TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TimesformerForVideoClassification",
"TimesformerModel",
"TimesformerPreTrainedModel",
]
)
_import_structure["models.trajectory_transformer"].extend(
[
"TRAJECTORY_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TrajectoryTransformerModel",
"TrajectoryTransformerPreTrainedModel",
]
)
_import_structure["models.transfo_xl"].extend(
[
"TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST",
"AdaptiveEmbedding",
"TransfoXLForSequenceClassification",
"TransfoXLLMHeadModel",
"TransfoXLModel",
"TransfoXLPreTrainedModel",
"load_tf_weights_in_transfo_xl",
]
)
_import_structure["models.trocr"].extend(
["TROCR_PRETRAINED_MODEL_ARCHIVE_LIST", "TrOCRForCausalLM", "TrOCRPreTrainedModel"]
)
_import_structure["models.unispeech"].extend(
[
"UNISPEECH_PRETRAINED_MODEL_ARCHIVE_LIST",
"UniSpeechForCTC",
"UniSpeechForPreTraining",
"UniSpeechForSequenceClassification",
"UniSpeechModel",
"UniSpeechPreTrainedModel",
]
)
_import_structure["models.unispeech_sat"].extend(
[
"UNISPEECH_SAT_PRETRAINED_MODEL_ARCHIVE_LIST",
"UniSpeechSatForAudioFrameClassification",
"UniSpeechSatForCTC",
"UniSpeechSatForPreTraining",
"UniSpeechSatForSequenceClassification",
"UniSpeechSatForXVector",
"UniSpeechSatModel",
"UniSpeechSatPreTrainedModel",
]
)
_import_structure["models.van"].extend(
[
"VAN_PRETRAINED_MODEL_ARCHIVE_LIST",
"VanForImageClassification",
"VanModel",
"VanPreTrainedModel",
]
)
_import_structure["models.videomae"].extend(
[
"VIDEOMAE_PRETRAINED_MODEL_ARCHIVE_LIST",
"VideoMAEForPreTraining",
"VideoMAEForVideoClassification",
"VideoMAEModel",
"VideoMAEPreTrainedModel",
]
)
_import_structure["models.vilt"].extend(
[
"VILT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViltForImageAndTextRetrieval",
"ViltForImagesAndTextClassification",
"ViltForMaskedLM",
"ViltForQuestionAnswering",
"ViltForTokenClassification",
"ViltLayer",
"ViltModel",
"ViltPreTrainedModel",
]
)
_import_structure["models.vision_encoder_decoder"].extend(["VisionEncoderDecoderModel"])
_import_structure["models.vision_text_dual_encoder"].extend(["VisionTextDualEncoderModel"])
_import_structure["models.visual_bert"].extend(
[
"VISUAL_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"VisualBertForMultipleChoice",
"VisualBertForPreTraining",
"VisualBertForQuestionAnswering",
"VisualBertForRegionToPhraseAlignment",
"VisualBertForVisualReasoning",
"VisualBertLayer",
"VisualBertModel",
"VisualBertPreTrainedModel",
]
)
_import_structure["models.vit"].extend(
[
"VIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTForImageClassification",
"ViTForMaskedImageModeling",
"ViTModel",
"ViTPreTrainedModel",
]
)
_import_structure["models.vit_hybrid"].extend(
[
"VIT_HYBRID_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTHybridForImageClassification",
"ViTHybridModel",
"ViTHybridPreTrainedModel",
]
)
_import_structure["models.vit_mae"].extend(
[
"VIT_MAE_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTMAEForPreTraining",
"ViTMAELayer",
"ViTMAEModel",
"ViTMAEPreTrainedModel",
]
)
_import_structure["models.vit_msn"].extend(
[
"VIT_MSN_PRETRAINED_MODEL_ARCHIVE_LIST",
"ViTMSNForImageClassification",
"ViTMSNModel",
"ViTMSNPreTrainedModel",
]
)
_import_structure["models.wav2vec2"].extend(
[
"WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST",
"Wav2Vec2ForAudioFrameClassification",
"Wav2Vec2ForCTC",
"Wav2Vec2ForMaskedLM",
"Wav2Vec2ForPreTraining",
"Wav2Vec2ForSequenceClassification",
"Wav2Vec2ForXVector",
"Wav2Vec2Model",
"Wav2Vec2PreTrainedModel",
]
)
_import_structure["models.wav2vec2_conformer"].extend(
[
"WAV2VEC2_CONFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"Wav2Vec2ConformerForAudioFrameClassification",
"Wav2Vec2ConformerForCTC",
"Wav2Vec2ConformerForPreTraining",
"Wav2Vec2ConformerForSequenceClassification",
"Wav2Vec2ConformerForXVector",
"Wav2Vec2ConformerModel",
"Wav2Vec2ConformerPreTrainedModel",
]
)
_import_structure["models.wavlm"].extend(
[
"WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"WavLMForAudioFrameClassification",
"WavLMForCTC",
"WavLMForSequenceClassification",
"WavLMForXVector",
"WavLMModel",
"WavLMPreTrainedModel",
]
)
_import_structure["models.whisper"].extend(
[
"WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST",
"WhisperForConditionalGeneration",
"WhisperModel",
"WhisperPreTrainedModel",
]
)
_import_structure["models.x_clip"].extend(
[
"XCLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"XCLIPModel",
"XCLIPPreTrainedModel",
"XCLIPTextModel",
"XCLIPVisionModel",
]
)
_import_structure["models.xglm"].extend(
[
"XGLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"XGLMForCausalLM",
"XGLMModel",
"XGLMPreTrainedModel",
]
)
_import_structure["models.xlm"].extend(
[
"XLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMForMultipleChoice",
"XLMForQuestionAnswering",
"XLMForQuestionAnsweringSimple",
"XLMForSequenceClassification",
"XLMForTokenClassification",
"XLMModel",
"XLMPreTrainedModel",
"XLMWithLMHeadModel",
]
)
_import_structure["models.xlm_prophetnet"].extend(
[
"XLM_PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMProphetNetDecoder",
"XLMProphetNetEncoder",
"XLMProphetNetForCausalLM",
"XLMProphetNetForConditionalGeneration",
"XLMProphetNetModel",
"XLMProphetNetPreTrainedModel",
]
)
_import_structure["models.xlm_roberta"].extend(
[
"XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMRobertaForCausalLM",
"XLMRobertaForMaskedLM",
"XLMRobertaForMultipleChoice",
"XLMRobertaForQuestionAnswering",
"XLMRobertaForSequenceClassification",
"XLMRobertaForTokenClassification",
"XLMRobertaModel",
"XLMRobertaPreTrainedModel",
]
)
_import_structure["models.xlm_roberta_xl"].extend(
[
"XLM_ROBERTA_XL_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLMRobertaXLForCausalLM",
"XLMRobertaXLForMaskedLM",
"XLMRobertaXLForMultipleChoice",
"XLMRobertaXLForQuestionAnswering",
"XLMRobertaXLForSequenceClassification",
"XLMRobertaXLForTokenClassification",
"XLMRobertaXLModel",
"XLMRobertaXLPreTrainedModel",
]
)
_import_structure["models.xlnet"].extend(
[
"XLNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"XLNetForMultipleChoice",
"XLNetForQuestionAnswering",
"XLNetForQuestionAnsweringSimple",
"XLNetForSequenceClassification",
"XLNetForTokenClassification",
"XLNetLMHeadModel",
"XLNetModel",
"XLNetPreTrainedModel",
"load_tf_weights_in_xlnet",
]
)
_import_structure["models.yolos"].extend(
[
"YOLOS_PRETRAINED_MODEL_ARCHIVE_LIST",
"YolosForObjectDetection",
"YolosModel",
"YolosPreTrainedModel",
]
)
_import_structure["models.yoso"].extend(
[
"YOSO_PRETRAINED_MODEL_ARCHIVE_LIST",
"YosoForMaskedLM",
"YosoForMultipleChoice",
"YosoForQuestionAnswering",
"YosoForSequenceClassification",
"YosoForTokenClassification",
"YosoLayer",
"YosoModel",
"YosoPreTrainedModel",
]
)
_import_structure["optimization"] = [
"Adafactor",
"AdamW",
"get_constant_schedule",
"get_constant_schedule_with_warmup",
"get_cosine_schedule_with_warmup",
"get_cosine_with_hard_restarts_schedule_with_warmup",
"get_linear_schedule_with_warmup",
"get_polynomial_decay_schedule_with_warmup",
"get_scheduler",
]
_import_structure["pytorch_utils"] = ["Conv1D", "apply_chunking_to_forward", "prune_layer"]
_import_structure["sagemaker"] = []
_import_structure["trainer"] = ["Trainer"]
_import_structure["trainer_pt_utils"] = ["torch_distributed_zero_first"]
_import_structure["trainer_seq2seq"] = ["Seq2SeqTrainer"]
# TensorFlow-backed objects
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_tf_objects
_import_structure["utils.dummy_tf_objects"] = [name for name in dir(dummy_tf_objects) if not name.startswith("_")]
else:
_import_structure["activations_tf"] = []
_import_structure["benchmark.benchmark_args_tf"] = ["TensorFlowBenchmarkArguments"]
_import_structure["benchmark.benchmark_tf"] = ["TensorFlowBenchmark"]
_import_structure["generation"].extend(
[
"TFForcedBOSTokenLogitsProcessor",
"TFForcedEOSTokenLogitsProcessor",
"TFGenerationMixin",
"TFLogitsProcessor",
"TFLogitsProcessorList",
"TFLogitsWarper",
"TFMinLengthLogitsProcessor",
"TFNoBadWordsLogitsProcessor",
"TFNoRepeatNGramLogitsProcessor",
"TFRepetitionPenaltyLogitsProcessor",
"TFTemperatureLogitsWarper",
"TFTopKLogitsWarper",
"TFTopPLogitsWarper",
"tf_top_k_top_p_filtering",
]
)
_import_structure["generation_tf_utils"] = []
_import_structure["keras_callbacks"] = ["KerasMetricCallback", "PushToHubCallback"]
_import_structure["modeling_tf_outputs"] = []
_import_structure["modeling_tf_utils"] = [
"TFPreTrainedModel",
"TFSequenceSummary",
"TFSharedEmbeddings",
"shape_list",
]
# TensorFlow models structure
_import_structure["models.albert"].extend(
[
"TF_ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFAlbertForMaskedLM",
"TFAlbertForMultipleChoice",
"TFAlbertForPreTraining",
"TFAlbertForQuestionAnswering",
"TFAlbertForSequenceClassification",
"TFAlbertForTokenClassification",
"TFAlbertMainLayer",
"TFAlbertModel",
"TFAlbertPreTrainedModel",
]
)
_import_structure["models.auto"].extend(
[
"TF_MODEL_FOR_CAUSAL_LM_MAPPING",
"TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING",
"TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING",
"TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING",
"TF_MODEL_FOR_MASKED_LM_MAPPING",
"TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING",
"TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING",
"TF_MODEL_FOR_PRETRAINING_MAPPING",
"TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING",
"TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING",
"TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING",
"TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING",
"TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING",
"TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING",
"TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING",
"TF_MODEL_FOR_VISION_2_SEQ_MAPPING",
"TF_MODEL_MAPPING",
"TF_MODEL_WITH_LM_HEAD_MAPPING",
"TFAutoModel",
"TFAutoModelForCausalLM",
"TFAutoModelForDocumentQuestionAnswering",
"TFAutoModelForImageClassification",
"TFAutoModelForMaskedLM",
"TFAutoModelForMultipleChoice",
"TFAutoModelForNextSentencePrediction",
"TFAutoModelForPreTraining",
"TFAutoModelForQuestionAnswering",
"TFAutoModelForSemanticSegmentation",
"TFAutoModelForSeq2SeqLM",
"TFAutoModelForSequenceClassification",
"TFAutoModelForSpeechSeq2Seq",
"TFAutoModelForTableQuestionAnswering",
"TFAutoModelForTokenClassification",
"TFAutoModelForVision2Seq",
"TFAutoModelWithLMHead",
]
)
_import_structure["models.bart"].extend(
["TFBartForConditionalGeneration", "TFBartForSequenceClassification", "TFBartModel", "TFBartPretrainedModel"]
)
_import_structure["models.bert"].extend(
[
"TF_BERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFBertEmbeddings",
"TFBertForMaskedLM",
"TFBertForMultipleChoice",
"TFBertForNextSentencePrediction",
"TFBertForPreTraining",
"TFBertForQuestionAnswering",
"TFBertForSequenceClassification",
"TFBertForTokenClassification",
"TFBertLMHeadModel",
"TFBertMainLayer",
"TFBertModel",
"TFBertPreTrainedModel",
]
)
_import_structure["models.blenderbot"].extend(
["TFBlenderbotForConditionalGeneration", "TFBlenderbotModel", "TFBlenderbotPreTrainedModel"]
)
_import_structure["models.blenderbot_small"].extend(
["TFBlenderbotSmallForConditionalGeneration", "TFBlenderbotSmallModel", "TFBlenderbotSmallPreTrainedModel"]
)
_import_structure["models.camembert"].extend(
[
"TF_CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCamembertForCausalLM",
"TFCamembertForMaskedLM",
"TFCamembertForMultipleChoice",
"TFCamembertForQuestionAnswering",
"TFCamembertForSequenceClassification",
"TFCamembertForTokenClassification",
"TFCamembertModel",
"TFCamembertPreTrainedModel",
]
)
_import_structure["models.clip"].extend(
[
"TF_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCLIPModel",
"TFCLIPPreTrainedModel",
"TFCLIPTextModel",
"TFCLIPVisionModel",
]
)
_import_structure["models.convbert"].extend(
[
"TF_CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFConvBertForMaskedLM",
"TFConvBertForMultipleChoice",
"TFConvBertForQuestionAnswering",
"TFConvBertForSequenceClassification",
"TFConvBertForTokenClassification",
"TFConvBertLayer",
"TFConvBertModel",
"TFConvBertPreTrainedModel",
]
)
_import_structure["models.convnext"].extend(
[
"TFConvNextForImageClassification",
"TFConvNextModel",
"TFConvNextPreTrainedModel",
]
)
_import_structure["models.ctrl"].extend(
[
"TF_CTRL_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCTRLForSequenceClassification",
"TFCTRLLMHeadModel",
"TFCTRLModel",
"TFCTRLPreTrainedModel",
]
)
_import_structure["models.cvt"].extend(
[
"TF_CVT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFCvtForImageClassification",
"TFCvtModel",
"TFCvtPreTrainedModel",
]
)
_import_structure["models.data2vec"].extend(
[
"TFData2VecVisionForImageClassification",
"TFData2VecVisionForSemanticSegmentation",
"TFData2VecVisionModel",
"TFData2VecVisionPreTrainedModel",
]
)
_import_structure["models.deberta"].extend(
[
"TF_DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDebertaForMaskedLM",
"TFDebertaForQuestionAnswering",
"TFDebertaForSequenceClassification",
"TFDebertaForTokenClassification",
"TFDebertaModel",
"TFDebertaPreTrainedModel",
]
)
_import_structure["models.deberta_v2"].extend(
[
"TF_DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDebertaV2ForMaskedLM",
"TFDebertaV2ForQuestionAnswering",
"TFDebertaV2ForSequenceClassification",
"TFDebertaV2ForTokenClassification",
"TFDebertaV2Model",
"TFDebertaV2PreTrainedModel",
]
)
_import_structure["models.deit"].extend(
[
"TF_DEIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDeiTForImageClassification",
"TFDeiTForImageClassificationWithTeacher",
"TFDeiTForMaskedImageModeling",
"TFDeiTModel",
"TFDeiTPreTrainedModel",
]
)
_import_structure["models.distilbert"].extend(
[
"TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDistilBertForMaskedLM",
"TFDistilBertForMultipleChoice",
"TFDistilBertForQuestionAnswering",
"TFDistilBertForSequenceClassification",
"TFDistilBertForTokenClassification",
"TFDistilBertMainLayer",
"TFDistilBertModel",
"TFDistilBertPreTrainedModel",
]
)
_import_structure["models.dpr"].extend(
[
"TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TF_DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TF_DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFDPRContextEncoder",
"TFDPRPretrainedContextEncoder",
"TFDPRPretrainedQuestionEncoder",
"TFDPRPretrainedReader",
"TFDPRQuestionEncoder",
"TFDPRReader",
]
)
_import_structure["models.electra"].extend(
[
"TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFElectraForMaskedLM",
"TFElectraForMultipleChoice",
"TFElectraForPreTraining",
"TFElectraForQuestionAnswering",
"TFElectraForSequenceClassification",
"TFElectraForTokenClassification",
"TFElectraModel",
"TFElectraPreTrainedModel",
]
)
_import_structure["models.encoder_decoder"].append("TFEncoderDecoderModel")
_import_structure["models.esm"].extend(
[
"ESM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFEsmForMaskedLM",
"TFEsmForSequenceClassification",
"TFEsmForTokenClassification",
"TFEsmModel",
"TFEsmPreTrainedModel",
]
)
_import_structure["models.flaubert"].extend(
[
"TF_FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFFlaubertForMultipleChoice",
"TFFlaubertForQuestionAnsweringSimple",
"TFFlaubertForSequenceClassification",
"TFFlaubertForTokenClassification",
"TFFlaubertModel",
"TFFlaubertPreTrainedModel",
"TFFlaubertWithLMHeadModel",
]
)
_import_structure["models.funnel"].extend(
[
"TF_FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFFunnelBaseModel",
"TFFunnelForMaskedLM",
"TFFunnelForMultipleChoice",
"TFFunnelForPreTraining",
"TFFunnelForQuestionAnswering",
"TFFunnelForSequenceClassification",
"TFFunnelForTokenClassification",
"TFFunnelModel",
"TFFunnelPreTrainedModel",
]
)
_import_structure["models.gpt2"].extend(
[
"TF_GPT2_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFGPT2DoubleHeadsModel",
"TFGPT2ForSequenceClassification",
"TFGPT2LMHeadModel",
"TFGPT2MainLayer",
"TFGPT2Model",
"TFGPT2PreTrainedModel",
]
)
_import_structure["models.gptj"].extend(
[
"TFGPTJForCausalLM",
"TFGPTJForQuestionAnswering",
"TFGPTJForSequenceClassification",
"TFGPTJModel",
"TFGPTJPreTrainedModel",
]
)
_import_structure["models.groupvit"].extend(
[
"TF_GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFGroupViTModel",
"TFGroupViTPreTrainedModel",
"TFGroupViTTextModel",
"TFGroupViTVisionModel",
]
)
_import_structure["models.hubert"].extend(
[
"TF_HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFHubertForCTC",
"TFHubertModel",
"TFHubertPreTrainedModel",
]
)
_import_structure["models.layoutlm"].extend(
[
"TF_LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLayoutLMForMaskedLM",
"TFLayoutLMForQuestionAnswering",
"TFLayoutLMForSequenceClassification",
"TFLayoutLMForTokenClassification",
"TFLayoutLMMainLayer",
"TFLayoutLMModel",
"TFLayoutLMPreTrainedModel",
]
)
_import_structure["models.layoutlmv3"].extend(
[
"TF_LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLayoutLMv3ForQuestionAnswering",
"TFLayoutLMv3ForSequenceClassification",
"TFLayoutLMv3ForTokenClassification",
"TFLayoutLMv3Model",
"TFLayoutLMv3PreTrainedModel",
]
)
_import_structure["models.led"].extend(["TFLEDForConditionalGeneration", "TFLEDModel", "TFLEDPreTrainedModel"])
_import_structure["models.longformer"].extend(
[
"TF_LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLongformerForMaskedLM",
"TFLongformerForMultipleChoice",
"TFLongformerForQuestionAnswering",
"TFLongformerForSequenceClassification",
"TFLongformerForTokenClassification",
"TFLongformerModel",
"TFLongformerPreTrainedModel",
"TFLongformerSelfAttention",
]
)
_import_structure["models.lxmert"].extend(
[
"TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFLxmertForPreTraining",
"TFLxmertMainLayer",
"TFLxmertModel",
"TFLxmertPreTrainedModel",
"TFLxmertVisualFeatureEncoder",
]
)
_import_structure["models.marian"].extend(["TFMarianModel", "TFMarianMTModel", "TFMarianPreTrainedModel"])
_import_structure["models.mbart"].extend(
["TFMBartForConditionalGeneration", "TFMBartModel", "TFMBartPreTrainedModel"]
)
_import_structure["models.mobilebert"].extend(
[
"TF_MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFMobileBertForMaskedLM",
"TFMobileBertForMultipleChoice",
"TFMobileBertForNextSentencePrediction",
"TFMobileBertForPreTraining",
"TFMobileBertForQuestionAnswering",
"TFMobileBertForSequenceClassification",
"TFMobileBertForTokenClassification",
"TFMobileBertMainLayer",
"TFMobileBertModel",
"TFMobileBertPreTrainedModel",
]
)
_import_structure["models.mobilevit"].extend(
[
"TF_MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFMobileViTForImageClassification",
"TFMobileViTForSemanticSegmentation",
"TFMobileViTModel",
"TFMobileViTPreTrainedModel",
]
)
_import_structure["models.mpnet"].extend(
[
"TF_MPNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFMPNetForMaskedLM",
"TFMPNetForMultipleChoice",
"TFMPNetForQuestionAnswering",
"TFMPNetForSequenceClassification",
"TFMPNetForTokenClassification",
"TFMPNetMainLayer",
"TFMPNetModel",
"TFMPNetPreTrainedModel",
]
)
_import_structure["models.mt5"].extend(["TFMT5EncoderModel", "TFMT5ForConditionalGeneration", "TFMT5Model"])
_import_structure["models.openai"].extend(
[
"TF_OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFOpenAIGPTDoubleHeadsModel",
"TFOpenAIGPTForSequenceClassification",
"TFOpenAIGPTLMHeadModel",
"TFOpenAIGPTMainLayer",
"TFOpenAIGPTModel",
"TFOpenAIGPTPreTrainedModel",
]
)
_import_structure["models.opt"].extend(
[
"TFOPTForCausalLM",
"TFOPTModel",
"TFOPTPreTrainedModel",
]
)
_import_structure["models.pegasus"].extend(
["TFPegasusForConditionalGeneration", "TFPegasusModel", "TFPegasusPreTrainedModel"]
)
_import_structure["models.rag"].extend(
[
"TFRagModel",
"TFRagPreTrainedModel",
"TFRagSequenceForGeneration",
"TFRagTokenForGeneration",
]
)
_import_structure["models.regnet"].extend(
[
"TF_REGNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRegNetForImageClassification",
"TFRegNetModel",
"TFRegNetPreTrainedModel",
]
)
_import_structure["models.rembert"].extend(
[
"TF_REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRemBertForCausalLM",
"TFRemBertForMaskedLM",
"TFRemBertForMultipleChoice",
"TFRemBertForQuestionAnswering",
"TFRemBertForSequenceClassification",
"TFRemBertForTokenClassification",
"TFRemBertLayer",
"TFRemBertModel",
"TFRemBertPreTrainedModel",
]
)
_import_structure["models.resnet"].extend(
[
"TF_RESNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFResNetForImageClassification",
"TFResNetModel",
"TFResNetPreTrainedModel",
]
)
_import_structure["models.roberta"].extend(
[
"TF_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRobertaForCausalLM",
"TFRobertaForMaskedLM",
"TFRobertaForMultipleChoice",
"TFRobertaForQuestionAnswering",
"TFRobertaForSequenceClassification",
"TFRobertaForTokenClassification",
"TFRobertaMainLayer",
"TFRobertaModel",
"TFRobertaPreTrainedModel",
]
)
_import_structure["models.roformer"].extend(
[
"TF_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFRoFormerForCausalLM",
"TFRoFormerForMaskedLM",
"TFRoFormerForMultipleChoice",
"TFRoFormerForQuestionAnswering",
"TFRoFormerForSequenceClassification",
"TFRoFormerForTokenClassification",
"TFRoFormerLayer",
"TFRoFormerModel",
"TFRoFormerPreTrainedModel",
]
)
_import_structure["models.segformer"].extend(
[
"TF_SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFSegformerDecodeHead",
"TFSegformerForImageClassification",
"TFSegformerForSemanticSegmentation",
"TFSegformerModel",
"TFSegformerPreTrainedModel",
]
)
_import_structure["models.speech_to_text"].extend(
[
"TF_SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFSpeech2TextForConditionalGeneration",
"TFSpeech2TextModel",
"TFSpeech2TextPreTrainedModel",
]
)
_import_structure["models.swin"].extend(
[
"TF_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFSwinForImageClassification",
"TFSwinForMaskedImageModeling",
"TFSwinModel",
"TFSwinPreTrainedModel",
]
)
_import_structure["models.t5"].extend(
[
"TF_T5_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFT5EncoderModel",
"TFT5ForConditionalGeneration",
"TFT5Model",
"TFT5PreTrainedModel",
]
)
_import_structure["models.tapas"].extend(
[
"TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFTapasForMaskedLM",
"TFTapasForQuestionAnswering",
"TFTapasForSequenceClassification",
"TFTapasModel",
"TFTapasPreTrainedModel",
]
)
_import_structure["models.transfo_xl"].extend(
[
"TF_TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFAdaptiveEmbedding",
"TFTransfoXLForSequenceClassification",
"TFTransfoXLLMHeadModel",
"TFTransfoXLMainLayer",
"TFTransfoXLModel",
"TFTransfoXLPreTrainedModel",
]
)
_import_structure["models.vision_encoder_decoder"].extend(["TFVisionEncoderDecoderModel"])
_import_structure["models.vit"].extend(
[
"TFViTForImageClassification",
"TFViTModel",
"TFViTPreTrainedModel",
]
)
_import_structure["models.vit_mae"].extend(
[
"TFViTMAEForPreTraining",
"TFViTMAEModel",
"TFViTMAEPreTrainedModel",
]
)
_import_structure["models.wav2vec2"].extend(
[
"TF_WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFWav2Vec2ForCTC",
"TFWav2Vec2Model",
"TFWav2Vec2PreTrainedModel",
]
)
_import_structure["models.whisper"].extend(
[
"TF_WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFWhisperForConditionalGeneration",
"TFWhisperModel",
"TFWhisperPreTrainedModel",
]
)
_import_structure["models.xglm"].extend(
[
"TF_XGLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXGLMForCausalLM",
"TFXGLMModel",
"TFXGLMPreTrainedModel",
]
)
_import_structure["models.xlm"].extend(
[
"TF_XLM_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXLMForMultipleChoice",
"TFXLMForQuestionAnsweringSimple",
"TFXLMForSequenceClassification",
"TFXLMForTokenClassification",
"TFXLMMainLayer",
"TFXLMModel",
"TFXLMPreTrainedModel",
"TFXLMWithLMHeadModel",
]
)
_import_structure["models.xlm_roberta"].extend(
[
"TF_XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXLMRobertaForMaskedLM",
"TFXLMRobertaForMultipleChoice",
"TFXLMRobertaForQuestionAnswering",
"TFXLMRobertaForSequenceClassification",
"TFXLMRobertaForTokenClassification",
"TFXLMRobertaModel",
]
)
_import_structure["models.xlnet"].extend(
[
"TF_XLNET_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFXLNetForMultipleChoice",
"TFXLNetForQuestionAnsweringSimple",
"TFXLNetForSequenceClassification",
"TFXLNetForTokenClassification",
"TFXLNetLMHeadModel",
"TFXLNetMainLayer",
"TFXLNetModel",
"TFXLNetPreTrainedModel",
]
)
_import_structure["optimization_tf"] = ["AdamWeightDecay", "GradientAccumulator", "WarmUp", "create_optimizer"]
_import_structure["tf_utils"] = []
_import_structure["trainer_tf"] = ["TFTrainer"]
# FLAX-backed objects
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils import dummy_flax_objects
_import_structure["utils.dummy_flax_objects"] = [
name for name in dir(dummy_flax_objects) if not name.startswith("_")
]
else:
_import_structure["generation"].extend(
[
"FlaxForcedBOSTokenLogitsProcessor",
"FlaxForcedEOSTokenLogitsProcessor",
"FlaxGenerationMixin",
"FlaxLogitsProcessor",
"FlaxLogitsProcessorList",
"FlaxLogitsWarper",
"FlaxMinLengthLogitsProcessor",
"FlaxTemperatureLogitsWarper",
"FlaxTopKLogitsWarper",
"FlaxTopPLogitsWarper",
]
)
_import_structure["generation_flax_utils"] = []
_import_structure["modeling_flax_outputs"] = []
_import_structure["modeling_flax_utils"] = ["FlaxPreTrainedModel"]
_import_structure["models.albert"].extend(
[
"FlaxAlbertForMaskedLM",
"FlaxAlbertForMultipleChoice",
"FlaxAlbertForPreTraining",
"FlaxAlbertForQuestionAnswering",
"FlaxAlbertForSequenceClassification",
"FlaxAlbertForTokenClassification",
"FlaxAlbertModel",
"FlaxAlbertPreTrainedModel",
]
)
_import_structure["models.auto"].extend(
[
"FLAX_MODEL_FOR_CAUSAL_LM_MAPPING",
"FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING",
"FLAX_MODEL_FOR_MASKED_LM_MAPPING",
"FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING",
"FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING",
"FLAX_MODEL_FOR_PRETRAINING_MAPPING",
"FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING",
"FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING",
"FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING",
"FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING",
"FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING",
"FLAX_MODEL_MAPPING",
"FlaxAutoModel",
"FlaxAutoModelForCausalLM",
"FlaxAutoModelForImageClassification",
"FlaxAutoModelForMaskedLM",
"FlaxAutoModelForMultipleChoice",
"FlaxAutoModelForNextSentencePrediction",
"FlaxAutoModelForPreTraining",
"FlaxAutoModelForQuestionAnswering",
"FlaxAutoModelForSeq2SeqLM",
"FlaxAutoModelForSequenceClassification",
"FlaxAutoModelForTokenClassification",
"FlaxAutoModelForVision2Seq",
]
)
# Flax models structure
_import_structure["models.bart"].extend(
[
"FlaxBartDecoderPreTrainedModel",
"FlaxBartForCausalLM",
"FlaxBartForConditionalGeneration",
"FlaxBartForQuestionAnswering",
"FlaxBartForSequenceClassification",
"FlaxBartModel",
"FlaxBartPreTrainedModel",
]
)
_import_structure["models.beit"].extend(
[
"FlaxBeitForImageClassification",
"FlaxBeitForMaskedImageModeling",
"FlaxBeitModel",
"FlaxBeitPreTrainedModel",
]
)
_import_structure["models.bert"].extend(
[
"FlaxBertForCausalLM",
"FlaxBertForMaskedLM",
"FlaxBertForMultipleChoice",
"FlaxBertForNextSentencePrediction",
"FlaxBertForPreTraining",
"FlaxBertForQuestionAnswering",
"FlaxBertForSequenceClassification",
"FlaxBertForTokenClassification",
"FlaxBertModel",
"FlaxBertPreTrainedModel",
]
)
_import_structure["models.big_bird"].extend(
[
"FlaxBigBirdForCausalLM",
"FlaxBigBirdForMaskedLM",
"FlaxBigBirdForMultipleChoice",
"FlaxBigBirdForPreTraining",
"FlaxBigBirdForQuestionAnswering",
"FlaxBigBirdForSequenceClassification",
"FlaxBigBirdForTokenClassification",
"FlaxBigBirdModel",
"FlaxBigBirdPreTrainedModel",
]
)
_import_structure["models.blenderbot"].extend(
["FlaxBlenderbotForConditionalGeneration", "FlaxBlenderbotModel", "FlaxBlenderbotPreTrainedModel"]
)
_import_structure["models.blenderbot_small"].extend(
[
"FlaxBlenderbotSmallForConditionalGeneration",
"FlaxBlenderbotSmallModel",
"FlaxBlenderbotSmallPreTrainedModel",
]
)
_import_structure["models.clip"].extend(
[
"FlaxCLIPModel",
"FlaxCLIPPreTrainedModel",
"FlaxCLIPTextModel",
"FlaxCLIPTextPreTrainedModel",
"FlaxCLIPVisionModel",
"FlaxCLIPVisionPreTrainedModel",
]
)
_import_structure["models.distilbert"].extend(
[
"FlaxDistilBertForMaskedLM",
"FlaxDistilBertForMultipleChoice",
"FlaxDistilBertForQuestionAnswering",
"FlaxDistilBertForSequenceClassification",
"FlaxDistilBertForTokenClassification",
"FlaxDistilBertModel",
"FlaxDistilBertPreTrainedModel",
]
)
_import_structure["models.electra"].extend(
[
"FlaxElectraForCausalLM",
"FlaxElectraForMaskedLM",
"FlaxElectraForMultipleChoice",
"FlaxElectraForPreTraining",
"FlaxElectraForQuestionAnswering",
"FlaxElectraForSequenceClassification",
"FlaxElectraForTokenClassification",
"FlaxElectraModel",
"FlaxElectraPreTrainedModel",
]
)
_import_structure["models.encoder_decoder"].append("FlaxEncoderDecoderModel")
_import_structure["models.gpt2"].extend(["FlaxGPT2LMHeadModel", "FlaxGPT2Model", "FlaxGPT2PreTrainedModel"])
_import_structure["models.gpt_neo"].extend(
["FlaxGPTNeoForCausalLM", "FlaxGPTNeoModel", "FlaxGPTNeoPreTrainedModel"]
)
_import_structure["models.gptj"].extend(["FlaxGPTJForCausalLM", "FlaxGPTJModel", "FlaxGPTJPreTrainedModel"])
_import_structure["models.longt5"].extend(
["FlaxLongT5ForConditionalGeneration", "FlaxLongT5Model", "FlaxLongT5PreTrainedModel"]
)
_import_structure["models.marian"].extend(
[
"FlaxMarianModel",
"FlaxMarianMTModel",
"FlaxMarianPreTrainedModel",
]
)
_import_structure["models.mbart"].extend(
[
"FlaxMBartForConditionalGeneration",
"FlaxMBartForQuestionAnswering",
"FlaxMBartForSequenceClassification",
"FlaxMBartModel",
"FlaxMBartPreTrainedModel",
]
)
_import_structure["models.mt5"].extend(["FlaxMT5EncoderModel", "FlaxMT5ForConditionalGeneration", "FlaxMT5Model"])
_import_structure["models.opt"].extend(
[
"FlaxOPTForCausalLM",
"FlaxOPTModel",
"FlaxOPTPreTrainedModel",
]
)
_import_structure["models.pegasus"].extend(
[
"FlaxPegasusForConditionalGeneration",
"FlaxPegasusModel",
"FlaxPegasusPreTrainedModel",
]
)
_import_structure["models.roberta"].extend(
[
"FlaxRobertaForCausalLM",
"FlaxRobertaForMaskedLM",
"FlaxRobertaForMultipleChoice",
"FlaxRobertaForQuestionAnswering",
"FlaxRobertaForSequenceClassification",
"FlaxRobertaForTokenClassification",
"FlaxRobertaModel",
"FlaxRobertaPreTrainedModel",
]
)
_import_structure["models.roformer"].extend(
[
"FlaxRoFormerForMaskedLM",
"FlaxRoFormerForMultipleChoice",
"FlaxRoFormerForQuestionAnswering",
"FlaxRoFormerForSequenceClassification",
"FlaxRoFormerForTokenClassification",
"FlaxRoFormerModel",
"FlaxRoFormerPreTrainedModel",
]
)
_import_structure["models.speech_encoder_decoder"].append("FlaxSpeechEncoderDecoderModel")
_import_structure["models.t5"].extend(
["FlaxT5EncoderModel", "FlaxT5ForConditionalGeneration", "FlaxT5Model", "FlaxT5PreTrainedModel"]
)
_import_structure["models.vision_encoder_decoder"].append("FlaxVisionEncoderDecoderModel")
_import_structure["models.vision_text_dual_encoder"].extend(["FlaxVisionTextDualEncoderModel"])
_import_structure["models.vit"].extend(["FlaxViTForImageClassification", "FlaxViTModel", "FlaxViTPreTrainedModel"])
_import_structure["models.wav2vec2"].extend(
["FlaxWav2Vec2ForCTC", "FlaxWav2Vec2ForPreTraining", "FlaxWav2Vec2Model", "FlaxWav2Vec2PreTrainedModel"]
)
_import_structure["models.xglm"].extend(
[
"FlaxXGLMForCausalLM",
"FlaxXGLMModel",
"FlaxXGLMPreTrainedModel",
]
)
_import_structure["models.xlm_roberta"].extend(
[
"FlaxXLMRobertaForMaskedLM",
"FlaxXLMRobertaForMultipleChoice",
"FlaxXLMRobertaForQuestionAnswering",
"FlaxXLMRobertaForSequenceClassification",
"FlaxXLMRobertaForTokenClassification",
"FlaxXLMRobertaModel",
]
)
# Direct imports for type-checking
if TYPE_CHECKING:
# Configuration
from .configuration_utils import PretrainedConfig
# Data
from .data import (
DataProcessor,
InputExample,
InputFeatures,
SingleSentenceClassificationProcessor,
SquadExample,
SquadFeatures,
SquadV1Processor,
SquadV2Processor,
glue_compute_metrics,
glue_convert_examples_to_features,
glue_output_modes,
glue_processors,
glue_tasks_num_labels,
squad_convert_examples_to_features,
xnli_compute_metrics,
xnli_output_modes,
xnli_processors,
xnli_tasks_num_labels,
)
from .data.data_collator import (
DataCollator,
DataCollatorForLanguageModeling,
DataCollatorForPermutationLanguageModeling,
DataCollatorForSeq2Seq,
DataCollatorForSOP,
DataCollatorForTokenClassification,
DataCollatorForWholeWordMask,
DataCollatorWithPadding,
DefaultDataCollator,
default_data_collator,
)
from .feature_extraction_sequence_utils import SequenceFeatureExtractor
# Feature Extractor
from .feature_extraction_utils import BatchFeature, FeatureExtractionMixin
# Generation
from .generation import GenerationConfig
from .hf_argparser import HfArgumentParser
# Integrations
from .integrations import (
is_clearml_available,
is_comet_available,
is_neptune_available,
is_optuna_available,
is_ray_available,
is_ray_tune_available,
is_sigopt_available,
is_tensorboard_available,
is_wandb_available,
)
# Model Cards
from .modelcard import ModelCard
# TF 2.0 <=> PyTorch conversion utilities
from .modeling_tf_pytorch_utils import (
convert_tf_weight_name_to_pt_weight_name,
load_pytorch_checkpoint_in_tf2_model,
load_pytorch_model_in_tf2_model,
load_pytorch_weights_in_tf2_model,
load_tf2_checkpoint_in_pytorch_model,
load_tf2_model_in_pytorch_model,
load_tf2_weights_in_pytorch_model,
)
from .models.albert import ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, AlbertConfig
from .models.audio_spectrogram_transformer import (
AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
ASTConfig,
)
from .models.auto import (
ALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
CONFIG_MAPPING,
FEATURE_EXTRACTOR_MAPPING,
IMAGE_PROCESSOR_MAPPING,
MODEL_NAMES_MAPPING,
PROCESSOR_MAPPING,
TOKENIZER_MAPPING,
AutoConfig,
AutoFeatureExtractor,
AutoImageProcessor,
AutoProcessor,
AutoTokenizer,
)
from .models.bart import BartConfig, BartTokenizer
from .models.beit import BEIT_PRETRAINED_CONFIG_ARCHIVE_MAP, BeitConfig
from .models.bert import (
BERT_PRETRAINED_CONFIG_ARCHIVE_MAP,
BasicTokenizer,
BertConfig,
BertTokenizer,
WordpieceTokenizer,
)
from .models.bert_generation import BertGenerationConfig
from .models.bert_japanese import BertJapaneseTokenizer, CharacterTokenizer, MecabTokenizer
from .models.bertweet import BertweetTokenizer
from .models.big_bird import BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP, BigBirdConfig
from .models.bigbird_pegasus import BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP, BigBirdPegasusConfig
from .models.biogpt import BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, BioGptConfig, BioGptTokenizer
from .models.bit import BIT_PRETRAINED_CONFIG_ARCHIVE_MAP, BitConfig
from .models.blenderbot import BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP, BlenderbotConfig, BlenderbotTokenizer
from .models.blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP,
BlenderbotSmallConfig,
BlenderbotSmallTokenizer,
)
from .models.bloom import BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP, BloomConfig
from .models.byt5 import ByT5Tokenizer
from .models.camembert import CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, CamembertConfig
from .models.canine import CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP, CanineConfig, CanineTokenizer
from .models.chinese_clip import (
CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
ChineseCLIPConfig,
ChineseCLIPProcessor,
ChineseCLIPTextConfig,
ChineseCLIPVisionConfig,
)
from .models.clip import (
CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
CLIPConfig,
CLIPProcessor,
CLIPTextConfig,
CLIPTokenizer,
CLIPVisionConfig,
)
from .models.clipseg import (
CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP,
CLIPSegConfig,
CLIPSegProcessor,
CLIPSegTextConfig,
CLIPSegVisionConfig,
)
from .models.codegen import CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP, CodeGenConfig, CodeGenTokenizer
from .models.conditional_detr import CONDITIONAL_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP, ConditionalDetrConfig
from .models.convbert import CONVBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, ConvBertConfig, ConvBertTokenizer
from .models.convnext import CONVNEXT_PRETRAINED_CONFIG_ARCHIVE_MAP, ConvNextConfig
from .models.ctrl import CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP, CTRLConfig, CTRLTokenizer
from .models.cvt import CVT_PRETRAINED_CONFIG_ARCHIVE_MAP, CvtConfig
from .models.data2vec import (
DATA2VEC_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP,
DATA2VEC_VISION_PRETRAINED_CONFIG_ARCHIVE_MAP,
Data2VecAudioConfig,
Data2VecTextConfig,
Data2VecVisionConfig,
)
from .models.deberta import DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaConfig, DebertaTokenizer
from .models.deberta_v2 import DEBERTA_V2_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaV2Config
from .models.decision_transformer import (
DECISION_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
DecisionTransformerConfig,
)
from .models.deformable_detr import DEFORMABLE_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP, DeformableDetrConfig
from .models.deit import DEIT_PRETRAINED_CONFIG_ARCHIVE_MAP, DeiTConfig
from .models.detr import DETR_PRETRAINED_CONFIG_ARCHIVE_MAP, DetrConfig
from .models.dinat import DINAT_PRETRAINED_CONFIG_ARCHIVE_MAP, DinatConfig
from .models.distilbert import DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, DistilBertConfig, DistilBertTokenizer
from .models.donut import DONUT_SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP, DonutProcessor, DonutSwinConfig
from .models.dpr import (
DPR_PRETRAINED_CONFIG_ARCHIVE_MAP,
DPRConfig,
DPRContextEncoderTokenizer,
DPRQuestionEncoderTokenizer,
DPRReaderOutput,
DPRReaderTokenizer,
)
from .models.dpt import DPT_PRETRAINED_CONFIG_ARCHIVE_MAP, DPTConfig
from .models.electra import ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, ElectraConfig, ElectraTokenizer
from .models.encoder_decoder import EncoderDecoderConfig
from .models.ernie import ERNIE_PRETRAINED_CONFIG_ARCHIVE_MAP, ErnieConfig
from .models.esm import ESM_PRETRAINED_CONFIG_ARCHIVE_MAP, EsmConfig, EsmTokenizer
from .models.flaubert import FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, FlaubertConfig, FlaubertTokenizer
from .models.flava import (
FLAVA_PRETRAINED_CONFIG_ARCHIVE_MAP,
FlavaConfig,
FlavaImageCodebookConfig,
FlavaImageConfig,
FlavaMultimodalConfig,
FlavaTextConfig,
)
from .models.fnet import FNET_PRETRAINED_CONFIG_ARCHIVE_MAP, FNetConfig
from .models.fsmt import FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP, FSMTConfig, FSMTTokenizer
from .models.funnel import FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP, FunnelConfig, FunnelTokenizer
from .models.glpn import GLPN_PRETRAINED_CONFIG_ARCHIVE_MAP, GLPNConfig
from .models.gpt2 import GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP, GPT2Config, GPT2Tokenizer
from .models.gpt_neo import GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoConfig
from .models.gpt_neox import GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXConfig
from .models.gpt_neox_japanese import GPT_NEOX_JAPANESE_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXJapaneseConfig
from .models.gptj import GPTJ_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTJConfig
from .models.groupvit import (
GROUPVIT_PRETRAINED_CONFIG_ARCHIVE_MAP,
GroupViTConfig,
GroupViTTextConfig,
GroupViTVisionConfig,
)
from .models.herbert import HerbertTokenizer
from .models.hubert import HUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, HubertConfig
from .models.ibert import IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, IBertConfig
from .models.imagegpt import IMAGEGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, ImageGPTConfig
from .models.jukebox import (
JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP,
JukeboxConfig,
JukeboxPriorConfig,
JukeboxTokenizer,
JukeboxVQVAEConfig,
)
from .models.layoutlm import LAYOUTLM_PRETRAINED_CONFIG_ARCHIVE_MAP, LayoutLMConfig, LayoutLMTokenizer
from .models.layoutlmv2 import (
LAYOUTLMV2_PRETRAINED_CONFIG_ARCHIVE_MAP,
LayoutLMv2Config,
LayoutLMv2FeatureExtractor,
LayoutLMv2ImageProcessor,
LayoutLMv2Processor,
LayoutLMv2Tokenizer,
)
from .models.layoutlmv3 import (
LAYOUTLMV3_PRETRAINED_CONFIG_ARCHIVE_MAP,
LayoutLMv3Config,
LayoutLMv3FeatureExtractor,
LayoutLMv3ImageProcessor,
LayoutLMv3Processor,
LayoutLMv3Tokenizer,
)
from .models.layoutxlm import LayoutXLMProcessor
from .models.led import LED_PRETRAINED_CONFIG_ARCHIVE_MAP, LEDConfig, LEDTokenizer
from .models.levit import LEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP, LevitConfig
from .models.lilt import LILT_PRETRAINED_CONFIG_ARCHIVE_MAP, LiltConfig
from .models.longformer import LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, LongformerConfig, LongformerTokenizer
from .models.longt5 import LONGT5_PRETRAINED_CONFIG_ARCHIVE_MAP, LongT5Config
from .models.luke import LUKE_PRETRAINED_CONFIG_ARCHIVE_MAP, LukeConfig, LukeTokenizer
from .models.lxmert import LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP, LxmertConfig, LxmertTokenizer
from .models.m2m_100 import M2M_100_PRETRAINED_CONFIG_ARCHIVE_MAP, M2M100Config
from .models.marian import MarianConfig
from .models.markuplm import (
MARKUPLM_PRETRAINED_CONFIG_ARCHIVE_MAP,
MarkupLMConfig,
MarkupLMFeatureExtractor,
MarkupLMProcessor,
MarkupLMTokenizer,
)
from .models.maskformer import MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskFormerConfig, MaskFormerSwinConfig
from .models.mbart import MBartConfig
from .models.mctct import MCTCT_PRETRAINED_CONFIG_ARCHIVE_MAP, MCTCTConfig, MCTCTProcessor
from .models.megatron_bert import MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, MegatronBertConfig
from .models.mmbt import MMBTConfig
from .models.mobilebert import MOBILEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileBertConfig, MobileBertTokenizer
from .models.mobilenet_v1 import MOBILENET_V1_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileNetV1Config
from .models.mobilenet_v2 import MOBILENET_V2_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileNetV2Config
from .models.mobilevit import MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileViTConfig
from .models.mpnet import MPNET_PRETRAINED_CONFIG_ARCHIVE_MAP, MPNetConfig, MPNetTokenizer
from .models.mt5 import MT5Config
from .models.mvp import MvpConfig, MvpTokenizer
from .models.nat import NAT_PRETRAINED_CONFIG_ARCHIVE_MAP, NatConfig
from .models.nezha import NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP, NezhaConfig
from .models.nystromformer import NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, NystromformerConfig
from .models.openai import OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP, OpenAIGPTConfig, OpenAIGPTTokenizer
from .models.opt import OPTConfig
from .models.owlvit import (
OWLVIT_PRETRAINED_CONFIG_ARCHIVE_MAP,
OwlViTConfig,
OwlViTProcessor,
OwlViTTextConfig,
OwlViTVisionConfig,
)
from .models.pegasus import PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP, PegasusConfig, PegasusTokenizer
from .models.pegasus_x import PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP, PegasusXConfig
from .models.perceiver import PERCEIVER_PRETRAINED_CONFIG_ARCHIVE_MAP, PerceiverConfig, PerceiverTokenizer
from .models.phobert import PhobertTokenizer
from .models.plbart import PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP, PLBartConfig
from .models.poolformer import POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, PoolFormerConfig
from .models.prophetnet import PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ProphetNetConfig, ProphetNetTokenizer
from .models.qdqbert import QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, QDQBertConfig
from .models.rag import RagConfig, RagRetriever, RagTokenizer
from .models.realm import REALM_PRETRAINED_CONFIG_ARCHIVE_MAP, RealmConfig, RealmTokenizer
from .models.reformer import REFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, ReformerConfig
from .models.regnet import REGNET_PRETRAINED_CONFIG_ARCHIVE_MAP, RegNetConfig
from .models.rembert import REMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, RemBertConfig
from .models.resnet import RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ResNetConfig
from .models.retribert import RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, RetriBertConfig, RetriBertTokenizer
from .models.roberta import ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, RobertaConfig, RobertaTokenizer
from .models.roc_bert import ROC_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, RoCBertConfig, RoCBertTokenizer
from .models.roformer import ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, RoFormerConfig, RoFormerTokenizer
from .models.segformer import SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, SegformerConfig
from .models.sew import SEW_PRETRAINED_CONFIG_ARCHIVE_MAP, SEWConfig
from .models.sew_d import SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP, SEWDConfig
from .models.speech_encoder_decoder import SpeechEncoderDecoderConfig
from .models.speech_to_text import SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP, Speech2TextConfig
from .models.speech_to_text_2 import (
SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP,
Speech2Text2Config,
Speech2Text2Processor,
Speech2Text2Tokenizer,
)
from .models.splinter import SPLINTER_PRETRAINED_CONFIG_ARCHIVE_MAP, SplinterConfig, SplinterTokenizer
from .models.squeezebert import SQUEEZEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, SqueezeBertConfig, SqueezeBertTokenizer
from .models.swin import SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP, SwinConfig
from .models.swinv2 import SWINV2_PRETRAINED_CONFIG_ARCHIVE_MAP, Swinv2Config
from .models.switch_transformers import SWITCH_TRANSFORMERS_PRETRAINED_CONFIG_ARCHIVE_MAP, SwitchTransformersConfig
from .models.t5 import T5_PRETRAINED_CONFIG_ARCHIVE_MAP, T5Config
from .models.table_transformer import TABLE_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, TableTransformerConfig
from .models.tapas import TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP, TapasConfig, TapasTokenizer
from .models.tapex import TapexTokenizer
from .models.time_series_transformer import (
TIME_SERIES_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
TimeSeriesTransformerConfig,
)
from .models.timesformer import TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, TimesformerConfig
from .models.trajectory_transformer import (
TRAJECTORY_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
TrajectoryTransformerConfig,
)
from .models.transfo_xl import (
TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP,
TransfoXLConfig,
TransfoXLCorpus,
TransfoXLTokenizer,
)
from .models.trocr import TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP, TrOCRConfig, TrOCRProcessor
from .models.unispeech import UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP, UniSpeechConfig
from .models.unispeech_sat import UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP, UniSpeechSatConfig
from .models.van import VAN_PRETRAINED_CONFIG_ARCHIVE_MAP, VanConfig
from .models.videomae import VIDEOMAE_PRETRAINED_CONFIG_ARCHIVE_MAP, VideoMAEConfig
from .models.vilt import (
VILT_PRETRAINED_CONFIG_ARCHIVE_MAP,
ViltConfig,
ViltFeatureExtractor,
ViltImageProcessor,
ViltProcessor,
)
from .models.vision_encoder_decoder import VisionEncoderDecoderConfig
from .models.vision_text_dual_encoder import VisionTextDualEncoderConfig, VisionTextDualEncoderProcessor
from .models.visual_bert import VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, VisualBertConfig
from .models.vit import VIT_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTConfig
from .models.vit_hybrid import VIT_HYBRID_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTHybridConfig
from .models.vit_mae import VIT_MAE_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTMAEConfig
from .models.vit_msn import VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTMSNConfig
from .models.wav2vec2 import (
WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP,
Wav2Vec2Config,
Wav2Vec2CTCTokenizer,
Wav2Vec2FeatureExtractor,
Wav2Vec2Processor,
Wav2Vec2Tokenizer,
)
from .models.wav2vec2_conformer import WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, Wav2Vec2ConformerConfig
from .models.wav2vec2_phoneme import Wav2Vec2PhonemeCTCTokenizer
from .models.wav2vec2_with_lm import Wav2Vec2ProcessorWithLM
from .models.wavlm import WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP, WavLMConfig
from .models.whisper import (
WHISPER_PRETRAINED_CONFIG_ARCHIVE_MAP,
WhisperConfig,
WhisperFeatureExtractor,
WhisperProcessor,
WhisperTokenizer,
)
from .models.x_clip import (
XCLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
XCLIPConfig,
XCLIPProcessor,
XCLIPTextConfig,
XCLIPVisionConfig,
)
from .models.xglm import XGLM_PRETRAINED_CONFIG_ARCHIVE_MAP, XGLMConfig
from .models.xlm import XLM_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMConfig, XLMTokenizer
from .models.xlm_prophetnet import XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMProphetNetConfig
from .models.xlm_roberta import XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMRobertaConfig
from .models.xlm_roberta_xl import XLM_ROBERTA_XL_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMRobertaXLConfig
from .models.xlnet import XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP, XLNetConfig
from .models.yolos import YOLOS_PRETRAINED_CONFIG_ARCHIVE_MAP, YolosConfig
from .models.yoso import YOSO_PRETRAINED_CONFIG_ARCHIVE_MAP, YosoConfig
# Pipelines
from .pipelines import (
AudioClassificationPipeline,
AutomaticSpeechRecognitionPipeline,
Conversation,
ConversationalPipeline,
CsvPipelineDataFormat,
DepthEstimationPipeline,
DocumentQuestionAnsweringPipeline,
FeatureExtractionPipeline,
FillMaskPipeline,
ImageClassificationPipeline,
ImageSegmentationPipeline,
ImageToTextPipeline,
JsonPipelineDataFormat,
NerPipeline,
ObjectDetectionPipeline,
PipedPipelineDataFormat,
Pipeline,
PipelineDataFormat,
QuestionAnsweringPipeline,
SummarizationPipeline,
TableQuestionAnsweringPipeline,
Text2TextGenerationPipeline,
TextClassificationPipeline,
TextGenerationPipeline,
TokenClassificationPipeline,
TranslationPipeline,
VideoClassificationPipeline,
VisualQuestionAnsweringPipeline,
ZeroShotClassificationPipeline,
ZeroShotImageClassificationPipeline,
ZeroShotObjectDetectionPipeline,
pipeline,
)
from .processing_utils import ProcessorMixin
# Tokenization
from .tokenization_utils import PreTrainedTokenizer
from .tokenization_utils_base import (
AddedToken,
BatchEncoding,
CharSpan,
PreTrainedTokenizerBase,
SpecialTokensMixin,
TokenSpan,
)
# Trainer
from .trainer_callback import (
DefaultFlowCallback,
EarlyStoppingCallback,
PrinterCallback,
ProgressCallback,
TrainerCallback,
TrainerControl,
TrainerState,
)
from .trainer_utils import EvalPrediction, IntervalStrategy, SchedulerType, enable_full_determinism, set_seed
from .training_args import TrainingArguments
from .training_args_seq2seq import Seq2SeqTrainingArguments
from .training_args_tf import TFTrainingArguments
# Files and general utilities
from .utils import (
CONFIG_NAME,
MODEL_CARD_NAME,
PYTORCH_PRETRAINED_BERT_CACHE,
PYTORCH_TRANSFORMERS_CACHE,
SPIECE_UNDERLINE,
TF2_WEIGHTS_NAME,
TF_WEIGHTS_NAME,
TRANSFORMERS_CACHE,
WEIGHTS_NAME,
TensorType,
add_end_docstrings,
add_start_docstrings,
is_apex_available,
is_datasets_available,
is_decord_available,
is_faiss_available,
is_flax_available,
is_keras_nlp_available,
is_phonemizer_available,
is_psutil_available,
is_py3nvml_available,
is_pyctcdecode_available,
is_safetensors_available,
is_scipy_available,
is_sentencepiece_available,
is_sklearn_available,
is_speech_available,
is_tensorflow_text_available,
is_tf_available,
is_timm_available,
is_tokenizers_available,
is_torch_available,
is_torch_tpu_available,
is_vision_available,
logging,
)
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_sentencepiece_objects import *
else:
from .models.albert import AlbertTokenizer
from .models.barthez import BarthezTokenizer
from .models.bartpho import BartphoTokenizer
from .models.bert_generation import BertGenerationTokenizer
from .models.big_bird import BigBirdTokenizer
from .models.camembert import CamembertTokenizer
from .models.cpm import CpmTokenizer
from .models.deberta_v2 import DebertaV2Tokenizer
from .models.fnet import FNetTokenizer
from .models.gpt_sw3 import GPTSw3Tokenizer
from .models.layoutxlm import LayoutXLMTokenizer
from .models.m2m_100 import M2M100Tokenizer
from .models.marian import MarianTokenizer
from .models.mbart import MBart50Tokenizer, MBartTokenizer
from .models.mluke import MLukeTokenizer
from .models.mt5 import MT5Tokenizer
from .models.nllb import NllbTokenizer
from .models.pegasus import PegasusTokenizer
from .models.plbart import PLBartTokenizer
from .models.reformer import ReformerTokenizer
from .models.rembert import RemBertTokenizer
from .models.speech_to_text import Speech2TextTokenizer
from .models.t5 import T5Tokenizer
from .models.xglm import XGLMTokenizer
from .models.xlm_prophetnet import XLMProphetNetTokenizer
from .models.xlm_roberta import XLMRobertaTokenizer
from .models.xlnet import XLNetTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_tokenizers_objects import *
else:
# Fast tokenizers imports
from .models.albert import AlbertTokenizerFast
from .models.bart import BartTokenizerFast
from .models.barthez import BarthezTokenizerFast
from .models.bert import BertTokenizerFast
from .models.big_bird import BigBirdTokenizerFast
from .models.blenderbot import BlenderbotTokenizerFast
from .models.blenderbot_small import BlenderbotSmallTokenizerFast
from .models.bloom import BloomTokenizerFast
from .models.camembert import CamembertTokenizerFast
from .models.clip import CLIPTokenizerFast
from .models.codegen import CodeGenTokenizerFast
from .models.convbert import ConvBertTokenizerFast
from .models.cpm import CpmTokenizerFast
from .models.deberta import DebertaTokenizerFast
from .models.deberta_v2 import DebertaV2TokenizerFast
from .models.distilbert import DistilBertTokenizerFast
from .models.dpr import DPRContextEncoderTokenizerFast, DPRQuestionEncoderTokenizerFast, DPRReaderTokenizerFast
from .models.electra import ElectraTokenizerFast
from .models.fnet import FNetTokenizerFast
from .models.funnel import FunnelTokenizerFast
from .models.gpt2 import GPT2TokenizerFast
from .models.gpt_neox import GPTNeoXTokenizerFast
from .models.gpt_neox_japanese import GPTNeoXJapaneseTokenizer
from .models.herbert import HerbertTokenizerFast
from .models.layoutlm import LayoutLMTokenizerFast
from .models.layoutlmv2 import LayoutLMv2TokenizerFast
from .models.layoutlmv3 import LayoutLMv3TokenizerFast
from .models.layoutxlm import LayoutXLMTokenizerFast
from .models.led import LEDTokenizerFast
from .models.longformer import LongformerTokenizerFast
from .models.lxmert import LxmertTokenizerFast
from .models.markuplm import MarkupLMTokenizerFast
from .models.mbart import MBartTokenizerFast
from .models.mbart50 import MBart50TokenizerFast
from .models.mobilebert import MobileBertTokenizerFast
from .models.mpnet import MPNetTokenizerFast
from .models.mt5 import MT5TokenizerFast
from .models.mvp import MvpTokenizerFast
from .models.nllb import NllbTokenizerFast
from .models.openai import OpenAIGPTTokenizerFast
from .models.pegasus import PegasusTokenizerFast
from .models.realm import RealmTokenizerFast
from .models.reformer import ReformerTokenizerFast
from .models.rembert import RemBertTokenizerFast
from .models.retribert import RetriBertTokenizerFast
from .models.roberta import RobertaTokenizerFast
from .models.roformer import RoFormerTokenizerFast
from .models.splinter import SplinterTokenizerFast
from .models.squeezebert import SqueezeBertTokenizerFast
from .models.t5 import T5TokenizerFast
from .models.xglm import XGLMTokenizerFast
from .models.xlm_roberta import XLMRobertaTokenizerFast
from .models.xlnet import XLNetTokenizerFast
from .tokenization_utils_fast import PreTrainedTokenizerFast
try:
if not (is_sentencepiece_available() and is_tokenizers_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummies_sentencepiece_and_tokenizers_objects import *
else:
from .convert_slow_tokenizer import SLOW_TO_FAST_CONVERTERS, convert_slow_tokenizer
try:
if not is_speech_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_speech_objects import *
else:
from .models.audio_spectrogram_transformer import ASTFeatureExtractor
from .models.mctct import MCTCTFeatureExtractor
from .models.speech_to_text import Speech2TextFeatureExtractor
try:
if not is_tensorflow_text_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_tensorflow_text_objects import *
else:
from .models.bert import TFBertTokenizer
try:
if not is_keras_nlp_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_keras_nlp_objects import *
else:
from .models.gpt2 import TFGPT2Tokenizer
try:
if not (is_speech_available() and is_sentencepiece_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_sentencepiece_and_speech_objects import *
else:
from .models.speech_to_text import Speech2TextProcessor
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_vision_objects import *
else:
from .image_processing_utils import ImageProcessingMixin
from .image_transforms import rescale, resize, to_pil_image
from .image_utils import ImageFeatureExtractionMixin
from .models.beit import BeitFeatureExtractor, BeitImageProcessor
from .models.bit import BitImageProcessor
from .models.chinese_clip import ChineseCLIPFeatureExtractor, ChineseCLIPImageProcessor
from .models.clip import CLIPFeatureExtractor, CLIPImageProcessor
from .models.conditional_detr import ConditionalDetrFeatureExtractor, ConditionalDetrImageProcessor
from .models.convnext import ConvNextFeatureExtractor, ConvNextImageProcessor
from .models.deformable_detr import DeformableDetrFeatureExtractor, DeformableDetrImageProcessor
from .models.deit import DeiTFeatureExtractor, DeiTImageProcessor
from .models.detr import DetrFeatureExtractor, DetrImageProcessor
from .models.donut import DonutFeatureExtractor, DonutImageProcessor
from .models.dpt import DPTFeatureExtractor, DPTImageProcessor
from .models.flava import FlavaFeatureExtractor, FlavaImageProcessor, FlavaProcessor
from .models.glpn import GLPNFeatureExtractor, GLPNImageProcessor
from .models.imagegpt import ImageGPTFeatureExtractor, ImageGPTImageProcessor
from .models.layoutlmv2 import LayoutLMv2FeatureExtractor, LayoutLMv2ImageProcessor
from .models.layoutlmv3 import LayoutLMv3FeatureExtractor, LayoutLMv3ImageProcessor
from .models.levit import LevitFeatureExtractor, LevitImageProcessor
from .models.maskformer import MaskFormerFeatureExtractor, MaskFormerImageProcessor
from .models.mobilenet_v1 import MobileNetV1FeatureExtractor, MobileNetV1ImageProcessor
from .models.mobilenet_v2 import MobileNetV2FeatureExtractor, MobileNetV2ImageProcessor
from .models.mobilevit import MobileViTFeatureExtractor, MobileViTImageProcessor
from .models.owlvit import OwlViTFeatureExtractor, OwlViTImageProcessor
from .models.perceiver import PerceiverFeatureExtractor, PerceiverImageProcessor
from .models.poolformer import PoolFormerFeatureExtractor, PoolFormerImageProcessor
from .models.segformer import SegformerFeatureExtractor, SegformerImageProcessor
from .models.videomae import VideoMAEFeatureExtractor, VideoMAEImageProcessor
from .models.vilt import ViltFeatureExtractor, ViltImageProcessor, ViltProcessor
from .models.vit import ViTFeatureExtractor, ViTImageProcessor
from .models.vit_hybrid import ViTHybridImageProcessor
from .models.yolos import YolosFeatureExtractor, YolosImageProcessor
# Modeling
try:
if not (is_timm_available() and is_vision_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_timm_and_vision_objects import *
else:
from .models.conditional_detr import (
CONDITIONAL_DETR_PRETRAINED_MODEL_ARCHIVE_LIST,
ConditionalDetrForObjectDetection,
ConditionalDetrForSegmentation,
ConditionalDetrModel,
ConditionalDetrPreTrainedModel,
)
from .models.deformable_detr import (
DEFORMABLE_DETR_PRETRAINED_MODEL_ARCHIVE_LIST,
DeformableDetrForObjectDetection,
DeformableDetrModel,
DeformableDetrPreTrainedModel,
)
from .models.detr import (
DETR_PRETRAINED_MODEL_ARCHIVE_LIST,
DetrForObjectDetection,
DetrForSegmentation,
DetrModel,
DetrPreTrainedModel,
)
from .models.table_transformer import (
TABLE_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TableTransformerForObjectDetection,
TableTransformerModel,
TableTransformerPreTrainedModel,
)
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from .utils.dummy_pt_objects import *
else:
# Benchmarks
from .benchmark.benchmark import PyTorchBenchmark
from .benchmark.benchmark_args import PyTorchBenchmarkArguments
from .data.datasets import (
GlueDataset,
GlueDataTrainingArguments,
LineByLineTextDataset,
LineByLineWithRefDataset,
LineByLineWithSOPTextDataset,
SquadDataset,
SquadDataTrainingArguments,
TextDataset,
TextDatasetForNextSentencePrediction,
)
from .generation import (
BeamScorer,
BeamSearchScorer,
ConstrainedBeamSearchScorer,
Constraint,
ConstraintListState,
DisjunctiveConstraint,
ForcedBOSTokenLogitsProcessor,
ForcedEOSTokenLogitsProcessor,
GenerationMixin,
HammingDiversityLogitsProcessor,
InfNanRemoveLogitsProcessor,
LogitsProcessor,
LogitsProcessorList,
LogitsWarper,
MaxLengthCriteria,
MaxTimeCriteria,
MinLengthLogitsProcessor,
NoBadWordsLogitsProcessor,
NoRepeatNGramLogitsProcessor,
PhrasalConstraint,
PrefixConstrainedLogitsProcessor,
RepetitionPenaltyLogitsProcessor,
StoppingCriteria,
StoppingCriteriaList,
TemperatureLogitsWarper,
TopKLogitsWarper,
TopPLogitsWarper,
TypicalLogitsWarper,
top_k_top_p_filtering,
)
from .modeling_utils import PreTrainedModel
# PyTorch model imports
from .models.albert import (
ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
AlbertForMaskedLM,
AlbertForMultipleChoice,
AlbertForPreTraining,
AlbertForQuestionAnswering,
AlbertForSequenceClassification,
AlbertForTokenClassification,
AlbertModel,
AlbertPreTrainedModel,
load_tf_weights_in_albert,
)
from .models.audio_spectrogram_transformer import (
AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
ASTForAudioClassification,
ASTModel,
ASTPreTrainedModel,
)
from .models.auto import (
MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING,
MODEL_FOR_AUDIO_XVECTOR_MAPPING,
MODEL_FOR_BACKBONE_MAPPING,
MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING,
MODEL_FOR_CAUSAL_LM_MAPPING,
MODEL_FOR_CTC_MAPPING,
MODEL_FOR_DEPTH_ESTIMATION_MAPPING,
MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
MODEL_FOR_IMAGE_SEGMENTATION_MAPPING,
MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING,
MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
MODEL_FOR_MULTIPLE_CHOICE_MAPPING,
MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING,
MODEL_FOR_OBJECT_DETECTION_MAPPING,
MODEL_FOR_PRETRAINING_MAPPING,
MODEL_FOR_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING,
MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING,
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING,
MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
MODEL_FOR_VISION_2_SEQ_MAPPING,
MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING,
MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING,
MODEL_MAPPING,
MODEL_WITH_LM_HEAD_MAPPING,
AutoBackbone,
AutoModel,
AutoModelForAudioClassification,
AutoModelForAudioFrameClassification,
AutoModelForAudioXVector,
AutoModelForCausalLM,
AutoModelForCTC,
AutoModelForDepthEstimation,
AutoModelForDocumentQuestionAnswering,
AutoModelForImageClassification,
AutoModelForImageSegmentation,
AutoModelForInstanceSegmentation,
AutoModelForMaskedImageModeling,
AutoModelForMaskedLM,
AutoModelForMultipleChoice,
AutoModelForNextSentencePrediction,
AutoModelForObjectDetection,
AutoModelForPreTraining,
AutoModelForQuestionAnswering,
AutoModelForSemanticSegmentation,
AutoModelForSeq2SeqLM,
AutoModelForSequenceClassification,
AutoModelForSpeechSeq2Seq,
AutoModelForTableQuestionAnswering,
AutoModelForTokenClassification,
AutoModelForVideoClassification,
AutoModelForVision2Seq,
AutoModelForVisualQuestionAnswering,
AutoModelForZeroShotObjectDetection,
AutoModelWithLMHead,
)
from .models.bart import (
BART_PRETRAINED_MODEL_ARCHIVE_LIST,
BartForCausalLM,
BartForConditionalGeneration,
BartForQuestionAnswering,
BartForSequenceClassification,
BartModel,
BartPretrainedModel,
PretrainedBartModel,
)
from .models.beit import (
BEIT_PRETRAINED_MODEL_ARCHIVE_LIST,
BeitForImageClassification,
BeitForMaskedImageModeling,
BeitForSemanticSegmentation,
BeitModel,
BeitPreTrainedModel,
)
from .models.bert import (
BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
BertForMaskedLM,
BertForMultipleChoice,
BertForNextSentencePrediction,
BertForPreTraining,
BertForQuestionAnswering,
BertForSequenceClassification,
BertForTokenClassification,
BertLayer,
BertLMHeadModel,
BertModel,
BertPreTrainedModel,
load_tf_weights_in_bert,
)
from .models.bert_generation import (
BertGenerationDecoder,
BertGenerationEncoder,
BertGenerationPreTrainedModel,
load_tf_weights_in_bert_generation,
)
from .models.big_bird import (
BIG_BIRD_PRETRAINED_MODEL_ARCHIVE_LIST,
BigBirdForCausalLM,
BigBirdForMaskedLM,
BigBirdForMultipleChoice,
BigBirdForPreTraining,
BigBirdForQuestionAnswering,
BigBirdForSequenceClassification,
BigBirdForTokenClassification,
BigBirdLayer,
BigBirdModel,
BigBirdPreTrainedModel,
load_tf_weights_in_big_bird,
)
from .models.bigbird_pegasus import (
BIGBIRD_PEGASUS_PRETRAINED_MODEL_ARCHIVE_LIST,
BigBirdPegasusForCausalLM,
BigBirdPegasusForConditionalGeneration,
BigBirdPegasusForQuestionAnswering,
BigBirdPegasusForSequenceClassification,
BigBirdPegasusModel,
BigBirdPegasusPreTrainedModel,
)
from .models.biogpt import (
BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST,
BioGptForCausalLM,
BioGptModel,
BioGptPreTrainedModel,
)
from .models.bit import (
BIT_PRETRAINED_MODEL_ARCHIVE_LIST,
BitBackbone,
BitForImageClassification,
BitModel,
BitPreTrainedModel,
)
from .models.blenderbot import (
BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotForCausalLM,
BlenderbotForConditionalGeneration,
BlenderbotModel,
BlenderbotPreTrainedModel,
)
from .models.blenderbot_small import (
BLENDERBOT_SMALL_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotSmallForCausalLM,
BlenderbotSmallForConditionalGeneration,
BlenderbotSmallModel,
BlenderbotSmallPreTrainedModel,
)
from .models.bloom import (
BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST,
BloomForCausalLM,
BloomForQuestionAnswering,
BloomForSequenceClassification,
BloomForTokenClassification,
BloomModel,
BloomPreTrainedModel,
)
from .models.camembert import (
CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
CamembertForCausalLM,
CamembertForMaskedLM,
CamembertForMultipleChoice,
CamembertForQuestionAnswering,
CamembertForSequenceClassification,
CamembertForTokenClassification,
CamembertModel,
CamembertPreTrainedModel,
)
from .models.canine import (
CANINE_PRETRAINED_MODEL_ARCHIVE_LIST,
CanineForMultipleChoice,
CanineForQuestionAnswering,
CanineForSequenceClassification,
CanineForTokenClassification,
CanineLayer,
CanineModel,
CaninePreTrainedModel,
load_tf_weights_in_canine,
)
from .models.chinese_clip import (
CHINESE_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
ChineseCLIPModel,
ChineseCLIPPreTrainedModel,
ChineseCLIPTextModel,
ChineseCLIPVisionModel,
)
from .models.clip import (
CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
CLIPModel,
CLIPPreTrainedModel,
CLIPTextModel,
CLIPTextModelWithProjection,
CLIPVisionModel,
CLIPVisionModelWithProjection,
)
from .models.clipseg import (
CLIPSEG_PRETRAINED_MODEL_ARCHIVE_LIST,
CLIPSegForImageSegmentation,
CLIPSegModel,
CLIPSegPreTrainedModel,
CLIPSegTextModel,
CLIPSegVisionModel,
)
from .models.codegen import (
CODEGEN_PRETRAINED_MODEL_ARCHIVE_LIST,
CodeGenForCausalLM,
CodeGenModel,
CodeGenPreTrainedModel,
)
from .models.convbert import (
CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
ConvBertForMaskedLM,
ConvBertForMultipleChoice,
ConvBertForQuestionAnswering,
ConvBertForSequenceClassification,
ConvBertForTokenClassification,
ConvBertLayer,
ConvBertModel,
ConvBertPreTrainedModel,
load_tf_weights_in_convbert,
)
from .models.convnext import (
CONVNEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
ConvNextForImageClassification,
ConvNextModel,
ConvNextPreTrainedModel,
)
from .models.ctrl import (
CTRL_PRETRAINED_MODEL_ARCHIVE_LIST,
CTRLForSequenceClassification,
CTRLLMHeadModel,
CTRLModel,
CTRLPreTrainedModel,
)
from .models.cvt import (
CVT_PRETRAINED_MODEL_ARCHIVE_LIST,
CvtForImageClassification,
CvtModel,
CvtPreTrainedModel,
)
from .models.data2vec import (
DATA2VEC_AUDIO_PRETRAINED_MODEL_ARCHIVE_LIST,
DATA2VEC_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
DATA2VEC_VISION_PRETRAINED_MODEL_ARCHIVE_LIST,
Data2VecAudioForAudioFrameClassification,
Data2VecAudioForCTC,
Data2VecAudioForSequenceClassification,
Data2VecAudioForXVector,
Data2VecAudioModel,
Data2VecAudioPreTrainedModel,
Data2VecTextForCausalLM,
Data2VecTextForMaskedLM,
Data2VecTextForMultipleChoice,
Data2VecTextForQuestionAnswering,
Data2VecTextForSequenceClassification,
Data2VecTextForTokenClassification,
Data2VecTextModel,
Data2VecTextPreTrainedModel,
Data2VecVisionForImageClassification,
Data2VecVisionForSemanticSegmentation,
Data2VecVisionModel,
Data2VecVisionPreTrainedModel,
)
from .models.deberta import (
DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
DebertaForMaskedLM,
DebertaForQuestionAnswering,
DebertaForSequenceClassification,
DebertaForTokenClassification,
DebertaModel,
DebertaPreTrainedModel,
)
from .models.deberta_v2 import (
DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST,
DebertaV2ForMaskedLM,
DebertaV2ForMultipleChoice,
DebertaV2ForQuestionAnswering,
DebertaV2ForSequenceClassification,
DebertaV2ForTokenClassification,
DebertaV2Model,
DebertaV2PreTrainedModel,
)
from .models.decision_transformer import (
DECISION_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
DecisionTransformerGPT2Model,
DecisionTransformerGPT2PreTrainedModel,
DecisionTransformerModel,
DecisionTransformerPreTrainedModel,
)
from .models.deit import (
DEIT_PRETRAINED_MODEL_ARCHIVE_LIST,
DeiTForImageClassification,
DeiTForImageClassificationWithTeacher,
DeiTForMaskedImageModeling,
DeiTModel,
DeiTPreTrainedModel,
)
from .models.dinat import (
DINAT_PRETRAINED_MODEL_ARCHIVE_LIST,
DinatForImageClassification,
DinatModel,
DinatPreTrainedModel,
)
from .models.distilbert import (
DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
DistilBertForMaskedLM,
DistilBertForMultipleChoice,
DistilBertForQuestionAnswering,
DistilBertForSequenceClassification,
DistilBertForTokenClassification,
DistilBertModel,
DistilBertPreTrainedModel,
)
from .models.donut import DONUT_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST, DonutSwinModel, DonutSwinPreTrainedModel
from .models.dpr import (
DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST,
DPRContextEncoder,
DPRPretrainedContextEncoder,
DPRPreTrainedModel,
DPRPretrainedQuestionEncoder,
DPRPretrainedReader,
DPRQuestionEncoder,
DPRReader,
)
from .models.dpt import (
DPT_PRETRAINED_MODEL_ARCHIVE_LIST,
DPTForDepthEstimation,
DPTForSemanticSegmentation,
DPTModel,
DPTPreTrainedModel,
)
from .models.electra import (
ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
ElectraForCausalLM,
ElectraForMaskedLM,
ElectraForMultipleChoice,
ElectraForPreTraining,
ElectraForQuestionAnswering,
ElectraForSequenceClassification,
ElectraForTokenClassification,
ElectraModel,
ElectraPreTrainedModel,
load_tf_weights_in_electra,
)
from .models.encoder_decoder import EncoderDecoderModel
from .models.ernie import (
ERNIE_PRETRAINED_MODEL_ARCHIVE_LIST,
ErnieForCausalLM,
ErnieForMaskedLM,
ErnieForMultipleChoice,
ErnieForNextSentencePrediction,
ErnieForPreTraining,
ErnieForQuestionAnswering,
ErnieForSequenceClassification,
ErnieForTokenClassification,
ErnieModel,
ErniePreTrainedModel,
)
from .models.esm import (
ESM_PRETRAINED_MODEL_ARCHIVE_LIST,
EsmFoldPreTrainedModel,
EsmForMaskedLM,
EsmForProteinFolding,
EsmForSequenceClassification,
EsmForTokenClassification,
EsmModel,
EsmPreTrainedModel,
)
from .models.flaubert import (
FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
FlaubertForMultipleChoice,
FlaubertForQuestionAnswering,
FlaubertForQuestionAnsweringSimple,
FlaubertForSequenceClassification,
FlaubertForTokenClassification,
FlaubertModel,
FlaubertPreTrainedModel,
FlaubertWithLMHeadModel,
)
from .models.flava import (
FLAVA_PRETRAINED_MODEL_ARCHIVE_LIST,
FlavaForPreTraining,
FlavaImageCodebook,
FlavaImageModel,
FlavaModel,
FlavaMultimodalModel,
FlavaPreTrainedModel,
FlavaTextModel,
)
from .models.fnet import (
FNET_PRETRAINED_MODEL_ARCHIVE_LIST,
FNetForMaskedLM,
FNetForMultipleChoice,
FNetForNextSentencePrediction,
FNetForPreTraining,
FNetForQuestionAnswering,
FNetForSequenceClassification,
FNetForTokenClassification,
FNetLayer,
FNetModel,
FNetPreTrainedModel,
)
from .models.fsmt import FSMTForConditionalGeneration, FSMTModel, PretrainedFSMTModel
from .models.funnel import (
FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST,
FunnelBaseModel,
FunnelForMaskedLM,
FunnelForMultipleChoice,
FunnelForPreTraining,
FunnelForQuestionAnswering,
FunnelForSequenceClassification,
FunnelForTokenClassification,
FunnelModel,
FunnelPreTrainedModel,
load_tf_weights_in_funnel,
)
from .models.glpn import (
GLPN_PRETRAINED_MODEL_ARCHIVE_LIST,
GLPNForDepthEstimation,
GLPNModel,
GLPNPreTrainedModel,
)
from .models.gpt2 import (
GPT2_PRETRAINED_MODEL_ARCHIVE_LIST,
GPT2DoubleHeadsModel,
GPT2ForSequenceClassification,
GPT2ForTokenClassification,
GPT2LMHeadModel,
GPT2Model,
GPT2PreTrainedModel,
load_tf_weights_in_gpt2,
)
from .models.gpt_neo import (
GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoForCausalLM,
GPTNeoForSequenceClassification,
GPTNeoModel,
GPTNeoPreTrainedModel,
load_tf_weights_in_gpt_neo,
)
from .models.gpt_neox import (
GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoXForCausalLM,
GPTNeoXLayer,
GPTNeoXModel,
GPTNeoXPreTrainedModel,
)
from .models.gpt_neox_japanese import (
GPT_NEOX_JAPANESE_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoXJapaneseForCausalLM,
GPTNeoXJapaneseLayer,
GPTNeoXJapaneseModel,
GPTNeoXJapanesePreTrainedModel,
)
from .models.gptj import (
GPTJ_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTJForCausalLM,
GPTJForQuestionAnswering,
GPTJForSequenceClassification,
GPTJModel,
GPTJPreTrainedModel,
)
from .models.groupvit import (
GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
GroupViTModel,
GroupViTPreTrainedModel,
GroupViTTextModel,
GroupViTVisionModel,
)
from .models.hubert import (
HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
HubertForCTC,
HubertForSequenceClassification,
HubertModel,
HubertPreTrainedModel,
)
from .models.ibert import (
IBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
IBertForMaskedLM,
IBertForMultipleChoice,
IBertForQuestionAnswering,
IBertForSequenceClassification,
IBertForTokenClassification,
IBertModel,
IBertPreTrainedModel,
)
from .models.imagegpt import (
IMAGEGPT_PRETRAINED_MODEL_ARCHIVE_LIST,
ImageGPTForCausalImageModeling,
ImageGPTForImageClassification,
ImageGPTModel,
ImageGPTPreTrainedModel,
load_tf_weights_in_imagegpt,
)
from .models.jukebox import (
JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST,
JukeboxModel,
JukeboxPreTrainedModel,
JukeboxPrior,
JukeboxVQVAE,
)
from .models.layoutlm import (
LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST,
LayoutLMForMaskedLM,
LayoutLMForQuestionAnswering,
LayoutLMForSequenceClassification,
LayoutLMForTokenClassification,
LayoutLMModel,
LayoutLMPreTrainedModel,
)
from .models.layoutlmv2 import (
LAYOUTLMV2_PRETRAINED_MODEL_ARCHIVE_LIST,
LayoutLMv2ForQuestionAnswering,
LayoutLMv2ForSequenceClassification,
LayoutLMv2ForTokenClassification,
LayoutLMv2Model,
LayoutLMv2PreTrainedModel,
)
from .models.layoutlmv3 import (
LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST,
LayoutLMv3ForQuestionAnswering,
LayoutLMv3ForSequenceClassification,
LayoutLMv3ForTokenClassification,
LayoutLMv3Model,
LayoutLMv3PreTrainedModel,
)
from .models.led import (
LED_PRETRAINED_MODEL_ARCHIVE_LIST,
LEDForConditionalGeneration,
LEDForQuestionAnswering,
LEDForSequenceClassification,
LEDModel,
LEDPreTrainedModel,
)
from .models.levit import (
LEVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
LevitForImageClassification,
LevitForImageClassificationWithTeacher,
LevitModel,
LevitPreTrainedModel,
)
from .models.lilt import (
LILT_PRETRAINED_MODEL_ARCHIVE_LIST,
LiltForQuestionAnswering,
LiltForSequenceClassification,
LiltForTokenClassification,
LiltModel,
LiltPreTrainedModel,
)
from .models.longformer import (
LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
LongformerForMaskedLM,
LongformerForMultipleChoice,
LongformerForQuestionAnswering,
LongformerForSequenceClassification,
LongformerForTokenClassification,
LongformerModel,
LongformerPreTrainedModel,
LongformerSelfAttention,
)
from .models.longt5 import (
LONGT5_PRETRAINED_MODEL_ARCHIVE_LIST,
LongT5EncoderModel,
LongT5ForConditionalGeneration,
LongT5Model,
LongT5PreTrainedModel,
)
from .models.luke import (
LUKE_PRETRAINED_MODEL_ARCHIVE_LIST,
LukeForEntityClassification,
LukeForEntityPairClassification,
LukeForEntitySpanClassification,
LukeForMaskedLM,
LukeForMultipleChoice,
LukeForQuestionAnswering,
LukeForSequenceClassification,
LukeForTokenClassification,
LukeModel,
LukePreTrainedModel,
)
from .models.lxmert import (
LxmertEncoder,
LxmertForPreTraining,
LxmertForQuestionAnswering,
LxmertModel,
LxmertPreTrainedModel,
LxmertVisualFeatureEncoder,
LxmertXLayer,
)
from .models.m2m_100 import (
M2M_100_PRETRAINED_MODEL_ARCHIVE_LIST,
M2M100ForConditionalGeneration,
M2M100Model,
M2M100PreTrainedModel,
)
from .models.marian import MarianForCausalLM, MarianModel, MarianMTModel
from .models.markuplm import (
MARKUPLM_PRETRAINED_MODEL_ARCHIVE_LIST,
MarkupLMForQuestionAnswering,
MarkupLMForSequenceClassification,
MarkupLMForTokenClassification,
MarkupLMModel,
MarkupLMPreTrainedModel,
)
from .models.maskformer import (
MASKFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
MaskFormerForInstanceSegmentation,
MaskFormerModel,
MaskFormerPreTrainedModel,
MaskFormerSwinBackbone,
)
from .models.mbart import (
MBartForCausalLM,
MBartForConditionalGeneration,
MBartForQuestionAnswering,
MBartForSequenceClassification,
MBartModel,
MBartPreTrainedModel,
)
from .models.mctct import MCTCT_PRETRAINED_MODEL_ARCHIVE_LIST, MCTCTForCTC, MCTCTModel, MCTCTPreTrainedModel
from .models.megatron_bert import (
MEGATRON_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
MegatronBertForCausalLM,
MegatronBertForMaskedLM,
MegatronBertForMultipleChoice,
MegatronBertForNextSentencePrediction,
MegatronBertForPreTraining,
MegatronBertForQuestionAnswering,
MegatronBertForSequenceClassification,
MegatronBertForTokenClassification,
MegatronBertModel,
MegatronBertPreTrainedModel,
)
from .models.mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
from .models.mobilebert import (
MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
MobileBertLayer,
MobileBertModel,
MobileBertPreTrainedModel,
load_tf_weights_in_mobilebert,
)
from .models.mobilenet_v1 import (
MOBILENET_V1_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileNetV1ForImageClassification,
MobileNetV1Model,
MobileNetV1PreTrainedModel,
load_tf_weights_in_mobilenet_v1,
)
from .models.mobilenet_v2 import (
MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileNetV2ForImageClassification,
MobileNetV2ForSemanticSegmentation,
MobileNetV2Model,
MobileNetV2PreTrainedModel,
load_tf_weights_in_mobilenet_v2,
)
from .models.mobilevit import (
MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
MobileViTForImageClassification,
MobileViTForSemanticSegmentation,
MobileViTModel,
MobileViTPreTrainedModel,
)
from .models.mpnet import (
MPNET_PRETRAINED_MODEL_ARCHIVE_LIST,
MPNetForMaskedLM,
MPNetForMultipleChoice,
MPNetForQuestionAnswering,
MPNetForSequenceClassification,
MPNetForTokenClassification,
MPNetLayer,
MPNetModel,
MPNetPreTrainedModel,
)
from .models.mt5 import MT5EncoderModel, MT5ForConditionalGeneration, MT5Model
from .models.mvp import (
MVP_PRETRAINED_MODEL_ARCHIVE_LIST,
MvpForCausalLM,
MvpForConditionalGeneration,
MvpForQuestionAnswering,
MvpForSequenceClassification,
MvpModel,
MvpPreTrainedModel,
)
from .models.nat import (
NAT_PRETRAINED_MODEL_ARCHIVE_LIST,
NatForImageClassification,
NatModel,
NatPreTrainedModel,
)
from .models.nezha import (
NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST,
NezhaForMaskedLM,
NezhaForMultipleChoice,
NezhaForNextSentencePrediction,
NezhaForPreTraining,
NezhaForQuestionAnswering,
NezhaForSequenceClassification,
NezhaForTokenClassification,
NezhaModel,
NezhaPreTrainedModel,
)
from .models.nystromformer import (
NYSTROMFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
NystromformerForMaskedLM,
NystromformerForMultipleChoice,
NystromformerForQuestionAnswering,
NystromformerForSequenceClassification,
NystromformerForTokenClassification,
NystromformerLayer,
NystromformerModel,
NystromformerPreTrainedModel,
)
from .models.openai import (
OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
OpenAIGPTDoubleHeadsModel,
OpenAIGPTForSequenceClassification,
OpenAIGPTLMHeadModel,
OpenAIGPTModel,
OpenAIGPTPreTrainedModel,
load_tf_weights_in_openai_gpt,
)
from .models.opt import (
OPT_PRETRAINED_MODEL_ARCHIVE_LIST,
OPTForCausalLM,
OPTForQuestionAnswering,
OPTForSequenceClassification,
OPTModel,
OPTPreTrainedModel,
)
from .models.owlvit import (
OWLVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
OwlViTForObjectDetection,
OwlViTModel,
OwlViTPreTrainedModel,
OwlViTTextModel,
OwlViTVisionModel,
)
from .models.pegasus import (
PegasusForCausalLM,
PegasusForConditionalGeneration,
PegasusModel,
PegasusPreTrainedModel,
)
from .models.pegasus_x import (
PEGASUS_X_PRETRAINED_MODEL_ARCHIVE_LIST,
PegasusXForConditionalGeneration,
PegasusXModel,
PegasusXPreTrainedModel,
)
from .models.perceiver import (
PERCEIVER_PRETRAINED_MODEL_ARCHIVE_LIST,
PerceiverForImageClassificationConvProcessing,
PerceiverForImageClassificationFourier,
PerceiverForImageClassificationLearned,
PerceiverForMaskedLM,
PerceiverForMultimodalAutoencoding,
PerceiverForOpticalFlow,
PerceiverForSequenceClassification,
PerceiverLayer,
PerceiverModel,
PerceiverPreTrainedModel,
)
from .models.plbart import (
PLBART_PRETRAINED_MODEL_ARCHIVE_LIST,
PLBartForCausalLM,
PLBartForConditionalGeneration,
PLBartForSequenceClassification,
PLBartModel,
PLBartPreTrainedModel,
)
from .models.poolformer import (
POOLFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
PoolFormerForImageClassification,
PoolFormerModel,
PoolFormerPreTrainedModel,
)
from .models.prophetnet import (
PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST,
ProphetNetDecoder,
ProphetNetEncoder,
ProphetNetForCausalLM,
ProphetNetForConditionalGeneration,
ProphetNetModel,
ProphetNetPreTrainedModel,
)
from .models.qdqbert import (
QDQBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
QDQBertForMaskedLM,
QDQBertForMultipleChoice,
QDQBertForNextSentencePrediction,
QDQBertForQuestionAnswering,
QDQBertForSequenceClassification,
QDQBertForTokenClassification,
QDQBertLayer,
QDQBertLMHeadModel,
QDQBertModel,
QDQBertPreTrainedModel,
load_tf_weights_in_qdqbert,
)
from .models.rag import RagModel, RagPreTrainedModel, RagSequenceForGeneration, RagTokenForGeneration
from .models.realm import (
REALM_PRETRAINED_MODEL_ARCHIVE_LIST,
RealmEmbedder,
RealmForOpenQA,
RealmKnowledgeAugEncoder,
RealmPreTrainedModel,
RealmReader,
RealmRetriever,
RealmScorer,
load_tf_weights_in_realm,
)
from .models.reformer import (
REFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
ReformerAttention,
ReformerForMaskedLM,
ReformerForQuestionAnswering,
ReformerForSequenceClassification,
ReformerLayer,
ReformerModel,
ReformerModelWithLMHead,
ReformerPreTrainedModel,
)
from .models.regnet import (
REGNET_PRETRAINED_MODEL_ARCHIVE_LIST,
RegNetForImageClassification,
RegNetModel,
RegNetPreTrainedModel,
)
from .models.rembert import (
REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
RemBertForCausalLM,
RemBertForMaskedLM,
RemBertForMultipleChoice,
RemBertForQuestionAnswering,
RemBertForSequenceClassification,
RemBertForTokenClassification,
RemBertLayer,
RemBertModel,
RemBertPreTrainedModel,
load_tf_weights_in_rembert,
)
from .models.resnet import (
RESNET_PRETRAINED_MODEL_ARCHIVE_LIST,
ResNetBackbone,
ResNetForImageClassification,
ResNetModel,
ResNetPreTrainedModel,
)
from .models.retribert import RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST, RetriBertModel, RetriBertPreTrainedModel
from .models.roberta import (
ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
RobertaForCausalLM,
RobertaForMaskedLM,
RobertaForMultipleChoice,
RobertaForQuestionAnswering,
RobertaForSequenceClassification,
RobertaForTokenClassification,
RobertaModel,
RobertaPreTrainedModel,
)
from .models.roc_bert import (
ROC_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
RoCBertForCausalLM,
RoCBertForMaskedLM,
RoCBertForMultipleChoice,
RoCBertForPreTraining,
RoCBertForQuestionAnswering,
RoCBertForSequenceClassification,
RoCBertForTokenClassification,
RoCBertLayer,
RoCBertModel,
RoCBertPreTrainedModel,
load_tf_weights_in_roc_bert,
)
from .models.roformer import (
ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
RoFormerForCausalLM,
RoFormerForMaskedLM,
RoFormerForMultipleChoice,
RoFormerForQuestionAnswering,
RoFormerForSequenceClassification,
RoFormerForTokenClassification,
RoFormerLayer,
RoFormerModel,
RoFormerPreTrainedModel,
load_tf_weights_in_roformer,
)
from .models.segformer import (
SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
SegformerDecodeHead,
SegformerForImageClassification,
SegformerForSemanticSegmentation,
SegformerLayer,
SegformerModel,
SegformerPreTrainedModel,
)
from .models.sew import (
SEW_PRETRAINED_MODEL_ARCHIVE_LIST,
SEWForCTC,
SEWForSequenceClassification,
SEWModel,
SEWPreTrainedModel,
)
from .models.sew_d import (
SEW_D_PRETRAINED_MODEL_ARCHIVE_LIST,
SEWDForCTC,
SEWDForSequenceClassification,
SEWDModel,
SEWDPreTrainedModel,
)
from .models.speech_encoder_decoder import SpeechEncoderDecoderModel
from .models.speech_to_text import (
SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
Speech2TextForConditionalGeneration,
Speech2TextModel,
Speech2TextPreTrainedModel,
)
from .models.speech_to_text_2 import Speech2Text2ForCausalLM, Speech2Text2PreTrainedModel
from .models.splinter import (
SPLINTER_PRETRAINED_MODEL_ARCHIVE_LIST,
SplinterForPreTraining,
SplinterForQuestionAnswering,
SplinterLayer,
SplinterModel,
SplinterPreTrainedModel,
)
from .models.squeezebert import (
SQUEEZEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
SqueezeBertForMaskedLM,
SqueezeBertForMultipleChoice,
SqueezeBertForQuestionAnswering,
SqueezeBertForSequenceClassification,
SqueezeBertForTokenClassification,
SqueezeBertModel,
SqueezeBertModule,
SqueezeBertPreTrainedModel,
)
from .models.swin import (
SWIN_PRETRAINED_MODEL_ARCHIVE_LIST,
SwinForImageClassification,
SwinForMaskedImageModeling,
SwinModel,
SwinPreTrainedModel,
)
from .models.swinv2 import (
SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST,
Swinv2ForImageClassification,
Swinv2ForMaskedImageModeling,
Swinv2Model,
Swinv2PreTrainedModel,
)
from .models.switch_transformers import (
SWITCH_TRANSFORMERS_PRETRAINED_MODEL_ARCHIVE_LIST,
SwitchTransformersEncoderModel,
SwitchTransformersForConditionalGeneration,
SwitchTransformersModel,
SwitchTransformersPreTrainedModel,
SwitchTransformersSparseMLP,
SwitchTransformersTop1Router,
)
from .models.t5 import (
T5_PRETRAINED_MODEL_ARCHIVE_LIST,
T5EncoderModel,
T5ForConditionalGeneration,
T5Model,
T5PreTrainedModel,
load_tf_weights_in_t5,
)
from .models.tapas import (
TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TapasForMaskedLM,
TapasForQuestionAnswering,
TapasForSequenceClassification,
TapasModel,
TapasPreTrainedModel,
load_tf_weights_in_tapas,
)
from .models.time_series_transformer import (
TIME_SERIES_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TimeSeriesTransformerForPrediction,
TimeSeriesTransformerModel,
TimeSeriesTransformerPreTrainedModel,
)
from .models.timesformer import (
TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TimesformerForVideoClassification,
TimesformerModel,
TimesformerPreTrainedModel,
)
from .models.trajectory_transformer import (
TRAJECTORY_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TrajectoryTransformerModel,
TrajectoryTransformerPreTrainedModel,
)
from .models.transfo_xl import (
TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST,
AdaptiveEmbedding,
TransfoXLForSequenceClassification,
TransfoXLLMHeadModel,
TransfoXLModel,
TransfoXLPreTrainedModel,
load_tf_weights_in_transfo_xl,
)
from .models.trocr import TROCR_PRETRAINED_MODEL_ARCHIVE_LIST, TrOCRForCausalLM, TrOCRPreTrainedModel
from .models.unispeech import (
UNISPEECH_PRETRAINED_MODEL_ARCHIVE_LIST,
UniSpeechForCTC,
UniSpeechForPreTraining,
UniSpeechForSequenceClassification,
UniSpeechModel,
UniSpeechPreTrainedModel,
)
from .models.unispeech_sat import (
UNISPEECH_SAT_PRETRAINED_MODEL_ARCHIVE_LIST,
UniSpeechSatForAudioFrameClassification,
UniSpeechSatForCTC,
UniSpeechSatForPreTraining,
UniSpeechSatForSequenceClassification,
UniSpeechSatForXVector,
UniSpeechSatModel,
UniSpeechSatPreTrainedModel,
)
from .models.van import (
VAN_PRETRAINED_MODEL_ARCHIVE_LIST,
VanForImageClassification,
VanModel,
VanPreTrainedModel,
)
from .models.videomae import (
VIDEOMAE_PRETRAINED_MODEL_ARCHIVE_LIST,
VideoMAEForPreTraining,
VideoMAEForVideoClassification,
VideoMAEModel,
VideoMAEPreTrainedModel,
)
from .models.vilt import (
VILT_PRETRAINED_MODEL_ARCHIVE_LIST,
ViltForImageAndTextRetrieval,
ViltForImagesAndTextClassification,
ViltForMaskedLM,
ViltForQuestionAnswering,
ViltForTokenClassification,
ViltLayer,
ViltModel,
ViltPreTrainedModel,
)
from .models.vision_encoder_decoder import VisionEncoderDecoderModel
from .models.vision_text_dual_encoder import VisionTextDualEncoderModel
from .models.visual_bert import (
VISUAL_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
VisualBertForMultipleChoice,
VisualBertForPreTraining,
VisualBertForQuestionAnswering,
VisualBertForRegionToPhraseAlignment,
VisualBertForVisualReasoning,
VisualBertLayer,
VisualBertModel,
VisualBertPreTrainedModel,
)
from .models.vit import (
VIT_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTForImageClassification,
ViTForMaskedImageModeling,
ViTModel,
ViTPreTrainedModel,
)
from .models.vit_hybrid import (
VIT_HYBRID_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTHybridForImageClassification,
ViTHybridModel,
ViTHybridPreTrainedModel,
)
from .models.vit_mae import (
VIT_MAE_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTMAEForPreTraining,
ViTMAELayer,
ViTMAEModel,
ViTMAEPreTrainedModel,
)
from .models.vit_msn import (
VIT_MSN_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTMSNForImageClassification,
ViTMSNModel,
ViTMSNPreTrainedModel,
)
from .models.wav2vec2 import (
WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST,
Wav2Vec2ForAudioFrameClassification,
Wav2Vec2ForCTC,
Wav2Vec2ForMaskedLM,
Wav2Vec2ForPreTraining,
Wav2Vec2ForSequenceClassification,
Wav2Vec2ForXVector,
Wav2Vec2Model,
Wav2Vec2PreTrainedModel,
)
from .models.wav2vec2_conformer import (
WAV2VEC2_CONFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
Wav2Vec2ConformerForAudioFrameClassification,
Wav2Vec2ConformerForCTC,
Wav2Vec2ConformerForPreTraining,
Wav2Vec2ConformerForSequenceClassification,
Wav2Vec2ConformerForXVector,
Wav2Vec2ConformerModel,
Wav2Vec2ConformerPreTrainedModel,
)
from .models.wavlm import (
WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST,
WavLMForAudioFrameClassification,
WavLMForCTC,
WavLMForSequenceClassification,
WavLMForXVector,
WavLMModel,
WavLMPreTrainedModel,
)
from .models.whisper import (
WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST,
WhisperForConditionalGeneration,
WhisperModel,
WhisperPreTrainedModel,
)
from .models.x_clip import (
XCLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
XCLIPModel,
XCLIPPreTrainedModel,
XCLIPTextModel,
XCLIPVisionModel,
)
from .models.xglm import XGLM_PRETRAINED_MODEL_ARCHIVE_LIST, XGLMForCausalLM, XGLMModel, XGLMPreTrainedModel
from .models.xlm import (
XLM_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMForMultipleChoice,
XLMForQuestionAnswering,
XLMForQuestionAnsweringSimple,
XLMForSequenceClassification,
XLMForTokenClassification,
XLMModel,
XLMPreTrainedModel,
XLMWithLMHeadModel,
)
from .models.xlm_prophetnet import (
XLM_PROPHETNET_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMProphetNetDecoder,
XLMProphetNetEncoder,
XLMProphetNetForCausalLM,
XLMProphetNetForConditionalGeneration,
XLMProphetNetModel,
XLMProphetNetPreTrainedModel,
)
from .models.xlm_roberta import (
XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMRobertaForCausalLM,
XLMRobertaForMaskedLM,
XLMRobertaForMultipleChoice,
XLMRobertaForQuestionAnswering,
XLMRobertaForSequenceClassification,
XLMRobertaForTokenClassification,
XLMRobertaModel,
XLMRobertaPreTrainedModel,
)
from .models.xlm_roberta_xl import (
XLM_ROBERTA_XL_PRETRAINED_MODEL_ARCHIVE_LIST,
XLMRobertaXLForCausalLM,
XLMRobertaXLForMaskedLM,
XLMRobertaXLForMultipleChoice,
XLMRobertaXLForQuestionAnswering,
XLMRobertaXLForSequenceClassification,
XLMRobertaXLForTokenClassification,
XLMRobertaXLModel,
XLMRobertaXLPreTrainedModel,
)
from .models.xlnet import (
XLNET_PRETRAINED_MODEL_ARCHIVE_LIST,
XLNetForMultipleChoice,
XLNetForQuestionAnswering,
XLNetForQuestionAnsweringSimple,
XLNetForSequenceClassification,
XLNetForTokenClassification,
XLNetLMHeadModel,
XLNetModel,
XLNetPreTrainedModel,
load_tf_weights_in_xlnet,
)
from .models.yolos import (
YOLOS_PRETRAINED_MODEL_ARCHIVE_LIST,
YolosForObjectDetection,
YolosModel,
YolosPreTrainedModel,
)
from .models.yoso import (
YOSO_PRETRAINED_MODEL_ARCHIVE_LIST,
YosoForMaskedLM,
YosoForMultipleChoice,
YosoForQuestionAnswering,
YosoForSequenceClassification,
YosoForTokenClassification,
YosoLayer,
YosoModel,
YosoPreTrainedModel,
)
# Optimization
from .optimization import (
Adafactor,
AdamW,
get_constant_schedule,
get_constant_schedule_with_warmup,
get_cosine_schedule_with_warmup,
get_cosine_with_hard_restarts_schedule_with_warmup,
get_linear_schedule_with_warmup,
get_polynomial_decay_schedule_with_warmup,
get_scheduler,
)
from .pytorch_utils import Conv1D, apply_chunking_to_forward, prune_layer
# Trainer
from .trainer import Trainer
from .trainer_pt_utils import torch_distributed_zero_first
from .trainer_seq2seq import Seq2SeqTrainer
# TensorFlow
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
# Import the same objects as dummies to get them in the namespace.
# They will raise an import error if the user tries to instantiate / use them.
from .utils.dummy_tf_objects import *
else:
from .benchmark.benchmark_args_tf import TensorFlowBenchmarkArguments
# Benchmarks
from .benchmark.benchmark_tf import TensorFlowBenchmark
from .generation import (
TFForcedBOSTokenLogitsProcessor,
TFForcedEOSTokenLogitsProcessor,
TFGenerationMixin,
TFLogitsProcessor,
TFLogitsProcessorList,
TFLogitsWarper,
TFMinLengthLogitsProcessor,
TFNoBadWordsLogitsProcessor,
TFNoRepeatNGramLogitsProcessor,
TFRepetitionPenaltyLogitsProcessor,
TFTemperatureLogitsWarper,
TFTopKLogitsWarper,
TFTopPLogitsWarper,
tf_top_k_top_p_filtering,
)
from .keras_callbacks import KerasMetricCallback, PushToHubCallback
from .modeling_tf_layoutlm import (
TF_LAYOUTLM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLayoutLMForMaskedLM,
TFLayoutLMForQuestionAnswering,
TFLayoutLMForSequenceClassification,
TFLayoutLMForTokenClassification,
TFLayoutLMMainLayer,
TFLayoutLMModel,
TFLayoutLMPreTrainedModel,
)
from .modeling_tf_utils import TFPreTrainedModel, TFSequenceSummary, TFSharedEmbeddings, shape_list
# TensorFlow model imports
from .models.albert import (
TF_ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFAlbertForMaskedLM,
TFAlbertForMultipleChoice,
TFAlbertForPreTraining,
TFAlbertForQuestionAnswering,
TFAlbertForSequenceClassification,
TFAlbertForTokenClassification,
TFAlbertMainLayer,
TFAlbertModel,
TFAlbertPreTrainedModel,
)
from .models.auto import (
TF_MODEL_FOR_CAUSAL_LM_MAPPING,
TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING,
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING,
TF_MODEL_FOR_MASKED_LM_MAPPING,
TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING,
TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING,
TF_MODEL_FOR_PRETRAINING_MAPPING,
TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING,
TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING,
TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING,
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING,
TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING,
TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_VISION_2_SEQ_MAPPING,
TF_MODEL_MAPPING,
TF_MODEL_WITH_LM_HEAD_MAPPING,
TFAutoModel,
TFAutoModelForCausalLM,
TFAutoModelForDocumentQuestionAnswering,
TFAutoModelForImageClassification,
TFAutoModelForMaskedLM,
TFAutoModelForMultipleChoice,
TFAutoModelForNextSentencePrediction,
TFAutoModelForPreTraining,
TFAutoModelForQuestionAnswering,
TFAutoModelForSemanticSegmentation,
TFAutoModelForSeq2SeqLM,
TFAutoModelForSequenceClassification,
TFAutoModelForSpeechSeq2Seq,
TFAutoModelForTableQuestionAnswering,
TFAutoModelForTokenClassification,
TFAutoModelForVision2Seq,
TFAutoModelWithLMHead,
)
from .models.bart import (
TFBartForConditionalGeneration,
TFBartForSequenceClassification,
TFBartModel,
TFBartPretrainedModel,
)
from .models.bert import (
TF_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFBertEmbeddings,
TFBertForMaskedLM,
TFBertForMultipleChoice,
TFBertForNextSentencePrediction,
TFBertForPreTraining,
TFBertForQuestionAnswering,
TFBertForSequenceClassification,
TFBertForTokenClassification,
TFBertLMHeadModel,
TFBertMainLayer,
TFBertModel,
TFBertPreTrainedModel,
)
from .models.blenderbot import (
TFBlenderbotForConditionalGeneration,
TFBlenderbotModel,
TFBlenderbotPreTrainedModel,
)
from .models.blenderbot_small import (
TFBlenderbotSmallForConditionalGeneration,
TFBlenderbotSmallModel,
TFBlenderbotSmallPreTrainedModel,
)
from .models.camembert import (
TF_CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCamembertForCausalLM,
TFCamembertForMaskedLM,
TFCamembertForMultipleChoice,
TFCamembertForQuestionAnswering,
TFCamembertForSequenceClassification,
TFCamembertForTokenClassification,
TFCamembertModel,
TFCamembertPreTrainedModel,
)
from .models.clip import (
TF_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCLIPModel,
TFCLIPPreTrainedModel,
TFCLIPTextModel,
TFCLIPVisionModel,
)
from .models.convbert import (
TF_CONVBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFConvBertForMaskedLM,
TFConvBertForMultipleChoice,
TFConvBertForQuestionAnswering,
TFConvBertForSequenceClassification,
TFConvBertForTokenClassification,
TFConvBertLayer,
TFConvBertModel,
TFConvBertPreTrainedModel,
)
from .models.convnext import TFConvNextForImageClassification, TFConvNextModel, TFConvNextPreTrainedModel
from .models.ctrl import (
TF_CTRL_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCTRLForSequenceClassification,
TFCTRLLMHeadModel,
TFCTRLModel,
TFCTRLPreTrainedModel,
)
from .models.cvt import (
TF_CVT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCvtForImageClassification,
TFCvtModel,
TFCvtPreTrainedModel,
)
from .models.data2vec import (
TFData2VecVisionForImageClassification,
TFData2VecVisionForSemanticSegmentation,
TFData2VecVisionModel,
TFData2VecVisionPreTrainedModel,
)
from .models.deberta import (
TF_DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDebertaForMaskedLM,
TFDebertaForQuestionAnswering,
TFDebertaForSequenceClassification,
TFDebertaForTokenClassification,
TFDebertaModel,
TFDebertaPreTrainedModel,
)
from .models.deberta_v2 import (
TF_DEBERTA_V2_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDebertaV2ForMaskedLM,
TFDebertaV2ForQuestionAnswering,
TFDebertaV2ForSequenceClassification,
TFDebertaV2ForTokenClassification,
TFDebertaV2Model,
TFDebertaV2PreTrainedModel,
)
from .models.deit import (
TF_DEIT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDeiTForImageClassification,
TFDeiTForImageClassificationWithTeacher,
TFDeiTForMaskedImageModeling,
TFDeiTModel,
TFDeiTPreTrainedModel,
)
from .models.distilbert import (
TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDistilBertForMaskedLM,
TFDistilBertForMultipleChoice,
TFDistilBertForQuestionAnswering,
TFDistilBertForSequenceClassification,
TFDistilBertForTokenClassification,
TFDistilBertMainLayer,
TFDistilBertModel,
TFDistilBertPreTrainedModel,
)
from .models.dpr import (
TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDPRContextEncoder,
TFDPRPretrainedContextEncoder,
TFDPRPretrainedQuestionEncoder,
TFDPRPretrainedReader,
TFDPRQuestionEncoder,
TFDPRReader,
)
from .models.electra import (
TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFElectraForMaskedLM,
TFElectraForMultipleChoice,
TFElectraForPreTraining,
TFElectraForQuestionAnswering,
TFElectraForSequenceClassification,
TFElectraForTokenClassification,
TFElectraModel,
TFElectraPreTrainedModel,
)
from .models.encoder_decoder import TFEncoderDecoderModel
from .models.esm import (
ESM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFEsmForMaskedLM,
TFEsmForSequenceClassification,
TFEsmForTokenClassification,
TFEsmModel,
TFEsmPreTrainedModel,
)
from .models.flaubert import (
TF_FLAUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFFlaubertForMultipleChoice,
TFFlaubertForQuestionAnsweringSimple,
TFFlaubertForSequenceClassification,
TFFlaubertForTokenClassification,
TFFlaubertModel,
TFFlaubertPreTrainedModel,
TFFlaubertWithLMHeadModel,
)
from .models.funnel import (
TF_FUNNEL_PRETRAINED_MODEL_ARCHIVE_LIST,
TFFunnelBaseModel,
TFFunnelForMaskedLM,
TFFunnelForMultipleChoice,
TFFunnelForPreTraining,
TFFunnelForQuestionAnswering,
TFFunnelForSequenceClassification,
TFFunnelForTokenClassification,
TFFunnelModel,
TFFunnelPreTrainedModel,
)
from .models.gpt2 import (
TF_GPT2_PRETRAINED_MODEL_ARCHIVE_LIST,
TFGPT2DoubleHeadsModel,
TFGPT2ForSequenceClassification,
TFGPT2LMHeadModel,
TFGPT2MainLayer,
TFGPT2Model,
TFGPT2PreTrainedModel,
)
from .models.gptj import (
TFGPTJForCausalLM,
TFGPTJForQuestionAnswering,
TFGPTJForSequenceClassification,
TFGPTJModel,
TFGPTJPreTrainedModel,
)
from .models.groupvit import (
TF_GROUPVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFGroupViTModel,
TFGroupViTPreTrainedModel,
TFGroupViTTextModel,
TFGroupViTVisionModel,
)
from .models.hubert import (
TF_HUBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFHubertForCTC,
TFHubertModel,
TFHubertPreTrainedModel,
)
from .models.layoutlmv3 import (
TF_LAYOUTLMV3_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLayoutLMv3ForQuestionAnswering,
TFLayoutLMv3ForSequenceClassification,
TFLayoutLMv3ForTokenClassification,
TFLayoutLMv3Model,
TFLayoutLMv3PreTrainedModel,
)
from .models.led import TFLEDForConditionalGeneration, TFLEDModel, TFLEDPreTrainedModel
from .models.longformer import (
TF_LONGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLongformerForMaskedLM,
TFLongformerForMultipleChoice,
TFLongformerForQuestionAnswering,
TFLongformerForSequenceClassification,
TFLongformerForTokenClassification,
TFLongformerModel,
TFLongformerPreTrainedModel,
TFLongformerSelfAttention,
)
from .models.lxmert import (
TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLxmertForPreTraining,
TFLxmertMainLayer,
TFLxmertModel,
TFLxmertPreTrainedModel,
TFLxmertVisualFeatureEncoder,
)
from .models.marian import TFMarianModel, TFMarianMTModel, TFMarianPreTrainedModel
from .models.mbart import TFMBartForConditionalGeneration, TFMBartModel, TFMBartPreTrainedModel
from .models.mobilebert import (
TF_MOBILEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFMobileBertForMaskedLM,
TFMobileBertForMultipleChoice,
TFMobileBertForNextSentencePrediction,
TFMobileBertForPreTraining,
TFMobileBertForQuestionAnswering,
TFMobileBertForSequenceClassification,
TFMobileBertForTokenClassification,
TFMobileBertMainLayer,
TFMobileBertModel,
TFMobileBertPreTrainedModel,
TFMobileViTForImageClassification,
TFMobileViTForSemanticSegmentation,
TFMobileViTModel,
TFMobileViTPreTrainedModel,
)
from .models.mpnet import (
TF_MPNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFMPNetForMaskedLM,
TFMPNetForMultipleChoice,
TFMPNetForQuestionAnswering,
TFMPNetForSequenceClassification,
TFMPNetForTokenClassification,
TFMPNetMainLayer,
TFMPNetModel,
TFMPNetPreTrainedModel,
)
from .models.mt5 import TFMT5EncoderModel, TFMT5ForConditionalGeneration, TFMT5Model
from .models.openai import (
TF_OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFOpenAIGPTDoubleHeadsModel,
TFOpenAIGPTForSequenceClassification,
TFOpenAIGPTLMHeadModel,
TFOpenAIGPTMainLayer,
TFOpenAIGPTModel,
TFOpenAIGPTPreTrainedModel,
)
from .models.opt import TFOPTForCausalLM, TFOPTModel, TFOPTPreTrainedModel
from .models.pegasus import TFPegasusForConditionalGeneration, TFPegasusModel, TFPegasusPreTrainedModel
from .models.rag import TFRagModel, TFRagPreTrainedModel, TFRagSequenceForGeneration, TFRagTokenForGeneration
from .models.regnet import (
TF_REGNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRegNetForImageClassification,
TFRegNetModel,
TFRegNetPreTrainedModel,
)
from .models.rembert import (
TF_REMBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRemBertForCausalLM,
TFRemBertForMaskedLM,
TFRemBertForMultipleChoice,
TFRemBertForQuestionAnswering,
TFRemBertForSequenceClassification,
TFRemBertForTokenClassification,
TFRemBertLayer,
TFRemBertModel,
TFRemBertPreTrainedModel,
)
from .models.resnet import (
TF_RESNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFResNetForImageClassification,
TFResNetModel,
TFResNetPreTrainedModel,
)
from .models.roberta import (
TF_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRobertaForCausalLM,
TFRobertaForMaskedLM,
TFRobertaForMultipleChoice,
TFRobertaForQuestionAnswering,
TFRobertaForSequenceClassification,
TFRobertaForTokenClassification,
TFRobertaMainLayer,
TFRobertaModel,
TFRobertaPreTrainedModel,
)
from .models.roformer import (
TF_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRoFormerForCausalLM,
TFRoFormerForMaskedLM,
TFRoFormerForMultipleChoice,
TFRoFormerForQuestionAnswering,
TFRoFormerForSequenceClassification,
TFRoFormerForTokenClassification,
TFRoFormerLayer,
TFRoFormerModel,
TFRoFormerPreTrainedModel,
)
from .models.segformer import (
TF_SEGFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFSegformerDecodeHead,
TFSegformerForImageClassification,
TFSegformerForSemanticSegmentation,
TFSegformerModel,
TFSegformerPreTrainedModel,
)
from .models.speech_to_text import (
TF_SPEECH_TO_TEXT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFSpeech2TextForConditionalGeneration,
TFSpeech2TextModel,
TFSpeech2TextPreTrainedModel,
)
from .models.swin import (
TF_SWIN_PRETRAINED_MODEL_ARCHIVE_LIST,
TFSwinForImageClassification,
TFSwinForMaskedImageModeling,
TFSwinModel,
TFSwinPreTrainedModel,
)
from .models.t5 import (
TF_T5_PRETRAINED_MODEL_ARCHIVE_LIST,
TFT5EncoderModel,
TFT5ForConditionalGeneration,
TFT5Model,
TFT5PreTrainedModel,
)
from .models.tapas import (
TF_TAPAS_PRETRAINED_MODEL_ARCHIVE_LIST,
TFTapasForMaskedLM,
TFTapasForQuestionAnswering,
TFTapasForSequenceClassification,
TFTapasModel,
TFTapasPreTrainedModel,
)
from .models.transfo_xl import (
TF_TRANSFO_XL_PRETRAINED_MODEL_ARCHIVE_LIST,
TFAdaptiveEmbedding,
TFTransfoXLForSequenceClassification,
TFTransfoXLLMHeadModel,
TFTransfoXLMainLayer,
TFTransfoXLModel,
TFTransfoXLPreTrainedModel,
)
from .models.vision_encoder_decoder import TFVisionEncoderDecoderModel
from .models.vit import TFViTForImageClassification, TFViTModel, TFViTPreTrainedModel
from .models.vit_mae import TFViTMAEForPreTraining, TFViTMAEModel, TFViTMAEPreTrainedModel
from .models.wav2vec2 import (
TF_WAV_2_VEC_2_PRETRAINED_MODEL_ARCHIVE_LIST,
TFWav2Vec2ForCTC,
TFWav2Vec2Model,
TFWav2Vec2PreTrainedModel,
)
from .models.whisper import (
TF_WHISPER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFWhisperForConditionalGeneration,
TFWhisperModel,
TFWhisperPreTrainedModel,
)
from .models.xglm import (
TF_XGLM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXGLMForCausalLM,
TFXGLMModel,
TFXGLMPreTrainedModel,
)
from .models.xlm import (
TF_XLM_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXLMForMultipleChoice,
TFXLMForQuestionAnsweringSimple,
TFXLMForSequenceClassification,
TFXLMForTokenClassification,
TFXLMMainLayer,
TFXLMModel,
TFXLMPreTrainedModel,
TFXLMWithLMHeadModel,
)
from .models.xlm_roberta import (
TF_XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXLMRobertaForMaskedLM,
TFXLMRobertaForMultipleChoice,
TFXLMRobertaForQuestionAnswering,
TFXLMRobertaForSequenceClassification,
TFXLMRobertaForTokenClassification,
TFXLMRobertaModel,
)
from .models.xlnet import (
TF_XLNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFXLNetForMultipleChoice,
TFXLNetForQuestionAnsweringSimple,
TFXLNetForSequenceClassification,
TFXLNetForTokenClassification,
TFXLNetLMHeadModel,
TFXLNetMainLayer,
TFXLNetModel,
TFXLNetPreTrainedModel,
)
# Optimization
from .optimization_tf import AdamWeightDecay, GradientAccumulator, WarmUp, create_optimizer
# Trainer
from .trainer_tf import TFTrainer
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
# Import the same objects as dummies to get them in the namespace.
# They will raise an import error if the user tries to instantiate / use them.
from .utils.dummy_flax_objects import *
else:
from .generation import (
FlaxForcedBOSTokenLogitsProcessor,
FlaxForcedEOSTokenLogitsProcessor,
FlaxGenerationMixin,
FlaxLogitsProcessor,
FlaxLogitsProcessorList,
FlaxLogitsWarper,
FlaxMinLengthLogitsProcessor,
FlaxTemperatureLogitsWarper,
FlaxTopKLogitsWarper,
FlaxTopPLogitsWarper,
)
from .modeling_flax_utils import FlaxPreTrainedModel
# Flax model imports
from .models.albert import (
FlaxAlbertForMaskedLM,
FlaxAlbertForMultipleChoice,
FlaxAlbertForPreTraining,
FlaxAlbertForQuestionAnswering,
FlaxAlbertForSequenceClassification,
FlaxAlbertForTokenClassification,
FlaxAlbertModel,
FlaxAlbertPreTrainedModel,
)
from .models.auto import (
FLAX_MODEL_FOR_CAUSAL_LM_MAPPING,
FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
FLAX_MODEL_FOR_MASKED_LM_MAPPING,
FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING,
FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING,
FLAX_MODEL_FOR_PRETRAINING_MAPPING,
FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING,
FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING,
FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING,
FLAX_MODEL_MAPPING,
FlaxAutoModel,
FlaxAutoModelForCausalLM,
FlaxAutoModelForImageClassification,
FlaxAutoModelForMaskedLM,
FlaxAutoModelForMultipleChoice,
FlaxAutoModelForNextSentencePrediction,
FlaxAutoModelForPreTraining,
FlaxAutoModelForQuestionAnswering,
FlaxAutoModelForSeq2SeqLM,
FlaxAutoModelForSequenceClassification,
FlaxAutoModelForTokenClassification,
FlaxAutoModelForVision2Seq,
)
from .models.bart import (
FlaxBartDecoderPreTrainedModel,
FlaxBartForCausalLM,
FlaxBartForConditionalGeneration,
FlaxBartForQuestionAnswering,
FlaxBartForSequenceClassification,
FlaxBartModel,
FlaxBartPreTrainedModel,
)
from .models.beit import (
FlaxBeitForImageClassification,
FlaxBeitForMaskedImageModeling,
FlaxBeitModel,
FlaxBeitPreTrainedModel,
)
from .models.bert import (
FlaxBertForCausalLM,
FlaxBertForMaskedLM,
FlaxBertForMultipleChoice,
FlaxBertForNextSentencePrediction,
FlaxBertForPreTraining,
FlaxBertForQuestionAnswering,
FlaxBertForSequenceClassification,
FlaxBertForTokenClassification,
FlaxBertModel,
FlaxBertPreTrainedModel,
)
from .models.big_bird import (
FlaxBigBirdForCausalLM,
FlaxBigBirdForMaskedLM,
FlaxBigBirdForMultipleChoice,
FlaxBigBirdForPreTraining,
FlaxBigBirdForQuestionAnswering,
FlaxBigBirdForSequenceClassification,
FlaxBigBirdForTokenClassification,
FlaxBigBirdModel,
FlaxBigBirdPreTrainedModel,
)
from .models.blenderbot import (
FlaxBlenderbotForConditionalGeneration,
FlaxBlenderbotModel,
FlaxBlenderbotPreTrainedModel,
)
from .models.blenderbot_small import (
FlaxBlenderbotSmallForConditionalGeneration,
FlaxBlenderbotSmallModel,
FlaxBlenderbotSmallPreTrainedModel,
)
from .models.clip import (
FlaxCLIPModel,
FlaxCLIPPreTrainedModel,
FlaxCLIPTextModel,
FlaxCLIPTextPreTrainedModel,
FlaxCLIPVisionModel,
FlaxCLIPVisionPreTrainedModel,
)
from .models.distilbert import (
FlaxDistilBertForMaskedLM,
FlaxDistilBertForMultipleChoice,
FlaxDistilBertForQuestionAnswering,
FlaxDistilBertForSequenceClassification,
FlaxDistilBertForTokenClassification,
FlaxDistilBertModel,
FlaxDistilBertPreTrainedModel,
)
from .models.electra import (
FlaxElectraForCausalLM,
FlaxElectraForMaskedLM,
FlaxElectraForMultipleChoice,
FlaxElectraForPreTraining,
FlaxElectraForQuestionAnswering,
FlaxElectraForSequenceClassification,
FlaxElectraForTokenClassification,
FlaxElectraModel,
FlaxElectraPreTrainedModel,
)
from .models.encoder_decoder import FlaxEncoderDecoderModel
from .models.gpt2 import FlaxGPT2LMHeadModel, FlaxGPT2Model, FlaxGPT2PreTrainedModel
from .models.gpt_neo import FlaxGPTNeoForCausalLM, FlaxGPTNeoModel, FlaxGPTNeoPreTrainedModel
from .models.gptj import FlaxGPTJForCausalLM, FlaxGPTJModel, FlaxGPTJPreTrainedModel
from .models.longt5 import FlaxLongT5ForConditionalGeneration, FlaxLongT5Model, FlaxLongT5PreTrainedModel
from .models.marian import FlaxMarianModel, FlaxMarianMTModel, FlaxMarianPreTrainedModel
from .models.mbart import (
FlaxMBartForConditionalGeneration,
FlaxMBartForQuestionAnswering,
FlaxMBartForSequenceClassification,
FlaxMBartModel,
FlaxMBartPreTrainedModel,
)
from .models.mt5 import FlaxMT5EncoderModel, FlaxMT5ForConditionalGeneration, FlaxMT5Model
from .models.opt import FlaxOPTForCausalLM, FlaxOPTModel, FlaxOPTPreTrainedModel
from .models.pegasus import FlaxPegasusForConditionalGeneration, FlaxPegasusModel, FlaxPegasusPreTrainedModel
from .models.roberta import (
FlaxRobertaForCausalLM,
FlaxRobertaForMaskedLM,
FlaxRobertaForMultipleChoice,
FlaxRobertaForQuestionAnswering,
FlaxRobertaForSequenceClassification,
FlaxRobertaForTokenClassification,
FlaxRobertaModel,
FlaxRobertaPreTrainedModel,
)
from .models.roformer import (
FlaxRoFormerForMaskedLM,
FlaxRoFormerForMultipleChoice,
FlaxRoFormerForQuestionAnswering,
FlaxRoFormerForSequenceClassification,
FlaxRoFormerForTokenClassification,
FlaxRoFormerModel,
FlaxRoFormerPreTrainedModel,
)
from .models.speech_encoder_decoder import FlaxSpeechEncoderDecoderModel
from .models.t5 import FlaxT5EncoderModel, FlaxT5ForConditionalGeneration, FlaxT5Model, FlaxT5PreTrainedModel
from .models.vision_encoder_decoder import FlaxVisionEncoderDecoderModel
from .models.vision_text_dual_encoder import FlaxVisionTextDualEncoderModel
from .models.vit import FlaxViTForImageClassification, FlaxViTModel, FlaxViTPreTrainedModel
from .models.wav2vec2 import (
FlaxWav2Vec2ForCTC,
FlaxWav2Vec2ForPreTraining,
FlaxWav2Vec2Model,
FlaxWav2Vec2PreTrainedModel,
)
from .models.xglm import FlaxXGLMForCausalLM, FlaxXGLMModel, FlaxXGLMPreTrainedModel
from .models.xlm_roberta import (
FlaxXLMRobertaForMaskedLM,
FlaxXLMRobertaForMultipleChoice,
FlaxXLMRobertaForQuestionAnswering,
FlaxXLMRobertaForSequenceClassification,
FlaxXLMRobertaForTokenClassification,
FlaxXLMRobertaModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(
__name__,
globals()["__file__"],
_import_structure,
module_spec=__spec__,
extra_objects={"__version__": __version__},
)
if not is_tf_available() and not is_torch_available() and not is_flax_available():
logger.warning(
"None of PyTorch, TensorFlow >= 2.0, or Flax have been found. "
"Models won't be available and only tokenizers, configuration "
"and file/data utilities can be used."
)
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/__init__.py | # flake8: noqa
# There's no way to ignore "F401 '...' imported but unused" warnings in this
# module, but to preserve other warnings. So, don't check this module at all.
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from . import (
albert,
audio_spectrogram_transformer,
auto,
bart,
barthez,
bartpho,
beit,
bert,
bert_generation,
bert_japanese,
bertweet,
big_bird,
bigbird_pegasus,
biogpt,
bit,
blenderbot,
blenderbot_small,
bloom,
bort,
byt5,
camembert,
canine,
chinese_clip,
clip,
clipseg,
codegen,
conditional_detr,
convbert,
convnext,
cpm,
ctrl,
cvt,
data2vec,
deberta,
deberta_v2,
decision_transformer,
deformable_detr,
deit,
detr,
dialogpt,
dinat,
distilbert,
dit,
donut,
dpr,
dpt,
electra,
encoder_decoder,
ernie,
esm,
flaubert,
flava,
fnet,
fsmt,
funnel,
glpn,
gpt2,
gpt_neo,
gpt_neox,
gpt_neox_japanese,
gptj,
groupvit,
herbert,
hubert,
ibert,
imagegpt,
jukebox,
layoutlm,
layoutlmv2,
layoutlmv3,
layoutxlm,
led,
levit,
lilt,
longformer,
longt5,
luke,
lxmert,
m2m_100,
marian,
markuplm,
maskformer,
mbart,
mbart50,
mctct,
megatron_bert,
megatron_gpt2,
mluke,
mmbt,
mobilebert,
mobilenet_v1,
mobilenet_v2,
mobilevit,
mpnet,
mt5,
mvp,
nat,
nezha,
nllb,
nystromformer,
openai,
opt,
owlvit,
pegasus,
pegasus_x,
perceiver,
phobert,
plbart,
poolformer,
prophetnet,
qdqbert,
rag,
realm,
reformer,
regnet,
rembert,
resnet,
retribert,
roberta,
roc_bert,
roformer,
segformer,
sew,
sew_d,
speech_encoder_decoder,
speech_to_text,
speech_to_text_2,
splinter,
squeezebert,
swin,
swinv2,
switch_transformers,
t5,
table_transformer,
tapas,
tapex,
time_series_transformer,
timesformer,
trajectory_transformer,
transfo_xl,
trocr,
unispeech,
unispeech_sat,
van,
videomae,
vilt,
vision_encoder_decoder,
vision_text_dual_encoder,
visual_bert,
vit,
vit_hybrid,
vit_mae,
vit_msn,
wav2vec2,
wav2vec2_conformer,
wav2vec2_phoneme,
wav2vec2_with_lm,
wavlm,
whisper,
x_clip,
xglm,
xlm,
xlm_prophetnet,
xlm_roberta,
xlm_roberta_xl,
xlnet,
yolos,
yoso,
)
| # flake8: noqa
# There's no way to ignore "F401 '...' imported but unused" warnings in this
# module, but to preserve other warnings. So, don't check this module at all.
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from . import (
albert,
audio_spectrogram_transformer,
auto,
bart,
barthez,
bartpho,
beit,
bert,
bert_generation,
bert_japanese,
bertweet,
big_bird,
bigbird_pegasus,
biogpt,
bit,
blenderbot,
blenderbot_small,
bloom,
bort,
byt5,
camembert,
canine,
chinese_clip,
clip,
clipseg,
codegen,
conditional_detr,
convbert,
convnext,
cpm,
ctrl,
cvt,
data2vec,
deberta,
deberta_v2,
decision_transformer,
deformable_detr,
deit,
detr,
dialogpt,
dinat,
distilbert,
dit,
donut,
dpr,
dpt,
electra,
encoder_decoder,
ernie,
esm,
flaubert,
flava,
fnet,
fsmt,
funnel,
glpn,
gpt2,
gpt_neo,
gpt_neox,
gpt_neox_japanese,
gpt_sw3,
gptj,
groupvit,
herbert,
hubert,
ibert,
imagegpt,
jukebox,
layoutlm,
layoutlmv2,
layoutlmv3,
layoutxlm,
led,
levit,
lilt,
longformer,
longt5,
luke,
lxmert,
m2m_100,
marian,
markuplm,
maskformer,
mbart,
mbart50,
mctct,
megatron_bert,
megatron_gpt2,
mluke,
mmbt,
mobilebert,
mobilenet_v1,
mobilenet_v2,
mobilevit,
mpnet,
mt5,
mvp,
nat,
nezha,
nllb,
nystromformer,
openai,
opt,
owlvit,
pegasus,
pegasus_x,
perceiver,
phobert,
plbart,
poolformer,
prophetnet,
qdqbert,
rag,
realm,
reformer,
regnet,
rembert,
resnet,
retribert,
roberta,
roc_bert,
roformer,
segformer,
sew,
sew_d,
speech_encoder_decoder,
speech_to_text,
speech_to_text_2,
splinter,
squeezebert,
swin,
swinv2,
switch_transformers,
t5,
table_transformer,
tapas,
tapex,
time_series_transformer,
timesformer,
trajectory_transformer,
transfo_xl,
trocr,
unispeech,
unispeech_sat,
van,
videomae,
vilt,
vision_encoder_decoder,
vision_text_dual_encoder,
visual_bert,
vit,
vit_hybrid,
vit_mae,
vit_msn,
wav2vec2,
wav2vec2_conformer,
wav2vec2_phoneme,
wav2vec2_with_lm,
wavlm,
whisper,
x_clip,
xglm,
xlm,
xlm_prophetnet,
xlm_roberta,
xlm_roberta_xl,
xlnet,
yolos,
yoso,
)
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/auto/configuration_auto.py | # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Config class."""
import importlib
import re
import warnings
from collections import OrderedDict
from typing import List, Union
from ...configuration_utils import PretrainedConfig
from ...dynamic_module_utils import get_class_from_dynamic_module
from ...utils import CONFIG_NAME, logging
logger = logging.get_logger(__name__)
CONFIG_MAPPING_NAMES = OrderedDict(
[
# Add configs here
("albert", "AlbertConfig"),
("audio-spectrogram-transformer", "ASTConfig"),
("bart", "BartConfig"),
("beit", "BeitConfig"),
("bert", "BertConfig"),
("bert-generation", "BertGenerationConfig"),
("big_bird", "BigBirdConfig"),
("bigbird_pegasus", "BigBirdPegasusConfig"),
("biogpt", "BioGptConfig"),
("bit", "BitConfig"),
("blenderbot", "BlenderbotConfig"),
("blenderbot-small", "BlenderbotSmallConfig"),
("bloom", "BloomConfig"),
("camembert", "CamembertConfig"),
("canine", "CanineConfig"),
("chinese_clip", "ChineseCLIPConfig"),
("clip", "CLIPConfig"),
("clipseg", "CLIPSegConfig"),
("codegen", "CodeGenConfig"),
("conditional_detr", "ConditionalDetrConfig"),
("convbert", "ConvBertConfig"),
("convnext", "ConvNextConfig"),
("ctrl", "CTRLConfig"),
("cvt", "CvtConfig"),
("data2vec-audio", "Data2VecAudioConfig"),
("data2vec-text", "Data2VecTextConfig"),
("data2vec-vision", "Data2VecVisionConfig"),
("deberta", "DebertaConfig"),
("deberta-v2", "DebertaV2Config"),
("decision_transformer", "DecisionTransformerConfig"),
("deformable_detr", "DeformableDetrConfig"),
("deit", "DeiTConfig"),
("detr", "DetrConfig"),
("dinat", "DinatConfig"),
("distilbert", "DistilBertConfig"),
("donut-swin", "DonutSwinConfig"),
("dpr", "DPRConfig"),
("dpt", "DPTConfig"),
("electra", "ElectraConfig"),
("encoder-decoder", "EncoderDecoderConfig"),
("ernie", "ErnieConfig"),
("esm", "EsmConfig"),
("flaubert", "FlaubertConfig"),
("flava", "FlavaConfig"),
("fnet", "FNetConfig"),
("fsmt", "FSMTConfig"),
("funnel", "FunnelConfig"),
("glpn", "GLPNConfig"),
("gpt2", "GPT2Config"),
("gpt_neo", "GPTNeoConfig"),
("gpt_neox", "GPTNeoXConfig"),
("gpt_neox_japanese", "GPTNeoXJapaneseConfig"),
("gptj", "GPTJConfig"),
("groupvit", "GroupViTConfig"),
("hubert", "HubertConfig"),
("ibert", "IBertConfig"),
("imagegpt", "ImageGPTConfig"),
("jukebox", "JukeboxConfig"),
("layoutlm", "LayoutLMConfig"),
("layoutlmv2", "LayoutLMv2Config"),
("layoutlmv3", "LayoutLMv3Config"),
("led", "LEDConfig"),
("levit", "LevitConfig"),
("lilt", "LiltConfig"),
("longformer", "LongformerConfig"),
("longt5", "LongT5Config"),
("luke", "LukeConfig"),
("lxmert", "LxmertConfig"),
("m2m_100", "M2M100Config"),
("marian", "MarianConfig"),
("markuplm", "MarkupLMConfig"),
("maskformer", "MaskFormerConfig"),
("maskformer-swin", "MaskFormerSwinConfig"),
("mbart", "MBartConfig"),
("mctct", "MCTCTConfig"),
("megatron-bert", "MegatronBertConfig"),
("mobilebert", "MobileBertConfig"),
("mobilenet_v1", "MobileNetV1Config"),
("mobilenet_v2", "MobileNetV2Config"),
("mobilevit", "MobileViTConfig"),
("mpnet", "MPNetConfig"),
("mt5", "MT5Config"),
("mvp", "MvpConfig"),
("nat", "NatConfig"),
("nezha", "NezhaConfig"),
("nystromformer", "NystromformerConfig"),
("openai-gpt", "OpenAIGPTConfig"),
("opt", "OPTConfig"),
("owlvit", "OwlViTConfig"),
("pegasus", "PegasusConfig"),
("pegasus_x", "PegasusXConfig"),
("perceiver", "PerceiverConfig"),
("plbart", "PLBartConfig"),
("poolformer", "PoolFormerConfig"),
("prophetnet", "ProphetNetConfig"),
("qdqbert", "QDQBertConfig"),
("rag", "RagConfig"),
("realm", "RealmConfig"),
("reformer", "ReformerConfig"),
("regnet", "RegNetConfig"),
("rembert", "RemBertConfig"),
("resnet", "ResNetConfig"),
("retribert", "RetriBertConfig"),
("roberta", "RobertaConfig"),
("roc_bert", "RoCBertConfig"),
("roformer", "RoFormerConfig"),
("segformer", "SegformerConfig"),
("sew", "SEWConfig"),
("sew-d", "SEWDConfig"),
("speech-encoder-decoder", "SpeechEncoderDecoderConfig"),
("speech_to_text", "Speech2TextConfig"),
("speech_to_text_2", "Speech2Text2Config"),
("splinter", "SplinterConfig"),
("squeezebert", "SqueezeBertConfig"),
("swin", "SwinConfig"),
("swinv2", "Swinv2Config"),
("switch_transformers", "SwitchTransformersConfig"),
("t5", "T5Config"),
("table-transformer", "TableTransformerConfig"),
("tapas", "TapasConfig"),
("time_series_transformer", "TimeSeriesTransformerConfig"),
("timesformer", "TimesformerConfig"),
("trajectory_transformer", "TrajectoryTransformerConfig"),
("transfo-xl", "TransfoXLConfig"),
("trocr", "TrOCRConfig"),
("unispeech", "UniSpeechConfig"),
("unispeech-sat", "UniSpeechSatConfig"),
("van", "VanConfig"),
("videomae", "VideoMAEConfig"),
("vilt", "ViltConfig"),
("vision-encoder-decoder", "VisionEncoderDecoderConfig"),
("vision-text-dual-encoder", "VisionTextDualEncoderConfig"),
("visual_bert", "VisualBertConfig"),
("vit", "ViTConfig"),
("vit_hybrid", "ViTHybridConfig"),
("vit_mae", "ViTMAEConfig"),
("vit_msn", "ViTMSNConfig"),
("wav2vec2", "Wav2Vec2Config"),
("wav2vec2-conformer", "Wav2Vec2ConformerConfig"),
("wavlm", "WavLMConfig"),
("whisper", "WhisperConfig"),
("xclip", "XCLIPConfig"),
("xglm", "XGLMConfig"),
("xlm", "XLMConfig"),
("xlm-prophetnet", "XLMProphetNetConfig"),
("xlm-roberta", "XLMRobertaConfig"),
("xlm-roberta-xl", "XLMRobertaXLConfig"),
("xlnet", "XLNetConfig"),
("yolos", "YolosConfig"),
("yoso", "YosoConfig"),
]
)
CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
[
# Add archive maps here)
("albert", "ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("audio-spectrogram-transformer", "AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bart", "BART_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("beit", "BEIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bert", "BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("big_bird", "BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bigbird_pegasus", "BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("biogpt", "BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bit", "BIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("blenderbot", "BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("blenderbot-small", "BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bloom", "BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("camembert", "CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("canine", "CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("chinese_clip", "CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("clip", "CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("clipseg", "CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("codegen", "CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("conditional_detr", "CONDITIONAL_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("convbert", "CONVBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("convnext", "CONVNEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("ctrl", "CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("cvt", "CVT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("data2vec-audio", "DATA2VEC_AUDIO_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("data2vec-text", "DATA2VEC_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("data2vec-vision", "DATA2VEC_VISION_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deberta", "DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deberta-v2", "DEBERTA_V2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deformable_detr", "DEFORMABLE_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deit", "DEIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("detr", "DETR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("dinat", "DINAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("distilbert", "DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("donut-swin", "DONUT_SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("dpr", "DPR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("dpt", "DPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("electra", "ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("ernie", "ERNIE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("esm", "ESM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("flaubert", "FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("flava", "FLAVA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("fnet", "FNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("fsmt", "FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("funnel", "FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("glpn", "GLPN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt2", "GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt_neo", "GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt_neox", "GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt_neox_japanese", "GPT_NEOX_JAPANESE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gptj", "GPTJ_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("groupvit", "GROUPVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("hubert", "HUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("ibert", "IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("imagegpt", "IMAGEGPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("jukebox", "JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("layoutlm", "LAYOUTLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("layoutlmv2", "LAYOUTLMV2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("layoutlmv3", "LAYOUTLMV3_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("led", "LED_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("levit", "LEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("lilt", "LILT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("longformer", "LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("longt5", "LONGT5_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("luke", "LUKE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("lxmert", "LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("m2m_100", "M2M_100_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("markuplm", "MARKUPLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("maskformer", "MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mbart", "MBART_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mctct", "MCTCT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("megatron-bert", "MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mobilenet_v1", "MOBILENET_V1_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mobilenet_v2", "MOBILENET_V2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mobilevit", "MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mpnet", "MPNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mvp", "MVP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("nat", "NAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("nezha", "NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("nystromformer", "NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("openai-gpt", "OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("opt", "OPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("owlvit", "OWLVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("pegasus", "PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("pegasus_x", "PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("perceiver", "PERCEIVER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("plbart", "PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("poolformer", "POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("prophetnet", "PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("qdqbert", "QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("realm", "REALM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("regnet", "REGNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("rembert", "REMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("resnet", "RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("retribert", "RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("roberta", "ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("roc_bert", "ROC_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("roformer", "ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("segformer", "SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("sew", "SEW_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("sew-d", "SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("speech_to_text", "SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("speech_to_text_2", "SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("splinter", "SPLINTER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("squeezebert", "SQUEEZEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("swin", "SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("swinv2", "SWINV2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("switch_transformers", "SWITCH_TRANSFORMERS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("t5", "T5_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("table-transformer", "TABLE_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("tapas", "TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("time_series_transformer", "TIME_SERIES_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("timesformer", "TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("transfo-xl", "TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("unispeech", "UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("unispeech-sat", "UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("van", "VAN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("videomae", "VIDEOMAE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vilt", "VILT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("visual_bert", "VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit", "VIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit_hybrid", "VIT_HYBRID_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit_mae", "VIT_MAE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit_msn", "VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("wav2vec2", "WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("wav2vec2-conformer", "WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("whisper", "WHISPER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xclip", "X_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xglm", "XGLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlm", "XLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlm-prophetnet", "XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlm-roberta", "XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlnet", "XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("yolos", "YOLOS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("yoso", "YOSO_PRETRAINED_CONFIG_ARCHIVE_MAP"),
]
)
MODEL_NAMES_MAPPING = OrderedDict(
[
# Add full (and cased) model names here
("albert", "ALBERT"),
("audio-spectrogram-transformer", "Audio Spectrogram Transformer"),
("bart", "BART"),
("barthez", "BARThez"),
("bartpho", "BARTpho"),
("beit", "BEiT"),
("bert", "BERT"),
("bert-generation", "Bert Generation"),
("bert-japanese", "BertJapanese"),
("bertweet", "BERTweet"),
("big_bird", "BigBird"),
("bigbird_pegasus", "BigBird-Pegasus"),
("biogpt", "BioGpt"),
("bit", "BiT"),
("blenderbot", "Blenderbot"),
("blenderbot-small", "BlenderbotSmall"),
("bloom", "BLOOM"),
("bort", "BORT"),
("byt5", "ByT5"),
("camembert", "CamemBERT"),
("canine", "CANINE"),
("chinese_clip", "Chinese-CLIP"),
("clip", "CLIP"),
("clipseg", "CLIPSeg"),
("codegen", "CodeGen"),
("conditional_detr", "Conditional DETR"),
("convbert", "ConvBERT"),
("convnext", "ConvNeXT"),
("cpm", "CPM"),
("ctrl", "CTRL"),
("cvt", "CvT"),
("data2vec-audio", "Data2VecAudio"),
("data2vec-text", "Data2VecText"),
("data2vec-vision", "Data2VecVision"),
("deberta", "DeBERTa"),
("deberta-v2", "DeBERTa-v2"),
("decision_transformer", "Decision Transformer"),
("deformable_detr", "Deformable DETR"),
("deit", "DeiT"),
("detr", "DETR"),
("dialogpt", "DialoGPT"),
("dinat", "DiNAT"),
("distilbert", "DistilBERT"),
("dit", "DiT"),
("donut-swin", "DonutSwin"),
("dpr", "DPR"),
("dpt", "DPT"),
("electra", "ELECTRA"),
("encoder-decoder", "Encoder decoder"),
("ernie", "ERNIE"),
("esm", "ESM"),
("flan-t5", "FLAN-T5"),
("flaubert", "FlauBERT"),
("flava", "FLAVA"),
("fnet", "FNet"),
("fsmt", "FairSeq Machine-Translation"),
("funnel", "Funnel Transformer"),
("glpn", "GLPN"),
("gpt2", "OpenAI GPT-2"),
("gpt_neo", "GPT Neo"),
("gpt_neox", "GPT NeoX"),
("gpt_neox_japanese", "GPT NeoX Japanese"),
("gptj", "GPT-J"),
("groupvit", "GroupViT"),
("herbert", "HerBERT"),
("hubert", "Hubert"),
("ibert", "I-BERT"),
("imagegpt", "ImageGPT"),
("jukebox", "Jukebox"),
("layoutlm", "LayoutLM"),
("layoutlmv2", "LayoutLMv2"),
("layoutlmv3", "LayoutLMv3"),
("layoutxlm", "LayoutXLM"),
("led", "LED"),
("levit", "LeViT"),
("lilt", "LiLT"),
("longformer", "Longformer"),
("longt5", "LongT5"),
("luke", "LUKE"),
("lxmert", "LXMERT"),
("m2m_100", "M2M100"),
("marian", "Marian"),
("markuplm", "MarkupLM"),
("maskformer", "MaskFormer"),
("maskformer-swin", "MaskFormerSwin"),
("mbart", "mBART"),
("mbart50", "mBART-50"),
("mctct", "M-CTC-T"),
("megatron-bert", "Megatron-BERT"),
("megatron_gpt2", "Megatron-GPT2"),
("mluke", "mLUKE"),
("mobilebert", "MobileBERT"),
("mobilenet_v1", "MobileNetV1"),
("mobilenet_v2", "MobileNetV2"),
("mobilevit", "MobileViT"),
("mpnet", "MPNet"),
("mt5", "MT5"),
("mvp", "MVP"),
("nat", "NAT"),
("nezha", "Nezha"),
("nllb", "NLLB"),
("nystromformer", "Nyströmformer"),
("openai-gpt", "OpenAI GPT"),
("opt", "OPT"),
("owlvit", "OWL-ViT"),
("pegasus", "Pegasus"),
("pegasus_x", "PEGASUS-X"),
("perceiver", "Perceiver"),
("phobert", "PhoBERT"),
("plbart", "PLBart"),
("poolformer", "PoolFormer"),
("prophetnet", "ProphetNet"),
("qdqbert", "QDQBert"),
("rag", "RAG"),
("realm", "REALM"),
("reformer", "Reformer"),
("regnet", "RegNet"),
("rembert", "RemBERT"),
("resnet", "ResNet"),
("retribert", "RetriBERT"),
("roberta", "RoBERTa"),
("roc_bert", "RoCBert"),
("roformer", "RoFormer"),
("segformer", "SegFormer"),
("sew", "SEW"),
("sew-d", "SEW-D"),
("speech-encoder-decoder", "Speech Encoder decoder"),
("speech_to_text", "Speech2Text"),
("speech_to_text_2", "Speech2Text2"),
("splinter", "Splinter"),
("squeezebert", "SqueezeBERT"),
("swin", "Swin Transformer"),
("swinv2", "Swin Transformer V2"),
("switch_transformers", "SwitchTransformers"),
("t5", "T5"),
("t5v1.1", "T5v1.1"),
("table-transformer", "Table Transformer"),
("tapas", "TAPAS"),
("tapex", "TAPEX"),
("time_series_transformer", "Time Series Transformer"),
("timesformer", "TimeSformer"),
("trajectory_transformer", "Trajectory Transformer"),
("transfo-xl", "Transformer-XL"),
("trocr", "TrOCR"),
("ul2", "UL2"),
("unispeech", "UniSpeech"),
("unispeech-sat", "UniSpeechSat"),
("van", "VAN"),
("videomae", "VideoMAE"),
("vilt", "ViLT"),
("vision-encoder-decoder", "Vision Encoder decoder"),
("vision-text-dual-encoder", "VisionTextDualEncoder"),
("visual_bert", "VisualBERT"),
("vit", "ViT"),
("vit_hybrid", "ViT Hybrid"),
("vit_mae", "ViTMAE"),
("vit_msn", "ViTMSN"),
("wav2vec2", "Wav2Vec2"),
("wav2vec2-conformer", "Wav2Vec2-Conformer"),
("wav2vec2_phoneme", "Wav2Vec2Phoneme"),
("wavlm", "WavLM"),
("whisper", "Whisper"),
("xclip", "X-CLIP"),
("xglm", "XGLM"),
("xlm", "XLM"),
("xlm-prophetnet", "XLM-ProphetNet"),
("xlm-roberta", "XLM-RoBERTa"),
("xlm-roberta-xl", "XLM-RoBERTa-XL"),
("xlnet", "XLNet"),
("xls_r", "XLS-R"),
("xlsr_wav2vec2", "XLSR-Wav2Vec2"),
("yolos", "YOLOS"),
("yoso", "YOSO"),
]
)
SPECIAL_MODEL_TYPE_TO_MODULE_NAME = OrderedDict(
[
("openai-gpt", "openai"),
("data2vec-audio", "data2vec"),
("data2vec-text", "data2vec"),
("data2vec-vision", "data2vec"),
("donut-swin", "donut"),
("maskformer-swin", "maskformer"),
("xclip", "x_clip"),
]
)
def model_type_to_module_name(key):
"""Converts a config key to the corresponding module."""
# Special treatment
if key in SPECIAL_MODEL_TYPE_TO_MODULE_NAME:
return SPECIAL_MODEL_TYPE_TO_MODULE_NAME[key]
return key.replace("-", "_")
def config_class_to_model_type(config):
"""Converts a config class name to the corresponding model type"""
for key, cls in CONFIG_MAPPING_NAMES.items():
if cls == config:
return key
return None
class _LazyConfigMapping(OrderedDict):
"""
A dictionary that lazily load its values when they are requested.
"""
def __init__(self, mapping):
self._mapping = mapping
self._extra_content = {}
self._modules = {}
def __getitem__(self, key):
if key in self._extra_content:
return self._extra_content[key]
if key not in self._mapping:
raise KeyError(key)
value = self._mapping[key]
module_name = model_type_to_module_name(key)
if module_name not in self._modules:
self._modules[module_name] = importlib.import_module(f".{module_name}", "transformers.models")
if hasattr(self._modules[module_name], value):
return getattr(self._modules[module_name], value)
# Some of the mappings have entries model_type -> config of another model type. In that case we try to grab the
# object at the top level.
transformers_module = importlib.import_module("transformers")
return getattr(transformers_module, value)
def keys(self):
return list(self._mapping.keys()) + list(self._extra_content.keys())
def values(self):
return [self[k] for k in self._mapping.keys()] + list(self._extra_content.values())
def items(self):
return [(k, self[k]) for k in self._mapping.keys()] + list(self._extra_content.items())
def __iter__(self):
return iter(list(self._mapping.keys()) + list(self._extra_content.keys()))
def __contains__(self, item):
return item in self._mapping or item in self._extra_content
def register(self, key, value):
"""
Register a new configuration in this mapping.
"""
if key in self._mapping.keys():
raise ValueError(f"'{key}' is already used by a Transformers config, pick another name.")
self._extra_content[key] = value
CONFIG_MAPPING = _LazyConfigMapping(CONFIG_MAPPING_NAMES)
class _LazyLoadAllMappings(OrderedDict):
"""
A mapping that will load all pairs of key values at the first access (either by indexing, requestions keys, values,
etc.)
Args:
mapping: The mapping to load.
"""
def __init__(self, mapping):
self._mapping = mapping
self._initialized = False
self._data = {}
def _initialize(self):
if self._initialized:
return
warnings.warn(
"ALL_PRETRAINED_CONFIG_ARCHIVE_MAP is deprecated and will be removed in v5 of Transformers. "
"It does not contain all available model checkpoints, far from it. Checkout hf.co/models for that.",
FutureWarning,
)
for model_type, map_name in self._mapping.items():
module_name = model_type_to_module_name(model_type)
module = importlib.import_module(f".{module_name}", "transformers.models")
mapping = getattr(module, map_name)
self._data.update(mapping)
self._initialized = True
def __getitem__(self, key):
self._initialize()
return self._data[key]
def keys(self):
self._initialize()
return self._data.keys()
def values(self):
self._initialize()
return self._data.values()
def items(self):
self._initialize()
return self._data.keys()
def __iter__(self):
self._initialize()
return iter(self._data)
def __contains__(self, item):
self._initialize()
return item in self._data
ALL_PRETRAINED_CONFIG_ARCHIVE_MAP = _LazyLoadAllMappings(CONFIG_ARCHIVE_MAP_MAPPING_NAMES)
def _get_class_name(model_class: Union[str, List[str]]):
if isinstance(model_class, (list, tuple)):
return " or ".join([f"[`{c}`]" for c in model_class if c is not None])
return f"[`{model_class}`]"
def _list_model_options(indent, config_to_class=None, use_model_types=True):
if config_to_class is None and not use_model_types:
raise ValueError("Using `use_model_types=False` requires a `config_to_class` dictionary.")
if use_model_types:
if config_to_class is None:
model_type_to_name = {model_type: f"[`{config}`]" for model_type, config in CONFIG_MAPPING_NAMES.items()}
else:
model_type_to_name = {
model_type: _get_class_name(model_class)
for model_type, model_class in config_to_class.items()
if model_type in MODEL_NAMES_MAPPING
}
lines = [
f"{indent}- **{model_type}** -- {model_type_to_name[model_type]} ({MODEL_NAMES_MAPPING[model_type]} model)"
for model_type in sorted(model_type_to_name.keys())
]
else:
config_to_name = {
CONFIG_MAPPING_NAMES[config]: _get_class_name(clas)
for config, clas in config_to_class.items()
if config in CONFIG_MAPPING_NAMES
}
config_to_model_name = {
config: MODEL_NAMES_MAPPING[model_type] for model_type, config in CONFIG_MAPPING_NAMES.items()
}
lines = [
f"{indent}- [`{config_name}`] configuration class:"
f" {config_to_name[config_name]} ({config_to_model_name[config_name]} model)"
for config_name in sorted(config_to_name.keys())
]
return "\n".join(lines)
def replace_list_option_in_docstrings(config_to_class=None, use_model_types=True):
def docstring_decorator(fn):
docstrings = fn.__doc__
lines = docstrings.split("\n")
i = 0
while i < len(lines) and re.search(r"^(\s*)List options\s*$", lines[i]) is None:
i += 1
if i < len(lines):
indent = re.search(r"^(\s*)List options\s*$", lines[i]).groups()[0]
if use_model_types:
indent = f"{indent} "
lines[i] = _list_model_options(indent, config_to_class=config_to_class, use_model_types=use_model_types)
docstrings = "\n".join(lines)
else:
raise ValueError(
f"The function {fn} should have an empty 'List options' in its docstring as placeholder, current"
f" docstring is:\n{docstrings}"
)
fn.__doc__ = docstrings
return fn
return docstring_decorator
class AutoConfig:
r"""
This is a generic configuration class that will be instantiated as one of the configuration classes of the library
when created with the [`~AutoConfig.from_pretrained`] class method.
This class cannot be instantiated directly using `__init__()` (throws an error).
"""
def __init__(self):
raise EnvironmentError(
"AutoConfig is designed to be instantiated "
"using the `AutoConfig.from_pretrained(pretrained_model_name_or_path)` method."
)
@classmethod
def for_model(cls, model_type: str, *args, **kwargs):
if model_type in CONFIG_MAPPING:
config_class = CONFIG_MAPPING[model_type]
return config_class(*args, **kwargs)
raise ValueError(
f"Unrecognized model identifier: {model_type}. Should contain one of {', '.join(CONFIG_MAPPING.keys())}"
)
@classmethod
@replace_list_option_in_docstrings()
def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
r"""
Instantiate one of the configuration classes of the library from a pretrained model configuration.
The configuration class to instantiate is selected based on the `model_type` property of the config object that
is loaded, or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`:
List options
Args:
pretrained_model_name_or_path (`str` or `os.PathLike`):
Can be either:
- A string, the *model id* of a pretrained model configuration hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or
namespaced under a user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing a configuration file saved using the
[`~PretrainedConfig.save_pretrained`] method, or the [`~PreTrainedModel.save_pretrained`] method,
e.g., `./my_model_directory/`.
- A path or url to a saved configuration JSON *file*, e.g.,
`./my_model_directory/configuration.json`.
cache_dir (`str` or `os.PathLike`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the
standard cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force the (re-)download the model weights and configuration files and override the
cached versions if they exist.
resume_download (`bool`, *optional*, defaults to `False`):
Whether or not to delete incompletely received files. Will attempt to resume the download if such a
file exists.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request.
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
return_unused_kwargs (`bool`, *optional*, defaults to `False`):
If `False`, then this function returns just the final configuration object.
If `True`, then this functions returns a `Tuple(config, unused_kwargs)` where *unused_kwargs* is a
dictionary consisting of the key/value pairs whose keys are not configuration attributes: i.e., the
part of `kwargs` which has not been used to update `config` and is otherwise ignored.
trust_remote_code (`bool`, *optional*, defaults to `False`):
Whether or not to allow for custom models defined on the Hub in their own modeling files. This option
should only be set to `True` for repositories you trust and in which you have read the code, as it will
execute code present on the Hub on your local machine.
kwargs(additional keyword arguments, *optional*):
The values in kwargs of any keys which are configuration attributes will be used to override the loaded
values. Behavior concerning key/value pairs whose keys are *not* configuration attributes is controlled
by the `return_unused_kwargs` keyword parameter.
Examples:
```python
>>> from transformers import AutoConfig
>>> # Download configuration from huggingface.co and cache.
>>> config = AutoConfig.from_pretrained("bert-base-uncased")
>>> # Download configuration from huggingface.co (user-uploaded) and cache.
>>> config = AutoConfig.from_pretrained("dbmdz/bert-base-german-cased")
>>> # If configuration file is in a directory (e.g., was saved using *save_pretrained('./test/saved_model/')*).
>>> config = AutoConfig.from_pretrained("./test/bert_saved_model/")
>>> # Load a specific configuration file.
>>> config = AutoConfig.from_pretrained("./test/bert_saved_model/my_configuration.json")
>>> # Change some config attributes when loading a pretrained config.
>>> config = AutoConfig.from_pretrained("bert-base-uncased", output_attentions=True, foo=False)
>>> config.output_attentions
True
>>> config, unused_kwargs = AutoConfig.from_pretrained(
... "bert-base-uncased", output_attentions=True, foo=False, return_unused_kwargs=True
... )
>>> config.output_attentions
True
>>> unused_kwargs
{'foo': False}
```"""
kwargs["_from_auto"] = True
kwargs["name_or_path"] = pretrained_model_name_or_path
trust_remote_code = kwargs.pop("trust_remote_code", False)
config_dict, unused_kwargs = PretrainedConfig.get_config_dict(pretrained_model_name_or_path, **kwargs)
if "auto_map" in config_dict and "AutoConfig" in config_dict["auto_map"]:
if not trust_remote_code:
raise ValueError(
f"Loading {pretrained_model_name_or_path} requires you to execute the configuration file in that"
" repo on your local machine. Make sure you have read the code there to avoid malicious use, then"
" set the option `trust_remote_code=True` to remove this error."
)
if kwargs.get("revision", None) is None:
logger.warning(
"Explicitly passing a `revision` is encouraged when loading a configuration with custom code to "
"ensure no malicious code has been contributed in a newer revision."
)
class_ref = config_dict["auto_map"]["AutoConfig"]
module_file, class_name = class_ref.split(".")
config_class = get_class_from_dynamic_module(
pretrained_model_name_or_path, module_file + ".py", class_name, **kwargs
)
return config_class.from_pretrained(pretrained_model_name_or_path, **kwargs)
elif "model_type" in config_dict:
config_class = CONFIG_MAPPING[config_dict["model_type"]]
return config_class.from_dict(config_dict, **unused_kwargs)
else:
# Fallback: use pattern matching on the string.
# We go from longer names to shorter names to catch roberta before bert (for instance)
for pattern in sorted(CONFIG_MAPPING.keys(), key=len, reverse=True):
if pattern in str(pretrained_model_name_or_path):
return CONFIG_MAPPING[pattern].from_dict(config_dict, **unused_kwargs)
raise ValueError(
f"Unrecognized model in {pretrained_model_name_or_path}. "
f"Should have a `model_type` key in its {CONFIG_NAME}, or contain one of the following strings "
f"in its name: {', '.join(CONFIG_MAPPING.keys())}"
)
@staticmethod
def register(model_type, config):
"""
Register a new configuration for this class.
Args:
model_type (`str`): The model type like "bert" or "gpt".
config ([`PretrainedConfig`]): The config to register.
"""
if issubclass(config, PretrainedConfig) and config.model_type != model_type:
raise ValueError(
"The config you are passing has a `model_type` attribute that is not consistent with the model type "
f"you passed (config has {config.model_type} and you passed {model_type}. Fix one of those so they "
"match!"
)
CONFIG_MAPPING.register(model_type, config)
| # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Config class."""
import importlib
import re
import warnings
from collections import OrderedDict
from typing import List, Union
from ...configuration_utils import PretrainedConfig
from ...dynamic_module_utils import get_class_from_dynamic_module
from ...utils import CONFIG_NAME, logging
logger = logging.get_logger(__name__)
CONFIG_MAPPING_NAMES = OrderedDict(
[
# Add configs here
("albert", "AlbertConfig"),
("audio-spectrogram-transformer", "ASTConfig"),
("bart", "BartConfig"),
("beit", "BeitConfig"),
("bert", "BertConfig"),
("bert-generation", "BertGenerationConfig"),
("big_bird", "BigBirdConfig"),
("bigbird_pegasus", "BigBirdPegasusConfig"),
("biogpt", "BioGptConfig"),
("bit", "BitConfig"),
("blenderbot", "BlenderbotConfig"),
("blenderbot-small", "BlenderbotSmallConfig"),
("bloom", "BloomConfig"),
("camembert", "CamembertConfig"),
("canine", "CanineConfig"),
("chinese_clip", "ChineseCLIPConfig"),
("clip", "CLIPConfig"),
("clipseg", "CLIPSegConfig"),
("codegen", "CodeGenConfig"),
("conditional_detr", "ConditionalDetrConfig"),
("convbert", "ConvBertConfig"),
("convnext", "ConvNextConfig"),
("ctrl", "CTRLConfig"),
("cvt", "CvtConfig"),
("data2vec-audio", "Data2VecAudioConfig"),
("data2vec-text", "Data2VecTextConfig"),
("data2vec-vision", "Data2VecVisionConfig"),
("deberta", "DebertaConfig"),
("deberta-v2", "DebertaV2Config"),
("decision_transformer", "DecisionTransformerConfig"),
("deformable_detr", "DeformableDetrConfig"),
("deit", "DeiTConfig"),
("detr", "DetrConfig"),
("dinat", "DinatConfig"),
("distilbert", "DistilBertConfig"),
("donut-swin", "DonutSwinConfig"),
("dpr", "DPRConfig"),
("dpt", "DPTConfig"),
("electra", "ElectraConfig"),
("encoder-decoder", "EncoderDecoderConfig"),
("ernie", "ErnieConfig"),
("esm", "EsmConfig"),
("flaubert", "FlaubertConfig"),
("flava", "FlavaConfig"),
("fnet", "FNetConfig"),
("fsmt", "FSMTConfig"),
("funnel", "FunnelConfig"),
("glpn", "GLPNConfig"),
("gpt-sw3", "GPT2Config"),
("gpt2", "GPT2Config"),
("gpt_neo", "GPTNeoConfig"),
("gpt_neox", "GPTNeoXConfig"),
("gpt_neox_japanese", "GPTNeoXJapaneseConfig"),
("gptj", "GPTJConfig"),
("groupvit", "GroupViTConfig"),
("hubert", "HubertConfig"),
("ibert", "IBertConfig"),
("imagegpt", "ImageGPTConfig"),
("jukebox", "JukeboxConfig"),
("layoutlm", "LayoutLMConfig"),
("layoutlmv2", "LayoutLMv2Config"),
("layoutlmv3", "LayoutLMv3Config"),
("led", "LEDConfig"),
("levit", "LevitConfig"),
("lilt", "LiltConfig"),
("longformer", "LongformerConfig"),
("longt5", "LongT5Config"),
("luke", "LukeConfig"),
("lxmert", "LxmertConfig"),
("m2m_100", "M2M100Config"),
("marian", "MarianConfig"),
("markuplm", "MarkupLMConfig"),
("maskformer", "MaskFormerConfig"),
("maskformer-swin", "MaskFormerSwinConfig"),
("mbart", "MBartConfig"),
("mctct", "MCTCTConfig"),
("megatron-bert", "MegatronBertConfig"),
("mobilebert", "MobileBertConfig"),
("mobilenet_v1", "MobileNetV1Config"),
("mobilenet_v2", "MobileNetV2Config"),
("mobilevit", "MobileViTConfig"),
("mpnet", "MPNetConfig"),
("mt5", "MT5Config"),
("mvp", "MvpConfig"),
("nat", "NatConfig"),
("nezha", "NezhaConfig"),
("nystromformer", "NystromformerConfig"),
("openai-gpt", "OpenAIGPTConfig"),
("opt", "OPTConfig"),
("owlvit", "OwlViTConfig"),
("pegasus", "PegasusConfig"),
("pegasus_x", "PegasusXConfig"),
("perceiver", "PerceiverConfig"),
("plbart", "PLBartConfig"),
("poolformer", "PoolFormerConfig"),
("prophetnet", "ProphetNetConfig"),
("qdqbert", "QDQBertConfig"),
("rag", "RagConfig"),
("realm", "RealmConfig"),
("reformer", "ReformerConfig"),
("regnet", "RegNetConfig"),
("rembert", "RemBertConfig"),
("resnet", "ResNetConfig"),
("retribert", "RetriBertConfig"),
("roberta", "RobertaConfig"),
("roc_bert", "RoCBertConfig"),
("roformer", "RoFormerConfig"),
("segformer", "SegformerConfig"),
("sew", "SEWConfig"),
("sew-d", "SEWDConfig"),
("speech-encoder-decoder", "SpeechEncoderDecoderConfig"),
("speech_to_text", "Speech2TextConfig"),
("speech_to_text_2", "Speech2Text2Config"),
("splinter", "SplinterConfig"),
("squeezebert", "SqueezeBertConfig"),
("swin", "SwinConfig"),
("swinv2", "Swinv2Config"),
("switch_transformers", "SwitchTransformersConfig"),
("t5", "T5Config"),
("table-transformer", "TableTransformerConfig"),
("tapas", "TapasConfig"),
("time_series_transformer", "TimeSeriesTransformerConfig"),
("timesformer", "TimesformerConfig"),
("trajectory_transformer", "TrajectoryTransformerConfig"),
("transfo-xl", "TransfoXLConfig"),
("trocr", "TrOCRConfig"),
("unispeech", "UniSpeechConfig"),
("unispeech-sat", "UniSpeechSatConfig"),
("van", "VanConfig"),
("videomae", "VideoMAEConfig"),
("vilt", "ViltConfig"),
("vision-encoder-decoder", "VisionEncoderDecoderConfig"),
("vision-text-dual-encoder", "VisionTextDualEncoderConfig"),
("visual_bert", "VisualBertConfig"),
("vit", "ViTConfig"),
("vit_hybrid", "ViTHybridConfig"),
("vit_mae", "ViTMAEConfig"),
("vit_msn", "ViTMSNConfig"),
("wav2vec2", "Wav2Vec2Config"),
("wav2vec2-conformer", "Wav2Vec2ConformerConfig"),
("wavlm", "WavLMConfig"),
("whisper", "WhisperConfig"),
("xclip", "XCLIPConfig"),
("xglm", "XGLMConfig"),
("xlm", "XLMConfig"),
("xlm-prophetnet", "XLMProphetNetConfig"),
("xlm-roberta", "XLMRobertaConfig"),
("xlm-roberta-xl", "XLMRobertaXLConfig"),
("xlnet", "XLNetConfig"),
("yolos", "YolosConfig"),
("yoso", "YosoConfig"),
]
)
CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
[
# Add archive maps here)
("albert", "ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("audio-spectrogram-transformer", "AUDIO_SPECTROGRAM_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bart", "BART_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("beit", "BEIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bert", "BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("big_bird", "BIG_BIRD_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bigbird_pegasus", "BIGBIRD_PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("biogpt", "BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bit", "BIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("blenderbot", "BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("blenderbot-small", "BLENDERBOT_SMALL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("bloom", "BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("camembert", "CAMEMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("canine", "CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("chinese_clip", "CHINESE_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("clip", "CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("clipseg", "CLIPSEG_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("codegen", "CODEGEN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("conditional_detr", "CONDITIONAL_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("convbert", "CONVBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("convnext", "CONVNEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("ctrl", "CTRL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("cvt", "CVT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("data2vec-audio", "DATA2VEC_AUDIO_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("data2vec-text", "DATA2VEC_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("data2vec-vision", "DATA2VEC_VISION_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deberta", "DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deberta-v2", "DEBERTA_V2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deformable_detr", "DEFORMABLE_DETR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("deit", "DEIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("detr", "DETR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("dinat", "DINAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("distilbert", "DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("donut-swin", "DONUT_SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("dpr", "DPR_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("dpt", "DPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("electra", "ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("ernie", "ERNIE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("esm", "ESM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("flaubert", "FLAUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("flava", "FLAVA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("fnet", "FNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("fsmt", "FSMT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("funnel", "FUNNEL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("glpn", "GLPN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt2", "GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt_neo", "GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt_neox", "GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gpt_neox_japanese", "GPT_NEOX_JAPANESE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("gptj", "GPTJ_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("groupvit", "GROUPVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("hubert", "HUBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("ibert", "IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("imagegpt", "IMAGEGPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("jukebox", "JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("layoutlm", "LAYOUTLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("layoutlmv2", "LAYOUTLMV2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("layoutlmv3", "LAYOUTLMV3_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("led", "LED_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("levit", "LEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("lilt", "LILT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("longformer", "LONGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("longt5", "LONGT5_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("luke", "LUKE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("lxmert", "LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("m2m_100", "M2M_100_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("markuplm", "MARKUPLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("maskformer", "MASKFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mbart", "MBART_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mctct", "MCTCT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("megatron-bert", "MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mobilenet_v1", "MOBILENET_V1_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mobilenet_v2", "MOBILENET_V2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mobilevit", "MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mpnet", "MPNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("mvp", "MVP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("nat", "NAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("nezha", "NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("nystromformer", "NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("openai-gpt", "OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("opt", "OPT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("owlvit", "OWLVIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("pegasus", "PEGASUS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("pegasus_x", "PEGASUS_X_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("perceiver", "PERCEIVER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("plbart", "PLBART_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("poolformer", "POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("prophetnet", "PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("qdqbert", "QDQBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("realm", "REALM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("regnet", "REGNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("rembert", "REMBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("resnet", "RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("retribert", "RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("roberta", "ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("roc_bert", "ROC_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("roformer", "ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("segformer", "SEGFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("sew", "SEW_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("sew-d", "SEW_D_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("speech_to_text", "SPEECH_TO_TEXT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("speech_to_text_2", "SPEECH_TO_TEXT_2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("splinter", "SPLINTER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("squeezebert", "SQUEEZEBERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("swin", "SWIN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("swinv2", "SWINV2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("switch_transformers", "SWITCH_TRANSFORMERS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("t5", "T5_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("table-transformer", "TABLE_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("tapas", "TAPAS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("time_series_transformer", "TIME_SERIES_TRANSFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("timesformer", "TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("transfo-xl", "TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("unispeech", "UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("unispeech-sat", "UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("van", "VAN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("videomae", "VIDEOMAE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vilt", "VILT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("visual_bert", "VISUAL_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit", "VIT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit_hybrid", "VIT_HYBRID_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit_mae", "VIT_MAE_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("vit_msn", "VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("wav2vec2", "WAV_2_VEC_2_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("wav2vec2-conformer", "WAV2VEC2_CONFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("whisper", "WHISPER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xclip", "X_CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xglm", "XGLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlm", "XLM_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlm-prophetnet", "XLM_PROPHETNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlm-roberta", "XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("xlnet", "XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("yolos", "YOLOS_PRETRAINED_CONFIG_ARCHIVE_MAP"),
("yoso", "YOSO_PRETRAINED_CONFIG_ARCHIVE_MAP"),
]
)
MODEL_NAMES_MAPPING = OrderedDict(
[
# Add full (and cased) model names here
("albert", "ALBERT"),
("audio-spectrogram-transformer", "Audio Spectrogram Transformer"),
("bart", "BART"),
("barthez", "BARThez"),
("bartpho", "BARTpho"),
("beit", "BEiT"),
("bert", "BERT"),
("bert-generation", "Bert Generation"),
("bert-japanese", "BertJapanese"),
("bertweet", "BERTweet"),
("big_bird", "BigBird"),
("bigbird_pegasus", "BigBird-Pegasus"),
("biogpt", "BioGpt"),
("bit", "BiT"),
("blenderbot", "Blenderbot"),
("blenderbot-small", "BlenderbotSmall"),
("bloom", "BLOOM"),
("bort", "BORT"),
("byt5", "ByT5"),
("camembert", "CamemBERT"),
("canine", "CANINE"),
("chinese_clip", "Chinese-CLIP"),
("clip", "CLIP"),
("clipseg", "CLIPSeg"),
("codegen", "CodeGen"),
("conditional_detr", "Conditional DETR"),
("convbert", "ConvBERT"),
("convnext", "ConvNeXT"),
("cpm", "CPM"),
("ctrl", "CTRL"),
("cvt", "CvT"),
("data2vec-audio", "Data2VecAudio"),
("data2vec-text", "Data2VecText"),
("data2vec-vision", "Data2VecVision"),
("deberta", "DeBERTa"),
("deberta-v2", "DeBERTa-v2"),
("decision_transformer", "Decision Transformer"),
("deformable_detr", "Deformable DETR"),
("deit", "DeiT"),
("detr", "DETR"),
("dialogpt", "DialoGPT"),
("dinat", "DiNAT"),
("distilbert", "DistilBERT"),
("dit", "DiT"),
("donut-swin", "DonutSwin"),
("dpr", "DPR"),
("dpt", "DPT"),
("electra", "ELECTRA"),
("encoder-decoder", "Encoder decoder"),
("ernie", "ERNIE"),
("esm", "ESM"),
("flan-t5", "FLAN-T5"),
("flaubert", "FlauBERT"),
("flava", "FLAVA"),
("fnet", "FNet"),
("fsmt", "FairSeq Machine-Translation"),
("funnel", "Funnel Transformer"),
("glpn", "GLPN"),
("gpt-sw3", "GPT-Sw3"),
("gpt2", "OpenAI GPT-2"),
("gpt_neo", "GPT Neo"),
("gpt_neox", "GPT NeoX"),
("gpt_neox_japanese", "GPT NeoX Japanese"),
("gptj", "GPT-J"),
("groupvit", "GroupViT"),
("herbert", "HerBERT"),
("hubert", "Hubert"),
("ibert", "I-BERT"),
("imagegpt", "ImageGPT"),
("jukebox", "Jukebox"),
("layoutlm", "LayoutLM"),
("layoutlmv2", "LayoutLMv2"),
("layoutlmv3", "LayoutLMv3"),
("layoutxlm", "LayoutXLM"),
("led", "LED"),
("levit", "LeViT"),
("lilt", "LiLT"),
("longformer", "Longformer"),
("longt5", "LongT5"),
("luke", "LUKE"),
("lxmert", "LXMERT"),
("m2m_100", "M2M100"),
("marian", "Marian"),
("markuplm", "MarkupLM"),
("maskformer", "MaskFormer"),
("maskformer-swin", "MaskFormerSwin"),
("mbart", "mBART"),
("mbart50", "mBART-50"),
("mctct", "M-CTC-T"),
("megatron-bert", "Megatron-BERT"),
("megatron_gpt2", "Megatron-GPT2"),
("mluke", "mLUKE"),
("mobilebert", "MobileBERT"),
("mobilenet_v1", "MobileNetV1"),
("mobilenet_v2", "MobileNetV2"),
("mobilevit", "MobileViT"),
("mpnet", "MPNet"),
("mt5", "MT5"),
("mvp", "MVP"),
("nat", "NAT"),
("nezha", "Nezha"),
("nllb", "NLLB"),
("nystromformer", "Nyströmformer"),
("openai-gpt", "OpenAI GPT"),
("opt", "OPT"),
("owlvit", "OWL-ViT"),
("pegasus", "Pegasus"),
("pegasus_x", "PEGASUS-X"),
("perceiver", "Perceiver"),
("phobert", "PhoBERT"),
("plbart", "PLBart"),
("poolformer", "PoolFormer"),
("prophetnet", "ProphetNet"),
("qdqbert", "QDQBert"),
("rag", "RAG"),
("realm", "REALM"),
("reformer", "Reformer"),
("regnet", "RegNet"),
("rembert", "RemBERT"),
("resnet", "ResNet"),
("retribert", "RetriBERT"),
("roberta", "RoBERTa"),
("roc_bert", "RoCBert"),
("roformer", "RoFormer"),
("segformer", "SegFormer"),
("sew", "SEW"),
("sew-d", "SEW-D"),
("speech-encoder-decoder", "Speech Encoder decoder"),
("speech_to_text", "Speech2Text"),
("speech_to_text_2", "Speech2Text2"),
("splinter", "Splinter"),
("squeezebert", "SqueezeBERT"),
("swin", "Swin Transformer"),
("swinv2", "Swin Transformer V2"),
("switch_transformers", "SwitchTransformers"),
("t5", "T5"),
("t5v1.1", "T5v1.1"),
("table-transformer", "Table Transformer"),
("tapas", "TAPAS"),
("tapex", "TAPEX"),
("time_series_transformer", "Time Series Transformer"),
("timesformer", "TimeSformer"),
("trajectory_transformer", "Trajectory Transformer"),
("transfo-xl", "Transformer-XL"),
("trocr", "TrOCR"),
("ul2", "UL2"),
("unispeech", "UniSpeech"),
("unispeech-sat", "UniSpeechSat"),
("van", "VAN"),
("videomae", "VideoMAE"),
("vilt", "ViLT"),
("vision-encoder-decoder", "Vision Encoder decoder"),
("vision-text-dual-encoder", "VisionTextDualEncoder"),
("visual_bert", "VisualBERT"),
("vit", "ViT"),
("vit_hybrid", "ViT Hybrid"),
("vit_mae", "ViTMAE"),
("vit_msn", "ViTMSN"),
("wav2vec2", "Wav2Vec2"),
("wav2vec2-conformer", "Wav2Vec2-Conformer"),
("wav2vec2_phoneme", "Wav2Vec2Phoneme"),
("wavlm", "WavLM"),
("whisper", "Whisper"),
("xclip", "X-CLIP"),
("xglm", "XGLM"),
("xlm", "XLM"),
("xlm-prophetnet", "XLM-ProphetNet"),
("xlm-roberta", "XLM-RoBERTa"),
("xlm-roberta-xl", "XLM-RoBERTa-XL"),
("xlnet", "XLNet"),
("xls_r", "XLS-R"),
("xlsr_wav2vec2", "XLSR-Wav2Vec2"),
("yolos", "YOLOS"),
("yoso", "YOSO"),
]
)
SPECIAL_MODEL_TYPE_TO_MODULE_NAME = OrderedDict(
[
("openai-gpt", "openai"),
("data2vec-audio", "data2vec"),
("data2vec-text", "data2vec"),
("data2vec-vision", "data2vec"),
("donut-swin", "donut"),
("maskformer-swin", "maskformer"),
("xclip", "x_clip"),
]
)
def model_type_to_module_name(key):
"""Converts a config key to the corresponding module."""
# Special treatment
if key in SPECIAL_MODEL_TYPE_TO_MODULE_NAME:
return SPECIAL_MODEL_TYPE_TO_MODULE_NAME[key]
return key.replace("-", "_")
def config_class_to_model_type(config):
"""Converts a config class name to the corresponding model type"""
for key, cls in CONFIG_MAPPING_NAMES.items():
if cls == config:
return key
return None
class _LazyConfigMapping(OrderedDict):
"""
A dictionary that lazily load its values when they are requested.
"""
def __init__(self, mapping):
self._mapping = mapping
self._extra_content = {}
self._modules = {}
def __getitem__(self, key):
if key in self._extra_content:
return self._extra_content[key]
if key not in self._mapping:
raise KeyError(key)
value = self._mapping[key]
module_name = model_type_to_module_name(key)
if module_name not in self._modules:
self._modules[module_name] = importlib.import_module(f".{module_name}", "transformers.models")
if hasattr(self._modules[module_name], value):
return getattr(self._modules[module_name], value)
# Some of the mappings have entries model_type -> config of another model type. In that case we try to grab the
# object at the top level.
transformers_module = importlib.import_module("transformers")
return getattr(transformers_module, value)
def keys(self):
return list(self._mapping.keys()) + list(self._extra_content.keys())
def values(self):
return [self[k] for k in self._mapping.keys()] + list(self._extra_content.values())
def items(self):
return [(k, self[k]) for k in self._mapping.keys()] + list(self._extra_content.items())
def __iter__(self):
return iter(list(self._mapping.keys()) + list(self._extra_content.keys()))
def __contains__(self, item):
return item in self._mapping or item in self._extra_content
def register(self, key, value):
"""
Register a new configuration in this mapping.
"""
if key in self._mapping.keys():
raise ValueError(f"'{key}' is already used by a Transformers config, pick another name.")
self._extra_content[key] = value
CONFIG_MAPPING = _LazyConfigMapping(CONFIG_MAPPING_NAMES)
class _LazyLoadAllMappings(OrderedDict):
"""
A mapping that will load all pairs of key values at the first access (either by indexing, requestions keys, values,
etc.)
Args:
mapping: The mapping to load.
"""
def __init__(self, mapping):
self._mapping = mapping
self._initialized = False
self._data = {}
def _initialize(self):
if self._initialized:
return
warnings.warn(
"ALL_PRETRAINED_CONFIG_ARCHIVE_MAP is deprecated and will be removed in v5 of Transformers. "
"It does not contain all available model checkpoints, far from it. Checkout hf.co/models for that.",
FutureWarning,
)
for model_type, map_name in self._mapping.items():
module_name = model_type_to_module_name(model_type)
module = importlib.import_module(f".{module_name}", "transformers.models")
mapping = getattr(module, map_name)
self._data.update(mapping)
self._initialized = True
def __getitem__(self, key):
self._initialize()
return self._data[key]
def keys(self):
self._initialize()
return self._data.keys()
def values(self):
self._initialize()
return self._data.values()
def items(self):
self._initialize()
return self._data.keys()
def __iter__(self):
self._initialize()
return iter(self._data)
def __contains__(self, item):
self._initialize()
return item in self._data
ALL_PRETRAINED_CONFIG_ARCHIVE_MAP = _LazyLoadAllMappings(CONFIG_ARCHIVE_MAP_MAPPING_NAMES)
def _get_class_name(model_class: Union[str, List[str]]):
if isinstance(model_class, (list, tuple)):
return " or ".join([f"[`{c}`]" for c in model_class if c is not None])
return f"[`{model_class}`]"
def _list_model_options(indent, config_to_class=None, use_model_types=True):
if config_to_class is None and not use_model_types:
raise ValueError("Using `use_model_types=False` requires a `config_to_class` dictionary.")
if use_model_types:
if config_to_class is None:
model_type_to_name = {model_type: f"[`{config}`]" for model_type, config in CONFIG_MAPPING_NAMES.items()}
else:
model_type_to_name = {
model_type: _get_class_name(model_class)
for model_type, model_class in config_to_class.items()
if model_type in MODEL_NAMES_MAPPING
}
lines = [
f"{indent}- **{model_type}** -- {model_type_to_name[model_type]} ({MODEL_NAMES_MAPPING[model_type]} model)"
for model_type in sorted(model_type_to_name.keys())
]
else:
config_to_name = {
CONFIG_MAPPING_NAMES[config]: _get_class_name(clas)
for config, clas in config_to_class.items()
if config in CONFIG_MAPPING_NAMES
}
config_to_model_name = {
config: MODEL_NAMES_MAPPING[model_type] for model_type, config in CONFIG_MAPPING_NAMES.items()
}
lines = [
f"{indent}- [`{config_name}`] configuration class:"
f" {config_to_name[config_name]} ({config_to_model_name[config_name]} model)"
for config_name in sorted(config_to_name.keys())
]
return "\n".join(lines)
def replace_list_option_in_docstrings(config_to_class=None, use_model_types=True):
def docstring_decorator(fn):
docstrings = fn.__doc__
lines = docstrings.split("\n")
i = 0
while i < len(lines) and re.search(r"^(\s*)List options\s*$", lines[i]) is None:
i += 1
if i < len(lines):
indent = re.search(r"^(\s*)List options\s*$", lines[i]).groups()[0]
if use_model_types:
indent = f"{indent} "
lines[i] = _list_model_options(indent, config_to_class=config_to_class, use_model_types=use_model_types)
docstrings = "\n".join(lines)
else:
raise ValueError(
f"The function {fn} should have an empty 'List options' in its docstring as placeholder, current"
f" docstring is:\n{docstrings}"
)
fn.__doc__ = docstrings
return fn
return docstring_decorator
class AutoConfig:
r"""
This is a generic configuration class that will be instantiated as one of the configuration classes of the library
when created with the [`~AutoConfig.from_pretrained`] class method.
This class cannot be instantiated directly using `__init__()` (throws an error).
"""
def __init__(self):
raise EnvironmentError(
"AutoConfig is designed to be instantiated "
"using the `AutoConfig.from_pretrained(pretrained_model_name_or_path)` method."
)
@classmethod
def for_model(cls, model_type: str, *args, **kwargs):
if model_type in CONFIG_MAPPING:
config_class = CONFIG_MAPPING[model_type]
return config_class(*args, **kwargs)
raise ValueError(
f"Unrecognized model identifier: {model_type}. Should contain one of {', '.join(CONFIG_MAPPING.keys())}"
)
@classmethod
@replace_list_option_in_docstrings()
def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
r"""
Instantiate one of the configuration classes of the library from a pretrained model configuration.
The configuration class to instantiate is selected based on the `model_type` property of the config object that
is loaded, or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`:
List options
Args:
pretrained_model_name_or_path (`str` or `os.PathLike`):
Can be either:
- A string, the *model id* of a pretrained model configuration hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or
namespaced under a user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing a configuration file saved using the
[`~PretrainedConfig.save_pretrained`] method, or the [`~PreTrainedModel.save_pretrained`] method,
e.g., `./my_model_directory/`.
- A path or url to a saved configuration JSON *file*, e.g.,
`./my_model_directory/configuration.json`.
cache_dir (`str` or `os.PathLike`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the
standard cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force the (re-)download the model weights and configuration files and override the
cached versions if they exist.
resume_download (`bool`, *optional*, defaults to `False`):
Whether or not to delete incompletely received files. Will attempt to resume the download if such a
file exists.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request.
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
return_unused_kwargs (`bool`, *optional*, defaults to `False`):
If `False`, then this function returns just the final configuration object.
If `True`, then this functions returns a `Tuple(config, unused_kwargs)` where *unused_kwargs* is a
dictionary consisting of the key/value pairs whose keys are not configuration attributes: i.e., the
part of `kwargs` which has not been used to update `config` and is otherwise ignored.
trust_remote_code (`bool`, *optional*, defaults to `False`):
Whether or not to allow for custom models defined on the Hub in their own modeling files. This option
should only be set to `True` for repositories you trust and in which you have read the code, as it will
execute code present on the Hub on your local machine.
kwargs(additional keyword arguments, *optional*):
The values in kwargs of any keys which are configuration attributes will be used to override the loaded
values. Behavior concerning key/value pairs whose keys are *not* configuration attributes is controlled
by the `return_unused_kwargs` keyword parameter.
Examples:
```python
>>> from transformers import AutoConfig
>>> # Download configuration from huggingface.co and cache.
>>> config = AutoConfig.from_pretrained("bert-base-uncased")
>>> # Download configuration from huggingface.co (user-uploaded) and cache.
>>> config = AutoConfig.from_pretrained("dbmdz/bert-base-german-cased")
>>> # If configuration file is in a directory (e.g., was saved using *save_pretrained('./test/saved_model/')*).
>>> config = AutoConfig.from_pretrained("./test/bert_saved_model/")
>>> # Load a specific configuration file.
>>> config = AutoConfig.from_pretrained("./test/bert_saved_model/my_configuration.json")
>>> # Change some config attributes when loading a pretrained config.
>>> config = AutoConfig.from_pretrained("bert-base-uncased", output_attentions=True, foo=False)
>>> config.output_attentions
True
>>> config, unused_kwargs = AutoConfig.from_pretrained(
... "bert-base-uncased", output_attentions=True, foo=False, return_unused_kwargs=True
... )
>>> config.output_attentions
True
>>> unused_kwargs
{'foo': False}
```"""
kwargs["_from_auto"] = True
kwargs["name_or_path"] = pretrained_model_name_or_path
trust_remote_code = kwargs.pop("trust_remote_code", False)
config_dict, unused_kwargs = PretrainedConfig.get_config_dict(pretrained_model_name_or_path, **kwargs)
if "auto_map" in config_dict and "AutoConfig" in config_dict["auto_map"]:
if not trust_remote_code:
raise ValueError(
f"Loading {pretrained_model_name_or_path} requires you to execute the configuration file in that"
" repo on your local machine. Make sure you have read the code there to avoid malicious use, then"
" set the option `trust_remote_code=True` to remove this error."
)
if kwargs.get("revision", None) is None:
logger.warning(
"Explicitly passing a `revision` is encouraged when loading a configuration with custom code to "
"ensure no malicious code has been contributed in a newer revision."
)
class_ref = config_dict["auto_map"]["AutoConfig"]
module_file, class_name = class_ref.split(".")
config_class = get_class_from_dynamic_module(
pretrained_model_name_or_path, module_file + ".py", class_name, **kwargs
)
return config_class.from_pretrained(pretrained_model_name_or_path, **kwargs)
elif "model_type" in config_dict:
config_class = CONFIG_MAPPING[config_dict["model_type"]]
return config_class.from_dict(config_dict, **unused_kwargs)
else:
# Fallback: use pattern matching on the string.
# We go from longer names to shorter names to catch roberta before bert (for instance)
for pattern in sorted(CONFIG_MAPPING.keys(), key=len, reverse=True):
if pattern in str(pretrained_model_name_or_path):
return CONFIG_MAPPING[pattern].from_dict(config_dict, **unused_kwargs)
raise ValueError(
f"Unrecognized model in {pretrained_model_name_or_path}. "
f"Should have a `model_type` key in its {CONFIG_NAME}, or contain one of the following strings "
f"in its name: {', '.join(CONFIG_MAPPING.keys())}"
)
@staticmethod
def register(model_type, config):
"""
Register a new configuration for this class.
Args:
model_type (`str`): The model type like "bert" or "gpt".
config ([`PretrainedConfig`]): The config to register.
"""
if issubclass(config, PretrainedConfig) and config.model_type != model_type:
raise ValueError(
"The config you are passing has a `model_type` attribute that is not consistent with the model type "
f"you passed (config has {config.model_type} and you passed {model_type}. Fix one of those so they "
"match!"
)
CONFIG_MAPPING.register(model_type, config)
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/auto/modeling_auto.py | # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Model class."""
import warnings
from collections import OrderedDict
from ...utils import logging
from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update
from .configuration_auto import CONFIG_MAPPING_NAMES
logger = logging.get_logger(__name__)
MODEL_MAPPING_NAMES = OrderedDict(
[
# Base model mapping
("albert", "AlbertModel"),
("audio-spectrogram-transformer", "ASTModel"),
("bart", "BartModel"),
("beit", "BeitModel"),
("bert", "BertModel"),
("bert-generation", "BertGenerationEncoder"),
("big_bird", "BigBirdModel"),
("bigbird_pegasus", "BigBirdPegasusModel"),
("biogpt", "BioGptModel"),
("bit", "BitModel"),
("blenderbot", "BlenderbotModel"),
("blenderbot-small", "BlenderbotSmallModel"),
("bloom", "BloomModel"),
("camembert", "CamembertModel"),
("canine", "CanineModel"),
("chinese_clip", "ChineseCLIPModel"),
("clip", "CLIPModel"),
("clipseg", "CLIPSegModel"),
("codegen", "CodeGenModel"),
("conditional_detr", "ConditionalDetrModel"),
("convbert", "ConvBertModel"),
("convnext", "ConvNextModel"),
("ctrl", "CTRLModel"),
("cvt", "CvtModel"),
("data2vec-audio", "Data2VecAudioModel"),
("data2vec-text", "Data2VecTextModel"),
("data2vec-vision", "Data2VecVisionModel"),
("deberta", "DebertaModel"),
("deberta-v2", "DebertaV2Model"),
("decision_transformer", "DecisionTransformerModel"),
("decision_transformer_gpt2", "DecisionTransformerGPT2Model"),
("deformable_detr", "DeformableDetrModel"),
("deit", "DeiTModel"),
("detr", "DetrModel"),
("dinat", "DinatModel"),
("distilbert", "DistilBertModel"),
("donut-swin", "DonutSwinModel"),
("dpr", "DPRQuestionEncoder"),
("dpt", "DPTModel"),
("electra", "ElectraModel"),
("ernie", "ErnieModel"),
("esm", "EsmModel"),
("flaubert", "FlaubertModel"),
("flava", "FlavaModel"),
("fnet", "FNetModel"),
("fsmt", "FSMTModel"),
("funnel", ("FunnelModel", "FunnelBaseModel")),
("glpn", "GLPNModel"),
("gpt2", "GPT2Model"),
("gpt_neo", "GPTNeoModel"),
("gpt_neox", "GPTNeoXModel"),
("gpt_neox_japanese", "GPTNeoXJapaneseModel"),
("gptj", "GPTJModel"),
("groupvit", "GroupViTModel"),
("hubert", "HubertModel"),
("ibert", "IBertModel"),
("imagegpt", "ImageGPTModel"),
("jukebox", "JukeboxModel"),
("layoutlm", "LayoutLMModel"),
("layoutlmv2", "LayoutLMv2Model"),
("layoutlmv3", "LayoutLMv3Model"),
("led", "LEDModel"),
("levit", "LevitModel"),
("lilt", "LiltModel"),
("longformer", "LongformerModel"),
("longt5", "LongT5Model"),
("luke", "LukeModel"),
("lxmert", "LxmertModel"),
("m2m_100", "M2M100Model"),
("marian", "MarianModel"),
("markuplm", "MarkupLMModel"),
("maskformer", "MaskFormerModel"),
("maskformer-swin", "MaskFormerSwinModel"),
("mbart", "MBartModel"),
("mctct", "MCTCTModel"),
("megatron-bert", "MegatronBertModel"),
("mobilebert", "MobileBertModel"),
("mobilenet_v1", "MobileNetV1Model"),
("mobilenet_v2", "MobileNetV2Model"),
("mobilevit", "MobileViTModel"),
("mpnet", "MPNetModel"),
("mt5", "MT5Model"),
("mvp", "MvpModel"),
("nat", "NatModel"),
("nezha", "NezhaModel"),
("nllb", "M2M100Model"),
("nystromformer", "NystromformerModel"),
("openai-gpt", "OpenAIGPTModel"),
("opt", "OPTModel"),
("owlvit", "OwlViTModel"),
("pegasus", "PegasusModel"),
("pegasus_x", "PegasusXModel"),
("perceiver", "PerceiverModel"),
("plbart", "PLBartModel"),
("poolformer", "PoolFormerModel"),
("prophetnet", "ProphetNetModel"),
("qdqbert", "QDQBertModel"),
("reformer", "ReformerModel"),
("regnet", "RegNetModel"),
("rembert", "RemBertModel"),
("resnet", "ResNetModel"),
("retribert", "RetriBertModel"),
("roberta", "RobertaModel"),
("roc_bert", "RoCBertModel"),
("roformer", "RoFormerModel"),
("segformer", "SegformerModel"),
("sew", "SEWModel"),
("sew-d", "SEWDModel"),
("speech_to_text", "Speech2TextModel"),
("splinter", "SplinterModel"),
("squeezebert", "SqueezeBertModel"),
("swin", "SwinModel"),
("swinv2", "Swinv2Model"),
("switch_transformers", "SwitchTransformersModel"),
("t5", "T5Model"),
("table-transformer", "TableTransformerModel"),
("tapas", "TapasModel"),
("time_series_transformer", "TimeSeriesTransformerModel"),
("timesformer", "TimesformerModel"),
("trajectory_transformer", "TrajectoryTransformerModel"),
("transfo-xl", "TransfoXLModel"),
("unispeech", "UniSpeechModel"),
("unispeech-sat", "UniSpeechSatModel"),
("van", "VanModel"),
("videomae", "VideoMAEModel"),
("vilt", "ViltModel"),
("vision-text-dual-encoder", "VisionTextDualEncoderModel"),
("visual_bert", "VisualBertModel"),
("vit", "ViTModel"),
("vit_hybrid", "ViTHybridModel"),
("vit_mae", "ViTMAEModel"),
("vit_msn", "ViTMSNModel"),
("wav2vec2", "Wav2Vec2Model"),
("wav2vec2-conformer", "Wav2Vec2ConformerModel"),
("wavlm", "WavLMModel"),
("whisper", "WhisperModel"),
("xclip", "XCLIPModel"),
("xglm", "XGLMModel"),
("xlm", "XLMModel"),
("xlm-prophetnet", "XLMProphetNetModel"),
("xlm-roberta", "XLMRobertaModel"),
("xlm-roberta-xl", "XLMRobertaXLModel"),
("xlnet", "XLNetModel"),
("yolos", "YolosModel"),
("yoso", "YosoModel"),
]
)
MODEL_FOR_PRETRAINING_MAPPING_NAMES = OrderedDict(
[
# Model for pre-training mapping
("albert", "AlbertForPreTraining"),
("bart", "BartForConditionalGeneration"),
("bert", "BertForPreTraining"),
("big_bird", "BigBirdForPreTraining"),
("bloom", "BloomForCausalLM"),
("camembert", "CamembertForMaskedLM"),
("ctrl", "CTRLLMHeadModel"),
("data2vec-text", "Data2VecTextForMaskedLM"),
("deberta", "DebertaForMaskedLM"),
("deberta-v2", "DebertaV2ForMaskedLM"),
("distilbert", "DistilBertForMaskedLM"),
("electra", "ElectraForPreTraining"),
("ernie", "ErnieForPreTraining"),
("flaubert", "FlaubertWithLMHeadModel"),
("flava", "FlavaForPreTraining"),
("fnet", "FNetForPreTraining"),
("fsmt", "FSMTForConditionalGeneration"),
("funnel", "FunnelForPreTraining"),
("gpt2", "GPT2LMHeadModel"),
("ibert", "IBertForMaskedLM"),
("layoutlm", "LayoutLMForMaskedLM"),
("longformer", "LongformerForMaskedLM"),
("luke", "LukeForMaskedLM"),
("lxmert", "LxmertForPreTraining"),
("megatron-bert", "MegatronBertForPreTraining"),
("mobilebert", "MobileBertForPreTraining"),
("mpnet", "MPNetForMaskedLM"),
("mvp", "MvpForConditionalGeneration"),
("nezha", "NezhaForPreTraining"),
("openai-gpt", "OpenAIGPTLMHeadModel"),
("retribert", "RetriBertModel"),
("roberta", "RobertaForMaskedLM"),
("roc_bert", "RoCBertForPreTraining"),
("splinter", "SplinterForPreTraining"),
("squeezebert", "SqueezeBertForMaskedLM"),
("switch_transformers", "SwitchTransformersForConditionalGeneration"),
("t5", "T5ForConditionalGeneration"),
("tapas", "TapasForMaskedLM"),
("transfo-xl", "TransfoXLLMHeadModel"),
("unispeech", "UniSpeechForPreTraining"),
("unispeech-sat", "UniSpeechSatForPreTraining"),
("videomae", "VideoMAEForPreTraining"),
("visual_bert", "VisualBertForPreTraining"),
("vit_mae", "ViTMAEForPreTraining"),
("wav2vec2", "Wav2Vec2ForPreTraining"),
("wav2vec2-conformer", "Wav2Vec2ConformerForPreTraining"),
("xlm", "XLMWithLMHeadModel"),
("xlm-roberta", "XLMRobertaForMaskedLM"),
("xlm-roberta-xl", "XLMRobertaXLForMaskedLM"),
("xlnet", "XLNetLMHeadModel"),
]
)
MODEL_WITH_LM_HEAD_MAPPING_NAMES = OrderedDict(
[
# Model with LM heads mapping
("albert", "AlbertForMaskedLM"),
("bart", "BartForConditionalGeneration"),
("bert", "BertForMaskedLM"),
("big_bird", "BigBirdForMaskedLM"),
("bigbird_pegasus", "BigBirdPegasusForConditionalGeneration"),
("blenderbot-small", "BlenderbotSmallForConditionalGeneration"),
("bloom", "BloomForCausalLM"),
("camembert", "CamembertForMaskedLM"),
("codegen", "CodeGenForCausalLM"),
("convbert", "ConvBertForMaskedLM"),
("ctrl", "CTRLLMHeadModel"),
("data2vec-text", "Data2VecTextForMaskedLM"),
("deberta", "DebertaForMaskedLM"),
("deberta-v2", "DebertaV2ForMaskedLM"),
("distilbert", "DistilBertForMaskedLM"),
("electra", "ElectraForMaskedLM"),
("encoder-decoder", "EncoderDecoderModel"),
("ernie", "ErnieForMaskedLM"),
("esm", "EsmForMaskedLM"),
("flaubert", "FlaubertWithLMHeadModel"),
("fnet", "FNetForMaskedLM"),
("fsmt", "FSMTForConditionalGeneration"),
("funnel", "FunnelForMaskedLM"),
("gpt2", "GPT2LMHeadModel"),
("gpt_neo", "GPTNeoForCausalLM"),
("gpt_neox", "GPTNeoXForCausalLM"),
("gpt_neox_japanese", "GPTNeoXJapaneseForCausalLM"),
("gptj", "GPTJForCausalLM"),
("ibert", "IBertForMaskedLM"),
("layoutlm", "LayoutLMForMaskedLM"),
("led", "LEDForConditionalGeneration"),
("longformer", "LongformerForMaskedLM"),
("longt5", "LongT5ForConditionalGeneration"),
("luke", "LukeForMaskedLM"),
("m2m_100", "M2M100ForConditionalGeneration"),
("marian", "MarianMTModel"),
("megatron-bert", "MegatronBertForCausalLM"),
("mobilebert", "MobileBertForMaskedLM"),
("mpnet", "MPNetForMaskedLM"),
("mvp", "MvpForConditionalGeneration"),
("nezha", "NezhaForMaskedLM"),
("nllb", "M2M100ForConditionalGeneration"),
("nystromformer", "NystromformerForMaskedLM"),
("openai-gpt", "OpenAIGPTLMHeadModel"),
("pegasus_x", "PegasusXForConditionalGeneration"),
("plbart", "PLBartForConditionalGeneration"),
("qdqbert", "QDQBertForMaskedLM"),
("reformer", "ReformerModelWithLMHead"),
("rembert", "RemBertForMaskedLM"),
("roberta", "RobertaForMaskedLM"),
("roc_bert", "RoCBertForMaskedLM"),
("roformer", "RoFormerForMaskedLM"),
("speech_to_text", "Speech2TextForConditionalGeneration"),
("squeezebert", "SqueezeBertForMaskedLM"),
("switch_transformers", "SwitchTransformersForConditionalGeneration"),
("t5", "T5ForConditionalGeneration"),
("tapas", "TapasForMaskedLM"),
("transfo-xl", "TransfoXLLMHeadModel"),
("wav2vec2", "Wav2Vec2ForMaskedLM"),
("whisper", "WhisperForConditionalGeneration"),
("xlm", "XLMWithLMHeadModel"),
("xlm-roberta", "XLMRobertaForMaskedLM"),
("xlm-roberta-xl", "XLMRobertaXLForMaskedLM"),
("xlnet", "XLNetLMHeadModel"),
("yoso", "YosoForMaskedLM"),
]
)
MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Causal LM mapping
("bart", "BartForCausalLM"),
("bert", "BertLMHeadModel"),
("bert-generation", "BertGenerationDecoder"),
("big_bird", "BigBirdForCausalLM"),
("bigbird_pegasus", "BigBirdPegasusForCausalLM"),
("biogpt", "BioGptForCausalLM"),
("blenderbot", "BlenderbotForCausalLM"),
("blenderbot-small", "BlenderbotSmallForCausalLM"),
("bloom", "BloomForCausalLM"),
("camembert", "CamembertForCausalLM"),
("codegen", "CodeGenForCausalLM"),
("ctrl", "CTRLLMHeadModel"),
("data2vec-text", "Data2VecTextForCausalLM"),
("electra", "ElectraForCausalLM"),
("ernie", "ErnieForCausalLM"),
("gpt2", "GPT2LMHeadModel"),
("gpt_neo", "GPTNeoForCausalLM"),
("gpt_neox", "GPTNeoXForCausalLM"),
("gpt_neox_japanese", "GPTNeoXJapaneseForCausalLM"),
("gptj", "GPTJForCausalLM"),
("marian", "MarianForCausalLM"),
("mbart", "MBartForCausalLM"),
("megatron-bert", "MegatronBertForCausalLM"),
("mvp", "MvpForCausalLM"),
("openai-gpt", "OpenAIGPTLMHeadModel"),
("opt", "OPTForCausalLM"),
("pegasus", "PegasusForCausalLM"),
("plbart", "PLBartForCausalLM"),
("prophetnet", "ProphetNetForCausalLM"),
("qdqbert", "QDQBertLMHeadModel"),
("reformer", "ReformerModelWithLMHead"),
("rembert", "RemBertForCausalLM"),
("roberta", "RobertaForCausalLM"),
("roc_bert", "RoCBertForCausalLM"),
("roformer", "RoFormerForCausalLM"),
("speech_to_text_2", "Speech2Text2ForCausalLM"),
("transfo-xl", "TransfoXLLMHeadModel"),
("trocr", "TrOCRForCausalLM"),
("xglm", "XGLMForCausalLM"),
("xlm", "XLMWithLMHeadModel"),
("xlm-prophetnet", "XLMProphetNetForCausalLM"),
("xlm-roberta", "XLMRobertaForCausalLM"),
("xlm-roberta-xl", "XLMRobertaXLForCausalLM"),
("xlnet", "XLNetLMHeadModel"),
]
)
MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
[
("deit", "DeiTForMaskedImageModeling"),
("swin", "SwinForMaskedImageModeling"),
("swinv2", "Swinv2ForMaskedImageModeling"),
("vit", "ViTForMaskedImageModeling"),
]
)
MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
# Model for Causal Image Modeling mapping
[
("imagegpt", "ImageGPTForCausalImageModeling"),
]
)
MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Image Classification mapping
("beit", "BeitForImageClassification"),
("bit", "BitForImageClassification"),
("convnext", "ConvNextForImageClassification"),
("cvt", "CvtForImageClassification"),
("data2vec-vision", "Data2VecVisionForImageClassification"),
("deit", ("DeiTForImageClassification", "DeiTForImageClassificationWithTeacher")),
("dinat", "DinatForImageClassification"),
("imagegpt", "ImageGPTForImageClassification"),
("levit", ("LevitForImageClassification", "LevitForImageClassificationWithTeacher")),
("mobilenet_v1", "MobileNetV1ForImageClassification"),
("mobilenet_v2", "MobileNetV2ForImageClassification"),
("mobilevit", "MobileViTForImageClassification"),
("nat", "NatForImageClassification"),
(
"perceiver",
(
"PerceiverForImageClassificationLearned",
"PerceiverForImageClassificationFourier",
"PerceiverForImageClassificationConvProcessing",
),
),
("poolformer", "PoolFormerForImageClassification"),
("regnet", "RegNetForImageClassification"),
("resnet", "ResNetForImageClassification"),
("segformer", "SegformerForImageClassification"),
("swin", "SwinForImageClassification"),
("swinv2", "Swinv2ForImageClassification"),
("van", "VanForImageClassification"),
("vit", "ViTForImageClassification"),
("vit_hybrid", "ViTHybridForImageClassification"),
("vit_msn", "ViTMSNForImageClassification"),
]
)
MODEL_FOR_IMAGE_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Do not add new models here, this class will be deprecated in the future.
# Model for Image Segmentation mapping
("detr", "DetrForSegmentation"),
]
)
MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Model for Semantic Segmentation mapping
("beit", "BeitForSemanticSegmentation"),
("data2vec-vision", "Data2VecVisionForSemanticSegmentation"),
("dpt", "DPTForSemanticSegmentation"),
("mobilenet_v2", "MobileNetV2ForSemanticSegmentation"),
("mobilevit", "MobileViTForSemanticSegmentation"),
("segformer", "SegformerForSemanticSegmentation"),
]
)
MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Model for Instance Segmentation mapping
("maskformer", "MaskFormerForInstanceSegmentation"),
]
)
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
("timesformer", "TimesformerForVideoClassification"),
("videomae", "VideoMAEForVideoClassification"),
]
)
MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("vision-encoder-decoder", "VisionEncoderDecoderModel"),
]
)
MODEL_FOR_MASKED_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Masked LM mapping
("albert", "AlbertForMaskedLM"),
("bart", "BartForConditionalGeneration"),
("bert", "BertForMaskedLM"),
("big_bird", "BigBirdForMaskedLM"),
("camembert", "CamembertForMaskedLM"),
("convbert", "ConvBertForMaskedLM"),
("data2vec-text", "Data2VecTextForMaskedLM"),
("deberta", "DebertaForMaskedLM"),
("deberta-v2", "DebertaV2ForMaskedLM"),
("distilbert", "DistilBertForMaskedLM"),
("electra", "ElectraForMaskedLM"),
("ernie", "ErnieForMaskedLM"),
("esm", "EsmForMaskedLM"),
("flaubert", "FlaubertWithLMHeadModel"),
("fnet", "FNetForMaskedLM"),
("funnel", "FunnelForMaskedLM"),
("ibert", "IBertForMaskedLM"),
("layoutlm", "LayoutLMForMaskedLM"),
("longformer", "LongformerForMaskedLM"),
("luke", "LukeForMaskedLM"),
("mbart", "MBartForConditionalGeneration"),
("megatron-bert", "MegatronBertForMaskedLM"),
("mobilebert", "MobileBertForMaskedLM"),
("mpnet", "MPNetForMaskedLM"),
("mvp", "MvpForConditionalGeneration"),
("nezha", "NezhaForMaskedLM"),
("nystromformer", "NystromformerForMaskedLM"),
("perceiver", "PerceiverForMaskedLM"),
("qdqbert", "QDQBertForMaskedLM"),
("reformer", "ReformerForMaskedLM"),
("rembert", "RemBertForMaskedLM"),
("roberta", "RobertaForMaskedLM"),
("roc_bert", "RoCBertForMaskedLM"),
("roformer", "RoFormerForMaskedLM"),
("squeezebert", "SqueezeBertForMaskedLM"),
("tapas", "TapasForMaskedLM"),
("wav2vec2", "Wav2Vec2ForMaskedLM"),
("xlm", "XLMWithLMHeadModel"),
("xlm-roberta", "XLMRobertaForMaskedLM"),
("xlm-roberta-xl", "XLMRobertaXLForMaskedLM"),
("yoso", "YosoForMaskedLM"),
]
)
MODEL_FOR_OBJECT_DETECTION_MAPPING_NAMES = OrderedDict(
[
# Model for Object Detection mapping
("conditional_detr", "ConditionalDetrForObjectDetection"),
("deformable_detr", "DeformableDetrForObjectDetection"),
("detr", "DetrForObjectDetection"),
("table-transformer", "TableTransformerForObjectDetection"),
("yolos", "YolosForObjectDetection"),
]
)
MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING_NAMES = OrderedDict(
[
# Model for Zero Shot Object Detection mapping
("owlvit", "OwlViTForObjectDetection")
]
)
MODEL_FOR_DEPTH_ESTIMATION_MAPPING_NAMES = OrderedDict(
[
# Model for depth estimation mapping
("dpt", "DPTForDepthEstimation"),
("glpn", "GLPNForDepthEstimation"),
]
)
MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Seq2Seq Causal LM mapping
("bart", "BartForConditionalGeneration"),
("bigbird_pegasus", "BigBirdPegasusForConditionalGeneration"),
("blenderbot", "BlenderbotForConditionalGeneration"),
("blenderbot-small", "BlenderbotSmallForConditionalGeneration"),
("encoder-decoder", "EncoderDecoderModel"),
("fsmt", "FSMTForConditionalGeneration"),
("led", "LEDForConditionalGeneration"),
("longt5", "LongT5ForConditionalGeneration"),
("m2m_100", "M2M100ForConditionalGeneration"),
("marian", "MarianMTModel"),
("mbart", "MBartForConditionalGeneration"),
("mt5", "MT5ForConditionalGeneration"),
("mvp", "MvpForConditionalGeneration"),
("nllb", "M2M100ForConditionalGeneration"),
("pegasus", "PegasusForConditionalGeneration"),
("pegasus_x", "PegasusXForConditionalGeneration"),
("plbart", "PLBartForConditionalGeneration"),
("prophetnet", "ProphetNetForConditionalGeneration"),
("switch_transformers", "SwitchTransformersForConditionalGeneration"),
("t5", "T5ForConditionalGeneration"),
("xlm-prophetnet", "XLMProphetNetForConditionalGeneration"),
]
)
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("speech-encoder-decoder", "SpeechEncoderDecoderModel"),
("speech_to_text", "Speech2TextForConditionalGeneration"),
("whisper", "WhisperForConditionalGeneration"),
]
)
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Sequence Classification mapping
("albert", "AlbertForSequenceClassification"),
("bart", "BartForSequenceClassification"),
("bert", "BertForSequenceClassification"),
("big_bird", "BigBirdForSequenceClassification"),
("bigbird_pegasus", "BigBirdPegasusForSequenceClassification"),
("bloom", "BloomForSequenceClassification"),
("camembert", "CamembertForSequenceClassification"),
("canine", "CanineForSequenceClassification"),
("convbert", "ConvBertForSequenceClassification"),
("ctrl", "CTRLForSequenceClassification"),
("data2vec-text", "Data2VecTextForSequenceClassification"),
("deberta", "DebertaForSequenceClassification"),
("deberta-v2", "DebertaV2ForSequenceClassification"),
("distilbert", "DistilBertForSequenceClassification"),
("electra", "ElectraForSequenceClassification"),
("ernie", "ErnieForSequenceClassification"),
("esm", "EsmForSequenceClassification"),
("flaubert", "FlaubertForSequenceClassification"),
("fnet", "FNetForSequenceClassification"),
("funnel", "FunnelForSequenceClassification"),
("gpt2", "GPT2ForSequenceClassification"),
("gpt_neo", "GPTNeoForSequenceClassification"),
("gptj", "GPTJForSequenceClassification"),
("ibert", "IBertForSequenceClassification"),
("layoutlm", "LayoutLMForSequenceClassification"),
("layoutlmv2", "LayoutLMv2ForSequenceClassification"),
("layoutlmv3", "LayoutLMv3ForSequenceClassification"),
("led", "LEDForSequenceClassification"),
("lilt", "LiltForSequenceClassification"),
("longformer", "LongformerForSequenceClassification"),
("luke", "LukeForSequenceClassification"),
("markuplm", "MarkupLMForSequenceClassification"),
("mbart", "MBartForSequenceClassification"),
("megatron-bert", "MegatronBertForSequenceClassification"),
("mobilebert", "MobileBertForSequenceClassification"),
("mpnet", "MPNetForSequenceClassification"),
("mvp", "MvpForSequenceClassification"),
("nezha", "NezhaForSequenceClassification"),
("nystromformer", "NystromformerForSequenceClassification"),
("openai-gpt", "OpenAIGPTForSequenceClassification"),
("opt", "OPTForSequenceClassification"),
("perceiver", "PerceiverForSequenceClassification"),
("plbart", "PLBartForSequenceClassification"),
("qdqbert", "QDQBertForSequenceClassification"),
("reformer", "ReformerForSequenceClassification"),
("rembert", "RemBertForSequenceClassification"),
("roberta", "RobertaForSequenceClassification"),
("roc_bert", "RoCBertForSequenceClassification"),
("roformer", "RoFormerForSequenceClassification"),
("squeezebert", "SqueezeBertForSequenceClassification"),
("tapas", "TapasForSequenceClassification"),
("transfo-xl", "TransfoXLForSequenceClassification"),
("xlm", "XLMForSequenceClassification"),
("xlm-roberta", "XLMRobertaForSequenceClassification"),
("xlm-roberta-xl", "XLMRobertaXLForSequenceClassification"),
("xlnet", "XLNetForSequenceClassification"),
("yoso", "YosoForSequenceClassification"),
]
)
MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Question Answering mapping
("albert", "AlbertForQuestionAnswering"),
("bart", "BartForQuestionAnswering"),
("bert", "BertForQuestionAnswering"),
("big_bird", "BigBirdForQuestionAnswering"),
("bigbird_pegasus", "BigBirdPegasusForQuestionAnswering"),
("bloom", "BloomForQuestionAnswering"),
("camembert", "CamembertForQuestionAnswering"),
("canine", "CanineForQuestionAnswering"),
("convbert", "ConvBertForQuestionAnswering"),
("data2vec-text", "Data2VecTextForQuestionAnswering"),
("deberta", "DebertaForQuestionAnswering"),
("deberta-v2", "DebertaV2ForQuestionAnswering"),
("distilbert", "DistilBertForQuestionAnswering"),
("electra", "ElectraForQuestionAnswering"),
("ernie", "ErnieForQuestionAnswering"),
("flaubert", "FlaubertForQuestionAnsweringSimple"),
("fnet", "FNetForQuestionAnswering"),
("funnel", "FunnelForQuestionAnswering"),
("gptj", "GPTJForQuestionAnswering"),
("ibert", "IBertForQuestionAnswering"),
("layoutlmv2", "LayoutLMv2ForQuestionAnswering"),
("layoutlmv3", "LayoutLMv3ForQuestionAnswering"),
("led", "LEDForQuestionAnswering"),
("lilt", "LiltForQuestionAnswering"),
("longformer", "LongformerForQuestionAnswering"),
("luke", "LukeForQuestionAnswering"),
("lxmert", "LxmertForQuestionAnswering"),
("markuplm", "MarkupLMForQuestionAnswering"),
("mbart", "MBartForQuestionAnswering"),
("megatron-bert", "MegatronBertForQuestionAnswering"),
("mobilebert", "MobileBertForQuestionAnswering"),
("mpnet", "MPNetForQuestionAnswering"),
("mvp", "MvpForQuestionAnswering"),
("nezha", "NezhaForQuestionAnswering"),
("nystromformer", "NystromformerForQuestionAnswering"),
("opt", "OPTForQuestionAnswering"),
("qdqbert", "QDQBertForQuestionAnswering"),
("reformer", "ReformerForQuestionAnswering"),
("rembert", "RemBertForQuestionAnswering"),
("roberta", "RobertaForQuestionAnswering"),
("roc_bert", "RoCBertForQuestionAnswering"),
("roformer", "RoFormerForQuestionAnswering"),
("splinter", "SplinterForQuestionAnswering"),
("squeezebert", "SqueezeBertForQuestionAnswering"),
("xlm", "XLMForQuestionAnsweringSimple"),
("xlm-roberta", "XLMRobertaForQuestionAnswering"),
("xlm-roberta-xl", "XLMRobertaXLForQuestionAnswering"),
("xlnet", "XLNetForQuestionAnsweringSimple"),
("yoso", "YosoForQuestionAnswering"),
]
)
MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Table Question Answering mapping
("tapas", "TapasForQuestionAnswering"),
]
)
MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
("vilt", "ViltForQuestionAnswering"),
]
)
MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
("layoutlm", "LayoutLMForQuestionAnswering"),
("layoutlmv2", "LayoutLMv2ForQuestionAnswering"),
("layoutlmv3", "LayoutLMv3ForQuestionAnswering"),
]
)
MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Token Classification mapping
("albert", "AlbertForTokenClassification"),
("bert", "BertForTokenClassification"),
("big_bird", "BigBirdForTokenClassification"),
("bloom", "BloomForTokenClassification"),
("camembert", "CamembertForTokenClassification"),
("canine", "CanineForTokenClassification"),
("convbert", "ConvBertForTokenClassification"),
("data2vec-text", "Data2VecTextForTokenClassification"),
("deberta", "DebertaForTokenClassification"),
("deberta-v2", "DebertaV2ForTokenClassification"),
("distilbert", "DistilBertForTokenClassification"),
("electra", "ElectraForTokenClassification"),
("ernie", "ErnieForTokenClassification"),
("esm", "EsmForTokenClassification"),
("flaubert", "FlaubertForTokenClassification"),
("fnet", "FNetForTokenClassification"),
("funnel", "FunnelForTokenClassification"),
("gpt2", "GPT2ForTokenClassification"),
("ibert", "IBertForTokenClassification"),
("layoutlm", "LayoutLMForTokenClassification"),
("layoutlmv2", "LayoutLMv2ForTokenClassification"),
("layoutlmv3", "LayoutLMv3ForTokenClassification"),
("lilt", "LiltForTokenClassification"),
("longformer", "LongformerForTokenClassification"),
("luke", "LukeForTokenClassification"),
("markuplm", "MarkupLMForTokenClassification"),
("megatron-bert", "MegatronBertForTokenClassification"),
("mobilebert", "MobileBertForTokenClassification"),
("mpnet", "MPNetForTokenClassification"),
("nezha", "NezhaForTokenClassification"),
("nystromformer", "NystromformerForTokenClassification"),
("qdqbert", "QDQBertForTokenClassification"),
("rembert", "RemBertForTokenClassification"),
("roberta", "RobertaForTokenClassification"),
("roc_bert", "RoCBertForTokenClassification"),
("roformer", "RoFormerForTokenClassification"),
("squeezebert", "SqueezeBertForTokenClassification"),
("xlm", "XLMForTokenClassification"),
("xlm-roberta", "XLMRobertaForTokenClassification"),
("xlm-roberta-xl", "XLMRobertaXLForTokenClassification"),
("xlnet", "XLNetForTokenClassification"),
("yoso", "YosoForTokenClassification"),
]
)
MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES = OrderedDict(
[
# Model for Multiple Choice mapping
("albert", "AlbertForMultipleChoice"),
("bert", "BertForMultipleChoice"),
("big_bird", "BigBirdForMultipleChoice"),
("camembert", "CamembertForMultipleChoice"),
("canine", "CanineForMultipleChoice"),
("convbert", "ConvBertForMultipleChoice"),
("data2vec-text", "Data2VecTextForMultipleChoice"),
("deberta-v2", "DebertaV2ForMultipleChoice"),
("distilbert", "DistilBertForMultipleChoice"),
("electra", "ElectraForMultipleChoice"),
("ernie", "ErnieForMultipleChoice"),
("flaubert", "FlaubertForMultipleChoice"),
("fnet", "FNetForMultipleChoice"),
("funnel", "FunnelForMultipleChoice"),
("ibert", "IBertForMultipleChoice"),
("longformer", "LongformerForMultipleChoice"),
("luke", "LukeForMultipleChoice"),
("megatron-bert", "MegatronBertForMultipleChoice"),
("mobilebert", "MobileBertForMultipleChoice"),
("mpnet", "MPNetForMultipleChoice"),
("nezha", "NezhaForMultipleChoice"),
("nystromformer", "NystromformerForMultipleChoice"),
("qdqbert", "QDQBertForMultipleChoice"),
("rembert", "RemBertForMultipleChoice"),
("roberta", "RobertaForMultipleChoice"),
("roc_bert", "RoCBertForMultipleChoice"),
("roformer", "RoFormerForMultipleChoice"),
("squeezebert", "SqueezeBertForMultipleChoice"),
("xlm", "XLMForMultipleChoice"),
("xlm-roberta", "XLMRobertaForMultipleChoice"),
("xlm-roberta-xl", "XLMRobertaXLForMultipleChoice"),
("xlnet", "XLNetForMultipleChoice"),
("yoso", "YosoForMultipleChoice"),
]
)
MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES = OrderedDict(
[
("bert", "BertForNextSentencePrediction"),
("ernie", "ErnieForNextSentencePrediction"),
("fnet", "FNetForNextSentencePrediction"),
("megatron-bert", "MegatronBertForNextSentencePrediction"),
("mobilebert", "MobileBertForNextSentencePrediction"),
("nezha", "NezhaForNextSentencePrediction"),
("qdqbert", "QDQBertForNextSentencePrediction"),
]
)
MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Audio Classification mapping
("audio-spectrogram-transformer", "ASTForAudioClassification"),
("data2vec-audio", "Data2VecAudioForSequenceClassification"),
("hubert", "HubertForSequenceClassification"),
("sew", "SEWForSequenceClassification"),
("sew-d", "SEWDForSequenceClassification"),
("unispeech", "UniSpeechForSequenceClassification"),
("unispeech-sat", "UniSpeechSatForSequenceClassification"),
("wav2vec2", "Wav2Vec2ForSequenceClassification"),
("wav2vec2-conformer", "Wav2Vec2ConformerForSequenceClassification"),
("wavlm", "WavLMForSequenceClassification"),
]
)
MODEL_FOR_CTC_MAPPING_NAMES = OrderedDict(
[
# Model for Connectionist temporal classification (CTC) mapping
("data2vec-audio", "Data2VecAudioForCTC"),
("hubert", "HubertForCTC"),
("mctct", "MCTCTForCTC"),
("sew", "SEWForCTC"),
("sew-d", "SEWDForCTC"),
("unispeech", "UniSpeechForCTC"),
("unispeech-sat", "UniSpeechSatForCTC"),
("wav2vec2", "Wav2Vec2ForCTC"),
("wav2vec2-conformer", "Wav2Vec2ConformerForCTC"),
("wavlm", "WavLMForCTC"),
]
)
MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Audio Classification mapping
("data2vec-audio", "Data2VecAudioForAudioFrameClassification"),
("unispeech-sat", "UniSpeechSatForAudioFrameClassification"),
("wav2vec2", "Wav2Vec2ForAudioFrameClassification"),
("wav2vec2-conformer", "Wav2Vec2ConformerForAudioFrameClassification"),
("wavlm", "WavLMForAudioFrameClassification"),
]
)
MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMES = OrderedDict(
[
# Model for Audio Classification mapping
("data2vec-audio", "Data2VecAudioForXVector"),
("unispeech-sat", "UniSpeechSatForXVector"),
("wav2vec2", "Wav2Vec2ForXVector"),
("wav2vec2-conformer", "Wav2Vec2ConformerForXVector"),
("wavlm", "WavLMForXVector"),
]
)
_MODEL_FOR_ZERO_SHOT_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Zero Shot Image Classification mapping
("chinese_clip", "ChineseCLIPModel"),
("clip", "CLIPModel"),
("clipseg", "CLIPSegModel"),
]
)
MODEL_FOR_BACKBONE_MAPPING_NAMES = OrderedDict(
[
# Backbone mapping
("bit", "BitBackbone"),
("maskformer-swin", "MaskFormerSwinBackbone"),
("resnet", "ResNetBackbone"),
]
)
MODEL_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_MAPPING_NAMES)
MODEL_FOR_PRETRAINING_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_PRETRAINING_MAPPING_NAMES)
MODEL_WITH_LM_HEAD_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_WITH_LM_HEAD_MAPPING_NAMES)
MODEL_FOR_CAUSAL_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_CAUSAL_LM_MAPPING_NAMES)
MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING_NAMES
)
MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_IMAGE_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_IMAGE_SEGMENTATION_MAPPING_NAMES
)
MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES
)
MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING_NAMES
)
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_VISION_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES)
MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_MASKED_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_MASKED_LM_MAPPING_NAMES)
MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES
)
MODEL_FOR_OBJECT_DETECTION_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_OBJECT_DETECTION_MAPPING_NAMES)
MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING_NAMES
)
MODEL_FOR_DEPTH_ESTIMATION_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_DEPTH_ESTIMATION_MAPPING_NAMES)
MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES
)
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_MULTIPLE_CHOICE_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES)
MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES
)
MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_CTC_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_CTC_MAPPING_NAMES)
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES)
MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_AUDIO_XVECTOR_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMES)
MODEL_FOR_BACKBONE_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_BACKBONE_MAPPING_NAMES)
class AutoModel(_BaseAutoModelClass):
_model_mapping = MODEL_MAPPING
AutoModel = auto_class_update(AutoModel)
class AutoModelForPreTraining(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_PRETRAINING_MAPPING
AutoModelForPreTraining = auto_class_update(AutoModelForPreTraining, head_doc="pretraining")
# Private on purpose, the public class will add the deprecation warnings.
class _AutoModelWithLMHead(_BaseAutoModelClass):
_model_mapping = MODEL_WITH_LM_HEAD_MAPPING
_AutoModelWithLMHead = auto_class_update(_AutoModelWithLMHead, head_doc="language modeling")
class AutoModelForCausalLM(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_CAUSAL_LM_MAPPING
AutoModelForCausalLM = auto_class_update(AutoModelForCausalLM, head_doc="causal language modeling")
class AutoModelForMaskedLM(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_MASKED_LM_MAPPING
AutoModelForMaskedLM = auto_class_update(AutoModelForMaskedLM, head_doc="masked language modeling")
class AutoModelForSeq2SeqLM(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING
AutoModelForSeq2SeqLM = auto_class_update(
AutoModelForSeq2SeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base"
)
class AutoModelForSequenceClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
AutoModelForSequenceClassification = auto_class_update(
AutoModelForSequenceClassification, head_doc="sequence classification"
)
class AutoModelForQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_QUESTION_ANSWERING_MAPPING
AutoModelForQuestionAnswering = auto_class_update(AutoModelForQuestionAnswering, head_doc="question answering")
class AutoModelForTableQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING
AutoModelForTableQuestionAnswering = auto_class_update(
AutoModelForTableQuestionAnswering,
head_doc="table question answering",
checkpoint_for_example="google/tapas-base-finetuned-wtq",
)
class AutoModelForVisualQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING
AutoModelForVisualQuestionAnswering = auto_class_update(
AutoModelForVisualQuestionAnswering,
head_doc="visual question answering",
checkpoint_for_example="dandelin/vilt-b32-finetuned-vqa",
)
class AutoModelForDocumentQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING
AutoModelForDocumentQuestionAnswering = auto_class_update(
AutoModelForDocumentQuestionAnswering,
head_doc="document question answering",
checkpoint_for_example='impira/layoutlm-document-qa", revision="52e01b3',
)
class AutoModelForTokenClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
AutoModelForTokenClassification = auto_class_update(AutoModelForTokenClassification, head_doc="token classification")
class AutoModelForMultipleChoice(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_MULTIPLE_CHOICE_MAPPING
AutoModelForMultipleChoice = auto_class_update(AutoModelForMultipleChoice, head_doc="multiple choice")
class AutoModelForNextSentencePrediction(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING
AutoModelForNextSentencePrediction = auto_class_update(
AutoModelForNextSentencePrediction, head_doc="next sentence prediction"
)
class AutoModelForImageClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
AutoModelForImageClassification = auto_class_update(AutoModelForImageClassification, head_doc="image classification")
class AutoModelForImageSegmentation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_IMAGE_SEGMENTATION_MAPPING
AutoModelForImageSegmentation = auto_class_update(AutoModelForImageSegmentation, head_doc="image segmentation")
class AutoModelForSemanticSegmentation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING
AutoModelForSemanticSegmentation = auto_class_update(
AutoModelForSemanticSegmentation, head_doc="semantic segmentation"
)
class AutoModelForInstanceSegmentation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING
AutoModelForInstanceSegmentation = auto_class_update(
AutoModelForInstanceSegmentation, head_doc="instance segmentation"
)
class AutoModelForObjectDetection(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_OBJECT_DETECTION_MAPPING
AutoModelForObjectDetection = auto_class_update(AutoModelForObjectDetection, head_doc="object detection")
class AutoModelForZeroShotObjectDetection(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING
AutoModelForZeroShotObjectDetection = auto_class_update(
AutoModelForZeroShotObjectDetection, head_doc="zero-shot object detection"
)
class AutoModelForDepthEstimation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_DEPTH_ESTIMATION_MAPPING
AutoModelForDepthEstimation = auto_class_update(AutoModelForDepthEstimation, head_doc="depth estimation")
class AutoModelForVideoClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING
AutoModelForVideoClassification = auto_class_update(AutoModelForVideoClassification, head_doc="video classification")
class AutoModelForVision2Seq(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_VISION_2_SEQ_MAPPING
AutoModelForVision2Seq = auto_class_update(AutoModelForVision2Seq, head_doc="vision-to-text modeling")
class AutoModelForAudioClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING
AutoModelForAudioClassification = auto_class_update(AutoModelForAudioClassification, head_doc="audio classification")
class AutoModelForCTC(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_CTC_MAPPING
AutoModelForCTC = auto_class_update(AutoModelForCTC, head_doc="connectionist temporal classification")
class AutoModelForSpeechSeq2Seq(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING
AutoModelForSpeechSeq2Seq = auto_class_update(
AutoModelForSpeechSeq2Seq, head_doc="sequence-to-sequence speech-to-text modeling"
)
class AutoModelForAudioFrameClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING
AutoModelForAudioFrameClassification = auto_class_update(
AutoModelForAudioFrameClassification, head_doc="audio frame (token) classification"
)
class AutoModelForAudioXVector(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_AUDIO_XVECTOR_MAPPING
class AutoBackbone(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_BACKBONE_MAPPING
AutoModelForAudioXVector = auto_class_update(AutoModelForAudioXVector, head_doc="audio retrieval via x-vector")
class AutoModelForMaskedImageModeling(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING
AutoModelForMaskedImageModeling = auto_class_update(AutoModelForMaskedImageModeling, head_doc="masked image modeling")
class AutoModelWithLMHead(_AutoModelWithLMHead):
@classmethod
def from_config(cls, config):
warnings.warn(
"The class `AutoModelWithLMHead` is deprecated and will be removed in a future version. Please use "
"`AutoModelForCausalLM` for causal language models, `AutoModelForMaskedLM` for masked language models and "
"`AutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_config(config)
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
warnings.warn(
"The class `AutoModelWithLMHead` is deprecated and will be removed in a future version. Please use "
"`AutoModelForCausalLM` for causal language models, `AutoModelForMaskedLM` for masked language models and "
"`AutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
| # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Model class."""
import warnings
from collections import OrderedDict
from ...utils import logging
from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update
from .configuration_auto import CONFIG_MAPPING_NAMES
logger = logging.get_logger(__name__)
MODEL_MAPPING_NAMES = OrderedDict(
[
# Base model mapping
("albert", "AlbertModel"),
("audio-spectrogram-transformer", "ASTModel"),
("bart", "BartModel"),
("beit", "BeitModel"),
("bert", "BertModel"),
("bert-generation", "BertGenerationEncoder"),
("big_bird", "BigBirdModel"),
("bigbird_pegasus", "BigBirdPegasusModel"),
("biogpt", "BioGptModel"),
("bit", "BitModel"),
("blenderbot", "BlenderbotModel"),
("blenderbot-small", "BlenderbotSmallModel"),
("bloom", "BloomModel"),
("camembert", "CamembertModel"),
("canine", "CanineModel"),
("chinese_clip", "ChineseCLIPModel"),
("clip", "CLIPModel"),
("clipseg", "CLIPSegModel"),
("codegen", "CodeGenModel"),
("conditional_detr", "ConditionalDetrModel"),
("convbert", "ConvBertModel"),
("convnext", "ConvNextModel"),
("ctrl", "CTRLModel"),
("cvt", "CvtModel"),
("data2vec-audio", "Data2VecAudioModel"),
("data2vec-text", "Data2VecTextModel"),
("data2vec-vision", "Data2VecVisionModel"),
("deberta", "DebertaModel"),
("deberta-v2", "DebertaV2Model"),
("decision_transformer", "DecisionTransformerModel"),
("decision_transformer_gpt2", "DecisionTransformerGPT2Model"),
("deformable_detr", "DeformableDetrModel"),
("deit", "DeiTModel"),
("detr", "DetrModel"),
("dinat", "DinatModel"),
("distilbert", "DistilBertModel"),
("donut-swin", "DonutSwinModel"),
("dpr", "DPRQuestionEncoder"),
("dpt", "DPTModel"),
("electra", "ElectraModel"),
("ernie", "ErnieModel"),
("esm", "EsmModel"),
("flaubert", "FlaubertModel"),
("flava", "FlavaModel"),
("fnet", "FNetModel"),
("fsmt", "FSMTModel"),
("funnel", ("FunnelModel", "FunnelBaseModel")),
("glpn", "GLPNModel"),
("gpt-sw3", "GPT2Model"),
("gpt2", "GPT2Model"),
("gpt_neo", "GPTNeoModel"),
("gpt_neox", "GPTNeoXModel"),
("gpt_neox_japanese", "GPTNeoXJapaneseModel"),
("gptj", "GPTJModel"),
("groupvit", "GroupViTModel"),
("hubert", "HubertModel"),
("ibert", "IBertModel"),
("imagegpt", "ImageGPTModel"),
("jukebox", "JukeboxModel"),
("layoutlm", "LayoutLMModel"),
("layoutlmv2", "LayoutLMv2Model"),
("layoutlmv3", "LayoutLMv3Model"),
("led", "LEDModel"),
("levit", "LevitModel"),
("lilt", "LiltModel"),
("longformer", "LongformerModel"),
("longt5", "LongT5Model"),
("luke", "LukeModel"),
("lxmert", "LxmertModel"),
("m2m_100", "M2M100Model"),
("marian", "MarianModel"),
("markuplm", "MarkupLMModel"),
("maskformer", "MaskFormerModel"),
("maskformer-swin", "MaskFormerSwinModel"),
("mbart", "MBartModel"),
("mctct", "MCTCTModel"),
("megatron-bert", "MegatronBertModel"),
("mobilebert", "MobileBertModel"),
("mobilenet_v1", "MobileNetV1Model"),
("mobilenet_v2", "MobileNetV2Model"),
("mobilevit", "MobileViTModel"),
("mpnet", "MPNetModel"),
("mt5", "MT5Model"),
("mvp", "MvpModel"),
("nat", "NatModel"),
("nezha", "NezhaModel"),
("nllb", "M2M100Model"),
("nystromformer", "NystromformerModel"),
("openai-gpt", "OpenAIGPTModel"),
("opt", "OPTModel"),
("owlvit", "OwlViTModel"),
("pegasus", "PegasusModel"),
("pegasus_x", "PegasusXModel"),
("perceiver", "PerceiverModel"),
("plbart", "PLBartModel"),
("poolformer", "PoolFormerModel"),
("prophetnet", "ProphetNetModel"),
("qdqbert", "QDQBertModel"),
("reformer", "ReformerModel"),
("regnet", "RegNetModel"),
("rembert", "RemBertModel"),
("resnet", "ResNetModel"),
("retribert", "RetriBertModel"),
("roberta", "RobertaModel"),
("roc_bert", "RoCBertModel"),
("roformer", "RoFormerModel"),
("segformer", "SegformerModel"),
("sew", "SEWModel"),
("sew-d", "SEWDModel"),
("speech_to_text", "Speech2TextModel"),
("splinter", "SplinterModel"),
("squeezebert", "SqueezeBertModel"),
("swin", "SwinModel"),
("swinv2", "Swinv2Model"),
("switch_transformers", "SwitchTransformersModel"),
("t5", "T5Model"),
("table-transformer", "TableTransformerModel"),
("tapas", "TapasModel"),
("time_series_transformer", "TimeSeriesTransformerModel"),
("timesformer", "TimesformerModel"),
("trajectory_transformer", "TrajectoryTransformerModel"),
("transfo-xl", "TransfoXLModel"),
("unispeech", "UniSpeechModel"),
("unispeech-sat", "UniSpeechSatModel"),
("van", "VanModel"),
("videomae", "VideoMAEModel"),
("vilt", "ViltModel"),
("vision-text-dual-encoder", "VisionTextDualEncoderModel"),
("visual_bert", "VisualBertModel"),
("vit", "ViTModel"),
("vit_hybrid", "ViTHybridModel"),
("vit_mae", "ViTMAEModel"),
("vit_msn", "ViTMSNModel"),
("wav2vec2", "Wav2Vec2Model"),
("wav2vec2-conformer", "Wav2Vec2ConformerModel"),
("wavlm", "WavLMModel"),
("whisper", "WhisperModel"),
("xclip", "XCLIPModel"),
("xglm", "XGLMModel"),
("xlm", "XLMModel"),
("xlm-prophetnet", "XLMProphetNetModel"),
("xlm-roberta", "XLMRobertaModel"),
("xlm-roberta-xl", "XLMRobertaXLModel"),
("xlnet", "XLNetModel"),
("yolos", "YolosModel"),
("yoso", "YosoModel"),
]
)
MODEL_FOR_PRETRAINING_MAPPING_NAMES = OrderedDict(
[
# Model for pre-training mapping
("albert", "AlbertForPreTraining"),
("bart", "BartForConditionalGeneration"),
("bert", "BertForPreTraining"),
("big_bird", "BigBirdForPreTraining"),
("bloom", "BloomForCausalLM"),
("camembert", "CamembertForMaskedLM"),
("ctrl", "CTRLLMHeadModel"),
("data2vec-text", "Data2VecTextForMaskedLM"),
("deberta", "DebertaForMaskedLM"),
("deberta-v2", "DebertaV2ForMaskedLM"),
("distilbert", "DistilBertForMaskedLM"),
("electra", "ElectraForPreTraining"),
("ernie", "ErnieForPreTraining"),
("flaubert", "FlaubertWithLMHeadModel"),
("flava", "FlavaForPreTraining"),
("fnet", "FNetForPreTraining"),
("fsmt", "FSMTForConditionalGeneration"),
("funnel", "FunnelForPreTraining"),
("gpt-sw3", "GPT2LMHeadModel"),
("gpt2", "GPT2LMHeadModel"),
("ibert", "IBertForMaskedLM"),
("layoutlm", "LayoutLMForMaskedLM"),
("longformer", "LongformerForMaskedLM"),
("luke", "LukeForMaskedLM"),
("lxmert", "LxmertForPreTraining"),
("megatron-bert", "MegatronBertForPreTraining"),
("mobilebert", "MobileBertForPreTraining"),
("mpnet", "MPNetForMaskedLM"),
("mvp", "MvpForConditionalGeneration"),
("nezha", "NezhaForPreTraining"),
("openai-gpt", "OpenAIGPTLMHeadModel"),
("retribert", "RetriBertModel"),
("roberta", "RobertaForMaskedLM"),
("roc_bert", "RoCBertForPreTraining"),
("splinter", "SplinterForPreTraining"),
("squeezebert", "SqueezeBertForMaskedLM"),
("switch_transformers", "SwitchTransformersForConditionalGeneration"),
("t5", "T5ForConditionalGeneration"),
("tapas", "TapasForMaskedLM"),
("transfo-xl", "TransfoXLLMHeadModel"),
("unispeech", "UniSpeechForPreTraining"),
("unispeech-sat", "UniSpeechSatForPreTraining"),
("videomae", "VideoMAEForPreTraining"),
("visual_bert", "VisualBertForPreTraining"),
("vit_mae", "ViTMAEForPreTraining"),
("wav2vec2", "Wav2Vec2ForPreTraining"),
("wav2vec2-conformer", "Wav2Vec2ConformerForPreTraining"),
("xlm", "XLMWithLMHeadModel"),
("xlm-roberta", "XLMRobertaForMaskedLM"),
("xlm-roberta-xl", "XLMRobertaXLForMaskedLM"),
("xlnet", "XLNetLMHeadModel"),
]
)
MODEL_WITH_LM_HEAD_MAPPING_NAMES = OrderedDict(
[
# Model with LM heads mapping
("albert", "AlbertForMaskedLM"),
("bart", "BartForConditionalGeneration"),
("bert", "BertForMaskedLM"),
("big_bird", "BigBirdForMaskedLM"),
("bigbird_pegasus", "BigBirdPegasusForConditionalGeneration"),
("blenderbot-small", "BlenderbotSmallForConditionalGeneration"),
("bloom", "BloomForCausalLM"),
("camembert", "CamembertForMaskedLM"),
("codegen", "CodeGenForCausalLM"),
("convbert", "ConvBertForMaskedLM"),
("ctrl", "CTRLLMHeadModel"),
("data2vec-text", "Data2VecTextForMaskedLM"),
("deberta", "DebertaForMaskedLM"),
("deberta-v2", "DebertaV2ForMaskedLM"),
("distilbert", "DistilBertForMaskedLM"),
("electra", "ElectraForMaskedLM"),
("encoder-decoder", "EncoderDecoderModel"),
("ernie", "ErnieForMaskedLM"),
("esm", "EsmForMaskedLM"),
("flaubert", "FlaubertWithLMHeadModel"),
("fnet", "FNetForMaskedLM"),
("fsmt", "FSMTForConditionalGeneration"),
("funnel", "FunnelForMaskedLM"),
("gpt-sw3", "GPT2LMHeadModel"),
("gpt2", "GPT2LMHeadModel"),
("gpt_neo", "GPTNeoForCausalLM"),
("gpt_neox", "GPTNeoXForCausalLM"),
("gpt_neox_japanese", "GPTNeoXJapaneseForCausalLM"),
("gptj", "GPTJForCausalLM"),
("ibert", "IBertForMaskedLM"),
("layoutlm", "LayoutLMForMaskedLM"),
("led", "LEDForConditionalGeneration"),
("longformer", "LongformerForMaskedLM"),
("longt5", "LongT5ForConditionalGeneration"),
("luke", "LukeForMaskedLM"),
("m2m_100", "M2M100ForConditionalGeneration"),
("marian", "MarianMTModel"),
("megatron-bert", "MegatronBertForCausalLM"),
("mobilebert", "MobileBertForMaskedLM"),
("mpnet", "MPNetForMaskedLM"),
("mvp", "MvpForConditionalGeneration"),
("nezha", "NezhaForMaskedLM"),
("nllb", "M2M100ForConditionalGeneration"),
("nystromformer", "NystromformerForMaskedLM"),
("openai-gpt", "OpenAIGPTLMHeadModel"),
("pegasus_x", "PegasusXForConditionalGeneration"),
("plbart", "PLBartForConditionalGeneration"),
("qdqbert", "QDQBertForMaskedLM"),
("reformer", "ReformerModelWithLMHead"),
("rembert", "RemBertForMaskedLM"),
("roberta", "RobertaForMaskedLM"),
("roc_bert", "RoCBertForMaskedLM"),
("roformer", "RoFormerForMaskedLM"),
("speech_to_text", "Speech2TextForConditionalGeneration"),
("squeezebert", "SqueezeBertForMaskedLM"),
("switch_transformers", "SwitchTransformersForConditionalGeneration"),
("t5", "T5ForConditionalGeneration"),
("tapas", "TapasForMaskedLM"),
("transfo-xl", "TransfoXLLMHeadModel"),
("wav2vec2", "Wav2Vec2ForMaskedLM"),
("whisper", "WhisperForConditionalGeneration"),
("xlm", "XLMWithLMHeadModel"),
("xlm-roberta", "XLMRobertaForMaskedLM"),
("xlm-roberta-xl", "XLMRobertaXLForMaskedLM"),
("xlnet", "XLNetLMHeadModel"),
("yoso", "YosoForMaskedLM"),
]
)
MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Causal LM mapping
("bart", "BartForCausalLM"),
("bert", "BertLMHeadModel"),
("bert-generation", "BertGenerationDecoder"),
("big_bird", "BigBirdForCausalLM"),
("bigbird_pegasus", "BigBirdPegasusForCausalLM"),
("biogpt", "BioGptForCausalLM"),
("blenderbot", "BlenderbotForCausalLM"),
("blenderbot-small", "BlenderbotSmallForCausalLM"),
("bloom", "BloomForCausalLM"),
("camembert", "CamembertForCausalLM"),
("codegen", "CodeGenForCausalLM"),
("ctrl", "CTRLLMHeadModel"),
("data2vec-text", "Data2VecTextForCausalLM"),
("electra", "ElectraForCausalLM"),
("ernie", "ErnieForCausalLM"),
("gpt-sw3", "GPT2LMHeadModel"),
("gpt2", "GPT2LMHeadModel"),
("gpt_neo", "GPTNeoForCausalLM"),
("gpt_neox", "GPTNeoXForCausalLM"),
("gpt_neox_japanese", "GPTNeoXJapaneseForCausalLM"),
("gptj", "GPTJForCausalLM"),
("marian", "MarianForCausalLM"),
("mbart", "MBartForCausalLM"),
("megatron-bert", "MegatronBertForCausalLM"),
("mvp", "MvpForCausalLM"),
("openai-gpt", "OpenAIGPTLMHeadModel"),
("opt", "OPTForCausalLM"),
("pegasus", "PegasusForCausalLM"),
("plbart", "PLBartForCausalLM"),
("prophetnet", "ProphetNetForCausalLM"),
("qdqbert", "QDQBertLMHeadModel"),
("reformer", "ReformerModelWithLMHead"),
("rembert", "RemBertForCausalLM"),
("roberta", "RobertaForCausalLM"),
("roc_bert", "RoCBertForCausalLM"),
("roformer", "RoFormerForCausalLM"),
("speech_to_text_2", "Speech2Text2ForCausalLM"),
("transfo-xl", "TransfoXLLMHeadModel"),
("trocr", "TrOCRForCausalLM"),
("xglm", "XGLMForCausalLM"),
("xlm", "XLMWithLMHeadModel"),
("xlm-prophetnet", "XLMProphetNetForCausalLM"),
("xlm-roberta", "XLMRobertaForCausalLM"),
("xlm-roberta-xl", "XLMRobertaXLForCausalLM"),
("xlnet", "XLNetLMHeadModel"),
]
)
MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
[
("deit", "DeiTForMaskedImageModeling"),
("swin", "SwinForMaskedImageModeling"),
("swinv2", "Swinv2ForMaskedImageModeling"),
("vit", "ViTForMaskedImageModeling"),
]
)
MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
# Model for Causal Image Modeling mapping
[
("imagegpt", "ImageGPTForCausalImageModeling"),
]
)
MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Image Classification mapping
("beit", "BeitForImageClassification"),
("bit", "BitForImageClassification"),
("convnext", "ConvNextForImageClassification"),
("cvt", "CvtForImageClassification"),
("data2vec-vision", "Data2VecVisionForImageClassification"),
("deit", ("DeiTForImageClassification", "DeiTForImageClassificationWithTeacher")),
("dinat", "DinatForImageClassification"),
("imagegpt", "ImageGPTForImageClassification"),
("levit", ("LevitForImageClassification", "LevitForImageClassificationWithTeacher")),
("mobilenet_v1", "MobileNetV1ForImageClassification"),
("mobilenet_v2", "MobileNetV2ForImageClassification"),
("mobilevit", "MobileViTForImageClassification"),
("nat", "NatForImageClassification"),
(
"perceiver",
(
"PerceiverForImageClassificationLearned",
"PerceiverForImageClassificationFourier",
"PerceiverForImageClassificationConvProcessing",
),
),
("poolformer", "PoolFormerForImageClassification"),
("regnet", "RegNetForImageClassification"),
("resnet", "ResNetForImageClassification"),
("segformer", "SegformerForImageClassification"),
("swin", "SwinForImageClassification"),
("swinv2", "Swinv2ForImageClassification"),
("van", "VanForImageClassification"),
("vit", "ViTForImageClassification"),
("vit_hybrid", "ViTHybridForImageClassification"),
("vit_msn", "ViTMSNForImageClassification"),
]
)
MODEL_FOR_IMAGE_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Do not add new models here, this class will be deprecated in the future.
# Model for Image Segmentation mapping
("detr", "DetrForSegmentation"),
]
)
MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Model for Semantic Segmentation mapping
("beit", "BeitForSemanticSegmentation"),
("data2vec-vision", "Data2VecVisionForSemanticSegmentation"),
("dpt", "DPTForSemanticSegmentation"),
("mobilenet_v2", "MobileNetV2ForSemanticSegmentation"),
("mobilevit", "MobileViTForSemanticSegmentation"),
("segformer", "SegformerForSemanticSegmentation"),
]
)
MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Model for Instance Segmentation mapping
("maskformer", "MaskFormerForInstanceSegmentation"),
]
)
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
("timesformer", "TimesformerForVideoClassification"),
("videomae", "VideoMAEForVideoClassification"),
]
)
MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("vision-encoder-decoder", "VisionEncoderDecoderModel"),
]
)
MODEL_FOR_MASKED_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Masked LM mapping
("albert", "AlbertForMaskedLM"),
("bart", "BartForConditionalGeneration"),
("bert", "BertForMaskedLM"),
("big_bird", "BigBirdForMaskedLM"),
("camembert", "CamembertForMaskedLM"),
("convbert", "ConvBertForMaskedLM"),
("data2vec-text", "Data2VecTextForMaskedLM"),
("deberta", "DebertaForMaskedLM"),
("deberta-v2", "DebertaV2ForMaskedLM"),
("distilbert", "DistilBertForMaskedLM"),
("electra", "ElectraForMaskedLM"),
("ernie", "ErnieForMaskedLM"),
("esm", "EsmForMaskedLM"),
("flaubert", "FlaubertWithLMHeadModel"),
("fnet", "FNetForMaskedLM"),
("funnel", "FunnelForMaskedLM"),
("ibert", "IBertForMaskedLM"),
("layoutlm", "LayoutLMForMaskedLM"),
("longformer", "LongformerForMaskedLM"),
("luke", "LukeForMaskedLM"),
("mbart", "MBartForConditionalGeneration"),
("megatron-bert", "MegatronBertForMaskedLM"),
("mobilebert", "MobileBertForMaskedLM"),
("mpnet", "MPNetForMaskedLM"),
("mvp", "MvpForConditionalGeneration"),
("nezha", "NezhaForMaskedLM"),
("nystromformer", "NystromformerForMaskedLM"),
("perceiver", "PerceiverForMaskedLM"),
("qdqbert", "QDQBertForMaskedLM"),
("reformer", "ReformerForMaskedLM"),
("rembert", "RemBertForMaskedLM"),
("roberta", "RobertaForMaskedLM"),
("roc_bert", "RoCBertForMaskedLM"),
("roformer", "RoFormerForMaskedLM"),
("squeezebert", "SqueezeBertForMaskedLM"),
("tapas", "TapasForMaskedLM"),
("wav2vec2", "Wav2Vec2ForMaskedLM"),
("xlm", "XLMWithLMHeadModel"),
("xlm-roberta", "XLMRobertaForMaskedLM"),
("xlm-roberta-xl", "XLMRobertaXLForMaskedLM"),
("yoso", "YosoForMaskedLM"),
]
)
MODEL_FOR_OBJECT_DETECTION_MAPPING_NAMES = OrderedDict(
[
# Model for Object Detection mapping
("conditional_detr", "ConditionalDetrForObjectDetection"),
("deformable_detr", "DeformableDetrForObjectDetection"),
("detr", "DetrForObjectDetection"),
("table-transformer", "TableTransformerForObjectDetection"),
("yolos", "YolosForObjectDetection"),
]
)
MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING_NAMES = OrderedDict(
[
# Model for Zero Shot Object Detection mapping
("owlvit", "OwlViTForObjectDetection")
]
)
MODEL_FOR_DEPTH_ESTIMATION_MAPPING_NAMES = OrderedDict(
[
# Model for depth estimation mapping
("dpt", "DPTForDepthEstimation"),
("glpn", "GLPNForDepthEstimation"),
]
)
MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Seq2Seq Causal LM mapping
("bart", "BartForConditionalGeneration"),
("bigbird_pegasus", "BigBirdPegasusForConditionalGeneration"),
("blenderbot", "BlenderbotForConditionalGeneration"),
("blenderbot-small", "BlenderbotSmallForConditionalGeneration"),
("encoder-decoder", "EncoderDecoderModel"),
("fsmt", "FSMTForConditionalGeneration"),
("led", "LEDForConditionalGeneration"),
("longt5", "LongT5ForConditionalGeneration"),
("m2m_100", "M2M100ForConditionalGeneration"),
("marian", "MarianMTModel"),
("mbart", "MBartForConditionalGeneration"),
("mt5", "MT5ForConditionalGeneration"),
("mvp", "MvpForConditionalGeneration"),
("nllb", "M2M100ForConditionalGeneration"),
("pegasus", "PegasusForConditionalGeneration"),
("pegasus_x", "PegasusXForConditionalGeneration"),
("plbart", "PLBartForConditionalGeneration"),
("prophetnet", "ProphetNetForConditionalGeneration"),
("switch_transformers", "SwitchTransformersForConditionalGeneration"),
("t5", "T5ForConditionalGeneration"),
("xlm-prophetnet", "XLMProphetNetForConditionalGeneration"),
]
)
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("speech-encoder-decoder", "SpeechEncoderDecoderModel"),
("speech_to_text", "Speech2TextForConditionalGeneration"),
("whisper", "WhisperForConditionalGeneration"),
]
)
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Sequence Classification mapping
("albert", "AlbertForSequenceClassification"),
("bart", "BartForSequenceClassification"),
("bert", "BertForSequenceClassification"),
("big_bird", "BigBirdForSequenceClassification"),
("bigbird_pegasus", "BigBirdPegasusForSequenceClassification"),
("bloom", "BloomForSequenceClassification"),
("camembert", "CamembertForSequenceClassification"),
("canine", "CanineForSequenceClassification"),
("convbert", "ConvBertForSequenceClassification"),
("ctrl", "CTRLForSequenceClassification"),
("data2vec-text", "Data2VecTextForSequenceClassification"),
("deberta", "DebertaForSequenceClassification"),
("deberta-v2", "DebertaV2ForSequenceClassification"),
("distilbert", "DistilBertForSequenceClassification"),
("electra", "ElectraForSequenceClassification"),
("ernie", "ErnieForSequenceClassification"),
("esm", "EsmForSequenceClassification"),
("flaubert", "FlaubertForSequenceClassification"),
("fnet", "FNetForSequenceClassification"),
("funnel", "FunnelForSequenceClassification"),
("gpt-sw3", "GPT2ForSequenceClassification"),
("gpt2", "GPT2ForSequenceClassification"),
("gpt_neo", "GPTNeoForSequenceClassification"),
("gptj", "GPTJForSequenceClassification"),
("ibert", "IBertForSequenceClassification"),
("layoutlm", "LayoutLMForSequenceClassification"),
("layoutlmv2", "LayoutLMv2ForSequenceClassification"),
("layoutlmv3", "LayoutLMv3ForSequenceClassification"),
("led", "LEDForSequenceClassification"),
("lilt", "LiltForSequenceClassification"),
("longformer", "LongformerForSequenceClassification"),
("luke", "LukeForSequenceClassification"),
("markuplm", "MarkupLMForSequenceClassification"),
("mbart", "MBartForSequenceClassification"),
("megatron-bert", "MegatronBertForSequenceClassification"),
("mobilebert", "MobileBertForSequenceClassification"),
("mpnet", "MPNetForSequenceClassification"),
("mvp", "MvpForSequenceClassification"),
("nezha", "NezhaForSequenceClassification"),
("nystromformer", "NystromformerForSequenceClassification"),
("openai-gpt", "OpenAIGPTForSequenceClassification"),
("opt", "OPTForSequenceClassification"),
("perceiver", "PerceiverForSequenceClassification"),
("plbart", "PLBartForSequenceClassification"),
("qdqbert", "QDQBertForSequenceClassification"),
("reformer", "ReformerForSequenceClassification"),
("rembert", "RemBertForSequenceClassification"),
("roberta", "RobertaForSequenceClassification"),
("roc_bert", "RoCBertForSequenceClassification"),
("roformer", "RoFormerForSequenceClassification"),
("squeezebert", "SqueezeBertForSequenceClassification"),
("tapas", "TapasForSequenceClassification"),
("transfo-xl", "TransfoXLForSequenceClassification"),
("xlm", "XLMForSequenceClassification"),
("xlm-roberta", "XLMRobertaForSequenceClassification"),
("xlm-roberta-xl", "XLMRobertaXLForSequenceClassification"),
("xlnet", "XLNetForSequenceClassification"),
("yoso", "YosoForSequenceClassification"),
]
)
MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Question Answering mapping
("albert", "AlbertForQuestionAnswering"),
("bart", "BartForQuestionAnswering"),
("bert", "BertForQuestionAnswering"),
("big_bird", "BigBirdForQuestionAnswering"),
("bigbird_pegasus", "BigBirdPegasusForQuestionAnswering"),
("bloom", "BloomForQuestionAnswering"),
("camembert", "CamembertForQuestionAnswering"),
("canine", "CanineForQuestionAnswering"),
("convbert", "ConvBertForQuestionAnswering"),
("data2vec-text", "Data2VecTextForQuestionAnswering"),
("deberta", "DebertaForQuestionAnswering"),
("deberta-v2", "DebertaV2ForQuestionAnswering"),
("distilbert", "DistilBertForQuestionAnswering"),
("electra", "ElectraForQuestionAnswering"),
("ernie", "ErnieForQuestionAnswering"),
("flaubert", "FlaubertForQuestionAnsweringSimple"),
("fnet", "FNetForQuestionAnswering"),
("funnel", "FunnelForQuestionAnswering"),
("gptj", "GPTJForQuestionAnswering"),
("ibert", "IBertForQuestionAnswering"),
("layoutlmv2", "LayoutLMv2ForQuestionAnswering"),
("layoutlmv3", "LayoutLMv3ForQuestionAnswering"),
("led", "LEDForQuestionAnswering"),
("lilt", "LiltForQuestionAnswering"),
("longformer", "LongformerForQuestionAnswering"),
("luke", "LukeForQuestionAnswering"),
("lxmert", "LxmertForQuestionAnswering"),
("markuplm", "MarkupLMForQuestionAnswering"),
("mbart", "MBartForQuestionAnswering"),
("megatron-bert", "MegatronBertForQuestionAnswering"),
("mobilebert", "MobileBertForQuestionAnswering"),
("mpnet", "MPNetForQuestionAnswering"),
("mvp", "MvpForQuestionAnswering"),
("nezha", "NezhaForQuestionAnswering"),
("nystromformer", "NystromformerForQuestionAnswering"),
("opt", "OPTForQuestionAnswering"),
("qdqbert", "QDQBertForQuestionAnswering"),
("reformer", "ReformerForQuestionAnswering"),
("rembert", "RemBertForQuestionAnswering"),
("roberta", "RobertaForQuestionAnswering"),
("roc_bert", "RoCBertForQuestionAnswering"),
("roformer", "RoFormerForQuestionAnswering"),
("splinter", "SplinterForQuestionAnswering"),
("squeezebert", "SqueezeBertForQuestionAnswering"),
("xlm", "XLMForQuestionAnsweringSimple"),
("xlm-roberta", "XLMRobertaForQuestionAnswering"),
("xlm-roberta-xl", "XLMRobertaXLForQuestionAnswering"),
("xlnet", "XLNetForQuestionAnsweringSimple"),
("yoso", "YosoForQuestionAnswering"),
]
)
MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Table Question Answering mapping
("tapas", "TapasForQuestionAnswering"),
]
)
MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
("vilt", "ViltForQuestionAnswering"),
]
)
MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
("layoutlm", "LayoutLMForQuestionAnswering"),
("layoutlmv2", "LayoutLMv2ForQuestionAnswering"),
("layoutlmv3", "LayoutLMv3ForQuestionAnswering"),
]
)
MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Token Classification mapping
("albert", "AlbertForTokenClassification"),
("bert", "BertForTokenClassification"),
("big_bird", "BigBirdForTokenClassification"),
("bloom", "BloomForTokenClassification"),
("camembert", "CamembertForTokenClassification"),
("canine", "CanineForTokenClassification"),
("convbert", "ConvBertForTokenClassification"),
("data2vec-text", "Data2VecTextForTokenClassification"),
("deberta", "DebertaForTokenClassification"),
("deberta-v2", "DebertaV2ForTokenClassification"),
("distilbert", "DistilBertForTokenClassification"),
("electra", "ElectraForTokenClassification"),
("ernie", "ErnieForTokenClassification"),
("esm", "EsmForTokenClassification"),
("flaubert", "FlaubertForTokenClassification"),
("fnet", "FNetForTokenClassification"),
("funnel", "FunnelForTokenClassification"),
("gpt-sw3", "GPT2ForTokenClassification"),
("gpt2", "GPT2ForTokenClassification"),
("ibert", "IBertForTokenClassification"),
("layoutlm", "LayoutLMForTokenClassification"),
("layoutlmv2", "LayoutLMv2ForTokenClassification"),
("layoutlmv3", "LayoutLMv3ForTokenClassification"),
("lilt", "LiltForTokenClassification"),
("longformer", "LongformerForTokenClassification"),
("luke", "LukeForTokenClassification"),
("markuplm", "MarkupLMForTokenClassification"),
("megatron-bert", "MegatronBertForTokenClassification"),
("mobilebert", "MobileBertForTokenClassification"),
("mpnet", "MPNetForTokenClassification"),
("nezha", "NezhaForTokenClassification"),
("nystromformer", "NystromformerForTokenClassification"),
("qdqbert", "QDQBertForTokenClassification"),
("rembert", "RemBertForTokenClassification"),
("roberta", "RobertaForTokenClassification"),
("roc_bert", "RoCBertForTokenClassification"),
("roformer", "RoFormerForTokenClassification"),
("squeezebert", "SqueezeBertForTokenClassification"),
("xlm", "XLMForTokenClassification"),
("xlm-roberta", "XLMRobertaForTokenClassification"),
("xlm-roberta-xl", "XLMRobertaXLForTokenClassification"),
("xlnet", "XLNetForTokenClassification"),
("yoso", "YosoForTokenClassification"),
]
)
MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES = OrderedDict(
[
# Model for Multiple Choice mapping
("albert", "AlbertForMultipleChoice"),
("bert", "BertForMultipleChoice"),
("big_bird", "BigBirdForMultipleChoice"),
("camembert", "CamembertForMultipleChoice"),
("canine", "CanineForMultipleChoice"),
("convbert", "ConvBertForMultipleChoice"),
("data2vec-text", "Data2VecTextForMultipleChoice"),
("deberta-v2", "DebertaV2ForMultipleChoice"),
("distilbert", "DistilBertForMultipleChoice"),
("electra", "ElectraForMultipleChoice"),
("ernie", "ErnieForMultipleChoice"),
("flaubert", "FlaubertForMultipleChoice"),
("fnet", "FNetForMultipleChoice"),
("funnel", "FunnelForMultipleChoice"),
("ibert", "IBertForMultipleChoice"),
("longformer", "LongformerForMultipleChoice"),
("luke", "LukeForMultipleChoice"),
("megatron-bert", "MegatronBertForMultipleChoice"),
("mobilebert", "MobileBertForMultipleChoice"),
("mpnet", "MPNetForMultipleChoice"),
("nezha", "NezhaForMultipleChoice"),
("nystromformer", "NystromformerForMultipleChoice"),
("qdqbert", "QDQBertForMultipleChoice"),
("rembert", "RemBertForMultipleChoice"),
("roberta", "RobertaForMultipleChoice"),
("roc_bert", "RoCBertForMultipleChoice"),
("roformer", "RoFormerForMultipleChoice"),
("squeezebert", "SqueezeBertForMultipleChoice"),
("xlm", "XLMForMultipleChoice"),
("xlm-roberta", "XLMRobertaForMultipleChoice"),
("xlm-roberta-xl", "XLMRobertaXLForMultipleChoice"),
("xlnet", "XLNetForMultipleChoice"),
("yoso", "YosoForMultipleChoice"),
]
)
MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES = OrderedDict(
[
("bert", "BertForNextSentencePrediction"),
("ernie", "ErnieForNextSentencePrediction"),
("fnet", "FNetForNextSentencePrediction"),
("megatron-bert", "MegatronBertForNextSentencePrediction"),
("mobilebert", "MobileBertForNextSentencePrediction"),
("nezha", "NezhaForNextSentencePrediction"),
("qdqbert", "QDQBertForNextSentencePrediction"),
]
)
MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Audio Classification mapping
("audio-spectrogram-transformer", "ASTForAudioClassification"),
("data2vec-audio", "Data2VecAudioForSequenceClassification"),
("hubert", "HubertForSequenceClassification"),
("sew", "SEWForSequenceClassification"),
("sew-d", "SEWDForSequenceClassification"),
("unispeech", "UniSpeechForSequenceClassification"),
("unispeech-sat", "UniSpeechSatForSequenceClassification"),
("wav2vec2", "Wav2Vec2ForSequenceClassification"),
("wav2vec2-conformer", "Wav2Vec2ConformerForSequenceClassification"),
("wavlm", "WavLMForSequenceClassification"),
]
)
MODEL_FOR_CTC_MAPPING_NAMES = OrderedDict(
[
# Model for Connectionist temporal classification (CTC) mapping
("data2vec-audio", "Data2VecAudioForCTC"),
("hubert", "HubertForCTC"),
("mctct", "MCTCTForCTC"),
("sew", "SEWForCTC"),
("sew-d", "SEWDForCTC"),
("unispeech", "UniSpeechForCTC"),
("unispeech-sat", "UniSpeechSatForCTC"),
("wav2vec2", "Wav2Vec2ForCTC"),
("wav2vec2-conformer", "Wav2Vec2ConformerForCTC"),
("wavlm", "WavLMForCTC"),
]
)
MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Audio Classification mapping
("data2vec-audio", "Data2VecAudioForAudioFrameClassification"),
("unispeech-sat", "UniSpeechSatForAudioFrameClassification"),
("wav2vec2", "Wav2Vec2ForAudioFrameClassification"),
("wav2vec2-conformer", "Wav2Vec2ConformerForAudioFrameClassification"),
("wavlm", "WavLMForAudioFrameClassification"),
]
)
MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMES = OrderedDict(
[
# Model for Audio Classification mapping
("data2vec-audio", "Data2VecAudioForXVector"),
("unispeech-sat", "UniSpeechSatForXVector"),
("wav2vec2", "Wav2Vec2ForXVector"),
("wav2vec2-conformer", "Wav2Vec2ConformerForXVector"),
("wavlm", "WavLMForXVector"),
]
)
_MODEL_FOR_ZERO_SHOT_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Zero Shot Image Classification mapping
("chinese_clip", "ChineseCLIPModel"),
("clip", "CLIPModel"),
("clipseg", "CLIPSegModel"),
]
)
MODEL_FOR_BACKBONE_MAPPING_NAMES = OrderedDict(
[
# Backbone mapping
("bit", "BitBackbone"),
("maskformer-swin", "MaskFormerSwinBackbone"),
("resnet", "ResNetBackbone"),
]
)
MODEL_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_MAPPING_NAMES)
MODEL_FOR_PRETRAINING_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_PRETRAINING_MAPPING_NAMES)
MODEL_WITH_LM_HEAD_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_WITH_LM_HEAD_MAPPING_NAMES)
MODEL_FOR_CAUSAL_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_CAUSAL_LM_MAPPING_NAMES)
MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING_NAMES
)
MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_IMAGE_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_IMAGE_SEGMENTATION_MAPPING_NAMES
)
MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES
)
MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING_NAMES
)
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_VISION_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES)
MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_MASKED_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_MASKED_LM_MAPPING_NAMES)
MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES
)
MODEL_FOR_OBJECT_DETECTION_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_OBJECT_DETECTION_MAPPING_NAMES)
MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING_NAMES
)
MODEL_FOR_DEPTH_ESTIMATION_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_DEPTH_ESTIMATION_MAPPING_NAMES)
MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES
)
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES
)
MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_MULTIPLE_CHOICE_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES)
MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES
)
MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_CTC_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_CTC_MAPPING_NAMES)
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES)
MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING_NAMES
)
MODEL_FOR_AUDIO_XVECTOR_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMES)
MODEL_FOR_BACKBONE_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, MODEL_FOR_BACKBONE_MAPPING_NAMES)
class AutoModel(_BaseAutoModelClass):
_model_mapping = MODEL_MAPPING
AutoModel = auto_class_update(AutoModel)
class AutoModelForPreTraining(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_PRETRAINING_MAPPING
AutoModelForPreTraining = auto_class_update(AutoModelForPreTraining, head_doc="pretraining")
# Private on purpose, the public class will add the deprecation warnings.
class _AutoModelWithLMHead(_BaseAutoModelClass):
_model_mapping = MODEL_WITH_LM_HEAD_MAPPING
_AutoModelWithLMHead = auto_class_update(_AutoModelWithLMHead, head_doc="language modeling")
class AutoModelForCausalLM(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_CAUSAL_LM_MAPPING
AutoModelForCausalLM = auto_class_update(AutoModelForCausalLM, head_doc="causal language modeling")
class AutoModelForMaskedLM(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_MASKED_LM_MAPPING
AutoModelForMaskedLM = auto_class_update(AutoModelForMaskedLM, head_doc="masked language modeling")
class AutoModelForSeq2SeqLM(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING
AutoModelForSeq2SeqLM = auto_class_update(
AutoModelForSeq2SeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base"
)
class AutoModelForSequenceClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
AutoModelForSequenceClassification = auto_class_update(
AutoModelForSequenceClassification, head_doc="sequence classification"
)
class AutoModelForQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_QUESTION_ANSWERING_MAPPING
AutoModelForQuestionAnswering = auto_class_update(AutoModelForQuestionAnswering, head_doc="question answering")
class AutoModelForTableQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING
AutoModelForTableQuestionAnswering = auto_class_update(
AutoModelForTableQuestionAnswering,
head_doc="table question answering",
checkpoint_for_example="google/tapas-base-finetuned-wtq",
)
class AutoModelForVisualQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING
AutoModelForVisualQuestionAnswering = auto_class_update(
AutoModelForVisualQuestionAnswering,
head_doc="visual question answering",
checkpoint_for_example="dandelin/vilt-b32-finetuned-vqa",
)
class AutoModelForDocumentQuestionAnswering(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING
AutoModelForDocumentQuestionAnswering = auto_class_update(
AutoModelForDocumentQuestionAnswering,
head_doc="document question answering",
checkpoint_for_example='impira/layoutlm-document-qa", revision="52e01b3',
)
class AutoModelForTokenClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
AutoModelForTokenClassification = auto_class_update(AutoModelForTokenClassification, head_doc="token classification")
class AutoModelForMultipleChoice(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_MULTIPLE_CHOICE_MAPPING
AutoModelForMultipleChoice = auto_class_update(AutoModelForMultipleChoice, head_doc="multiple choice")
class AutoModelForNextSentencePrediction(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING
AutoModelForNextSentencePrediction = auto_class_update(
AutoModelForNextSentencePrediction, head_doc="next sentence prediction"
)
class AutoModelForImageClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
AutoModelForImageClassification = auto_class_update(AutoModelForImageClassification, head_doc="image classification")
class AutoModelForImageSegmentation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_IMAGE_SEGMENTATION_MAPPING
AutoModelForImageSegmentation = auto_class_update(AutoModelForImageSegmentation, head_doc="image segmentation")
class AutoModelForSemanticSegmentation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING
AutoModelForSemanticSegmentation = auto_class_update(
AutoModelForSemanticSegmentation, head_doc="semantic segmentation"
)
class AutoModelForInstanceSegmentation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_INSTANCE_SEGMENTATION_MAPPING
AutoModelForInstanceSegmentation = auto_class_update(
AutoModelForInstanceSegmentation, head_doc="instance segmentation"
)
class AutoModelForObjectDetection(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_OBJECT_DETECTION_MAPPING
AutoModelForObjectDetection = auto_class_update(AutoModelForObjectDetection, head_doc="object detection")
class AutoModelForZeroShotObjectDetection(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING
AutoModelForZeroShotObjectDetection = auto_class_update(
AutoModelForZeroShotObjectDetection, head_doc="zero-shot object detection"
)
class AutoModelForDepthEstimation(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_DEPTH_ESTIMATION_MAPPING
AutoModelForDepthEstimation = auto_class_update(AutoModelForDepthEstimation, head_doc="depth estimation")
class AutoModelForVideoClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING
AutoModelForVideoClassification = auto_class_update(AutoModelForVideoClassification, head_doc="video classification")
class AutoModelForVision2Seq(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_VISION_2_SEQ_MAPPING
AutoModelForVision2Seq = auto_class_update(AutoModelForVision2Seq, head_doc="vision-to-text modeling")
class AutoModelForAudioClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING
AutoModelForAudioClassification = auto_class_update(AutoModelForAudioClassification, head_doc="audio classification")
class AutoModelForCTC(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_CTC_MAPPING
AutoModelForCTC = auto_class_update(AutoModelForCTC, head_doc="connectionist temporal classification")
class AutoModelForSpeechSeq2Seq(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING
AutoModelForSpeechSeq2Seq = auto_class_update(
AutoModelForSpeechSeq2Seq, head_doc="sequence-to-sequence speech-to-text modeling"
)
class AutoModelForAudioFrameClassification(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING
AutoModelForAudioFrameClassification = auto_class_update(
AutoModelForAudioFrameClassification, head_doc="audio frame (token) classification"
)
class AutoModelForAudioXVector(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_AUDIO_XVECTOR_MAPPING
class AutoBackbone(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_BACKBONE_MAPPING
AutoModelForAudioXVector = auto_class_update(AutoModelForAudioXVector, head_doc="audio retrieval via x-vector")
class AutoModelForMaskedImageModeling(_BaseAutoModelClass):
_model_mapping = MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING
AutoModelForMaskedImageModeling = auto_class_update(AutoModelForMaskedImageModeling, head_doc="masked image modeling")
class AutoModelWithLMHead(_AutoModelWithLMHead):
@classmethod
def from_config(cls, config):
warnings.warn(
"The class `AutoModelWithLMHead` is deprecated and will be removed in a future version. Please use "
"`AutoModelForCausalLM` for causal language models, `AutoModelForMaskedLM` for masked language models and "
"`AutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_config(config)
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
warnings.warn(
"The class `AutoModelWithLMHead` is deprecated and will be removed in a future version. Please use "
"`AutoModelForCausalLM` for causal language models, `AutoModelForMaskedLM` for masked language models and "
"`AutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/auto/modeling_flax_auto.py | # coding=utf-8
# Copyright 2018 The Google Flax Team Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Model class."""
from collections import OrderedDict
from ...utils import logging
from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update
from .configuration_auto import CONFIG_MAPPING_NAMES
logger = logging.get_logger(__name__)
FLAX_MODEL_MAPPING_NAMES = OrderedDict(
[
# Base model mapping
("albert", "FlaxAlbertModel"),
("bart", "FlaxBartModel"),
("beit", "FlaxBeitModel"),
("bert", "FlaxBertModel"),
("big_bird", "FlaxBigBirdModel"),
("blenderbot", "FlaxBlenderbotModel"),
("blenderbot-small", "FlaxBlenderbotSmallModel"),
("clip", "FlaxCLIPModel"),
("distilbert", "FlaxDistilBertModel"),
("electra", "FlaxElectraModel"),
("gpt2", "FlaxGPT2Model"),
("gpt_neo", "FlaxGPTNeoModel"),
("gptj", "FlaxGPTJModel"),
("longt5", "FlaxLongT5Model"),
("marian", "FlaxMarianModel"),
("mbart", "FlaxMBartModel"),
("mt5", "FlaxMT5Model"),
("opt", "FlaxOPTModel"),
("pegasus", "FlaxPegasusModel"),
("roberta", "FlaxRobertaModel"),
("roformer", "FlaxRoFormerModel"),
("t5", "FlaxT5Model"),
("vision-text-dual-encoder", "FlaxVisionTextDualEncoderModel"),
("vit", "FlaxViTModel"),
("wav2vec2", "FlaxWav2Vec2Model"),
("xglm", "FlaxXGLMModel"),
("xlm-roberta", "FlaxXLMRobertaModel"),
]
)
FLAX_MODEL_FOR_PRETRAINING_MAPPING_NAMES = OrderedDict(
[
# Model for pre-training mapping
("albert", "FlaxAlbertForPreTraining"),
("bart", "FlaxBartForConditionalGeneration"),
("bert", "FlaxBertForPreTraining"),
("big_bird", "FlaxBigBirdForPreTraining"),
("electra", "FlaxElectraForPreTraining"),
("longt5", "FlaxLongT5ForConditionalGeneration"),
("mbart", "FlaxMBartForConditionalGeneration"),
("mt5", "FlaxMT5ForConditionalGeneration"),
("roberta", "FlaxRobertaForMaskedLM"),
("roformer", "FlaxRoFormerForMaskedLM"),
("t5", "FlaxT5ForConditionalGeneration"),
("wav2vec2", "FlaxWav2Vec2ForPreTraining"),
("xlm-roberta", "FlaxXLMRobertaForMaskedLM"),
]
)
FLAX_MODEL_FOR_MASKED_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Masked LM mapping
("albert", "FlaxAlbertForMaskedLM"),
("bart", "FlaxBartForConditionalGeneration"),
("bert", "FlaxBertForMaskedLM"),
("big_bird", "FlaxBigBirdForMaskedLM"),
("distilbert", "FlaxDistilBertForMaskedLM"),
("electra", "FlaxElectraForMaskedLM"),
("mbart", "FlaxMBartForConditionalGeneration"),
("roberta", "FlaxRobertaForMaskedLM"),
("roformer", "FlaxRoFormerForMaskedLM"),
("xlm-roberta", "FlaxXLMRobertaForMaskedLM"),
]
)
FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Seq2Seq Causal LM mapping
("bart", "FlaxBartForConditionalGeneration"),
("blenderbot", "FlaxBlenderbotForConditionalGeneration"),
("blenderbot-small", "FlaxBlenderbotSmallForConditionalGeneration"),
("encoder-decoder", "FlaxEncoderDecoderModel"),
("longt5", "FlaxLongT5ForConditionalGeneration"),
("marian", "FlaxMarianMTModel"),
("mbart", "FlaxMBartForConditionalGeneration"),
("mt5", "FlaxMT5ForConditionalGeneration"),
("pegasus", "FlaxPegasusForConditionalGeneration"),
("t5", "FlaxT5ForConditionalGeneration"),
]
)
FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Image-classsification
("beit", "FlaxBeitForImageClassification"),
("vit", "FlaxViTForImageClassification"),
]
)
FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("vision-encoder-decoder", "FlaxVisionEncoderDecoderModel"),
]
)
FLAX_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Causal LM mapping
("bart", "FlaxBartForCausalLM"),
("bert", "FlaxBertForCausalLM"),
("big_bird", "FlaxBigBirdForCausalLM"),
("electra", "FlaxElectraForCausalLM"),
("gpt2", "FlaxGPT2LMHeadModel"),
("gpt_neo", "FlaxGPTNeoForCausalLM"),
("gptj", "FlaxGPTJForCausalLM"),
("opt", "FlaxOPTForCausalLM"),
("roberta", "FlaxRobertaForCausalLM"),
("xglm", "FlaxXGLMForCausalLM"),
]
)
FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Sequence Classification mapping
("albert", "FlaxAlbertForSequenceClassification"),
("bart", "FlaxBartForSequenceClassification"),
("bert", "FlaxBertForSequenceClassification"),
("big_bird", "FlaxBigBirdForSequenceClassification"),
("distilbert", "FlaxDistilBertForSequenceClassification"),
("electra", "FlaxElectraForSequenceClassification"),
("mbart", "FlaxMBartForSequenceClassification"),
("roberta", "FlaxRobertaForSequenceClassification"),
("roformer", "FlaxRoFormerForSequenceClassification"),
("xlm-roberta", "FlaxXLMRobertaForSequenceClassification"),
]
)
FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Question Answering mapping
("albert", "FlaxAlbertForQuestionAnswering"),
("bart", "FlaxBartForQuestionAnswering"),
("bert", "FlaxBertForQuestionAnswering"),
("big_bird", "FlaxBigBirdForQuestionAnswering"),
("distilbert", "FlaxDistilBertForQuestionAnswering"),
("electra", "FlaxElectraForQuestionAnswering"),
("mbart", "FlaxMBartForQuestionAnswering"),
("roberta", "FlaxRobertaForQuestionAnswering"),
("roformer", "FlaxRoFormerForQuestionAnswering"),
("xlm-roberta", "FlaxXLMRobertaForQuestionAnswering"),
]
)
FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Token Classification mapping
("albert", "FlaxAlbertForTokenClassification"),
("bert", "FlaxBertForTokenClassification"),
("big_bird", "FlaxBigBirdForTokenClassification"),
("distilbert", "FlaxDistilBertForTokenClassification"),
("electra", "FlaxElectraForTokenClassification"),
("roberta", "FlaxRobertaForTokenClassification"),
("roformer", "FlaxRoFormerForTokenClassification"),
("xlm-roberta", "FlaxXLMRobertaForTokenClassification"),
]
)
FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES = OrderedDict(
[
# Model for Multiple Choice mapping
("albert", "FlaxAlbertForMultipleChoice"),
("bert", "FlaxBertForMultipleChoice"),
("big_bird", "FlaxBigBirdForMultipleChoice"),
("distilbert", "FlaxDistilBertForMultipleChoice"),
("electra", "FlaxElectraForMultipleChoice"),
("roberta", "FlaxRobertaForMultipleChoice"),
("roformer", "FlaxRoFormerForMultipleChoice"),
("xlm-roberta", "FlaxXLMRobertaForMultipleChoice"),
]
)
FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES = OrderedDict(
[
("bert", "FlaxBertForNextSentencePrediction"),
]
)
FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("speech-encoder-decoder", "FlaxSpeechEncoderDecoderModel"),
]
)
FLAX_MODEL_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_MAPPING_NAMES)
FLAX_MODEL_FOR_PRETRAINING_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_PRETRAINING_MAPPING_NAMES)
FLAX_MODEL_FOR_MASKED_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_MASKED_LM_MAPPING_NAMES)
FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES
)
FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES
)
FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES)
FLAX_MODEL_FOR_CAUSAL_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES)
FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES
)
FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES
)
FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES
)
FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES
)
FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES
)
FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES
)
class FlaxAutoModel(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_MAPPING
FlaxAutoModel = auto_class_update(FlaxAutoModel)
class FlaxAutoModelForPreTraining(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_PRETRAINING_MAPPING
FlaxAutoModelForPreTraining = auto_class_update(FlaxAutoModelForPreTraining, head_doc="pretraining")
class FlaxAutoModelForCausalLM(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_CAUSAL_LM_MAPPING
FlaxAutoModelForCausalLM = auto_class_update(FlaxAutoModelForCausalLM, head_doc="causal language modeling")
class FlaxAutoModelForMaskedLM(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_MASKED_LM_MAPPING
FlaxAutoModelForMaskedLM = auto_class_update(FlaxAutoModelForMaskedLM, head_doc="masked language modeling")
class FlaxAutoModelForSeq2SeqLM(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING
FlaxAutoModelForSeq2SeqLM = auto_class_update(
FlaxAutoModelForSeq2SeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base"
)
class FlaxAutoModelForSequenceClassification(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
FlaxAutoModelForSequenceClassification = auto_class_update(
FlaxAutoModelForSequenceClassification, head_doc="sequence classification"
)
class FlaxAutoModelForQuestionAnswering(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING
FlaxAutoModelForQuestionAnswering = auto_class_update(FlaxAutoModelForQuestionAnswering, head_doc="question answering")
class FlaxAutoModelForTokenClassification(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
FlaxAutoModelForTokenClassification = auto_class_update(
FlaxAutoModelForTokenClassification, head_doc="token classification"
)
class FlaxAutoModelForMultipleChoice(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING
FlaxAutoModelForMultipleChoice = auto_class_update(FlaxAutoModelForMultipleChoice, head_doc="multiple choice")
class FlaxAutoModelForNextSentencePrediction(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING
FlaxAutoModelForNextSentencePrediction = auto_class_update(
FlaxAutoModelForNextSentencePrediction, head_doc="next sentence prediction"
)
class FlaxAutoModelForImageClassification(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
FlaxAutoModelForImageClassification = auto_class_update(
FlaxAutoModelForImageClassification, head_doc="image classification"
)
class FlaxAutoModelForVision2Seq(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING
FlaxAutoModelForVision2Seq = auto_class_update(FlaxAutoModelForVision2Seq, head_doc="vision-to-text modeling")
class FlaxAutoModelForSpeechSeq2Seq(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING
FlaxAutoModelForSpeechSeq2Seq = auto_class_update(
FlaxAutoModelForSpeechSeq2Seq, head_doc="sequence-to-sequence speech-to-text modeling"
)
| # coding=utf-8
# Copyright 2018 The Google Flax Team Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Model class."""
from collections import OrderedDict
from ...utils import logging
from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update
from .configuration_auto import CONFIG_MAPPING_NAMES
logger = logging.get_logger(__name__)
FLAX_MODEL_MAPPING_NAMES = OrderedDict(
[
# Base model mapping
("albert", "FlaxAlbertModel"),
("bart", "FlaxBartModel"),
("beit", "FlaxBeitModel"),
("bert", "FlaxBertModel"),
("big_bird", "FlaxBigBirdModel"),
("blenderbot", "FlaxBlenderbotModel"),
("blenderbot-small", "FlaxBlenderbotSmallModel"),
("clip", "FlaxCLIPModel"),
("distilbert", "FlaxDistilBertModel"),
("electra", "FlaxElectraModel"),
("gpt-sw3", "FlaxGPT2Model"),
("gpt2", "FlaxGPT2Model"),
("gpt_neo", "FlaxGPTNeoModel"),
("gptj", "FlaxGPTJModel"),
("longt5", "FlaxLongT5Model"),
("marian", "FlaxMarianModel"),
("mbart", "FlaxMBartModel"),
("mt5", "FlaxMT5Model"),
("opt", "FlaxOPTModel"),
("pegasus", "FlaxPegasusModel"),
("roberta", "FlaxRobertaModel"),
("roformer", "FlaxRoFormerModel"),
("t5", "FlaxT5Model"),
("vision-text-dual-encoder", "FlaxVisionTextDualEncoderModel"),
("vit", "FlaxViTModel"),
("wav2vec2", "FlaxWav2Vec2Model"),
("xglm", "FlaxXGLMModel"),
("xlm-roberta", "FlaxXLMRobertaModel"),
]
)
FLAX_MODEL_FOR_PRETRAINING_MAPPING_NAMES = OrderedDict(
[
# Model for pre-training mapping
("albert", "FlaxAlbertForPreTraining"),
("bart", "FlaxBartForConditionalGeneration"),
("bert", "FlaxBertForPreTraining"),
("big_bird", "FlaxBigBirdForPreTraining"),
("electra", "FlaxElectraForPreTraining"),
("longt5", "FlaxLongT5ForConditionalGeneration"),
("mbart", "FlaxMBartForConditionalGeneration"),
("mt5", "FlaxMT5ForConditionalGeneration"),
("roberta", "FlaxRobertaForMaskedLM"),
("roformer", "FlaxRoFormerForMaskedLM"),
("t5", "FlaxT5ForConditionalGeneration"),
("wav2vec2", "FlaxWav2Vec2ForPreTraining"),
("xlm-roberta", "FlaxXLMRobertaForMaskedLM"),
]
)
FLAX_MODEL_FOR_MASKED_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Masked LM mapping
("albert", "FlaxAlbertForMaskedLM"),
("bart", "FlaxBartForConditionalGeneration"),
("bert", "FlaxBertForMaskedLM"),
("big_bird", "FlaxBigBirdForMaskedLM"),
("distilbert", "FlaxDistilBertForMaskedLM"),
("electra", "FlaxElectraForMaskedLM"),
("mbart", "FlaxMBartForConditionalGeneration"),
("roberta", "FlaxRobertaForMaskedLM"),
("roformer", "FlaxRoFormerForMaskedLM"),
("xlm-roberta", "FlaxXLMRobertaForMaskedLM"),
]
)
FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Seq2Seq Causal LM mapping
("bart", "FlaxBartForConditionalGeneration"),
("blenderbot", "FlaxBlenderbotForConditionalGeneration"),
("blenderbot-small", "FlaxBlenderbotSmallForConditionalGeneration"),
("encoder-decoder", "FlaxEncoderDecoderModel"),
("longt5", "FlaxLongT5ForConditionalGeneration"),
("marian", "FlaxMarianMTModel"),
("mbart", "FlaxMBartForConditionalGeneration"),
("mt5", "FlaxMT5ForConditionalGeneration"),
("pegasus", "FlaxPegasusForConditionalGeneration"),
("t5", "FlaxT5ForConditionalGeneration"),
]
)
FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Image-classsification
("beit", "FlaxBeitForImageClassification"),
("vit", "FlaxViTForImageClassification"),
]
)
FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("vision-encoder-decoder", "FlaxVisionEncoderDecoderModel"),
]
)
FLAX_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Causal LM mapping
("bart", "FlaxBartForCausalLM"),
("bert", "FlaxBertForCausalLM"),
("big_bird", "FlaxBigBirdForCausalLM"),
("electra", "FlaxElectraForCausalLM"),
("gpt-sw3", "FlaxGPT2LMHeadModel"),
("gpt2", "FlaxGPT2LMHeadModel"),
("gpt_neo", "FlaxGPTNeoForCausalLM"),
("gptj", "FlaxGPTJForCausalLM"),
("opt", "FlaxOPTForCausalLM"),
("roberta", "FlaxRobertaForCausalLM"),
("xglm", "FlaxXGLMForCausalLM"),
]
)
FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Sequence Classification mapping
("albert", "FlaxAlbertForSequenceClassification"),
("bart", "FlaxBartForSequenceClassification"),
("bert", "FlaxBertForSequenceClassification"),
("big_bird", "FlaxBigBirdForSequenceClassification"),
("distilbert", "FlaxDistilBertForSequenceClassification"),
("electra", "FlaxElectraForSequenceClassification"),
("mbart", "FlaxMBartForSequenceClassification"),
("roberta", "FlaxRobertaForSequenceClassification"),
("roformer", "FlaxRoFormerForSequenceClassification"),
("xlm-roberta", "FlaxXLMRobertaForSequenceClassification"),
]
)
FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Question Answering mapping
("albert", "FlaxAlbertForQuestionAnswering"),
("bart", "FlaxBartForQuestionAnswering"),
("bert", "FlaxBertForQuestionAnswering"),
("big_bird", "FlaxBigBirdForQuestionAnswering"),
("distilbert", "FlaxDistilBertForQuestionAnswering"),
("electra", "FlaxElectraForQuestionAnswering"),
("mbart", "FlaxMBartForQuestionAnswering"),
("roberta", "FlaxRobertaForQuestionAnswering"),
("roformer", "FlaxRoFormerForQuestionAnswering"),
("xlm-roberta", "FlaxXLMRobertaForQuestionAnswering"),
]
)
FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Token Classification mapping
("albert", "FlaxAlbertForTokenClassification"),
("bert", "FlaxBertForTokenClassification"),
("big_bird", "FlaxBigBirdForTokenClassification"),
("distilbert", "FlaxDistilBertForTokenClassification"),
("electra", "FlaxElectraForTokenClassification"),
("roberta", "FlaxRobertaForTokenClassification"),
("roformer", "FlaxRoFormerForTokenClassification"),
("xlm-roberta", "FlaxXLMRobertaForTokenClassification"),
]
)
FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES = OrderedDict(
[
# Model for Multiple Choice mapping
("albert", "FlaxAlbertForMultipleChoice"),
("bert", "FlaxBertForMultipleChoice"),
("big_bird", "FlaxBigBirdForMultipleChoice"),
("distilbert", "FlaxDistilBertForMultipleChoice"),
("electra", "FlaxElectraForMultipleChoice"),
("roberta", "FlaxRobertaForMultipleChoice"),
("roformer", "FlaxRoFormerForMultipleChoice"),
("xlm-roberta", "FlaxXLMRobertaForMultipleChoice"),
]
)
FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES = OrderedDict(
[
("bert", "FlaxBertForNextSentencePrediction"),
]
)
FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("speech-encoder-decoder", "FlaxSpeechEncoderDecoderModel"),
]
)
FLAX_MODEL_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_MAPPING_NAMES)
FLAX_MODEL_FOR_PRETRAINING_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_PRETRAINING_MAPPING_NAMES)
FLAX_MODEL_FOR_MASKED_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_MASKED_LM_MAPPING_NAMES)
FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES
)
FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES
)
FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES)
FLAX_MODEL_FOR_CAUSAL_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES)
FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES
)
FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES
)
FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES
)
FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES
)
FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES
)
FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES
)
class FlaxAutoModel(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_MAPPING
FlaxAutoModel = auto_class_update(FlaxAutoModel)
class FlaxAutoModelForPreTraining(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_PRETRAINING_MAPPING
FlaxAutoModelForPreTraining = auto_class_update(FlaxAutoModelForPreTraining, head_doc="pretraining")
class FlaxAutoModelForCausalLM(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_CAUSAL_LM_MAPPING
FlaxAutoModelForCausalLM = auto_class_update(FlaxAutoModelForCausalLM, head_doc="causal language modeling")
class FlaxAutoModelForMaskedLM(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_MASKED_LM_MAPPING
FlaxAutoModelForMaskedLM = auto_class_update(FlaxAutoModelForMaskedLM, head_doc="masked language modeling")
class FlaxAutoModelForSeq2SeqLM(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING
FlaxAutoModelForSeq2SeqLM = auto_class_update(
FlaxAutoModelForSeq2SeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base"
)
class FlaxAutoModelForSequenceClassification(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
FlaxAutoModelForSequenceClassification = auto_class_update(
FlaxAutoModelForSequenceClassification, head_doc="sequence classification"
)
class FlaxAutoModelForQuestionAnswering(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING
FlaxAutoModelForQuestionAnswering = auto_class_update(FlaxAutoModelForQuestionAnswering, head_doc="question answering")
class FlaxAutoModelForTokenClassification(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
FlaxAutoModelForTokenClassification = auto_class_update(
FlaxAutoModelForTokenClassification, head_doc="token classification"
)
class FlaxAutoModelForMultipleChoice(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING
FlaxAutoModelForMultipleChoice = auto_class_update(FlaxAutoModelForMultipleChoice, head_doc="multiple choice")
class FlaxAutoModelForNextSentencePrediction(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING
FlaxAutoModelForNextSentencePrediction = auto_class_update(
FlaxAutoModelForNextSentencePrediction, head_doc="next sentence prediction"
)
class FlaxAutoModelForImageClassification(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
FlaxAutoModelForImageClassification = auto_class_update(
FlaxAutoModelForImageClassification, head_doc="image classification"
)
class FlaxAutoModelForVision2Seq(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING
FlaxAutoModelForVision2Seq = auto_class_update(FlaxAutoModelForVision2Seq, head_doc="vision-to-text modeling")
class FlaxAutoModelForSpeechSeq2Seq(_BaseAutoModelClass):
_model_mapping = FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING
FlaxAutoModelForSpeechSeq2Seq = auto_class_update(
FlaxAutoModelForSpeechSeq2Seq, head_doc="sequence-to-sequence speech-to-text modeling"
)
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/auto/modeling_tf_auto.py | # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Model class."""
import warnings
from collections import OrderedDict
from ...utils import logging
from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update
from .configuration_auto import CONFIG_MAPPING_NAMES
logger = logging.get_logger(__name__)
TF_MODEL_MAPPING_NAMES = OrderedDict(
[
# Base model mapping
("albert", "TFAlbertModel"),
("bart", "TFBartModel"),
("bert", "TFBertModel"),
("blenderbot", "TFBlenderbotModel"),
("blenderbot-small", "TFBlenderbotSmallModel"),
("camembert", "TFCamembertModel"),
("clip", "TFCLIPModel"),
("convbert", "TFConvBertModel"),
("convnext", "TFConvNextModel"),
("ctrl", "TFCTRLModel"),
("cvt", "TFCvtModel"),
("data2vec-vision", "TFData2VecVisionModel"),
("deberta", "TFDebertaModel"),
("deberta-v2", "TFDebertaV2Model"),
("deit", "TFDeiTModel"),
("distilbert", "TFDistilBertModel"),
("dpr", "TFDPRQuestionEncoder"),
("electra", "TFElectraModel"),
("esm", "TFEsmModel"),
("flaubert", "TFFlaubertModel"),
("funnel", ("TFFunnelModel", "TFFunnelBaseModel")),
("gpt2", "TFGPT2Model"),
("gptj", "TFGPTJModel"),
("groupvit", "TFGroupViTModel"),
("hubert", "TFHubertModel"),
("layoutlm", "TFLayoutLMModel"),
("layoutlmv3", "TFLayoutLMv3Model"),
("led", "TFLEDModel"),
("longformer", "TFLongformerModel"),
("lxmert", "TFLxmertModel"),
("marian", "TFMarianModel"),
("mbart", "TFMBartModel"),
("mobilebert", "TFMobileBertModel"),
("mobilevit", "TFMobileViTModel"),
("mpnet", "TFMPNetModel"),
("mt5", "TFMT5Model"),
("openai-gpt", "TFOpenAIGPTModel"),
("opt", "TFOPTModel"),
("pegasus", "TFPegasusModel"),
("regnet", "TFRegNetModel"),
("rembert", "TFRemBertModel"),
("resnet", "TFResNetModel"),
("roberta", "TFRobertaModel"),
("roformer", "TFRoFormerModel"),
("segformer", "TFSegformerModel"),
("speech_to_text", "TFSpeech2TextModel"),
("swin", "TFSwinModel"),
("t5", "TFT5Model"),
("tapas", "TFTapasModel"),
("transfo-xl", "TFTransfoXLModel"),
("vit", "TFViTModel"),
("vit_mae", "TFViTMAEModel"),
("wav2vec2", "TFWav2Vec2Model"),
("whisper", "TFWhisperModel"),
("xglm", "TFXGLMModel"),
("xlm", "TFXLMModel"),
("xlm-roberta", "TFXLMRobertaModel"),
("xlnet", "TFXLNetModel"),
]
)
TF_MODEL_FOR_PRETRAINING_MAPPING_NAMES = OrderedDict(
[
# Model for pre-training mapping
("albert", "TFAlbertForPreTraining"),
("bart", "TFBartForConditionalGeneration"),
("bert", "TFBertForPreTraining"),
("camembert", "TFCamembertForMaskedLM"),
("ctrl", "TFCTRLLMHeadModel"),
("distilbert", "TFDistilBertForMaskedLM"),
("electra", "TFElectraForPreTraining"),
("flaubert", "TFFlaubertWithLMHeadModel"),
("funnel", "TFFunnelForPreTraining"),
("gpt2", "TFGPT2LMHeadModel"),
("layoutlm", "TFLayoutLMForMaskedLM"),
("lxmert", "TFLxmertForPreTraining"),
("mobilebert", "TFMobileBertForPreTraining"),
("mpnet", "TFMPNetForMaskedLM"),
("openai-gpt", "TFOpenAIGPTLMHeadModel"),
("roberta", "TFRobertaForMaskedLM"),
("t5", "TFT5ForConditionalGeneration"),
("tapas", "TFTapasForMaskedLM"),
("transfo-xl", "TFTransfoXLLMHeadModel"),
("vit_mae", "TFViTMAEForPreTraining"),
("xlm", "TFXLMWithLMHeadModel"),
("xlm-roberta", "TFXLMRobertaForMaskedLM"),
("xlnet", "TFXLNetLMHeadModel"),
]
)
TF_MODEL_WITH_LM_HEAD_MAPPING_NAMES = OrderedDict(
[
# Model with LM heads mapping
("albert", "TFAlbertForMaskedLM"),
("bart", "TFBartForConditionalGeneration"),
("bert", "TFBertForMaskedLM"),
("camembert", "TFCamembertForMaskedLM"),
("convbert", "TFConvBertForMaskedLM"),
("ctrl", "TFCTRLLMHeadModel"),
("distilbert", "TFDistilBertForMaskedLM"),
("electra", "TFElectraForMaskedLM"),
("esm", "TFEsmForMaskedLM"),
("flaubert", "TFFlaubertWithLMHeadModel"),
("funnel", "TFFunnelForMaskedLM"),
("gpt2", "TFGPT2LMHeadModel"),
("gptj", "TFGPTJForCausalLM"),
("layoutlm", "TFLayoutLMForMaskedLM"),
("led", "TFLEDForConditionalGeneration"),
("longformer", "TFLongformerForMaskedLM"),
("marian", "TFMarianMTModel"),
("mobilebert", "TFMobileBertForMaskedLM"),
("mpnet", "TFMPNetForMaskedLM"),
("openai-gpt", "TFOpenAIGPTLMHeadModel"),
("rembert", "TFRemBertForMaskedLM"),
("roberta", "TFRobertaForMaskedLM"),
("roformer", "TFRoFormerForMaskedLM"),
("speech_to_text", "TFSpeech2TextForConditionalGeneration"),
("t5", "TFT5ForConditionalGeneration"),
("tapas", "TFTapasForMaskedLM"),
("transfo-xl", "TFTransfoXLLMHeadModel"),
("whisper", "TFWhisperForConditionalGeneration"),
("xlm", "TFXLMWithLMHeadModel"),
("xlm-roberta", "TFXLMRobertaForMaskedLM"),
("xlnet", "TFXLNetLMHeadModel"),
]
)
TF_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Causal LM mapping
("bert", "TFBertLMHeadModel"),
("camembert", "TFCamembertForCausalLM"),
("ctrl", "TFCTRLLMHeadModel"),
("gpt2", "TFGPT2LMHeadModel"),
("gptj", "TFGPTJForCausalLM"),
("openai-gpt", "TFOpenAIGPTLMHeadModel"),
("opt", "TFOPTForCausalLM"),
("rembert", "TFRemBertForCausalLM"),
("roberta", "TFRobertaForCausalLM"),
("roformer", "TFRoFormerForCausalLM"),
("transfo-xl", "TFTransfoXLLMHeadModel"),
("xglm", "TFXGLMForCausalLM"),
("xlm", "TFXLMWithLMHeadModel"),
("xlnet", "TFXLNetLMHeadModel"),
]
)
TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
[
("deit", "TFDeiTForMaskedImageModeling"),
("swin", "TFSwinForMaskedImageModeling"),
]
)
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Image-classsification
("convnext", "TFConvNextForImageClassification"),
("cvt", "TFCvtForImageClassification"),
("data2vec-vision", "TFData2VecVisionForImageClassification"),
("deit", ("TFDeiTForImageClassification", "TFDeiTForImageClassificationWithTeacher")),
("mobilevit", "TFMobileViTForImageClassification"),
("regnet", "TFRegNetForImageClassification"),
("resnet", "TFResNetForImageClassification"),
("segformer", "TFSegformerForImageClassification"),
("swin", "TFSwinForImageClassification"),
("vit", "TFViTForImageClassification"),
]
)
TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Model for Semantic Segmentation mapping
("data2vec-vision", "TFData2VecVisionForSemanticSegmentation"),
("mobilevit", "TFMobileViTForSemanticSegmentation"),
("segformer", "TFSegformerForSemanticSegmentation"),
]
)
TF_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("vision-encoder-decoder", "TFVisionEncoderDecoderModel"),
]
)
TF_MODEL_FOR_MASKED_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Masked LM mapping
("albert", "TFAlbertForMaskedLM"),
("bert", "TFBertForMaskedLM"),
("camembert", "TFCamembertForMaskedLM"),
("convbert", "TFConvBertForMaskedLM"),
("deberta", "TFDebertaForMaskedLM"),
("deberta-v2", "TFDebertaV2ForMaskedLM"),
("distilbert", "TFDistilBertForMaskedLM"),
("electra", "TFElectraForMaskedLM"),
("esm", "TFEsmForMaskedLM"),
("flaubert", "TFFlaubertWithLMHeadModel"),
("funnel", "TFFunnelForMaskedLM"),
("layoutlm", "TFLayoutLMForMaskedLM"),
("longformer", "TFLongformerForMaskedLM"),
("mobilebert", "TFMobileBertForMaskedLM"),
("mpnet", "TFMPNetForMaskedLM"),
("rembert", "TFRemBertForMaskedLM"),
("roberta", "TFRobertaForMaskedLM"),
("roformer", "TFRoFormerForMaskedLM"),
("tapas", "TFTapasForMaskedLM"),
("xlm", "TFXLMWithLMHeadModel"),
("xlm-roberta", "TFXLMRobertaForMaskedLM"),
]
)
TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Seq2Seq Causal LM mapping
("bart", "TFBartForConditionalGeneration"),
("blenderbot", "TFBlenderbotForConditionalGeneration"),
("blenderbot-small", "TFBlenderbotSmallForConditionalGeneration"),
("encoder-decoder", "TFEncoderDecoderModel"),
("led", "TFLEDForConditionalGeneration"),
("marian", "TFMarianMTModel"),
("mbart", "TFMBartForConditionalGeneration"),
("mt5", "TFMT5ForConditionalGeneration"),
("pegasus", "TFPegasusForConditionalGeneration"),
("t5", "TFT5ForConditionalGeneration"),
]
)
TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("speech_to_text", "TFSpeech2TextForConditionalGeneration"),
("whisper", "TFWhisperForConditionalGeneration"),
]
)
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Sequence Classification mapping
("albert", "TFAlbertForSequenceClassification"),
("bart", "TFBartForSequenceClassification"),
("bert", "TFBertForSequenceClassification"),
("camembert", "TFCamembertForSequenceClassification"),
("convbert", "TFConvBertForSequenceClassification"),
("ctrl", "TFCTRLForSequenceClassification"),
("deberta", "TFDebertaForSequenceClassification"),
("deberta-v2", "TFDebertaV2ForSequenceClassification"),
("distilbert", "TFDistilBertForSequenceClassification"),
("electra", "TFElectraForSequenceClassification"),
("esm", "TFEsmForSequenceClassification"),
("flaubert", "TFFlaubertForSequenceClassification"),
("funnel", "TFFunnelForSequenceClassification"),
("gpt2", "TFGPT2ForSequenceClassification"),
("gptj", "TFGPTJForSequenceClassification"),
("layoutlm", "TFLayoutLMForSequenceClassification"),
("layoutlmv3", "TFLayoutLMv3ForSequenceClassification"),
("longformer", "TFLongformerForSequenceClassification"),
("mobilebert", "TFMobileBertForSequenceClassification"),
("mpnet", "TFMPNetForSequenceClassification"),
("openai-gpt", "TFOpenAIGPTForSequenceClassification"),
("rembert", "TFRemBertForSequenceClassification"),
("roberta", "TFRobertaForSequenceClassification"),
("roformer", "TFRoFormerForSequenceClassification"),
("tapas", "TFTapasForSequenceClassification"),
("transfo-xl", "TFTransfoXLForSequenceClassification"),
("xlm", "TFXLMForSequenceClassification"),
("xlm-roberta", "TFXLMRobertaForSequenceClassification"),
("xlnet", "TFXLNetForSequenceClassification"),
]
)
TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Question Answering mapping
("albert", "TFAlbertForQuestionAnswering"),
("bert", "TFBertForQuestionAnswering"),
("camembert", "TFCamembertForQuestionAnswering"),
("convbert", "TFConvBertForQuestionAnswering"),
("deberta", "TFDebertaForQuestionAnswering"),
("deberta-v2", "TFDebertaV2ForQuestionAnswering"),
("distilbert", "TFDistilBertForQuestionAnswering"),
("electra", "TFElectraForQuestionAnswering"),
("flaubert", "TFFlaubertForQuestionAnsweringSimple"),
("funnel", "TFFunnelForQuestionAnswering"),
("gptj", "TFGPTJForQuestionAnswering"),
("layoutlmv3", "TFLayoutLMv3ForQuestionAnswering"),
("longformer", "TFLongformerForQuestionAnswering"),
("mobilebert", "TFMobileBertForQuestionAnswering"),
("mpnet", "TFMPNetForQuestionAnswering"),
("rembert", "TFRemBertForQuestionAnswering"),
("roberta", "TFRobertaForQuestionAnswering"),
("roformer", "TFRoFormerForQuestionAnswering"),
("xlm", "TFXLMForQuestionAnsweringSimple"),
("xlm-roberta", "TFXLMRobertaForQuestionAnswering"),
("xlnet", "TFXLNetForQuestionAnsweringSimple"),
]
)
TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
("layoutlm", "TFLayoutLMForQuestionAnswering"),
]
)
TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Table Question Answering mapping
("tapas", "TFTapasForQuestionAnswering"),
]
)
TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Token Classification mapping
("albert", "TFAlbertForTokenClassification"),
("bert", "TFBertForTokenClassification"),
("camembert", "TFCamembertForTokenClassification"),
("convbert", "TFConvBertForTokenClassification"),
("deberta", "TFDebertaForTokenClassification"),
("deberta-v2", "TFDebertaV2ForTokenClassification"),
("distilbert", "TFDistilBertForTokenClassification"),
("electra", "TFElectraForTokenClassification"),
("esm", "TFEsmForTokenClassification"),
("flaubert", "TFFlaubertForTokenClassification"),
("funnel", "TFFunnelForTokenClassification"),
("layoutlm", "TFLayoutLMForTokenClassification"),
("layoutlmv3", "TFLayoutLMv3ForTokenClassification"),
("longformer", "TFLongformerForTokenClassification"),
("mobilebert", "TFMobileBertForTokenClassification"),
("mpnet", "TFMPNetForTokenClassification"),
("rembert", "TFRemBertForTokenClassification"),
("roberta", "TFRobertaForTokenClassification"),
("roformer", "TFRoFormerForTokenClassification"),
("xlm", "TFXLMForTokenClassification"),
("xlm-roberta", "TFXLMRobertaForTokenClassification"),
("xlnet", "TFXLNetForTokenClassification"),
]
)
TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES = OrderedDict(
[
# Model for Multiple Choice mapping
("albert", "TFAlbertForMultipleChoice"),
("bert", "TFBertForMultipleChoice"),
("camembert", "TFCamembertForMultipleChoice"),
("convbert", "TFConvBertForMultipleChoice"),
("distilbert", "TFDistilBertForMultipleChoice"),
("electra", "TFElectraForMultipleChoice"),
("flaubert", "TFFlaubertForMultipleChoice"),
("funnel", "TFFunnelForMultipleChoice"),
("longformer", "TFLongformerForMultipleChoice"),
("mobilebert", "TFMobileBertForMultipleChoice"),
("mpnet", "TFMPNetForMultipleChoice"),
("rembert", "TFRemBertForMultipleChoice"),
("roberta", "TFRobertaForMultipleChoice"),
("roformer", "TFRoFormerForMultipleChoice"),
("xlm", "TFXLMForMultipleChoice"),
("xlm-roberta", "TFXLMRobertaForMultipleChoice"),
("xlnet", "TFXLNetForMultipleChoice"),
]
)
TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES = OrderedDict(
[
("bert", "TFBertForNextSentencePrediction"),
("mobilebert", "TFMobileBertForNextSentencePrediction"),
]
)
TF_MODEL_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_MAPPING_NAMES)
TF_MODEL_FOR_PRETRAINING_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_PRETRAINING_MAPPING_NAMES)
TF_MODEL_WITH_LM_HEAD_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_WITH_LM_HEAD_MAPPING_NAMES)
TF_MODEL_FOR_CAUSAL_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES)
TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES
)
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES
)
TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES
)
TF_MODEL_FOR_VISION_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES)
TF_MODEL_FOR_MASKED_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_MASKED_LM_MAPPING_NAMES)
TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES
)
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES
)
TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES
)
TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES
)
TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES
)
TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES
)
TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES
)
TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES
)
TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES
)
class TFAutoModel(_BaseAutoModelClass):
_model_mapping = TF_MODEL_MAPPING
TFAutoModel = auto_class_update(TFAutoModel)
class TFAutoModelForPreTraining(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_PRETRAINING_MAPPING
TFAutoModelForPreTraining = auto_class_update(TFAutoModelForPreTraining, head_doc="pretraining")
# Private on purpose, the public class will add the deprecation warnings.
class _TFAutoModelWithLMHead(_BaseAutoModelClass):
_model_mapping = TF_MODEL_WITH_LM_HEAD_MAPPING
_TFAutoModelWithLMHead = auto_class_update(_TFAutoModelWithLMHead, head_doc="language modeling")
class TFAutoModelForCausalLM(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_CAUSAL_LM_MAPPING
TFAutoModelForCausalLM = auto_class_update(TFAutoModelForCausalLM, head_doc="causal language modeling")
class TFAutoModelForMaskedImageModeling(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING
TFAutoModelForMaskedImageModeling = auto_class_update(
TFAutoModelForMaskedImageModeling, head_doc="masked image modeling"
)
class TFAutoModelForImageClassification(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
TFAutoModelForImageClassification = auto_class_update(
TFAutoModelForImageClassification, head_doc="image classification"
)
class TFAutoModelForSemanticSegmentation(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING
TF_AutoModelForSemanticSegmentation = auto_class_update(
TFAutoModelForSemanticSegmentation, head_doc="semantic segmentation"
)
class TFAutoModelForVision2Seq(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_VISION_2_SEQ_MAPPING
TFAutoModelForVision2Seq = auto_class_update(TFAutoModelForVision2Seq, head_doc="vision-to-text modeling")
class TFAutoModelForMaskedLM(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_MASKED_LM_MAPPING
TFAutoModelForMaskedLM = auto_class_update(TFAutoModelForMaskedLM, head_doc="masked language modeling")
class TFAutoModelForSeq2SeqLM(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING
TFAutoModelForSeq2SeqLM = auto_class_update(
TFAutoModelForSeq2SeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base"
)
class TFAutoModelForSequenceClassification(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
TFAutoModelForSequenceClassification = auto_class_update(
TFAutoModelForSequenceClassification, head_doc="sequence classification"
)
class TFAutoModelForQuestionAnswering(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING
TFAutoModelForQuestionAnswering = auto_class_update(TFAutoModelForQuestionAnswering, head_doc="question answering")
class TFAutoModelForDocumentQuestionAnswering(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING
TFAutoModelForDocumentQuestionAnswering = auto_class_update(
TFAutoModelForDocumentQuestionAnswering,
head_doc="document question answering",
checkpoint_for_example='impira/layoutlm-document-qa", revision="52e01b3',
)
class TFAutoModelForTableQuestionAnswering(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING
TFAutoModelForTableQuestionAnswering = auto_class_update(
TFAutoModelForTableQuestionAnswering,
head_doc="table question answering",
checkpoint_for_example="google/tapas-base-finetuned-wtq",
)
class TFAutoModelForTokenClassification(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
TFAutoModelForTokenClassification = auto_class_update(
TFAutoModelForTokenClassification, head_doc="token classification"
)
class TFAutoModelForMultipleChoice(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING
TFAutoModelForMultipleChoice = auto_class_update(TFAutoModelForMultipleChoice, head_doc="multiple choice")
class TFAutoModelForNextSentencePrediction(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING
TFAutoModelForNextSentencePrediction = auto_class_update(
TFAutoModelForNextSentencePrediction, head_doc="next sentence prediction"
)
class TFAutoModelForSpeechSeq2Seq(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING
TFAutoModelForSpeechSeq2Seq = auto_class_update(
TFAutoModelForSpeechSeq2Seq, head_doc="sequence-to-sequence speech-to-text modeling"
)
class TFAutoModelWithLMHead(_TFAutoModelWithLMHead):
@classmethod
def from_config(cls, config):
warnings.warn(
"The class `TFAutoModelWithLMHead` is deprecated and will be removed in a future version. Please use"
" `TFAutoModelForCausalLM` for causal language models, `TFAutoModelForMaskedLM` for masked language models"
" and `TFAutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_config(config)
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
warnings.warn(
"The class `TFAutoModelWithLMHead` is deprecated and will be removed in a future version. Please use"
" `TFAutoModelForCausalLM` for causal language models, `TFAutoModelForMaskedLM` for masked language models"
" and `TFAutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
| # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Model class."""
import warnings
from collections import OrderedDict
from ...utils import logging
from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update
from .configuration_auto import CONFIG_MAPPING_NAMES
logger = logging.get_logger(__name__)
TF_MODEL_MAPPING_NAMES = OrderedDict(
[
# Base model mapping
("albert", "TFAlbertModel"),
("bart", "TFBartModel"),
("bert", "TFBertModel"),
("blenderbot", "TFBlenderbotModel"),
("blenderbot-small", "TFBlenderbotSmallModel"),
("camembert", "TFCamembertModel"),
("clip", "TFCLIPModel"),
("convbert", "TFConvBertModel"),
("convnext", "TFConvNextModel"),
("ctrl", "TFCTRLModel"),
("cvt", "TFCvtModel"),
("data2vec-vision", "TFData2VecVisionModel"),
("deberta", "TFDebertaModel"),
("deberta-v2", "TFDebertaV2Model"),
("deit", "TFDeiTModel"),
("distilbert", "TFDistilBertModel"),
("dpr", "TFDPRQuestionEncoder"),
("electra", "TFElectraModel"),
("esm", "TFEsmModel"),
("flaubert", "TFFlaubertModel"),
("funnel", ("TFFunnelModel", "TFFunnelBaseModel")),
("gpt-sw3", "TFGPT2Model"),
("gpt2", "TFGPT2Model"),
("gptj", "TFGPTJModel"),
("groupvit", "TFGroupViTModel"),
("hubert", "TFHubertModel"),
("layoutlm", "TFLayoutLMModel"),
("layoutlmv3", "TFLayoutLMv3Model"),
("led", "TFLEDModel"),
("longformer", "TFLongformerModel"),
("lxmert", "TFLxmertModel"),
("marian", "TFMarianModel"),
("mbart", "TFMBartModel"),
("mobilebert", "TFMobileBertModel"),
("mobilevit", "TFMobileViTModel"),
("mpnet", "TFMPNetModel"),
("mt5", "TFMT5Model"),
("openai-gpt", "TFOpenAIGPTModel"),
("opt", "TFOPTModel"),
("pegasus", "TFPegasusModel"),
("regnet", "TFRegNetModel"),
("rembert", "TFRemBertModel"),
("resnet", "TFResNetModel"),
("roberta", "TFRobertaModel"),
("roformer", "TFRoFormerModel"),
("segformer", "TFSegformerModel"),
("speech_to_text", "TFSpeech2TextModel"),
("swin", "TFSwinModel"),
("t5", "TFT5Model"),
("tapas", "TFTapasModel"),
("transfo-xl", "TFTransfoXLModel"),
("vit", "TFViTModel"),
("vit_mae", "TFViTMAEModel"),
("wav2vec2", "TFWav2Vec2Model"),
("whisper", "TFWhisperModel"),
("xglm", "TFXGLMModel"),
("xlm", "TFXLMModel"),
("xlm-roberta", "TFXLMRobertaModel"),
("xlnet", "TFXLNetModel"),
]
)
TF_MODEL_FOR_PRETRAINING_MAPPING_NAMES = OrderedDict(
[
# Model for pre-training mapping
("albert", "TFAlbertForPreTraining"),
("bart", "TFBartForConditionalGeneration"),
("bert", "TFBertForPreTraining"),
("camembert", "TFCamembertForMaskedLM"),
("ctrl", "TFCTRLLMHeadModel"),
("distilbert", "TFDistilBertForMaskedLM"),
("electra", "TFElectraForPreTraining"),
("flaubert", "TFFlaubertWithLMHeadModel"),
("funnel", "TFFunnelForPreTraining"),
("gpt-sw3", "TFGPT2LMHeadModel"),
("gpt2", "TFGPT2LMHeadModel"),
("layoutlm", "TFLayoutLMForMaskedLM"),
("lxmert", "TFLxmertForPreTraining"),
("mobilebert", "TFMobileBertForPreTraining"),
("mpnet", "TFMPNetForMaskedLM"),
("openai-gpt", "TFOpenAIGPTLMHeadModel"),
("roberta", "TFRobertaForMaskedLM"),
("t5", "TFT5ForConditionalGeneration"),
("tapas", "TFTapasForMaskedLM"),
("transfo-xl", "TFTransfoXLLMHeadModel"),
("vit_mae", "TFViTMAEForPreTraining"),
("xlm", "TFXLMWithLMHeadModel"),
("xlm-roberta", "TFXLMRobertaForMaskedLM"),
("xlnet", "TFXLNetLMHeadModel"),
]
)
TF_MODEL_WITH_LM_HEAD_MAPPING_NAMES = OrderedDict(
[
# Model with LM heads mapping
("albert", "TFAlbertForMaskedLM"),
("bart", "TFBartForConditionalGeneration"),
("bert", "TFBertForMaskedLM"),
("camembert", "TFCamembertForMaskedLM"),
("convbert", "TFConvBertForMaskedLM"),
("ctrl", "TFCTRLLMHeadModel"),
("distilbert", "TFDistilBertForMaskedLM"),
("electra", "TFElectraForMaskedLM"),
("esm", "TFEsmForMaskedLM"),
("flaubert", "TFFlaubertWithLMHeadModel"),
("funnel", "TFFunnelForMaskedLM"),
("gpt-sw3", "TFGPT2LMHeadModel"),
("gpt2", "TFGPT2LMHeadModel"),
("gptj", "TFGPTJForCausalLM"),
("layoutlm", "TFLayoutLMForMaskedLM"),
("led", "TFLEDForConditionalGeneration"),
("longformer", "TFLongformerForMaskedLM"),
("marian", "TFMarianMTModel"),
("mobilebert", "TFMobileBertForMaskedLM"),
("mpnet", "TFMPNetForMaskedLM"),
("openai-gpt", "TFOpenAIGPTLMHeadModel"),
("rembert", "TFRemBertForMaskedLM"),
("roberta", "TFRobertaForMaskedLM"),
("roformer", "TFRoFormerForMaskedLM"),
("speech_to_text", "TFSpeech2TextForConditionalGeneration"),
("t5", "TFT5ForConditionalGeneration"),
("tapas", "TFTapasForMaskedLM"),
("transfo-xl", "TFTransfoXLLMHeadModel"),
("whisper", "TFWhisperForConditionalGeneration"),
("xlm", "TFXLMWithLMHeadModel"),
("xlm-roberta", "TFXLMRobertaForMaskedLM"),
("xlnet", "TFXLNetLMHeadModel"),
]
)
TF_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Causal LM mapping
("bert", "TFBertLMHeadModel"),
("camembert", "TFCamembertForCausalLM"),
("ctrl", "TFCTRLLMHeadModel"),
("gpt-sw3", "TFGPT2LMHeadModel"),
("gpt2", "TFGPT2LMHeadModel"),
("gptj", "TFGPTJForCausalLM"),
("openai-gpt", "TFOpenAIGPTLMHeadModel"),
("opt", "TFOPTForCausalLM"),
("rembert", "TFRemBertForCausalLM"),
("roberta", "TFRobertaForCausalLM"),
("roformer", "TFRoFormerForCausalLM"),
("transfo-xl", "TFTransfoXLLMHeadModel"),
("xglm", "TFXGLMForCausalLM"),
("xlm", "TFXLMWithLMHeadModel"),
("xlnet", "TFXLNetLMHeadModel"),
]
)
TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
[
("deit", "TFDeiTForMaskedImageModeling"),
("swin", "TFSwinForMaskedImageModeling"),
]
)
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Image-classsification
("convnext", "TFConvNextForImageClassification"),
("cvt", "TFCvtForImageClassification"),
("data2vec-vision", "TFData2VecVisionForImageClassification"),
("deit", ("TFDeiTForImageClassification", "TFDeiTForImageClassificationWithTeacher")),
("mobilevit", "TFMobileViTForImageClassification"),
("regnet", "TFRegNetForImageClassification"),
("resnet", "TFResNetForImageClassification"),
("segformer", "TFSegformerForImageClassification"),
("swin", "TFSwinForImageClassification"),
("vit", "TFViTForImageClassification"),
]
)
TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES = OrderedDict(
[
# Model for Semantic Segmentation mapping
("data2vec-vision", "TFData2VecVisionForSemanticSegmentation"),
("mobilevit", "TFMobileViTForSemanticSegmentation"),
("segformer", "TFSegformerForSemanticSegmentation"),
]
)
TF_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("vision-encoder-decoder", "TFVisionEncoderDecoderModel"),
]
)
TF_MODEL_FOR_MASKED_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Masked LM mapping
("albert", "TFAlbertForMaskedLM"),
("bert", "TFBertForMaskedLM"),
("camembert", "TFCamembertForMaskedLM"),
("convbert", "TFConvBertForMaskedLM"),
("deberta", "TFDebertaForMaskedLM"),
("deberta-v2", "TFDebertaV2ForMaskedLM"),
("distilbert", "TFDistilBertForMaskedLM"),
("electra", "TFElectraForMaskedLM"),
("esm", "TFEsmForMaskedLM"),
("flaubert", "TFFlaubertWithLMHeadModel"),
("funnel", "TFFunnelForMaskedLM"),
("layoutlm", "TFLayoutLMForMaskedLM"),
("longformer", "TFLongformerForMaskedLM"),
("mobilebert", "TFMobileBertForMaskedLM"),
("mpnet", "TFMPNetForMaskedLM"),
("rembert", "TFRemBertForMaskedLM"),
("roberta", "TFRobertaForMaskedLM"),
("roformer", "TFRoFormerForMaskedLM"),
("tapas", "TFTapasForMaskedLM"),
("xlm", "TFXLMWithLMHeadModel"),
("xlm-roberta", "TFXLMRobertaForMaskedLM"),
]
)
TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
[
# Model for Seq2Seq Causal LM mapping
("bart", "TFBartForConditionalGeneration"),
("blenderbot", "TFBlenderbotForConditionalGeneration"),
("blenderbot-small", "TFBlenderbotSmallForConditionalGeneration"),
("encoder-decoder", "TFEncoderDecoderModel"),
("led", "TFLEDForConditionalGeneration"),
("marian", "TFMarianMTModel"),
("mbart", "TFMBartForConditionalGeneration"),
("mt5", "TFMT5ForConditionalGeneration"),
("pegasus", "TFPegasusForConditionalGeneration"),
("t5", "TFT5ForConditionalGeneration"),
]
)
TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = OrderedDict(
[
("speech_to_text", "TFSpeech2TextForConditionalGeneration"),
("whisper", "TFWhisperForConditionalGeneration"),
]
)
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Sequence Classification mapping
("albert", "TFAlbertForSequenceClassification"),
("bart", "TFBartForSequenceClassification"),
("bert", "TFBertForSequenceClassification"),
("camembert", "TFCamembertForSequenceClassification"),
("convbert", "TFConvBertForSequenceClassification"),
("ctrl", "TFCTRLForSequenceClassification"),
("deberta", "TFDebertaForSequenceClassification"),
("deberta-v2", "TFDebertaV2ForSequenceClassification"),
("distilbert", "TFDistilBertForSequenceClassification"),
("electra", "TFElectraForSequenceClassification"),
("esm", "TFEsmForSequenceClassification"),
("flaubert", "TFFlaubertForSequenceClassification"),
("funnel", "TFFunnelForSequenceClassification"),
("gpt-sw3", "TFGPT2ForSequenceClassification"),
("gpt2", "TFGPT2ForSequenceClassification"),
("gptj", "TFGPTJForSequenceClassification"),
("layoutlm", "TFLayoutLMForSequenceClassification"),
("layoutlmv3", "TFLayoutLMv3ForSequenceClassification"),
("longformer", "TFLongformerForSequenceClassification"),
("mobilebert", "TFMobileBertForSequenceClassification"),
("mpnet", "TFMPNetForSequenceClassification"),
("openai-gpt", "TFOpenAIGPTForSequenceClassification"),
("rembert", "TFRemBertForSequenceClassification"),
("roberta", "TFRobertaForSequenceClassification"),
("roformer", "TFRoFormerForSequenceClassification"),
("tapas", "TFTapasForSequenceClassification"),
("transfo-xl", "TFTransfoXLForSequenceClassification"),
("xlm", "TFXLMForSequenceClassification"),
("xlm-roberta", "TFXLMRobertaForSequenceClassification"),
("xlnet", "TFXLNetForSequenceClassification"),
]
)
TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Question Answering mapping
("albert", "TFAlbertForQuestionAnswering"),
("bert", "TFBertForQuestionAnswering"),
("camembert", "TFCamembertForQuestionAnswering"),
("convbert", "TFConvBertForQuestionAnswering"),
("deberta", "TFDebertaForQuestionAnswering"),
("deberta-v2", "TFDebertaV2ForQuestionAnswering"),
("distilbert", "TFDistilBertForQuestionAnswering"),
("electra", "TFElectraForQuestionAnswering"),
("flaubert", "TFFlaubertForQuestionAnsweringSimple"),
("funnel", "TFFunnelForQuestionAnswering"),
("gptj", "TFGPTJForQuestionAnswering"),
("layoutlmv3", "TFLayoutLMv3ForQuestionAnswering"),
("longformer", "TFLongformerForQuestionAnswering"),
("mobilebert", "TFMobileBertForQuestionAnswering"),
("mpnet", "TFMPNetForQuestionAnswering"),
("rembert", "TFRemBertForQuestionAnswering"),
("roberta", "TFRobertaForQuestionAnswering"),
("roformer", "TFRoFormerForQuestionAnswering"),
("xlm", "TFXLMForQuestionAnsweringSimple"),
("xlm-roberta", "TFXLMRobertaForQuestionAnswering"),
("xlnet", "TFXLNetForQuestionAnsweringSimple"),
]
)
TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
("layoutlm", "TFLayoutLMForQuestionAnswering"),
]
)
TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
[
# Model for Table Question Answering mapping
("tapas", "TFTapasForQuestionAnswering"),
]
)
TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = OrderedDict(
[
# Model for Token Classification mapping
("albert", "TFAlbertForTokenClassification"),
("bert", "TFBertForTokenClassification"),
("camembert", "TFCamembertForTokenClassification"),
("convbert", "TFConvBertForTokenClassification"),
("deberta", "TFDebertaForTokenClassification"),
("deberta-v2", "TFDebertaV2ForTokenClassification"),
("distilbert", "TFDistilBertForTokenClassification"),
("electra", "TFElectraForTokenClassification"),
("esm", "TFEsmForTokenClassification"),
("flaubert", "TFFlaubertForTokenClassification"),
("funnel", "TFFunnelForTokenClassification"),
("layoutlm", "TFLayoutLMForTokenClassification"),
("layoutlmv3", "TFLayoutLMv3ForTokenClassification"),
("longformer", "TFLongformerForTokenClassification"),
("mobilebert", "TFMobileBertForTokenClassification"),
("mpnet", "TFMPNetForTokenClassification"),
("rembert", "TFRemBertForTokenClassification"),
("roberta", "TFRobertaForTokenClassification"),
("roformer", "TFRoFormerForTokenClassification"),
("xlm", "TFXLMForTokenClassification"),
("xlm-roberta", "TFXLMRobertaForTokenClassification"),
("xlnet", "TFXLNetForTokenClassification"),
]
)
TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES = OrderedDict(
[
# Model for Multiple Choice mapping
("albert", "TFAlbertForMultipleChoice"),
("bert", "TFBertForMultipleChoice"),
("camembert", "TFCamembertForMultipleChoice"),
("convbert", "TFConvBertForMultipleChoice"),
("distilbert", "TFDistilBertForMultipleChoice"),
("electra", "TFElectraForMultipleChoice"),
("flaubert", "TFFlaubertForMultipleChoice"),
("funnel", "TFFunnelForMultipleChoice"),
("longformer", "TFLongformerForMultipleChoice"),
("mobilebert", "TFMobileBertForMultipleChoice"),
("mpnet", "TFMPNetForMultipleChoice"),
("rembert", "TFRemBertForMultipleChoice"),
("roberta", "TFRobertaForMultipleChoice"),
("roformer", "TFRoFormerForMultipleChoice"),
("xlm", "TFXLMForMultipleChoice"),
("xlm-roberta", "TFXLMRobertaForMultipleChoice"),
("xlnet", "TFXLNetForMultipleChoice"),
]
)
TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES = OrderedDict(
[
("bert", "TFBertForNextSentencePrediction"),
("mobilebert", "TFMobileBertForNextSentencePrediction"),
]
)
TF_MODEL_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_MAPPING_NAMES)
TF_MODEL_FOR_PRETRAINING_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_PRETRAINING_MAPPING_NAMES)
TF_MODEL_WITH_LM_HEAD_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_WITH_LM_HEAD_MAPPING_NAMES)
TF_MODEL_FOR_CAUSAL_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES)
TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES
)
TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES
)
TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING_NAMES
)
TF_MODEL_FOR_VISION_2_SEQ_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES)
TF_MODEL_FOR_MASKED_LM_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TF_MODEL_FOR_MASKED_LM_MAPPING_NAMES)
TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES
)
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES
)
TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES
)
TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES
)
TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES
)
TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES
)
TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES
)
TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES
)
TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING = _LazyAutoMapping(
CONFIG_MAPPING_NAMES, TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES
)
class TFAutoModel(_BaseAutoModelClass):
_model_mapping = TF_MODEL_MAPPING
TFAutoModel = auto_class_update(TFAutoModel)
class TFAutoModelForPreTraining(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_PRETRAINING_MAPPING
TFAutoModelForPreTraining = auto_class_update(TFAutoModelForPreTraining, head_doc="pretraining")
# Private on purpose, the public class will add the deprecation warnings.
class _TFAutoModelWithLMHead(_BaseAutoModelClass):
_model_mapping = TF_MODEL_WITH_LM_HEAD_MAPPING
_TFAutoModelWithLMHead = auto_class_update(_TFAutoModelWithLMHead, head_doc="language modeling")
class TFAutoModelForCausalLM(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_CAUSAL_LM_MAPPING
TFAutoModelForCausalLM = auto_class_update(TFAutoModelForCausalLM, head_doc="causal language modeling")
class TFAutoModelForMaskedImageModeling(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING
TFAutoModelForMaskedImageModeling = auto_class_update(
TFAutoModelForMaskedImageModeling, head_doc="masked image modeling"
)
class TFAutoModelForImageClassification(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING
TFAutoModelForImageClassification = auto_class_update(
TFAutoModelForImageClassification, head_doc="image classification"
)
class TFAutoModelForSemanticSegmentation(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SEMANTIC_SEGMENTATION_MAPPING
TF_AutoModelForSemanticSegmentation = auto_class_update(
TFAutoModelForSemanticSegmentation, head_doc="semantic segmentation"
)
class TFAutoModelForVision2Seq(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_VISION_2_SEQ_MAPPING
TFAutoModelForVision2Seq = auto_class_update(TFAutoModelForVision2Seq, head_doc="vision-to-text modeling")
class TFAutoModelForMaskedLM(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_MASKED_LM_MAPPING
TFAutoModelForMaskedLM = auto_class_update(TFAutoModelForMaskedLM, head_doc="masked language modeling")
class TFAutoModelForSeq2SeqLM(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING
TFAutoModelForSeq2SeqLM = auto_class_update(
TFAutoModelForSeq2SeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base"
)
class TFAutoModelForSequenceClassification(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
TFAutoModelForSequenceClassification = auto_class_update(
TFAutoModelForSequenceClassification, head_doc="sequence classification"
)
class TFAutoModelForQuestionAnswering(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING
TFAutoModelForQuestionAnswering = auto_class_update(TFAutoModelForQuestionAnswering, head_doc="question answering")
class TFAutoModelForDocumentQuestionAnswering(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING
TFAutoModelForDocumentQuestionAnswering = auto_class_update(
TFAutoModelForDocumentQuestionAnswering,
head_doc="document question answering",
checkpoint_for_example='impira/layoutlm-document-qa", revision="52e01b3',
)
class TFAutoModelForTableQuestionAnswering(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING
TFAutoModelForTableQuestionAnswering = auto_class_update(
TFAutoModelForTableQuestionAnswering,
head_doc="table question answering",
checkpoint_for_example="google/tapas-base-finetuned-wtq",
)
class TFAutoModelForTokenClassification(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
TFAutoModelForTokenClassification = auto_class_update(
TFAutoModelForTokenClassification, head_doc="token classification"
)
class TFAutoModelForMultipleChoice(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_MULTIPLE_CHOICE_MAPPING
TFAutoModelForMultipleChoice = auto_class_update(TFAutoModelForMultipleChoice, head_doc="multiple choice")
class TFAutoModelForNextSentencePrediction(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING
TFAutoModelForNextSentencePrediction = auto_class_update(
TFAutoModelForNextSentencePrediction, head_doc="next sentence prediction"
)
class TFAutoModelForSpeechSeq2Seq(_BaseAutoModelClass):
_model_mapping = TF_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING
TFAutoModelForSpeechSeq2Seq = auto_class_update(
TFAutoModelForSpeechSeq2Seq, head_doc="sequence-to-sequence speech-to-text modeling"
)
class TFAutoModelWithLMHead(_TFAutoModelWithLMHead):
@classmethod
def from_config(cls, config):
warnings.warn(
"The class `TFAutoModelWithLMHead` is deprecated and will be removed in a future version. Please use"
" `TFAutoModelForCausalLM` for causal language models, `TFAutoModelForMaskedLM` for masked language models"
" and `TFAutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_config(config)
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
warnings.warn(
"The class `TFAutoModelWithLMHead` is deprecated and will be removed in a future version. Please use"
" `TFAutoModelForCausalLM` for causal language models, `TFAutoModelForMaskedLM` for masked language models"
" and `TFAutoModelForSeq2SeqLM` for encoder-decoder models.",
FutureWarning,
)
return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/auto/tokenization_auto.py | # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Tokenizer class."""
import importlib
import json
import os
from collections import OrderedDict
from typing import TYPE_CHECKING, Dict, Optional, Tuple, Union
from ...configuration_utils import PretrainedConfig
from ...dynamic_module_utils import get_class_from_dynamic_module
from ...tokenization_utils import PreTrainedTokenizer
from ...tokenization_utils_base import TOKENIZER_CONFIG_FILE
from ...tokenization_utils_fast import PreTrainedTokenizerFast
from ...utils import cached_file, extract_commit_hash, is_sentencepiece_available, is_tokenizers_available, logging
from ..encoder_decoder import EncoderDecoderConfig
from .auto_factory import _LazyAutoMapping
from .configuration_auto import (
CONFIG_MAPPING_NAMES,
AutoConfig,
config_class_to_model_type,
model_type_to_module_name,
replace_list_option_in_docstrings,
)
logger = logging.get_logger(__name__)
if TYPE_CHECKING:
# This significantly improves completion suggestion performance when
# the transformers package is used with Microsoft's Pylance language server.
TOKENIZER_MAPPING_NAMES: OrderedDict[str, Tuple[Optional[str], Optional[str]]] = OrderedDict()
else:
TOKENIZER_MAPPING_NAMES = OrderedDict(
[
(
"albert",
(
"AlbertTokenizer" if is_sentencepiece_available() else None,
"AlbertTokenizerFast" if is_tokenizers_available() else None,
),
),
("bart", ("BartTokenizer", "BartTokenizerFast")),
(
"barthez",
(
"BarthezTokenizer" if is_sentencepiece_available() else None,
"BarthezTokenizerFast" if is_tokenizers_available() else None,
),
),
("bartpho", ("BartphoTokenizer", None)),
("bert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("bert-generation", ("BertGenerationTokenizer" if is_sentencepiece_available() else None, None)),
("bert-japanese", ("BertJapaneseTokenizer", None)),
("bertweet", ("BertweetTokenizer", None)),
(
"big_bird",
(
"BigBirdTokenizer" if is_sentencepiece_available() else None,
"BigBirdTokenizerFast" if is_tokenizers_available() else None,
),
),
("bigbird_pegasus", ("PegasusTokenizer", "PegasusTokenizerFast" if is_tokenizers_available() else None)),
("biogpt", ("BioGptTokenizer", None)),
("blenderbot", ("BlenderbotTokenizer", "BlenderbotTokenizerFast")),
("blenderbot-small", ("BlenderbotSmallTokenizer", None)),
("bloom", (None, "BloomTokenizerFast" if is_tokenizers_available() else None)),
("byt5", ("ByT5Tokenizer", None)),
(
"camembert",
(
"CamembertTokenizer" if is_sentencepiece_available() else None,
"CamembertTokenizerFast" if is_tokenizers_available() else None,
),
),
("canine", ("CanineTokenizer", None)),
("chinese_clip", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
(
"clip",
(
"CLIPTokenizer",
"CLIPTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"clipseg",
(
"CLIPTokenizer",
"CLIPTokenizerFast" if is_tokenizers_available() else None,
),
),
("codegen", ("CodeGenTokenizer", "CodeGenTokenizerFast" if is_tokenizers_available() else None)),
("convbert", ("ConvBertTokenizer", "ConvBertTokenizerFast" if is_tokenizers_available() else None)),
(
"cpm",
(
"CpmTokenizer" if is_sentencepiece_available() else None,
"CpmTokenizerFast" if is_tokenizers_available() else None,
),
),
("ctrl", ("CTRLTokenizer", None)),
("data2vec-text", ("RobertaTokenizer", "RobertaTokenizerFast" if is_tokenizers_available() else None)),
("deberta", ("DebertaTokenizer", "DebertaTokenizerFast" if is_tokenizers_available() else None)),
(
"deberta-v2",
(
"DebertaV2Tokenizer" if is_sentencepiece_available() else None,
"DebertaV2TokenizerFast" if is_tokenizers_available() else None,
),
),
("distilbert", ("DistilBertTokenizer", "DistilBertTokenizerFast" if is_tokenizers_available() else None)),
(
"dpr",
(
"DPRQuestionEncoderTokenizer",
"DPRQuestionEncoderTokenizerFast" if is_tokenizers_available() else None,
),
),
("electra", ("ElectraTokenizer", "ElectraTokenizerFast" if is_tokenizers_available() else None)),
("ernie", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("esm", ("EsmTokenizer", None)),
("flaubert", ("FlaubertTokenizer", None)),
("fnet", ("FNetTokenizer", "FNetTokenizerFast" if is_tokenizers_available() else None)),
("fsmt", ("FSMTTokenizer", None)),
("funnel", ("FunnelTokenizer", "FunnelTokenizerFast" if is_tokenizers_available() else None)),
("gpt2", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),
("gpt_neo", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),
("gpt_neox", (None, "GPTNeoXTokenizerFast" if is_tokenizers_available() else None)),
("gpt_neox_japanese", ("GPTNeoXJapaneseTokenizer", None)),
("gptj", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),
("groupvit", ("CLIPTokenizer", "CLIPTokenizerFast" if is_tokenizers_available() else None)),
("herbert", ("HerbertTokenizer", "HerbertTokenizerFast" if is_tokenizers_available() else None)),
("hubert", ("Wav2Vec2CTCTokenizer", None)),
("ibert", ("RobertaTokenizer", "RobertaTokenizerFast" if is_tokenizers_available() else None)),
("jukebox", ("JukeboxTokenizer", None)),
("layoutlm", ("LayoutLMTokenizer", "LayoutLMTokenizerFast" if is_tokenizers_available() else None)),
("layoutlmv2", ("LayoutLMv2Tokenizer", "LayoutLMv2TokenizerFast" if is_tokenizers_available() else None)),
("layoutlmv3", ("LayoutLMv3Tokenizer", "LayoutLMv3TokenizerFast" if is_tokenizers_available() else None)),
("layoutxlm", ("LayoutXLMTokenizer", "LayoutXLMTokenizerFast" if is_tokenizers_available() else None)),
("led", ("LEDTokenizer", "LEDTokenizerFast" if is_tokenizers_available() else None)),
("lilt", ("LayoutLMv3Tokenizer", "LayoutLMv3TokenizerFast" if is_tokenizers_available() else None)),
("longformer", ("LongformerTokenizer", "LongformerTokenizerFast" if is_tokenizers_available() else None)),
(
"longt5",
(
"T5Tokenizer" if is_sentencepiece_available() else None,
"T5TokenizerFast" if is_tokenizers_available() else None,
),
),
("luke", ("LukeTokenizer", None)),
("lxmert", ("LxmertTokenizer", "LxmertTokenizerFast" if is_tokenizers_available() else None)),
("m2m_100", ("M2M100Tokenizer" if is_sentencepiece_available() else None, None)),
("marian", ("MarianTokenizer" if is_sentencepiece_available() else None, None)),
(
"mbart",
(
"MBartTokenizer" if is_sentencepiece_available() else None,
"MBartTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"mbart50",
(
"MBart50Tokenizer" if is_sentencepiece_available() else None,
"MBart50TokenizerFast" if is_tokenizers_available() else None,
),
),
("megatron-bert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("mluke", ("MLukeTokenizer" if is_sentencepiece_available() else None, None)),
("mobilebert", ("MobileBertTokenizer", "MobileBertTokenizerFast" if is_tokenizers_available() else None)),
("mpnet", ("MPNetTokenizer", "MPNetTokenizerFast" if is_tokenizers_available() else None)),
(
"mt5",
(
"MT5Tokenizer" if is_sentencepiece_available() else None,
"MT5TokenizerFast" if is_tokenizers_available() else None,
),
),
("mvp", ("MvpTokenizer", "MvpTokenizerFast" if is_tokenizers_available() else None)),
("nezha", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
(
"nllb",
(
"NllbTokenizer" if is_sentencepiece_available() else None,
"NllbTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"nystromformer",
(
"AlbertTokenizer" if is_sentencepiece_available() else None,
"AlbertTokenizerFast" if is_tokenizers_available() else None,
),
),
("openai-gpt", ("OpenAIGPTTokenizer", "OpenAIGPTTokenizerFast" if is_tokenizers_available() else None)),
("opt", ("GPT2Tokenizer", None)),
("owlvit", ("CLIPTokenizer", "CLIPTokenizerFast" if is_tokenizers_available() else None)),
(
"pegasus",
(
"PegasusTokenizer" if is_sentencepiece_available() else None,
"PegasusTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"pegasus_x",
(
"PegasusTokenizer" if is_sentencepiece_available() else None,
"PegasusTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"perceiver",
(
"PerceiverTokenizer",
None,
),
),
("phobert", ("PhobertTokenizer", None)),
("plbart", ("PLBartTokenizer" if is_sentencepiece_available() else None, None)),
("prophetnet", ("ProphetNetTokenizer", None)),
("qdqbert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("rag", ("RagTokenizer", None)),
("realm", ("RealmTokenizer", "RealmTokenizerFast" if is_tokenizers_available() else None)),
(
"reformer",
(
"ReformerTokenizer" if is_sentencepiece_available() else None,
"ReformerTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"rembert",
(
"RemBertTokenizer" if is_sentencepiece_available() else None,
"RemBertTokenizerFast" if is_tokenizers_available() else None,
),
),
("retribert", ("RetriBertTokenizer", "RetriBertTokenizerFast" if is_tokenizers_available() else None)),
("roberta", ("RobertaTokenizer", "RobertaTokenizerFast" if is_tokenizers_available() else None)),
("roc_bert", ("RoCBertTokenizer", None)),
("roformer", ("RoFormerTokenizer", "RoFormerTokenizerFast" if is_tokenizers_available() else None)),
("speech_to_text", ("Speech2TextTokenizer" if is_sentencepiece_available() else None, None)),
("speech_to_text_2", ("Speech2Text2Tokenizer", None)),
("splinter", ("SplinterTokenizer", "SplinterTokenizerFast")),
(
"squeezebert",
("SqueezeBertTokenizer", "SqueezeBertTokenizerFast" if is_tokenizers_available() else None),
),
(
"switch_transformers",
(
"T5Tokenizer" if is_sentencepiece_available() else None,
"T5TokenizerFast" if is_tokenizers_available() else None,
),
),
(
"t5",
(
"T5Tokenizer" if is_sentencepiece_available() else None,
"T5TokenizerFast" if is_tokenizers_available() else None,
),
),
("tapas", ("TapasTokenizer", None)),
("tapex", ("TapexTokenizer", None)),
("transfo-xl", ("TransfoXLTokenizer", None)),
("vilt", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("visual_bert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("wav2vec2", ("Wav2Vec2CTCTokenizer", None)),
("wav2vec2-conformer", ("Wav2Vec2CTCTokenizer", None)),
("wav2vec2_phoneme", ("Wav2Vec2PhonemeCTCTokenizer", None)),
("whisper", ("WhisperTokenizer" if is_sentencepiece_available() else None, None)),
("xclip", ("CLIPTokenizer", "CLIPTokenizerFast" if is_tokenizers_available() else None)),
(
"xglm",
(
"XGLMTokenizer" if is_sentencepiece_available() else None,
"XGLMTokenizerFast" if is_tokenizers_available() else None,
),
),
("xlm", ("XLMTokenizer", None)),
("xlm-prophetnet", ("XLMProphetNetTokenizer" if is_sentencepiece_available() else None, None)),
(
"xlm-roberta",
(
"XLMRobertaTokenizer" if is_sentencepiece_available() else None,
"XLMRobertaTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"xlm-roberta-xl",
(
"XLMRobertaTokenizer" if is_sentencepiece_available() else None,
"XLMRobertaTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"xlnet",
(
"XLNetTokenizer" if is_sentencepiece_available() else None,
"XLNetTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"yoso",
(
"AlbertTokenizer" if is_sentencepiece_available() else None,
"AlbertTokenizerFast" if is_tokenizers_available() else None,
),
),
]
)
TOKENIZER_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TOKENIZER_MAPPING_NAMES)
CONFIG_TO_TYPE = {v: k for k, v in CONFIG_MAPPING_NAMES.items()}
def tokenizer_class_from_name(class_name: str):
if class_name == "PreTrainedTokenizerFast":
return PreTrainedTokenizerFast
for module_name, tokenizers in TOKENIZER_MAPPING_NAMES.items():
if class_name in tokenizers:
module_name = model_type_to_module_name(module_name)
module = importlib.import_module(f".{module_name}", "transformers.models")
try:
return getattr(module, class_name)
except AttributeError:
continue
for config, tokenizers in TOKENIZER_MAPPING._extra_content.items():
for tokenizer in tokenizers:
if getattr(tokenizer, "__name__", None) == class_name:
return tokenizer
# We did not fine the class, but maybe it's because a dep is missing. In that case, the class will be in the main
# init and we return the proper dummy to get an appropriate error message.
main_module = importlib.import_module("transformers")
if hasattr(main_module, class_name):
return getattr(main_module, class_name)
return None
def get_tokenizer_config(
pretrained_model_name_or_path: Union[str, os.PathLike],
cache_dir: Optional[Union[str, os.PathLike]] = None,
force_download: bool = False,
resume_download: bool = False,
proxies: Optional[Dict[str, str]] = None,
use_auth_token: Optional[Union[bool, str]] = None,
revision: Optional[str] = None,
local_files_only: bool = False,
subfolder: str = "",
**kwargs,
):
"""
Loads the tokenizer configuration from a pretrained model tokenizer configuration.
Args:
pretrained_model_name_or_path (`str` or `os.PathLike`):
This can be either:
- a string, the *model id* of a pretrained model configuration hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced
under a user or organization name, like `dbmdz/bert-base-german-cased`.
- a path to a *directory* containing a configuration file saved using the
[`~PreTrainedTokenizer.save_pretrained`] method, e.g., `./my_model_directory/`.
cache_dir (`str` or `os.PathLike`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the standard
cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force to (re-)download the configuration files and override the cached versions if they
exist.
resume_download (`bool`, *optional*, defaults to `False`):
Whether or not to delete incompletely received file. Attempts to resume the download if such a file exists.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request.
use_auth_token (`str` or *bool*, *optional*):
The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated
when running `huggingface-cli login` (stored in `~/.huggingface`).
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
local_files_only (`bool`, *optional*, defaults to `False`):
If `True`, will only try to load the tokenizer configuration from local files.
subfolder (`str`, *optional*, defaults to `""`):
In case the tokenizer config is located inside a subfolder of the model repo on huggingface.co, you can
specify the folder name here.
<Tip>
Passing `use_auth_token=True` is required when you want to use a private model.
</Tip>
Returns:
`Dict`: The configuration of the tokenizer.
Examples:
```python
# Download configuration from huggingface.co and cache.
tokenizer_config = get_tokenizer_config("bert-base-uncased")
# This model does not have a tokenizer config so the result will be an empty dict.
tokenizer_config = get_tokenizer_config("xlm-roberta-base")
# Save a pretrained tokenizer locally and you can reload its config
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
tokenizer.save_pretrained("tokenizer-test")
tokenizer_config = get_tokenizer_config("tokenizer-test")
```"""
commit_hash = kwargs.get("_commit_hash", None)
resolved_config_file = cached_file(
pretrained_model_name_or_path,
TOKENIZER_CONFIG_FILE,
cache_dir=cache_dir,
force_download=force_download,
resume_download=resume_download,
proxies=proxies,
use_auth_token=use_auth_token,
revision=revision,
local_files_only=local_files_only,
subfolder=subfolder,
_raise_exceptions_for_missing_entries=False,
_raise_exceptions_for_connection_errors=False,
_commit_hash=commit_hash,
)
if resolved_config_file is None:
logger.info("Could not locate the tokenizer configuration file, will try to use the model config instead.")
return {}
commit_hash = extract_commit_hash(resolved_config_file, commit_hash)
with open(resolved_config_file, encoding="utf-8") as reader:
result = json.load(reader)
result["_commit_hash"] = commit_hash
return result
class AutoTokenizer:
r"""
This is a generic tokenizer class that will be instantiated as one of the tokenizer classes of the library when
created with the [`AutoTokenizer.from_pretrained`] class method.
This class cannot be instantiated directly using `__init__()` (throws an error).
"""
def __init__(self):
raise EnvironmentError(
"AutoTokenizer is designed to be instantiated "
"using the `AutoTokenizer.from_pretrained(pretrained_model_name_or_path)` method."
)
@classmethod
@replace_list_option_in_docstrings(TOKENIZER_MAPPING_NAMES)
def from_pretrained(cls, pretrained_model_name_or_path, *inputs, **kwargs):
r"""
Instantiate one of the tokenizer classes of the library from a pretrained model vocabulary.
The tokenizer class to instantiate is selected based on the `model_type` property of the config object (either
passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by
falling back to using pattern matching on `pretrained_model_name_or_path`:
List options
Params:
pretrained_model_name_or_path (`str` or `os.PathLike`):
Can be either:
- A string, the *model id* of a predefined tokenizer hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing vocabulary files required by the tokenizer, for instance saved
using the [`~PreTrainedTokenizer.save_pretrained`] method, e.g., `./my_model_directory/`.
- A path or url to a single saved vocabulary file if and only if the tokenizer only requires a
single vocabulary file (like Bert or XLNet), e.g.: `./my_model_directory/vocab.txt`. (Not
applicable to all derived classes)
inputs (additional positional arguments, *optional*):
Will be passed along to the Tokenizer `__init__()` method.
config ([`PretrainedConfig`], *optional*)
The configuration object used to dertermine the tokenizer class to instantiate.
cache_dir (`str` or `os.PathLike`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the
standard cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force the (re-)download the model weights and configuration files and override the
cached versions if they exist.
resume_download (`bool`, *optional*, defaults to `False`):
Whether or not to delete incompletely received files. Will attempt to resume the download if such a
file exists.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request.
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
subfolder (`str`, *optional*):
In case the relevant files are located inside a subfolder of the model repo on huggingface.co (e.g. for
facebook/rag-token-base), specify it here.
use_fast (`bool`, *optional*, defaults to `True`):
Whether or not to try to load the fast version of the tokenizer.
tokenizer_type (`str`, *optional*):
Tokenizer type to be loaded.
trust_remote_code (`bool`, *optional*, defaults to `False`):
Whether or not to allow for custom models defined on the Hub in their own modeling files. This option
should only be set to `True` for repositories you trust and in which you have read the code, as it will
execute code present on the Hub on your local machine.
kwargs (additional keyword arguments, *optional*):
Will be passed to the Tokenizer `__init__()` method. Can be used to set special tokens like
`bos_token`, `eos_token`, `unk_token`, `sep_token`, `pad_token`, `cls_token`, `mask_token`,
`additional_special_tokens`. See parameters in the `__init__()` for more details.
Examples:
```python
>>> from transformers import AutoTokenizer
>>> # Download vocabulary from huggingface.co and cache.
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> # Download vocabulary from huggingface.co (user-uploaded) and cache.
>>> tokenizer = AutoTokenizer.from_pretrained("dbmdz/bert-base-german-cased")
>>> # If vocabulary files are in a directory (e.g. tokenizer was saved using *save_pretrained('./test/saved_model/')*)
>>> tokenizer = AutoTokenizer.from_pretrained("./test/bert_saved_model/")
>>> # Download vocabulary from huggingface.co and define model-specific arguments
>>> tokenizer = AutoTokenizer.from_pretrained("roberta-base", add_prefix_space=True)
```"""
config = kwargs.pop("config", None)
kwargs["_from_auto"] = True
use_fast = kwargs.pop("use_fast", True)
tokenizer_type = kwargs.pop("tokenizer_type", None)
trust_remote_code = kwargs.pop("trust_remote_code", False)
# First, let's see whether the tokenizer_type is passed so that we can leverage it
if tokenizer_type is not None:
tokenizer_class = None
tokenizer_class_tuple = TOKENIZER_MAPPING_NAMES.get(tokenizer_type, None)
if tokenizer_class_tuple is None:
raise ValueError(
f"Passed `tokenizer_type` {tokenizer_type} does not exist. `tokenizer_type` should be one of "
f"{', '.join(c for c in TOKENIZER_MAPPING_NAMES.keys())}."
)
tokenizer_class_name, tokenizer_fast_class_name = tokenizer_class_tuple
if use_fast and tokenizer_fast_class_name is not None:
tokenizer_class = tokenizer_class_from_name(tokenizer_fast_class_name)
if tokenizer_class is None:
tokenizer_class = tokenizer_class_from_name(tokenizer_class_name)
if tokenizer_class is None:
raise ValueError(f"Tokenizer class {tokenizer_class_name} is not currently imported.")
return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
# Next, let's try to use the tokenizer_config file to get the tokenizer class.
tokenizer_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)
if "_commit_hash" in tokenizer_config:
kwargs["_commit_hash"] = tokenizer_config["_commit_hash"]
config_tokenizer_class = tokenizer_config.get("tokenizer_class")
tokenizer_auto_map = None
if "auto_map" in tokenizer_config:
if isinstance(tokenizer_config["auto_map"], (tuple, list)):
# Legacy format for dynamic tokenizers
tokenizer_auto_map = tokenizer_config["auto_map"]
else:
tokenizer_auto_map = tokenizer_config["auto_map"].get("AutoTokenizer", None)
# If that did not work, let's try to use the config.
if config_tokenizer_class is None:
if not isinstance(config, PretrainedConfig):
config = AutoConfig.from_pretrained(
pretrained_model_name_or_path, trust_remote_code=trust_remote_code, **kwargs
)
config_tokenizer_class = config.tokenizer_class
if hasattr(config, "auto_map") and "AutoTokenizer" in config.auto_map:
tokenizer_auto_map = config.auto_map["AutoTokenizer"]
# If we have the tokenizer class from the tokenizer config or the model config we're good!
if config_tokenizer_class is not None:
tokenizer_class = None
if tokenizer_auto_map is not None:
if not trust_remote_code:
raise ValueError(
f"Loading {pretrained_model_name_or_path} requires you to execute the tokenizer file in that"
" repo on your local machine. Make sure you have read the code there to avoid malicious use,"
" then set the option `trust_remote_code=True` to remove this error."
)
if kwargs.get("revision", None) is None:
logger.warning(
"Explicitly passing a `revision` is encouraged when loading a model with custom code to ensure"
" no malicious code has been contributed in a newer revision."
)
if use_fast and tokenizer_auto_map[1] is not None:
class_ref = tokenizer_auto_map[1]
else:
class_ref = tokenizer_auto_map[0]
module_file, class_name = class_ref.split(".")
tokenizer_class = get_class_from_dynamic_module(
pretrained_model_name_or_path, module_file + ".py", class_name, **kwargs
)
elif use_fast and not config_tokenizer_class.endswith("Fast"):
tokenizer_class_candidate = f"{config_tokenizer_class}Fast"
tokenizer_class = tokenizer_class_from_name(tokenizer_class_candidate)
if tokenizer_class is None:
tokenizer_class_candidate = config_tokenizer_class
tokenizer_class = tokenizer_class_from_name(tokenizer_class_candidate)
if tokenizer_class is None:
raise ValueError(
f"Tokenizer class {tokenizer_class_candidate} does not exist or is not currently imported."
)
return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
# Otherwise we have to be creative.
# if model is an encoder decoder, the encoder tokenizer class is used by default
if isinstance(config, EncoderDecoderConfig):
if type(config.decoder) is not type(config.encoder): # noqa: E721
logger.warning(
f"The encoder model config class: {config.encoder.__class__} is different from the decoder model "
f"config class: {config.decoder.__class__}. It is not recommended to use the "
"`AutoTokenizer.from_pretrained()` method in this case. Please use the encoder and decoder "
"specific tokenizer classes."
)
config = config.encoder
model_type = config_class_to_model_type(type(config).__name__)
if model_type is not None:
tokenizer_class_py, tokenizer_class_fast = TOKENIZER_MAPPING[type(config)]
if tokenizer_class_fast and (use_fast or tokenizer_class_py is None):
return tokenizer_class_fast.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
else:
if tokenizer_class_py is not None:
return tokenizer_class_py.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
else:
raise ValueError(
"This tokenizer cannot be instantiated. Please make sure you have `sentencepiece` installed "
"in order to use this tokenizer."
)
raise ValueError(
f"Unrecognized configuration class {config.__class__} to build an AutoTokenizer.\n"
f"Model type should be one of {', '.join(c.__name__ for c in TOKENIZER_MAPPING.keys())}."
)
def register(config_class, slow_tokenizer_class=None, fast_tokenizer_class=None):
"""
Register a new tokenizer in this mapping.
Args:
config_class ([`PretrainedConfig`]):
The configuration corresponding to the model to register.
slow_tokenizer_class ([`PretrainedTokenizer`], *optional*):
The slow tokenizer to register.
slow_tokenizer_class ([`PretrainedTokenizerFast`], *optional*):
The fast tokenizer to register.
"""
if slow_tokenizer_class is None and fast_tokenizer_class is None:
raise ValueError("You need to pass either a `slow_tokenizer_class` or a `fast_tokenizer_class")
if slow_tokenizer_class is not None and issubclass(slow_tokenizer_class, PreTrainedTokenizerFast):
raise ValueError("You passed a fast tokenizer in the `slow_tokenizer_class`.")
if fast_tokenizer_class is not None and issubclass(fast_tokenizer_class, PreTrainedTokenizer):
raise ValueError("You passed a slow tokenizer in the `fast_tokenizer_class`.")
if (
slow_tokenizer_class is not None
and fast_tokenizer_class is not None
and issubclass(fast_tokenizer_class, PreTrainedTokenizerFast)
and fast_tokenizer_class.slow_tokenizer_class != slow_tokenizer_class
):
raise ValueError(
"The fast tokenizer class you are passing has a `slow_tokenizer_class` attribute that is not "
"consistent with the slow tokenizer class you passed (fast tokenizer has "
f"{fast_tokenizer_class.slow_tokenizer_class} and you passed {slow_tokenizer_class}. Fix one of those "
"so they match!"
)
# Avoid resetting a set slow/fast tokenizer if we are passing just the other ones.
if config_class in TOKENIZER_MAPPING._extra_content:
existing_slow, existing_fast = TOKENIZER_MAPPING[config_class]
if slow_tokenizer_class is None:
slow_tokenizer_class = existing_slow
if fast_tokenizer_class is None:
fast_tokenizer_class = existing_fast
TOKENIZER_MAPPING.register(config_class, (slow_tokenizer_class, fast_tokenizer_class))
| # coding=utf-8
# Copyright 2018 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" Auto Tokenizer class."""
import importlib
import json
import os
from collections import OrderedDict
from typing import TYPE_CHECKING, Dict, Optional, Tuple, Union
from ...configuration_utils import PretrainedConfig
from ...dynamic_module_utils import get_class_from_dynamic_module
from ...tokenization_utils import PreTrainedTokenizer
from ...tokenization_utils_base import TOKENIZER_CONFIG_FILE
from ...tokenization_utils_fast import PreTrainedTokenizerFast
from ...utils import cached_file, extract_commit_hash, is_sentencepiece_available, is_tokenizers_available, logging
from ..encoder_decoder import EncoderDecoderConfig
from .auto_factory import _LazyAutoMapping
from .configuration_auto import (
CONFIG_MAPPING_NAMES,
AutoConfig,
config_class_to_model_type,
model_type_to_module_name,
replace_list_option_in_docstrings,
)
logger = logging.get_logger(__name__)
if TYPE_CHECKING:
# This significantly improves completion suggestion performance when
# the transformers package is used with Microsoft's Pylance language server.
TOKENIZER_MAPPING_NAMES: OrderedDict[str, Tuple[Optional[str], Optional[str]]] = OrderedDict()
else:
TOKENIZER_MAPPING_NAMES = OrderedDict(
[
(
"albert",
(
"AlbertTokenizer" if is_sentencepiece_available() else None,
"AlbertTokenizerFast" if is_tokenizers_available() else None,
),
),
("bart", ("BartTokenizer", "BartTokenizerFast")),
(
"barthez",
(
"BarthezTokenizer" if is_sentencepiece_available() else None,
"BarthezTokenizerFast" if is_tokenizers_available() else None,
),
),
("bartpho", ("BartphoTokenizer", None)),
("bert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("bert-generation", ("BertGenerationTokenizer" if is_sentencepiece_available() else None, None)),
("bert-japanese", ("BertJapaneseTokenizer", None)),
("bertweet", ("BertweetTokenizer", None)),
(
"big_bird",
(
"BigBirdTokenizer" if is_sentencepiece_available() else None,
"BigBirdTokenizerFast" if is_tokenizers_available() else None,
),
),
("bigbird_pegasus", ("PegasusTokenizer", "PegasusTokenizerFast" if is_tokenizers_available() else None)),
("biogpt", ("BioGptTokenizer", None)),
("blenderbot", ("BlenderbotTokenizer", "BlenderbotTokenizerFast")),
("blenderbot-small", ("BlenderbotSmallTokenizer", None)),
("bloom", (None, "BloomTokenizerFast" if is_tokenizers_available() else None)),
("byt5", ("ByT5Tokenizer", None)),
(
"camembert",
(
"CamembertTokenizer" if is_sentencepiece_available() else None,
"CamembertTokenizerFast" if is_tokenizers_available() else None,
),
),
("canine", ("CanineTokenizer", None)),
("chinese_clip", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
(
"clip",
(
"CLIPTokenizer",
"CLIPTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"clipseg",
(
"CLIPTokenizer",
"CLIPTokenizerFast" if is_tokenizers_available() else None,
),
),
("codegen", ("CodeGenTokenizer", "CodeGenTokenizerFast" if is_tokenizers_available() else None)),
("convbert", ("ConvBertTokenizer", "ConvBertTokenizerFast" if is_tokenizers_available() else None)),
(
"cpm",
(
"CpmTokenizer" if is_sentencepiece_available() else None,
"CpmTokenizerFast" if is_tokenizers_available() else None,
),
),
("ctrl", ("CTRLTokenizer", None)),
("data2vec-text", ("RobertaTokenizer", "RobertaTokenizerFast" if is_tokenizers_available() else None)),
("deberta", ("DebertaTokenizer", "DebertaTokenizerFast" if is_tokenizers_available() else None)),
(
"deberta-v2",
(
"DebertaV2Tokenizer" if is_sentencepiece_available() else None,
"DebertaV2TokenizerFast" if is_tokenizers_available() else None,
),
),
("distilbert", ("DistilBertTokenizer", "DistilBertTokenizerFast" if is_tokenizers_available() else None)),
(
"dpr",
(
"DPRQuestionEncoderTokenizer",
"DPRQuestionEncoderTokenizerFast" if is_tokenizers_available() else None,
),
),
("electra", ("ElectraTokenizer", "ElectraTokenizerFast" if is_tokenizers_available() else None)),
("ernie", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("esm", ("EsmTokenizer", None)),
("flaubert", ("FlaubertTokenizer", None)),
("fnet", ("FNetTokenizer", "FNetTokenizerFast" if is_tokenizers_available() else None)),
("fsmt", ("FSMTTokenizer", None)),
("funnel", ("FunnelTokenizer", "FunnelTokenizerFast" if is_tokenizers_available() else None)),
("gpt-sw3", ("GPTSw3Tokenizer" if is_sentencepiece_available() else None, None)),
("gpt2", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),
("gpt_neo", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),
("gpt_neox", (None, "GPTNeoXTokenizerFast" if is_tokenizers_available() else None)),
("gpt_neox_japanese", ("GPTNeoXJapaneseTokenizer", None)),
("gptj", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),
("groupvit", ("CLIPTokenizer", "CLIPTokenizerFast" if is_tokenizers_available() else None)),
("herbert", ("HerbertTokenizer", "HerbertTokenizerFast" if is_tokenizers_available() else None)),
("hubert", ("Wav2Vec2CTCTokenizer", None)),
("ibert", ("RobertaTokenizer", "RobertaTokenizerFast" if is_tokenizers_available() else None)),
("jukebox", ("JukeboxTokenizer", None)),
("layoutlm", ("LayoutLMTokenizer", "LayoutLMTokenizerFast" if is_tokenizers_available() else None)),
("layoutlmv2", ("LayoutLMv2Tokenizer", "LayoutLMv2TokenizerFast" if is_tokenizers_available() else None)),
("layoutlmv3", ("LayoutLMv3Tokenizer", "LayoutLMv3TokenizerFast" if is_tokenizers_available() else None)),
("layoutxlm", ("LayoutXLMTokenizer", "LayoutXLMTokenizerFast" if is_tokenizers_available() else None)),
("led", ("LEDTokenizer", "LEDTokenizerFast" if is_tokenizers_available() else None)),
("lilt", ("LayoutLMv3Tokenizer", "LayoutLMv3TokenizerFast" if is_tokenizers_available() else None)),
("longformer", ("LongformerTokenizer", "LongformerTokenizerFast" if is_tokenizers_available() else None)),
(
"longt5",
(
"T5Tokenizer" if is_sentencepiece_available() else None,
"T5TokenizerFast" if is_tokenizers_available() else None,
),
),
("luke", ("LukeTokenizer", None)),
("lxmert", ("LxmertTokenizer", "LxmertTokenizerFast" if is_tokenizers_available() else None)),
("m2m_100", ("M2M100Tokenizer" if is_sentencepiece_available() else None, None)),
("marian", ("MarianTokenizer" if is_sentencepiece_available() else None, None)),
(
"mbart",
(
"MBartTokenizer" if is_sentencepiece_available() else None,
"MBartTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"mbart50",
(
"MBart50Tokenizer" if is_sentencepiece_available() else None,
"MBart50TokenizerFast" if is_tokenizers_available() else None,
),
),
("megatron-bert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("mluke", ("MLukeTokenizer" if is_sentencepiece_available() else None, None)),
("mobilebert", ("MobileBertTokenizer", "MobileBertTokenizerFast" if is_tokenizers_available() else None)),
("mpnet", ("MPNetTokenizer", "MPNetTokenizerFast" if is_tokenizers_available() else None)),
(
"mt5",
(
"MT5Tokenizer" if is_sentencepiece_available() else None,
"MT5TokenizerFast" if is_tokenizers_available() else None,
),
),
("mvp", ("MvpTokenizer", "MvpTokenizerFast" if is_tokenizers_available() else None)),
("nezha", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
(
"nllb",
(
"NllbTokenizer" if is_sentencepiece_available() else None,
"NllbTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"nystromformer",
(
"AlbertTokenizer" if is_sentencepiece_available() else None,
"AlbertTokenizerFast" if is_tokenizers_available() else None,
),
),
("openai-gpt", ("OpenAIGPTTokenizer", "OpenAIGPTTokenizerFast" if is_tokenizers_available() else None)),
("opt", ("GPT2Tokenizer", None)),
("owlvit", ("CLIPTokenizer", "CLIPTokenizerFast" if is_tokenizers_available() else None)),
(
"pegasus",
(
"PegasusTokenizer" if is_sentencepiece_available() else None,
"PegasusTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"pegasus_x",
(
"PegasusTokenizer" if is_sentencepiece_available() else None,
"PegasusTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"perceiver",
(
"PerceiverTokenizer",
None,
),
),
("phobert", ("PhobertTokenizer", None)),
("plbart", ("PLBartTokenizer" if is_sentencepiece_available() else None, None)),
("prophetnet", ("ProphetNetTokenizer", None)),
("qdqbert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("rag", ("RagTokenizer", None)),
("realm", ("RealmTokenizer", "RealmTokenizerFast" if is_tokenizers_available() else None)),
(
"reformer",
(
"ReformerTokenizer" if is_sentencepiece_available() else None,
"ReformerTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"rembert",
(
"RemBertTokenizer" if is_sentencepiece_available() else None,
"RemBertTokenizerFast" if is_tokenizers_available() else None,
),
),
("retribert", ("RetriBertTokenizer", "RetriBertTokenizerFast" if is_tokenizers_available() else None)),
("roberta", ("RobertaTokenizer", "RobertaTokenizerFast" if is_tokenizers_available() else None)),
("roc_bert", ("RoCBertTokenizer", None)),
("roformer", ("RoFormerTokenizer", "RoFormerTokenizerFast" if is_tokenizers_available() else None)),
("speech_to_text", ("Speech2TextTokenizer" if is_sentencepiece_available() else None, None)),
("speech_to_text_2", ("Speech2Text2Tokenizer", None)),
("splinter", ("SplinterTokenizer", "SplinterTokenizerFast")),
(
"squeezebert",
("SqueezeBertTokenizer", "SqueezeBertTokenizerFast" if is_tokenizers_available() else None),
),
(
"switch_transformers",
(
"T5Tokenizer" if is_sentencepiece_available() else None,
"T5TokenizerFast" if is_tokenizers_available() else None,
),
),
(
"t5",
(
"T5Tokenizer" if is_sentencepiece_available() else None,
"T5TokenizerFast" if is_tokenizers_available() else None,
),
),
("tapas", ("TapasTokenizer", None)),
("tapex", ("TapexTokenizer", None)),
("transfo-xl", ("TransfoXLTokenizer", None)),
("vilt", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("visual_bert", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
("wav2vec2", ("Wav2Vec2CTCTokenizer", None)),
("wav2vec2-conformer", ("Wav2Vec2CTCTokenizer", None)),
("wav2vec2_phoneme", ("Wav2Vec2PhonemeCTCTokenizer", None)),
("whisper", ("WhisperTokenizer" if is_sentencepiece_available() else None, None)),
("xclip", ("CLIPTokenizer", "CLIPTokenizerFast" if is_tokenizers_available() else None)),
(
"xglm",
(
"XGLMTokenizer" if is_sentencepiece_available() else None,
"XGLMTokenizerFast" if is_tokenizers_available() else None,
),
),
("xlm", ("XLMTokenizer", None)),
("xlm-prophetnet", ("XLMProphetNetTokenizer" if is_sentencepiece_available() else None, None)),
(
"xlm-roberta",
(
"XLMRobertaTokenizer" if is_sentencepiece_available() else None,
"XLMRobertaTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"xlm-roberta-xl",
(
"XLMRobertaTokenizer" if is_sentencepiece_available() else None,
"XLMRobertaTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"xlnet",
(
"XLNetTokenizer" if is_sentencepiece_available() else None,
"XLNetTokenizerFast" if is_tokenizers_available() else None,
),
),
(
"yoso",
(
"AlbertTokenizer" if is_sentencepiece_available() else None,
"AlbertTokenizerFast" if is_tokenizers_available() else None,
),
),
]
)
TOKENIZER_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, TOKENIZER_MAPPING_NAMES)
CONFIG_TO_TYPE = {v: k for k, v in CONFIG_MAPPING_NAMES.items()}
def tokenizer_class_from_name(class_name: str):
if class_name == "PreTrainedTokenizerFast":
return PreTrainedTokenizerFast
for module_name, tokenizers in TOKENIZER_MAPPING_NAMES.items():
if class_name in tokenizers:
module_name = model_type_to_module_name(module_name)
module = importlib.import_module(f".{module_name}", "transformers.models")
try:
return getattr(module, class_name)
except AttributeError:
continue
for config, tokenizers in TOKENIZER_MAPPING._extra_content.items():
for tokenizer in tokenizers:
if getattr(tokenizer, "__name__", None) == class_name:
return tokenizer
# We did not fine the class, but maybe it's because a dep is missing. In that case, the class will be in the main
# init and we return the proper dummy to get an appropriate error message.
main_module = importlib.import_module("transformers")
if hasattr(main_module, class_name):
return getattr(main_module, class_name)
return None
def get_tokenizer_config(
pretrained_model_name_or_path: Union[str, os.PathLike],
cache_dir: Optional[Union[str, os.PathLike]] = None,
force_download: bool = False,
resume_download: bool = False,
proxies: Optional[Dict[str, str]] = None,
use_auth_token: Optional[Union[bool, str]] = None,
revision: Optional[str] = None,
local_files_only: bool = False,
subfolder: str = "",
**kwargs,
):
"""
Loads the tokenizer configuration from a pretrained model tokenizer configuration.
Args:
pretrained_model_name_or_path (`str` or `os.PathLike`):
This can be either:
- a string, the *model id* of a pretrained model configuration hosted inside a model repo on
huggingface.co. Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced
under a user or organization name, like `dbmdz/bert-base-german-cased`.
- a path to a *directory* containing a configuration file saved using the
[`~PreTrainedTokenizer.save_pretrained`] method, e.g., `./my_model_directory/`.
cache_dir (`str` or `os.PathLike`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the standard
cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force to (re-)download the configuration files and override the cached versions if they
exist.
resume_download (`bool`, *optional*, defaults to `False`):
Whether or not to delete incompletely received file. Attempts to resume the download if such a file exists.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request.
use_auth_token (`str` or *bool*, *optional*):
The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated
when running `huggingface-cli login` (stored in `~/.huggingface`).
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
local_files_only (`bool`, *optional*, defaults to `False`):
If `True`, will only try to load the tokenizer configuration from local files.
subfolder (`str`, *optional*, defaults to `""`):
In case the tokenizer config is located inside a subfolder of the model repo on huggingface.co, you can
specify the folder name here.
<Tip>
Passing `use_auth_token=True` is required when you want to use a private model.
</Tip>
Returns:
`Dict`: The configuration of the tokenizer.
Examples:
```python
# Download configuration from huggingface.co and cache.
tokenizer_config = get_tokenizer_config("bert-base-uncased")
# This model does not have a tokenizer config so the result will be an empty dict.
tokenizer_config = get_tokenizer_config("xlm-roberta-base")
# Save a pretrained tokenizer locally and you can reload its config
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
tokenizer.save_pretrained("tokenizer-test")
tokenizer_config = get_tokenizer_config("tokenizer-test")
```"""
commit_hash = kwargs.get("_commit_hash", None)
resolved_config_file = cached_file(
pretrained_model_name_or_path,
TOKENIZER_CONFIG_FILE,
cache_dir=cache_dir,
force_download=force_download,
resume_download=resume_download,
proxies=proxies,
use_auth_token=use_auth_token,
revision=revision,
local_files_only=local_files_only,
subfolder=subfolder,
_raise_exceptions_for_missing_entries=False,
_raise_exceptions_for_connection_errors=False,
_commit_hash=commit_hash,
)
if resolved_config_file is None:
logger.info("Could not locate the tokenizer configuration file, will try to use the model config instead.")
return {}
commit_hash = extract_commit_hash(resolved_config_file, commit_hash)
with open(resolved_config_file, encoding="utf-8") as reader:
result = json.load(reader)
result["_commit_hash"] = commit_hash
return result
class AutoTokenizer:
r"""
This is a generic tokenizer class that will be instantiated as one of the tokenizer classes of the library when
created with the [`AutoTokenizer.from_pretrained`] class method.
This class cannot be instantiated directly using `__init__()` (throws an error).
"""
def __init__(self):
raise EnvironmentError(
"AutoTokenizer is designed to be instantiated "
"using the `AutoTokenizer.from_pretrained(pretrained_model_name_or_path)` method."
)
@classmethod
@replace_list_option_in_docstrings(TOKENIZER_MAPPING_NAMES)
def from_pretrained(cls, pretrained_model_name_or_path, *inputs, **kwargs):
r"""
Instantiate one of the tokenizer classes of the library from a pretrained model vocabulary.
The tokenizer class to instantiate is selected based on the `model_type` property of the config object (either
passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by
falling back to using pattern matching on `pretrained_model_name_or_path`:
List options
Params:
pretrained_model_name_or_path (`str` or `os.PathLike`):
Can be either:
- A string, the *model id* of a predefined tokenizer hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing vocabulary files required by the tokenizer, for instance saved
using the [`~PreTrainedTokenizer.save_pretrained`] method, e.g., `./my_model_directory/`.
- A path or url to a single saved vocabulary file if and only if the tokenizer only requires a
single vocabulary file (like Bert or XLNet), e.g.: `./my_model_directory/vocab.txt`. (Not
applicable to all derived classes)
inputs (additional positional arguments, *optional*):
Will be passed along to the Tokenizer `__init__()` method.
config ([`PretrainedConfig`], *optional*)
The configuration object used to dertermine the tokenizer class to instantiate.
cache_dir (`str` or `os.PathLike`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the
standard cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force the (re-)download the model weights and configuration files and override the
cached versions if they exist.
resume_download (`bool`, *optional*, defaults to `False`):
Whether or not to delete incompletely received files. Will attempt to resume the download if such a
file exists.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request.
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
subfolder (`str`, *optional*):
In case the relevant files are located inside a subfolder of the model repo on huggingface.co (e.g. for
facebook/rag-token-base), specify it here.
use_fast (`bool`, *optional*, defaults to `True`):
Whether or not to try to load the fast version of the tokenizer.
tokenizer_type (`str`, *optional*):
Tokenizer type to be loaded.
trust_remote_code (`bool`, *optional*, defaults to `False`):
Whether or not to allow for custom models defined on the Hub in their own modeling files. This option
should only be set to `True` for repositories you trust and in which you have read the code, as it will
execute code present on the Hub on your local machine.
kwargs (additional keyword arguments, *optional*):
Will be passed to the Tokenizer `__init__()` method. Can be used to set special tokens like
`bos_token`, `eos_token`, `unk_token`, `sep_token`, `pad_token`, `cls_token`, `mask_token`,
`additional_special_tokens`. See parameters in the `__init__()` for more details.
Examples:
```python
>>> from transformers import AutoTokenizer
>>> # Download vocabulary from huggingface.co and cache.
>>> tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
>>> # Download vocabulary from huggingface.co (user-uploaded) and cache.
>>> tokenizer = AutoTokenizer.from_pretrained("dbmdz/bert-base-german-cased")
>>> # If vocabulary files are in a directory (e.g. tokenizer was saved using *save_pretrained('./test/saved_model/')*)
>>> tokenizer = AutoTokenizer.from_pretrained("./test/bert_saved_model/")
>>> # Download vocabulary from huggingface.co and define model-specific arguments
>>> tokenizer = AutoTokenizer.from_pretrained("roberta-base", add_prefix_space=True)
```"""
config = kwargs.pop("config", None)
kwargs["_from_auto"] = True
use_fast = kwargs.pop("use_fast", True)
tokenizer_type = kwargs.pop("tokenizer_type", None)
trust_remote_code = kwargs.pop("trust_remote_code", False)
# First, let's see whether the tokenizer_type is passed so that we can leverage it
if tokenizer_type is not None:
tokenizer_class = None
tokenizer_class_tuple = TOKENIZER_MAPPING_NAMES.get(tokenizer_type, None)
if tokenizer_class_tuple is None:
raise ValueError(
f"Passed `tokenizer_type` {tokenizer_type} does not exist. `tokenizer_type` should be one of "
f"{', '.join(c for c in TOKENIZER_MAPPING_NAMES.keys())}."
)
tokenizer_class_name, tokenizer_fast_class_name = tokenizer_class_tuple
if use_fast and tokenizer_fast_class_name is not None:
tokenizer_class = tokenizer_class_from_name(tokenizer_fast_class_name)
if tokenizer_class is None:
tokenizer_class = tokenizer_class_from_name(tokenizer_class_name)
if tokenizer_class is None:
raise ValueError(f"Tokenizer class {tokenizer_class_name} is not currently imported.")
return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
# Next, let's try to use the tokenizer_config file to get the tokenizer class.
tokenizer_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)
if "_commit_hash" in tokenizer_config:
kwargs["_commit_hash"] = tokenizer_config["_commit_hash"]
config_tokenizer_class = tokenizer_config.get("tokenizer_class")
tokenizer_auto_map = None
if "auto_map" in tokenizer_config:
if isinstance(tokenizer_config["auto_map"], (tuple, list)):
# Legacy format for dynamic tokenizers
tokenizer_auto_map = tokenizer_config["auto_map"]
else:
tokenizer_auto_map = tokenizer_config["auto_map"].get("AutoTokenizer", None)
# If that did not work, let's try to use the config.
if config_tokenizer_class is None:
if not isinstance(config, PretrainedConfig):
config = AutoConfig.from_pretrained(
pretrained_model_name_or_path, trust_remote_code=trust_remote_code, **kwargs
)
config_tokenizer_class = config.tokenizer_class
if hasattr(config, "auto_map") and "AutoTokenizer" in config.auto_map:
tokenizer_auto_map = config.auto_map["AutoTokenizer"]
# If we have the tokenizer class from the tokenizer config or the model config we're good!
if config_tokenizer_class is not None:
tokenizer_class = None
if tokenizer_auto_map is not None:
if not trust_remote_code:
raise ValueError(
f"Loading {pretrained_model_name_or_path} requires you to execute the tokenizer file in that"
" repo on your local machine. Make sure you have read the code there to avoid malicious use,"
" then set the option `trust_remote_code=True` to remove this error."
)
if kwargs.get("revision", None) is None:
logger.warning(
"Explicitly passing a `revision` is encouraged when loading a model with custom code to ensure"
" no malicious code has been contributed in a newer revision."
)
if use_fast and tokenizer_auto_map[1] is not None:
class_ref = tokenizer_auto_map[1]
else:
class_ref = tokenizer_auto_map[0]
module_file, class_name = class_ref.split(".")
tokenizer_class = get_class_from_dynamic_module(
pretrained_model_name_or_path, module_file + ".py", class_name, **kwargs
)
elif use_fast and not config_tokenizer_class.endswith("Fast"):
tokenizer_class_candidate = f"{config_tokenizer_class}Fast"
tokenizer_class = tokenizer_class_from_name(tokenizer_class_candidate)
if tokenizer_class is None:
tokenizer_class_candidate = config_tokenizer_class
tokenizer_class = tokenizer_class_from_name(tokenizer_class_candidate)
if tokenizer_class is None:
raise ValueError(
f"Tokenizer class {tokenizer_class_candidate} does not exist or is not currently imported."
)
return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
# Otherwise we have to be creative.
# if model is an encoder decoder, the encoder tokenizer class is used by default
if isinstance(config, EncoderDecoderConfig):
if type(config.decoder) is not type(config.encoder): # noqa: E721
logger.warning(
f"The encoder model config class: {config.encoder.__class__} is different from the decoder model "
f"config class: {config.decoder.__class__}. It is not recommended to use the "
"`AutoTokenizer.from_pretrained()` method in this case. Please use the encoder and decoder "
"specific tokenizer classes."
)
config = config.encoder
model_type = config_class_to_model_type(type(config).__name__)
if model_type is not None:
tokenizer_class_py, tokenizer_class_fast = TOKENIZER_MAPPING[type(config)]
if tokenizer_class_fast and (use_fast or tokenizer_class_py is None):
return tokenizer_class_fast.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
else:
if tokenizer_class_py is not None:
return tokenizer_class_py.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
else:
raise ValueError(
"This tokenizer cannot be instantiated. Please make sure you have `sentencepiece` installed "
"in order to use this tokenizer."
)
raise ValueError(
f"Unrecognized configuration class {config.__class__} to build an AutoTokenizer.\n"
f"Model type should be one of {', '.join(c.__name__ for c in TOKENIZER_MAPPING.keys())}."
)
def register(config_class, slow_tokenizer_class=None, fast_tokenizer_class=None):
"""
Register a new tokenizer in this mapping.
Args:
config_class ([`PretrainedConfig`]):
The configuration corresponding to the model to register.
slow_tokenizer_class ([`PretrainedTokenizer`], *optional*):
The slow tokenizer to register.
slow_tokenizer_class ([`PretrainedTokenizerFast`], *optional*):
The fast tokenizer to register.
"""
if slow_tokenizer_class is None and fast_tokenizer_class is None:
raise ValueError("You need to pass either a `slow_tokenizer_class` or a `fast_tokenizer_class")
if slow_tokenizer_class is not None and issubclass(slow_tokenizer_class, PreTrainedTokenizerFast):
raise ValueError("You passed a fast tokenizer in the `slow_tokenizer_class`.")
if fast_tokenizer_class is not None and issubclass(fast_tokenizer_class, PreTrainedTokenizer):
raise ValueError("You passed a slow tokenizer in the `fast_tokenizer_class`.")
if (
slow_tokenizer_class is not None
and fast_tokenizer_class is not None
and issubclass(fast_tokenizer_class, PreTrainedTokenizerFast)
and fast_tokenizer_class.slow_tokenizer_class != slow_tokenizer_class
):
raise ValueError(
"The fast tokenizer class you are passing has a `slow_tokenizer_class` attribute that is not "
"consistent with the slow tokenizer class you passed (fast tokenizer has "
f"{fast_tokenizer_class.slow_tokenizer_class} and you passed {slow_tokenizer_class}. Fix one of those "
"so they match!"
)
# Avoid resetting a set slow/fast tokenizer if we are passing just the other ones.
if config_class in TOKENIZER_MAPPING._extra_content:
existing_slow, existing_fast = TOKENIZER_MAPPING[config_class]
if slow_tokenizer_class is None:
slow_tokenizer_class = existing_slow
if fast_tokenizer_class is None:
fast_tokenizer_class = existing_fast
TOKENIZER_MAPPING.register(config_class, (slow_tokenizer_class, fast_tokenizer_class))
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/utils/dummy_sentencepiece_objects.py | # This file is autogenerated by the command `make fix-copies`, do not edit.
# flake8: noqa
from ..utils import DummyObject, requires_backends
class AlbertTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BarthezTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BartphoTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BertGenerationTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BigBirdTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class CamembertTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class CpmTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class DebertaV2Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class FNetTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class LayoutXLMTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class M2M100Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MarianTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MBart50Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MBartTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MLukeTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MT5Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class NllbTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class PegasusTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class PLBartTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class ReformerTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class RemBertTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class Speech2TextTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class T5Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XGLMTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XLMProphetNetTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XLMRobertaTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XLNetTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
| # This file is autogenerated by the command `make fix-copies`, do not edit.
# flake8: noqa
from ..utils import DummyObject, requires_backends
class AlbertTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BarthezTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BartphoTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BertGenerationTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class BigBirdTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class CamembertTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class CpmTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class DebertaV2Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class FNetTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class GPTSw3Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class LayoutXLMTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class M2M100Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MarianTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MBart50Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MBartTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MLukeTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class MT5Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class NllbTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class PegasusTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class PLBartTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class ReformerTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class RemBertTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class Speech2TextTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class T5Tokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XGLMTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XLMProphetNetTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XLMRobertaTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
class XLNetTokenizer(metaclass=DummyObject):
_backends = ["sentencepiece"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["sentencepiece"])
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./utils/check_repo.py | # coding=utf-8
# Copyright 2020 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import importlib
import inspect
import os
import re
import warnings
from collections import OrderedDict
from difflib import get_close_matches
from pathlib import Path
from transformers import is_flax_available, is_tf_available, is_torch_available
from transformers.models.auto import get_values
from transformers.utils import ENV_VARS_TRUE_VALUES
# All paths are set with the intent you should run this script from the root of the repo with the command
# python utils/check_repo.py
PATH_TO_TRANSFORMERS = "src/transformers"
PATH_TO_TESTS = "tests"
PATH_TO_DOC = "docs/source/en"
# Update this list with models that are supposed to be private.
PRIVATE_MODELS = [
"DPRSpanPredictor",
"LongT5Stack",
"RealmBertModel",
"T5Stack",
"SwitchTransformersStack",
"TFDPRSpanPredictor",
"MaskFormerSwinModel",
"MaskFormerSwinPreTrainedModel",
]
# Update this list for models that are not tested with a comment explaining the reason it should not be.
# Being in this list is an exception and should **not** be the rule.
IGNORE_NON_TESTED = PRIVATE_MODELS.copy() + [
# models to ignore for not tested
"CLIPSegDecoder", # Building part of bigger (tested) model.
"TableTransformerEncoder", # Building part of bigger (tested) model.
"TableTransformerDecoder", # Building part of bigger (tested) model.
"TimeSeriesTransformerEncoder", # Building part of bigger (tested) model.
"TimeSeriesTransformerDecoder", # Building part of bigger (tested) model.
"JukeboxVQVAE", # Building part of bigger (tested) model.
"JukeboxPrior", # Building part of bigger (tested) model.
"DeformableDetrEncoder", # Building part of bigger (tested) model.
"DeformableDetrDecoder", # Building part of bigger (tested) model.
"OPTDecoder", # Building part of bigger (tested) model.
"WhisperDecoder", # Building part of bigger (tested) model.
"WhisperEncoder", # Building part of bigger (tested) model.
"DecisionTransformerGPT2Model", # Building part of bigger (tested) model.
"SegformerDecodeHead", # Building part of bigger (tested) model.
"PLBartEncoder", # Building part of bigger (tested) model.
"PLBartDecoder", # Building part of bigger (tested) model.
"PLBartDecoderWrapper", # Building part of bigger (tested) model.
"BigBirdPegasusEncoder", # Building part of bigger (tested) model.
"BigBirdPegasusDecoder", # Building part of bigger (tested) model.
"BigBirdPegasusDecoderWrapper", # Building part of bigger (tested) model.
"DetrEncoder", # Building part of bigger (tested) model.
"DetrDecoder", # Building part of bigger (tested) model.
"DetrDecoderWrapper", # Building part of bigger (tested) model.
"ConditionalDetrEncoder", # Building part of bigger (tested) model.
"ConditionalDetrDecoder", # Building part of bigger (tested) model.
"M2M100Encoder", # Building part of bigger (tested) model.
"M2M100Decoder", # Building part of bigger (tested) model.
"MCTCTEncoder", # Building part of bigger (tested) model.
"Speech2TextEncoder", # Building part of bigger (tested) model.
"Speech2TextDecoder", # Building part of bigger (tested) model.
"LEDEncoder", # Building part of bigger (tested) model.
"LEDDecoder", # Building part of bigger (tested) model.
"BartDecoderWrapper", # Building part of bigger (tested) model.
"BartEncoder", # Building part of bigger (tested) model.
"BertLMHeadModel", # Needs to be setup as decoder.
"BlenderbotSmallEncoder", # Building part of bigger (tested) model.
"BlenderbotSmallDecoderWrapper", # Building part of bigger (tested) model.
"BlenderbotEncoder", # Building part of bigger (tested) model.
"BlenderbotDecoderWrapper", # Building part of bigger (tested) model.
"MBartEncoder", # Building part of bigger (tested) model.
"MBartDecoderWrapper", # Building part of bigger (tested) model.
"MegatronBertLMHeadModel", # Building part of bigger (tested) model.
"MegatronBertEncoder", # Building part of bigger (tested) model.
"MegatronBertDecoder", # Building part of bigger (tested) model.
"MegatronBertDecoderWrapper", # Building part of bigger (tested) model.
"MvpDecoderWrapper", # Building part of bigger (tested) model.
"MvpEncoder", # Building part of bigger (tested) model.
"PegasusEncoder", # Building part of bigger (tested) model.
"PegasusDecoderWrapper", # Building part of bigger (tested) model.
"PegasusXEncoder", # Building part of bigger (tested) model.
"PegasusXDecoder", # Building part of bigger (tested) model.
"PegasusXDecoderWrapper", # Building part of bigger (tested) model.
"DPREncoder", # Building part of bigger (tested) model.
"ProphetNetDecoderWrapper", # Building part of bigger (tested) model.
"RealmBertModel", # Building part of bigger (tested) model.
"RealmReader", # Not regular model.
"RealmScorer", # Not regular model.
"RealmForOpenQA", # Not regular model.
"ReformerForMaskedLM", # Needs to be setup as decoder.
"Speech2Text2DecoderWrapper", # Building part of bigger (tested) model.
"TFDPREncoder", # Building part of bigger (tested) model.
"TFElectraMainLayer", # Building part of bigger (tested) model (should it be a TFPreTrainedModel ?)
"TFRobertaForMultipleChoice", # TODO: fix
"TrOCRDecoderWrapper", # Building part of bigger (tested) model.
"TFWhisperEncoder", # Building part of bigger (tested) model.
"TFWhisperDecoder", # Building part of bigger (tested) model.
"SeparableConv1D", # Building part of bigger (tested) model.
"FlaxBartForCausalLM", # Building part of bigger (tested) model.
"FlaxBertForCausalLM", # Building part of bigger (tested) model. Tested implicitly through FlaxRobertaForCausalLM.
"OPTDecoderWrapper",
"TFSegformerDecodeHead", # Not a regular model.
]
# Update this list with test files that don't have a tester with a `all_model_classes` variable and which don't
# trigger the common tests.
TEST_FILES_WITH_NO_COMMON_TESTS = [
"models/decision_transformer/test_modeling_decision_transformer.py",
"models/camembert/test_modeling_camembert.py",
"models/mt5/test_modeling_flax_mt5.py",
"models/mbart/test_modeling_mbart.py",
"models/mt5/test_modeling_mt5.py",
"models/pegasus/test_modeling_pegasus.py",
"models/camembert/test_modeling_tf_camembert.py",
"models/mt5/test_modeling_tf_mt5.py",
"models/xlm_roberta/test_modeling_tf_xlm_roberta.py",
"models/xlm_roberta/test_modeling_flax_xlm_roberta.py",
"models/xlm_prophetnet/test_modeling_xlm_prophetnet.py",
"models/xlm_roberta/test_modeling_xlm_roberta.py",
"models/vision_text_dual_encoder/test_modeling_vision_text_dual_encoder.py",
"models/vision_text_dual_encoder/test_modeling_flax_vision_text_dual_encoder.py",
"models/decision_transformer/test_modeling_decision_transformer.py",
]
# Update this list for models that are not in any of the auto MODEL_XXX_MAPPING. Being in this list is an exception and
# should **not** be the rule.
IGNORE_NON_AUTO_CONFIGURED = PRIVATE_MODELS.copy() + [
# models to ignore for model xxx mapping
"CLIPSegForImageSegmentation",
"CLIPSegVisionModel",
"CLIPSegTextModel",
"EsmForProteinFolding",
"TimeSeriesTransformerForPrediction",
"JukeboxVQVAE",
"JukeboxPrior",
"PegasusXEncoder",
"PegasusXDecoder",
"PegasusXDecoderWrapper",
"PegasusXEncoder",
"PegasusXDecoder",
"PegasusXDecoderWrapper",
"DPTForDepthEstimation",
"DecisionTransformerGPT2Model",
"GLPNForDepthEstimation",
"ViltForImagesAndTextClassification",
"ViltForImageAndTextRetrieval",
"ViltForTokenClassification",
"ViltForMaskedLM",
"XGLMEncoder",
"XGLMDecoder",
"XGLMDecoderWrapper",
"PerceiverForMultimodalAutoencoding",
"PerceiverForOpticalFlow",
"SegformerDecodeHead",
"TFSegformerDecodeHead",
"FlaxBeitForMaskedImageModeling",
"PLBartEncoder",
"PLBartDecoder",
"PLBartDecoderWrapper",
"BeitForMaskedImageModeling",
"ChineseCLIPTextModel",
"ChineseCLIPVisionModel",
"CLIPTextModel",
"CLIPTextModelWithProjection",
"CLIPVisionModel",
"CLIPVisionModelWithProjection",
"GroupViTTextModel",
"GroupViTVisionModel",
"TFCLIPTextModel",
"TFCLIPVisionModel",
"TFGroupViTTextModel",
"TFGroupViTVisionModel",
"FlaxCLIPTextModel",
"FlaxCLIPVisionModel",
"FlaxWav2Vec2ForCTC",
"DetrForSegmentation",
"ConditionalDetrForSegmentation",
"DPRReader",
"FlaubertForQuestionAnswering",
"FlavaImageCodebook",
"FlavaTextModel",
"FlavaImageModel",
"FlavaMultimodalModel",
"GPT2DoubleHeadsModel",
"LayoutLMForQuestionAnswering",
"LukeForMaskedLM",
"LukeForEntityClassification",
"LukeForEntityPairClassification",
"LukeForEntitySpanClassification",
"OpenAIGPTDoubleHeadsModel",
"OwlViTTextModel",
"OwlViTVisionModel",
"OwlViTForObjectDetection",
"RagModel",
"RagSequenceForGeneration",
"RagTokenForGeneration",
"RealmEmbedder",
"RealmForOpenQA",
"RealmScorer",
"RealmReader",
"TFDPRReader",
"TFGPT2DoubleHeadsModel",
"TFLayoutLMForQuestionAnswering",
"TFOpenAIGPTDoubleHeadsModel",
"TFRagModel",
"TFRagSequenceForGeneration",
"TFRagTokenForGeneration",
"Wav2Vec2ForCTC",
"HubertForCTC",
"SEWForCTC",
"SEWDForCTC",
"XLMForQuestionAnswering",
"XLNetForQuestionAnswering",
"SeparableConv1D",
"VisualBertForRegionToPhraseAlignment",
"VisualBertForVisualReasoning",
"VisualBertForQuestionAnswering",
"VisualBertForMultipleChoice",
"TFWav2Vec2ForCTC",
"TFHubertForCTC",
"MaskFormerForInstanceSegmentation",
"XCLIPVisionModel",
"XCLIPTextModel",
]
# Update this list for models that have multiple model types for the same
# model doc
MODEL_TYPE_TO_DOC_MAPPING = OrderedDict(
[
("data2vec-text", "data2vec"),
("data2vec-audio", "data2vec"),
("data2vec-vision", "data2vec"),
("donut-swin", "donut"),
]
)
# This is to make sure the transformers module imported is the one in the repo.
spec = importlib.util.spec_from_file_location(
"transformers",
os.path.join(PATH_TO_TRANSFORMERS, "__init__.py"),
submodule_search_locations=[PATH_TO_TRANSFORMERS],
)
transformers = spec.loader.load_module()
def check_model_list():
"""Check the model list inside the transformers library."""
# Get the models from the directory structure of `src/transformers/models/`
models_dir = os.path.join(PATH_TO_TRANSFORMERS, "models")
_models = []
for model in os.listdir(models_dir):
model_dir = os.path.join(models_dir, model)
if os.path.isdir(model_dir) and "__init__.py" in os.listdir(model_dir):
_models.append(model)
# Get the models from the directory structure of `src/transformers/models/`
models = [model for model in dir(transformers.models) if not model.startswith("__")]
missing_models = sorted(list(set(_models).difference(models)))
if missing_models:
raise Exception(
f"The following models should be included in {models_dir}/__init__.py: {','.join(missing_models)}."
)
# If some modeling modules should be ignored for all checks, they should be added in the nested list
# _ignore_modules of this function.
def get_model_modules():
"""Get the model modules inside the transformers library."""
_ignore_modules = [
"modeling_auto",
"modeling_encoder_decoder",
"modeling_marian",
"modeling_mmbt",
"modeling_outputs",
"modeling_retribert",
"modeling_utils",
"modeling_flax_auto",
"modeling_flax_encoder_decoder",
"modeling_flax_utils",
"modeling_speech_encoder_decoder",
"modeling_flax_speech_encoder_decoder",
"modeling_flax_vision_encoder_decoder",
"modeling_transfo_xl_utilities",
"modeling_tf_auto",
"modeling_tf_encoder_decoder",
"modeling_tf_outputs",
"modeling_tf_pytorch_utils",
"modeling_tf_utils",
"modeling_tf_transfo_xl_utilities",
"modeling_tf_vision_encoder_decoder",
"modeling_vision_encoder_decoder",
]
modules = []
for model in dir(transformers.models):
# There are some magic dunder attributes in the dir, we ignore them
if not model.startswith("__"):
model_module = getattr(transformers.models, model)
for submodule in dir(model_module):
if submodule.startswith("modeling") and submodule not in _ignore_modules:
modeling_module = getattr(model_module, submodule)
if inspect.ismodule(modeling_module):
modules.append(modeling_module)
return modules
def get_models(module, include_pretrained=False):
"""Get the objects in module that are models."""
models = []
model_classes = (transformers.PreTrainedModel, transformers.TFPreTrainedModel, transformers.FlaxPreTrainedModel)
for attr_name in dir(module):
if not include_pretrained and ("Pretrained" in attr_name or "PreTrained" in attr_name):
continue
attr = getattr(module, attr_name)
if isinstance(attr, type) and issubclass(attr, model_classes) and attr.__module__ == module.__name__:
models.append((attr_name, attr))
return models
def is_a_private_model(model):
"""Returns True if the model should not be in the main init."""
if model in PRIVATE_MODELS:
return True
# Wrapper, Encoder and Decoder are all privates
if model.endswith("Wrapper"):
return True
if model.endswith("Encoder"):
return True
if model.endswith("Decoder"):
return True
return False
def check_models_are_in_init():
"""Checks all models defined in the library are in the main init."""
models_not_in_init = []
dir_transformers = dir(transformers)
for module in get_model_modules():
models_not_in_init += [
model[0] for model in get_models(module, include_pretrained=True) if model[0] not in dir_transformers
]
# Remove private models
models_not_in_init = [model for model in models_not_in_init if not is_a_private_model(model)]
if len(models_not_in_init) > 0:
raise Exception(f"The following models should be in the main init: {','.join(models_not_in_init)}.")
# If some test_modeling files should be ignored when checking models are all tested, they should be added in the
# nested list _ignore_files of this function.
def get_model_test_files():
"""Get the model test files.
The returned files should NOT contain the `tests` (i.e. `PATH_TO_TESTS` defined in this script). They will be
considered as paths relative to `tests`. A caller has to use `os.path.join(PATH_TO_TESTS, ...)` to access the files.
"""
_ignore_files = [
"test_modeling_common",
"test_modeling_encoder_decoder",
"test_modeling_flax_encoder_decoder",
"test_modeling_flax_speech_encoder_decoder",
"test_modeling_marian",
"test_modeling_tf_common",
"test_modeling_tf_encoder_decoder",
]
test_files = []
# Check both `PATH_TO_TESTS` and `PATH_TO_TESTS/models`
model_test_root = os.path.join(PATH_TO_TESTS, "models")
model_test_dirs = []
for x in os.listdir(model_test_root):
x = os.path.join(model_test_root, x)
if os.path.isdir(x):
model_test_dirs.append(x)
for target_dir in [PATH_TO_TESTS] + model_test_dirs:
for file_or_dir in os.listdir(target_dir):
path = os.path.join(target_dir, file_or_dir)
if os.path.isfile(path):
filename = os.path.split(path)[-1]
if "test_modeling" in filename and not os.path.splitext(filename)[0] in _ignore_files:
file = os.path.join(*path.split(os.sep)[1:])
test_files.append(file)
return test_files
# This is a bit hacky but I didn't find a way to import the test_file as a module and read inside the tester class
# for the all_model_classes variable.
def find_tested_models(test_file):
"""Parse the content of test_file to detect what's in all_model_classes"""
# This is a bit hacky but I didn't find a way to import the test_file as a module and read inside the class
with open(os.path.join(PATH_TO_TESTS, test_file), "r", encoding="utf-8", newline="\n") as f:
content = f.read()
all_models = re.findall(r"all_model_classes\s+=\s+\(\s*\(([^\)]*)\)", content)
# Check with one less parenthesis as well
all_models += re.findall(r"all_model_classes\s+=\s+\(([^\)]*)\)", content)
if len(all_models) > 0:
model_tested = []
for entry in all_models:
for line in entry.split(","):
name = line.strip()
if len(name) > 0:
model_tested.append(name)
return model_tested
def check_models_are_tested(module, test_file):
"""Check models defined in module are tested in test_file."""
# XxxPreTrainedModel are not tested
defined_models = get_models(module)
tested_models = find_tested_models(test_file)
if tested_models is None:
if test_file.replace(os.path.sep, "/") in TEST_FILES_WITH_NO_COMMON_TESTS:
return
return [
f"{test_file} should define `all_model_classes` to apply common tests to the models it tests. "
+ "If this intentional, add the test filename to `TEST_FILES_WITH_NO_COMMON_TESTS` in the file "
+ "`utils/check_repo.py`."
]
failures = []
for model_name, _ in defined_models:
if model_name not in tested_models and model_name not in IGNORE_NON_TESTED:
failures.append(
f"{model_name} is defined in {module.__name__} but is not tested in "
+ f"{os.path.join(PATH_TO_TESTS, test_file)}. Add it to the all_model_classes in that file."
+ "If common tests should not applied to that model, add its name to `IGNORE_NON_TESTED`"
+ "in the file `utils/check_repo.py`."
)
return failures
def check_all_models_are_tested():
"""Check all models are properly tested."""
modules = get_model_modules()
test_files = get_model_test_files()
failures = []
for module in modules:
test_file = [file for file in test_files if f"test_{module.__name__.split('.')[-1]}.py" in file]
if len(test_file) == 0:
failures.append(f"{module.__name__} does not have its corresponding test file {test_file}.")
elif len(test_file) > 1:
failures.append(f"{module.__name__} has several test files: {test_file}.")
else:
test_file = test_file[0]
new_failures = check_models_are_tested(module, test_file)
if new_failures is not None:
failures += new_failures
if len(failures) > 0:
raise Exception(f"There were {len(failures)} failures:\n" + "\n".join(failures))
def get_all_auto_configured_models():
"""Return the list of all models in at least one auto class."""
result = set() # To avoid duplicates we concatenate all model classes in a set.
if is_torch_available():
for attr_name in dir(transformers.models.auto.modeling_auto):
if attr_name.startswith("MODEL_") and attr_name.endswith("MAPPING_NAMES"):
result = result | set(get_values(getattr(transformers.models.auto.modeling_auto, attr_name)))
if is_tf_available():
for attr_name in dir(transformers.models.auto.modeling_tf_auto):
if attr_name.startswith("TF_MODEL_") and attr_name.endswith("MAPPING_NAMES"):
result = result | set(get_values(getattr(transformers.models.auto.modeling_tf_auto, attr_name)))
if is_flax_available():
for attr_name in dir(transformers.models.auto.modeling_flax_auto):
if attr_name.startswith("FLAX_MODEL_") and attr_name.endswith("MAPPING_NAMES"):
result = result | set(get_values(getattr(transformers.models.auto.modeling_flax_auto, attr_name)))
return [cls for cls in result]
def ignore_unautoclassed(model_name):
"""Rules to determine if `name` should be in an auto class."""
# Special white list
if model_name in IGNORE_NON_AUTO_CONFIGURED:
return True
# Encoder and Decoder should be ignored
if "Encoder" in model_name or "Decoder" in model_name:
return True
return False
def check_models_are_auto_configured(module, all_auto_models):
"""Check models defined in module are each in an auto class."""
defined_models = get_models(module)
failures = []
for model_name, _ in defined_models:
if model_name not in all_auto_models and not ignore_unautoclassed(model_name):
failures.append(
f"{model_name} is defined in {module.__name__} but is not present in any of the auto mapping. "
"If that is intended behavior, add its name to `IGNORE_NON_AUTO_CONFIGURED` in the file "
"`utils/check_repo.py`."
)
return failures
def check_all_models_are_auto_configured():
"""Check all models are each in an auto class."""
missing_backends = []
if not is_torch_available():
missing_backends.append("PyTorch")
if not is_tf_available():
missing_backends.append("TensorFlow")
if not is_flax_available():
missing_backends.append("Flax")
if len(missing_backends) > 0:
missing = ", ".join(missing_backends)
if os.getenv("TRANSFORMERS_IS_CI", "").upper() in ENV_VARS_TRUE_VALUES:
raise Exception(
"Full quality checks require all backends to be installed (with `pip install -e .[dev]` in the "
f"Transformers repo, the following are missing: {missing}."
)
else:
warnings.warn(
"Full quality checks require all backends to be installed (with `pip install -e .[dev]` in the "
f"Transformers repo, the following are missing: {missing}. While it's probably fine as long as you "
"didn't make any change in one of those backends modeling files, you should probably execute the "
"command above to be on the safe side."
)
modules = get_model_modules()
all_auto_models = get_all_auto_configured_models()
failures = []
for module in modules:
new_failures = check_models_are_auto_configured(module, all_auto_models)
if new_failures is not None:
failures += new_failures
if len(failures) > 0:
raise Exception(f"There were {len(failures)} failures:\n" + "\n".join(failures))
_re_decorator = re.compile(r"^\s*@(\S+)\s+$")
def check_decorator_order(filename):
"""Check that in the test file `filename` the slow decorator is always last."""
with open(filename, "r", encoding="utf-8", newline="\n") as f:
lines = f.readlines()
decorator_before = None
errors = []
for i, line in enumerate(lines):
search = _re_decorator.search(line)
if search is not None:
decorator_name = search.groups()[0]
if decorator_before is not None and decorator_name.startswith("parameterized"):
errors.append(i)
decorator_before = decorator_name
elif decorator_before is not None:
decorator_before = None
return errors
def check_all_decorator_order():
"""Check that in all test files, the slow decorator is always last."""
errors = []
for fname in os.listdir(PATH_TO_TESTS):
if fname.endswith(".py"):
filename = os.path.join(PATH_TO_TESTS, fname)
new_errors = check_decorator_order(filename)
errors += [f"- {filename}, line {i}" for i in new_errors]
if len(errors) > 0:
msg = "\n".join(errors)
raise ValueError(
"The parameterized decorator (and its variants) should always be first, but this is not the case in the"
f" following files:\n{msg}"
)
def find_all_documented_objects():
"""Parse the content of all doc files to detect which classes and functions it documents"""
documented_obj = []
for doc_file in Path(PATH_TO_DOC).glob("**/*.rst"):
with open(doc_file, "r", encoding="utf-8", newline="\n") as f:
content = f.read()
raw_doc_objs = re.findall(r"(?:autoclass|autofunction):: transformers.(\S+)\s+", content)
documented_obj += [obj.split(".")[-1] for obj in raw_doc_objs]
for doc_file in Path(PATH_TO_DOC).glob("**/*.mdx"):
with open(doc_file, "r", encoding="utf-8", newline="\n") as f:
content = f.read()
raw_doc_objs = re.findall("\[\[autodoc\]\]\s+(\S+)\s+", content)
documented_obj += [obj.split(".")[-1] for obj in raw_doc_objs]
return documented_obj
# One good reason for not being documented is to be deprecated. Put in this list deprecated objects.
DEPRECATED_OBJECTS = [
"AutoModelWithLMHead",
"BartPretrainedModel",
"DataCollator",
"DataCollatorForSOP",
"GlueDataset",
"GlueDataTrainingArguments",
"LineByLineTextDataset",
"LineByLineWithRefDataset",
"LineByLineWithSOPTextDataset",
"PretrainedBartModel",
"PretrainedFSMTModel",
"SingleSentenceClassificationProcessor",
"SquadDataTrainingArguments",
"SquadDataset",
"SquadExample",
"SquadFeatures",
"SquadV1Processor",
"SquadV2Processor",
"TFAutoModelWithLMHead",
"TFBartPretrainedModel",
"TextDataset",
"TextDatasetForNextSentencePrediction",
"Wav2Vec2ForMaskedLM",
"Wav2Vec2Tokenizer",
"glue_compute_metrics",
"glue_convert_examples_to_features",
"glue_output_modes",
"glue_processors",
"glue_tasks_num_labels",
"squad_convert_examples_to_features",
"xnli_compute_metrics",
"xnli_output_modes",
"xnli_processors",
"xnli_tasks_num_labels",
"TFTrainer",
"TFTrainingArguments",
]
# Exceptionally, some objects should not be documented after all rules passed.
# ONLY PUT SOMETHING IN THIS LIST AS A LAST RESORT!
UNDOCUMENTED_OBJECTS = [
"AddedToken", # This is a tokenizers class.
"BasicTokenizer", # Internal, should never have been in the main init.
"CharacterTokenizer", # Internal, should never have been in the main init.
"DPRPretrainedReader", # Like an Encoder.
"DummyObject", # Just picked by mistake sometimes.
"MecabTokenizer", # Internal, should never have been in the main init.
"ModelCard", # Internal type.
"SqueezeBertModule", # Internal building block (should have been called SqueezeBertLayer)
"TFDPRPretrainedReader", # Like an Encoder.
"TransfoXLCorpus", # Internal type.
"WordpieceTokenizer", # Internal, should never have been in the main init.
"absl", # External module
"add_end_docstrings", # Internal, should never have been in the main init.
"add_start_docstrings", # Internal, should never have been in the main init.
"convert_tf_weight_name_to_pt_weight_name", # Internal used to convert model weights
"logger", # Internal logger
"logging", # External module
"requires_backends", # Internal function
]
# This list should be empty. Objects in it should get their own doc page.
SHOULD_HAVE_THEIR_OWN_PAGE = [
# Benchmarks
"PyTorchBenchmark",
"PyTorchBenchmarkArguments",
"TensorFlowBenchmark",
"TensorFlowBenchmarkArguments",
"BitBackbone",
"MaskFormerSwinBackbone",
"ResNetBackbone",
"AutoBackbone",
"MaskFormerSwinConfig",
"MaskFormerSwinModel",
]
def ignore_undocumented(name):
"""Rules to determine if `name` should be undocumented."""
# NOT DOCUMENTED ON PURPOSE.
# Constants uppercase are not documented.
if name.isupper():
return True
# PreTrainedModels / Encoders / Decoders / Layers / Embeddings / Attention are not documented.
if (
name.endswith("PreTrainedModel")
or name.endswith("Decoder")
or name.endswith("Encoder")
or name.endswith("Layer")
or name.endswith("Embeddings")
or name.endswith("Attention")
):
return True
# Submodules are not documented.
if os.path.isdir(os.path.join(PATH_TO_TRANSFORMERS, name)) or os.path.isfile(
os.path.join(PATH_TO_TRANSFORMERS, f"{name}.py")
):
return True
# All load functions are not documented.
if name.startswith("load_tf") or name.startswith("load_pytorch"):
return True
# is_xxx_available functions are not documented.
if name.startswith("is_") and name.endswith("_available"):
return True
# Deprecated objects are not documented.
if name in DEPRECATED_OBJECTS or name in UNDOCUMENTED_OBJECTS:
return True
# MMBT model does not really work.
if name.startswith("MMBT"):
return True
if name in SHOULD_HAVE_THEIR_OWN_PAGE:
return True
return False
def check_all_objects_are_documented():
"""Check all models are properly documented."""
documented_objs = find_all_documented_objects()
modules = transformers._modules
objects = [c for c in dir(transformers) if c not in modules and not c.startswith("_")]
undocumented_objs = [c for c in objects if c not in documented_objs and not ignore_undocumented(c)]
if len(undocumented_objs) > 0:
raise Exception(
"The following objects are in the public init so should be documented:\n - "
+ "\n - ".join(undocumented_objs)
)
check_docstrings_are_in_md()
check_model_type_doc_match()
def check_model_type_doc_match():
"""Check all doc pages have a corresponding model type."""
model_doc_folder = Path(PATH_TO_DOC) / "model_doc"
model_docs = [m.stem for m in model_doc_folder.glob("*.mdx")]
model_types = list(transformers.models.auto.configuration_auto.MODEL_NAMES_MAPPING.keys())
model_types = [MODEL_TYPE_TO_DOC_MAPPING[m] if m in MODEL_TYPE_TO_DOC_MAPPING else m for m in model_types]
errors = []
for m in model_docs:
if m not in model_types and m != "auto":
close_matches = get_close_matches(m, model_types)
error_message = f"{m} is not a proper model identifier."
if len(close_matches) > 0:
close_matches = "/".join(close_matches)
error_message += f" Did you mean {close_matches}?"
errors.append(error_message)
if len(errors) > 0:
raise ValueError(
"Some model doc pages do not match any existing model type:\n"
+ "\n".join(errors)
+ "\nYou can add any missing model type to the `MODEL_NAMES_MAPPING` constant in "
"models/auto/configuration_auto.py."
)
# Re pattern to catch :obj:`xx`, :class:`xx`, :func:`xx` or :meth:`xx`.
_re_rst_special_words = re.compile(r":(?:obj|func|class|meth):`([^`]+)`")
# Re pattern to catch things between double backquotes.
_re_double_backquotes = re.compile(r"(^|[^`])``([^`]+)``([^`]|$)")
# Re pattern to catch example introduction.
_re_rst_example = re.compile(r"^\s*Example.*::\s*$", flags=re.MULTILINE)
def is_rst_docstring(docstring):
"""
Returns `True` if `docstring` is written in rst.
"""
if _re_rst_special_words.search(docstring) is not None:
return True
if _re_double_backquotes.search(docstring) is not None:
return True
if _re_rst_example.search(docstring) is not None:
return True
return False
def check_docstrings_are_in_md():
"""Check all docstrings are in md"""
files_with_rst = []
for file in Path(PATH_TO_TRANSFORMERS).glob("**/*.py"):
with open(file, encoding="utf-8") as f:
code = f.read()
docstrings = code.split('"""')
for idx, docstring in enumerate(docstrings):
if idx % 2 == 0 or not is_rst_docstring(docstring):
continue
files_with_rst.append(file)
break
if len(files_with_rst) > 0:
raise ValueError(
"The following files have docstrings written in rst:\n"
+ "\n".join([f"- {f}" for f in files_with_rst])
+ "\nTo fix this run `doc-builder convert path_to_py_file` after installing `doc-builder`\n"
"(`pip install git+https://github.com/huggingface/doc-builder`)"
)
def check_repo_quality():
"""Check all models are properly tested and documented."""
print("Checking all models are included.")
check_model_list()
print("Checking all models are public.")
check_models_are_in_init()
print("Checking all models are properly tested.")
check_all_decorator_order()
check_all_models_are_tested()
print("Checking all objects are properly documented.")
check_all_objects_are_documented()
print("Checking all models are in at least one auto class.")
check_all_models_are_auto_configured()
if __name__ == "__main__":
check_repo_quality()
| # coding=utf-8
# Copyright 2020 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import importlib
import inspect
import os
import re
import warnings
from collections import OrderedDict
from difflib import get_close_matches
from pathlib import Path
from transformers import is_flax_available, is_tf_available, is_torch_available
from transformers.models.auto import get_values
from transformers.utils import ENV_VARS_TRUE_VALUES
# All paths are set with the intent you should run this script from the root of the repo with the command
# python utils/check_repo.py
PATH_TO_TRANSFORMERS = "src/transformers"
PATH_TO_TESTS = "tests"
PATH_TO_DOC = "docs/source/en"
# Update this list with models that are supposed to be private.
PRIVATE_MODELS = [
"DPRSpanPredictor",
"LongT5Stack",
"RealmBertModel",
"T5Stack",
"SwitchTransformersStack",
"TFDPRSpanPredictor",
"MaskFormerSwinModel",
"MaskFormerSwinPreTrainedModel",
]
# Update this list for models that are not tested with a comment explaining the reason it should not be.
# Being in this list is an exception and should **not** be the rule.
IGNORE_NON_TESTED = PRIVATE_MODELS.copy() + [
# models to ignore for not tested
"CLIPSegDecoder", # Building part of bigger (tested) model.
"TableTransformerEncoder", # Building part of bigger (tested) model.
"TableTransformerDecoder", # Building part of bigger (tested) model.
"TimeSeriesTransformerEncoder", # Building part of bigger (tested) model.
"TimeSeriesTransformerDecoder", # Building part of bigger (tested) model.
"JukeboxVQVAE", # Building part of bigger (tested) model.
"JukeboxPrior", # Building part of bigger (tested) model.
"DeformableDetrEncoder", # Building part of bigger (tested) model.
"DeformableDetrDecoder", # Building part of bigger (tested) model.
"OPTDecoder", # Building part of bigger (tested) model.
"WhisperDecoder", # Building part of bigger (tested) model.
"WhisperEncoder", # Building part of bigger (tested) model.
"DecisionTransformerGPT2Model", # Building part of bigger (tested) model.
"SegformerDecodeHead", # Building part of bigger (tested) model.
"PLBartEncoder", # Building part of bigger (tested) model.
"PLBartDecoder", # Building part of bigger (tested) model.
"PLBartDecoderWrapper", # Building part of bigger (tested) model.
"BigBirdPegasusEncoder", # Building part of bigger (tested) model.
"BigBirdPegasusDecoder", # Building part of bigger (tested) model.
"BigBirdPegasusDecoderWrapper", # Building part of bigger (tested) model.
"DetrEncoder", # Building part of bigger (tested) model.
"DetrDecoder", # Building part of bigger (tested) model.
"DetrDecoderWrapper", # Building part of bigger (tested) model.
"ConditionalDetrEncoder", # Building part of bigger (tested) model.
"ConditionalDetrDecoder", # Building part of bigger (tested) model.
"M2M100Encoder", # Building part of bigger (tested) model.
"M2M100Decoder", # Building part of bigger (tested) model.
"MCTCTEncoder", # Building part of bigger (tested) model.
"Speech2TextEncoder", # Building part of bigger (tested) model.
"Speech2TextDecoder", # Building part of bigger (tested) model.
"LEDEncoder", # Building part of bigger (tested) model.
"LEDDecoder", # Building part of bigger (tested) model.
"BartDecoderWrapper", # Building part of bigger (tested) model.
"BartEncoder", # Building part of bigger (tested) model.
"BertLMHeadModel", # Needs to be setup as decoder.
"BlenderbotSmallEncoder", # Building part of bigger (tested) model.
"BlenderbotSmallDecoderWrapper", # Building part of bigger (tested) model.
"BlenderbotEncoder", # Building part of bigger (tested) model.
"BlenderbotDecoderWrapper", # Building part of bigger (tested) model.
"MBartEncoder", # Building part of bigger (tested) model.
"MBartDecoderWrapper", # Building part of bigger (tested) model.
"MegatronBertLMHeadModel", # Building part of bigger (tested) model.
"MegatronBertEncoder", # Building part of bigger (tested) model.
"MegatronBertDecoder", # Building part of bigger (tested) model.
"MegatronBertDecoderWrapper", # Building part of bigger (tested) model.
"MvpDecoderWrapper", # Building part of bigger (tested) model.
"MvpEncoder", # Building part of bigger (tested) model.
"PegasusEncoder", # Building part of bigger (tested) model.
"PegasusDecoderWrapper", # Building part of bigger (tested) model.
"PegasusXEncoder", # Building part of bigger (tested) model.
"PegasusXDecoder", # Building part of bigger (tested) model.
"PegasusXDecoderWrapper", # Building part of bigger (tested) model.
"DPREncoder", # Building part of bigger (tested) model.
"ProphetNetDecoderWrapper", # Building part of bigger (tested) model.
"RealmBertModel", # Building part of bigger (tested) model.
"RealmReader", # Not regular model.
"RealmScorer", # Not regular model.
"RealmForOpenQA", # Not regular model.
"ReformerForMaskedLM", # Needs to be setup as decoder.
"Speech2Text2DecoderWrapper", # Building part of bigger (tested) model.
"TFDPREncoder", # Building part of bigger (tested) model.
"TFElectraMainLayer", # Building part of bigger (tested) model (should it be a TFPreTrainedModel ?)
"TFRobertaForMultipleChoice", # TODO: fix
"TrOCRDecoderWrapper", # Building part of bigger (tested) model.
"TFWhisperEncoder", # Building part of bigger (tested) model.
"TFWhisperDecoder", # Building part of bigger (tested) model.
"SeparableConv1D", # Building part of bigger (tested) model.
"FlaxBartForCausalLM", # Building part of bigger (tested) model.
"FlaxBertForCausalLM", # Building part of bigger (tested) model. Tested implicitly through FlaxRobertaForCausalLM.
"OPTDecoderWrapper",
"TFSegformerDecodeHead", # Not a regular model.
]
# Update this list with test files that don't have a tester with a `all_model_classes` variable and which don't
# trigger the common tests.
TEST_FILES_WITH_NO_COMMON_TESTS = [
"models/decision_transformer/test_modeling_decision_transformer.py",
"models/camembert/test_modeling_camembert.py",
"models/mt5/test_modeling_flax_mt5.py",
"models/mbart/test_modeling_mbart.py",
"models/mt5/test_modeling_mt5.py",
"models/pegasus/test_modeling_pegasus.py",
"models/camembert/test_modeling_tf_camembert.py",
"models/mt5/test_modeling_tf_mt5.py",
"models/xlm_roberta/test_modeling_tf_xlm_roberta.py",
"models/xlm_roberta/test_modeling_flax_xlm_roberta.py",
"models/xlm_prophetnet/test_modeling_xlm_prophetnet.py",
"models/xlm_roberta/test_modeling_xlm_roberta.py",
"models/vision_text_dual_encoder/test_modeling_vision_text_dual_encoder.py",
"models/vision_text_dual_encoder/test_modeling_flax_vision_text_dual_encoder.py",
"models/decision_transformer/test_modeling_decision_transformer.py",
]
# Update this list for models that are not in any of the auto MODEL_XXX_MAPPING. Being in this list is an exception and
# should **not** be the rule.
IGNORE_NON_AUTO_CONFIGURED = PRIVATE_MODELS.copy() + [
# models to ignore for model xxx mapping
"CLIPSegForImageSegmentation",
"CLIPSegVisionModel",
"CLIPSegTextModel",
"EsmForProteinFolding",
"TimeSeriesTransformerForPrediction",
"JukeboxVQVAE",
"JukeboxPrior",
"PegasusXEncoder",
"PegasusXDecoder",
"PegasusXDecoderWrapper",
"PegasusXEncoder",
"PegasusXDecoder",
"PegasusXDecoderWrapper",
"DPTForDepthEstimation",
"DecisionTransformerGPT2Model",
"GLPNForDepthEstimation",
"ViltForImagesAndTextClassification",
"ViltForImageAndTextRetrieval",
"ViltForTokenClassification",
"ViltForMaskedLM",
"XGLMEncoder",
"XGLMDecoder",
"XGLMDecoderWrapper",
"PerceiverForMultimodalAutoencoding",
"PerceiverForOpticalFlow",
"SegformerDecodeHead",
"TFSegformerDecodeHead",
"FlaxBeitForMaskedImageModeling",
"PLBartEncoder",
"PLBartDecoder",
"PLBartDecoderWrapper",
"BeitForMaskedImageModeling",
"ChineseCLIPTextModel",
"ChineseCLIPVisionModel",
"CLIPTextModel",
"CLIPTextModelWithProjection",
"CLIPVisionModel",
"CLIPVisionModelWithProjection",
"GroupViTTextModel",
"GroupViTVisionModel",
"TFCLIPTextModel",
"TFCLIPVisionModel",
"TFGroupViTTextModel",
"TFGroupViTVisionModel",
"FlaxCLIPTextModel",
"FlaxCLIPVisionModel",
"FlaxWav2Vec2ForCTC",
"DetrForSegmentation",
"ConditionalDetrForSegmentation",
"DPRReader",
"FlaubertForQuestionAnswering",
"FlavaImageCodebook",
"FlavaTextModel",
"FlavaImageModel",
"FlavaMultimodalModel",
"GPT2DoubleHeadsModel",
"GPTSw3DoubleHeadsModel",
"LayoutLMForQuestionAnswering",
"LukeForMaskedLM",
"LukeForEntityClassification",
"LukeForEntityPairClassification",
"LukeForEntitySpanClassification",
"OpenAIGPTDoubleHeadsModel",
"OwlViTTextModel",
"OwlViTVisionModel",
"OwlViTForObjectDetection",
"RagModel",
"RagSequenceForGeneration",
"RagTokenForGeneration",
"RealmEmbedder",
"RealmForOpenQA",
"RealmScorer",
"RealmReader",
"TFDPRReader",
"TFGPT2DoubleHeadsModel",
"TFLayoutLMForQuestionAnswering",
"TFOpenAIGPTDoubleHeadsModel",
"TFRagModel",
"TFRagSequenceForGeneration",
"TFRagTokenForGeneration",
"Wav2Vec2ForCTC",
"HubertForCTC",
"SEWForCTC",
"SEWDForCTC",
"XLMForQuestionAnswering",
"XLNetForQuestionAnswering",
"SeparableConv1D",
"VisualBertForRegionToPhraseAlignment",
"VisualBertForVisualReasoning",
"VisualBertForQuestionAnswering",
"VisualBertForMultipleChoice",
"TFWav2Vec2ForCTC",
"TFHubertForCTC",
"MaskFormerForInstanceSegmentation",
"XCLIPVisionModel",
"XCLIPTextModel",
]
# Update this list for models that have multiple model types for the same
# model doc
MODEL_TYPE_TO_DOC_MAPPING = OrderedDict(
[
("data2vec-text", "data2vec"),
("data2vec-audio", "data2vec"),
("data2vec-vision", "data2vec"),
("donut-swin", "donut"),
]
)
# This is to make sure the transformers module imported is the one in the repo.
spec = importlib.util.spec_from_file_location(
"transformers",
os.path.join(PATH_TO_TRANSFORMERS, "__init__.py"),
submodule_search_locations=[PATH_TO_TRANSFORMERS],
)
transformers = spec.loader.load_module()
def check_model_list():
"""Check the model list inside the transformers library."""
# Get the models from the directory structure of `src/transformers/models/`
models_dir = os.path.join(PATH_TO_TRANSFORMERS, "models")
_models = []
for model in os.listdir(models_dir):
model_dir = os.path.join(models_dir, model)
if os.path.isdir(model_dir) and "__init__.py" in os.listdir(model_dir):
_models.append(model)
# Get the models from the directory structure of `src/transformers/models/`
models = [model for model in dir(transformers.models) if not model.startswith("__")]
missing_models = sorted(list(set(_models).difference(models)))
if missing_models:
raise Exception(
f"The following models should be included in {models_dir}/__init__.py: {','.join(missing_models)}."
)
# If some modeling modules should be ignored for all checks, they should be added in the nested list
# _ignore_modules of this function.
def get_model_modules():
"""Get the model modules inside the transformers library."""
_ignore_modules = [
"modeling_auto",
"modeling_encoder_decoder",
"modeling_marian",
"modeling_mmbt",
"modeling_outputs",
"modeling_retribert",
"modeling_utils",
"modeling_flax_auto",
"modeling_flax_encoder_decoder",
"modeling_flax_utils",
"modeling_speech_encoder_decoder",
"modeling_flax_speech_encoder_decoder",
"modeling_flax_vision_encoder_decoder",
"modeling_transfo_xl_utilities",
"modeling_tf_auto",
"modeling_tf_encoder_decoder",
"modeling_tf_outputs",
"modeling_tf_pytorch_utils",
"modeling_tf_utils",
"modeling_tf_transfo_xl_utilities",
"modeling_tf_vision_encoder_decoder",
"modeling_vision_encoder_decoder",
]
modules = []
for model in dir(transformers.models):
# There are some magic dunder attributes in the dir, we ignore them
if not model.startswith("__"):
model_module = getattr(transformers.models, model)
for submodule in dir(model_module):
if submodule.startswith("modeling") and submodule not in _ignore_modules:
modeling_module = getattr(model_module, submodule)
if inspect.ismodule(modeling_module):
modules.append(modeling_module)
return modules
def get_models(module, include_pretrained=False):
"""Get the objects in module that are models."""
models = []
model_classes = (transformers.PreTrainedModel, transformers.TFPreTrainedModel, transformers.FlaxPreTrainedModel)
for attr_name in dir(module):
if not include_pretrained and ("Pretrained" in attr_name or "PreTrained" in attr_name):
continue
attr = getattr(module, attr_name)
if isinstance(attr, type) and issubclass(attr, model_classes) and attr.__module__ == module.__name__:
models.append((attr_name, attr))
return models
def is_a_private_model(model):
"""Returns True if the model should not be in the main init."""
if model in PRIVATE_MODELS:
return True
# Wrapper, Encoder and Decoder are all privates
if model.endswith("Wrapper"):
return True
if model.endswith("Encoder"):
return True
if model.endswith("Decoder"):
return True
return False
def check_models_are_in_init():
"""Checks all models defined in the library are in the main init."""
models_not_in_init = []
dir_transformers = dir(transformers)
for module in get_model_modules():
models_not_in_init += [
model[0] for model in get_models(module, include_pretrained=True) if model[0] not in dir_transformers
]
# Remove private models
models_not_in_init = [model for model in models_not_in_init if not is_a_private_model(model)]
if len(models_not_in_init) > 0:
raise Exception(f"The following models should be in the main init: {','.join(models_not_in_init)}.")
# If some test_modeling files should be ignored when checking models are all tested, they should be added in the
# nested list _ignore_files of this function.
def get_model_test_files():
"""Get the model test files.
The returned files should NOT contain the `tests` (i.e. `PATH_TO_TESTS` defined in this script). They will be
considered as paths relative to `tests`. A caller has to use `os.path.join(PATH_TO_TESTS, ...)` to access the files.
"""
_ignore_files = [
"test_modeling_common",
"test_modeling_encoder_decoder",
"test_modeling_flax_encoder_decoder",
"test_modeling_flax_speech_encoder_decoder",
"test_modeling_marian",
"test_modeling_tf_common",
"test_modeling_tf_encoder_decoder",
]
test_files = []
# Check both `PATH_TO_TESTS` and `PATH_TO_TESTS/models`
model_test_root = os.path.join(PATH_TO_TESTS, "models")
model_test_dirs = []
for x in os.listdir(model_test_root):
x = os.path.join(model_test_root, x)
if os.path.isdir(x):
model_test_dirs.append(x)
for target_dir in [PATH_TO_TESTS] + model_test_dirs:
for file_or_dir in os.listdir(target_dir):
path = os.path.join(target_dir, file_or_dir)
if os.path.isfile(path):
filename = os.path.split(path)[-1]
if "test_modeling" in filename and not os.path.splitext(filename)[0] in _ignore_files:
file = os.path.join(*path.split(os.sep)[1:])
test_files.append(file)
return test_files
# This is a bit hacky but I didn't find a way to import the test_file as a module and read inside the tester class
# for the all_model_classes variable.
def find_tested_models(test_file):
"""Parse the content of test_file to detect what's in all_model_classes"""
# This is a bit hacky but I didn't find a way to import the test_file as a module and read inside the class
with open(os.path.join(PATH_TO_TESTS, test_file), "r", encoding="utf-8", newline="\n") as f:
content = f.read()
all_models = re.findall(r"all_model_classes\s+=\s+\(\s*\(([^\)]*)\)", content)
# Check with one less parenthesis as well
all_models += re.findall(r"all_model_classes\s+=\s+\(([^\)]*)\)", content)
if len(all_models) > 0:
model_tested = []
for entry in all_models:
for line in entry.split(","):
name = line.strip()
if len(name) > 0:
model_tested.append(name)
return model_tested
def check_models_are_tested(module, test_file):
"""Check models defined in module are tested in test_file."""
# XxxPreTrainedModel are not tested
defined_models = get_models(module)
tested_models = find_tested_models(test_file)
if tested_models is None:
if test_file.replace(os.path.sep, "/") in TEST_FILES_WITH_NO_COMMON_TESTS:
return
return [
f"{test_file} should define `all_model_classes` to apply common tests to the models it tests. "
+ "If this intentional, add the test filename to `TEST_FILES_WITH_NO_COMMON_TESTS` in the file "
+ "`utils/check_repo.py`."
]
failures = []
for model_name, _ in defined_models:
if model_name not in tested_models and model_name not in IGNORE_NON_TESTED:
failures.append(
f"{model_name} is defined in {module.__name__} but is not tested in "
+ f"{os.path.join(PATH_TO_TESTS, test_file)}. Add it to the all_model_classes in that file."
+ "If common tests should not applied to that model, add its name to `IGNORE_NON_TESTED`"
+ "in the file `utils/check_repo.py`."
)
return failures
def check_all_models_are_tested():
"""Check all models are properly tested."""
modules = get_model_modules()
test_files = get_model_test_files()
failures = []
for module in modules:
test_file = [file for file in test_files if f"test_{module.__name__.split('.')[-1]}.py" in file]
if len(test_file) == 0:
failures.append(f"{module.__name__} does not have its corresponding test file {test_file}.")
elif len(test_file) > 1:
failures.append(f"{module.__name__} has several test files: {test_file}.")
else:
test_file = test_file[0]
new_failures = check_models_are_tested(module, test_file)
if new_failures is not None:
failures += new_failures
if len(failures) > 0:
raise Exception(f"There were {len(failures)} failures:\n" + "\n".join(failures))
def get_all_auto_configured_models():
"""Return the list of all models in at least one auto class."""
result = set() # To avoid duplicates we concatenate all model classes in a set.
if is_torch_available():
for attr_name in dir(transformers.models.auto.modeling_auto):
if attr_name.startswith("MODEL_") and attr_name.endswith("MAPPING_NAMES"):
result = result | set(get_values(getattr(transformers.models.auto.modeling_auto, attr_name)))
if is_tf_available():
for attr_name in dir(transformers.models.auto.modeling_tf_auto):
if attr_name.startswith("TF_MODEL_") and attr_name.endswith("MAPPING_NAMES"):
result = result | set(get_values(getattr(transformers.models.auto.modeling_tf_auto, attr_name)))
if is_flax_available():
for attr_name in dir(transformers.models.auto.modeling_flax_auto):
if attr_name.startswith("FLAX_MODEL_") and attr_name.endswith("MAPPING_NAMES"):
result = result | set(get_values(getattr(transformers.models.auto.modeling_flax_auto, attr_name)))
return [cls for cls in result]
def ignore_unautoclassed(model_name):
"""Rules to determine if `name` should be in an auto class."""
# Special white list
if model_name in IGNORE_NON_AUTO_CONFIGURED:
return True
# Encoder and Decoder should be ignored
if "Encoder" in model_name or "Decoder" in model_name:
return True
return False
def check_models_are_auto_configured(module, all_auto_models):
"""Check models defined in module are each in an auto class."""
defined_models = get_models(module)
failures = []
for model_name, _ in defined_models:
if model_name not in all_auto_models and not ignore_unautoclassed(model_name):
failures.append(
f"{model_name} is defined in {module.__name__} but is not present in any of the auto mapping. "
"If that is intended behavior, add its name to `IGNORE_NON_AUTO_CONFIGURED` in the file "
"`utils/check_repo.py`."
)
return failures
def check_all_models_are_auto_configured():
"""Check all models are each in an auto class."""
missing_backends = []
if not is_torch_available():
missing_backends.append("PyTorch")
if not is_tf_available():
missing_backends.append("TensorFlow")
if not is_flax_available():
missing_backends.append("Flax")
if len(missing_backends) > 0:
missing = ", ".join(missing_backends)
if os.getenv("TRANSFORMERS_IS_CI", "").upper() in ENV_VARS_TRUE_VALUES:
raise Exception(
"Full quality checks require all backends to be installed (with `pip install -e .[dev]` in the "
f"Transformers repo, the following are missing: {missing}."
)
else:
warnings.warn(
"Full quality checks require all backends to be installed (with `pip install -e .[dev]` in the "
f"Transformers repo, the following are missing: {missing}. While it's probably fine as long as you "
"didn't make any change in one of those backends modeling files, you should probably execute the "
"command above to be on the safe side."
)
modules = get_model_modules()
all_auto_models = get_all_auto_configured_models()
failures = []
for module in modules:
new_failures = check_models_are_auto_configured(module, all_auto_models)
if new_failures is not None:
failures += new_failures
if len(failures) > 0:
raise Exception(f"There were {len(failures)} failures:\n" + "\n".join(failures))
_re_decorator = re.compile(r"^\s*@(\S+)\s+$")
def check_decorator_order(filename):
"""Check that in the test file `filename` the slow decorator is always last."""
with open(filename, "r", encoding="utf-8", newline="\n") as f:
lines = f.readlines()
decorator_before = None
errors = []
for i, line in enumerate(lines):
search = _re_decorator.search(line)
if search is not None:
decorator_name = search.groups()[0]
if decorator_before is not None and decorator_name.startswith("parameterized"):
errors.append(i)
decorator_before = decorator_name
elif decorator_before is not None:
decorator_before = None
return errors
def check_all_decorator_order():
"""Check that in all test files, the slow decorator is always last."""
errors = []
for fname in os.listdir(PATH_TO_TESTS):
if fname.endswith(".py"):
filename = os.path.join(PATH_TO_TESTS, fname)
new_errors = check_decorator_order(filename)
errors += [f"- {filename}, line {i}" for i in new_errors]
if len(errors) > 0:
msg = "\n".join(errors)
raise ValueError(
"The parameterized decorator (and its variants) should always be first, but this is not the case in the"
f" following files:\n{msg}"
)
def find_all_documented_objects():
"""Parse the content of all doc files to detect which classes and functions it documents"""
documented_obj = []
for doc_file in Path(PATH_TO_DOC).glob("**/*.rst"):
with open(doc_file, "r", encoding="utf-8", newline="\n") as f:
content = f.read()
raw_doc_objs = re.findall(r"(?:autoclass|autofunction):: transformers.(\S+)\s+", content)
documented_obj += [obj.split(".")[-1] for obj in raw_doc_objs]
for doc_file in Path(PATH_TO_DOC).glob("**/*.mdx"):
with open(doc_file, "r", encoding="utf-8", newline="\n") as f:
content = f.read()
raw_doc_objs = re.findall("\[\[autodoc\]\]\s+(\S+)\s+", content)
documented_obj += [obj.split(".")[-1] for obj in raw_doc_objs]
return documented_obj
# One good reason for not being documented is to be deprecated. Put in this list deprecated objects.
DEPRECATED_OBJECTS = [
"AutoModelWithLMHead",
"BartPretrainedModel",
"DataCollator",
"DataCollatorForSOP",
"GlueDataset",
"GlueDataTrainingArguments",
"LineByLineTextDataset",
"LineByLineWithRefDataset",
"LineByLineWithSOPTextDataset",
"PretrainedBartModel",
"PretrainedFSMTModel",
"SingleSentenceClassificationProcessor",
"SquadDataTrainingArguments",
"SquadDataset",
"SquadExample",
"SquadFeatures",
"SquadV1Processor",
"SquadV2Processor",
"TFAutoModelWithLMHead",
"TFBartPretrainedModel",
"TextDataset",
"TextDatasetForNextSentencePrediction",
"Wav2Vec2ForMaskedLM",
"Wav2Vec2Tokenizer",
"glue_compute_metrics",
"glue_convert_examples_to_features",
"glue_output_modes",
"glue_processors",
"glue_tasks_num_labels",
"squad_convert_examples_to_features",
"xnli_compute_metrics",
"xnli_output_modes",
"xnli_processors",
"xnli_tasks_num_labels",
"TFTrainer",
"TFTrainingArguments",
]
# Exceptionally, some objects should not be documented after all rules passed.
# ONLY PUT SOMETHING IN THIS LIST AS A LAST RESORT!
UNDOCUMENTED_OBJECTS = [
"AddedToken", # This is a tokenizers class.
"BasicTokenizer", # Internal, should never have been in the main init.
"CharacterTokenizer", # Internal, should never have been in the main init.
"DPRPretrainedReader", # Like an Encoder.
"DummyObject", # Just picked by mistake sometimes.
"MecabTokenizer", # Internal, should never have been in the main init.
"ModelCard", # Internal type.
"SqueezeBertModule", # Internal building block (should have been called SqueezeBertLayer)
"TFDPRPretrainedReader", # Like an Encoder.
"TransfoXLCorpus", # Internal type.
"WordpieceTokenizer", # Internal, should never have been in the main init.
"absl", # External module
"add_end_docstrings", # Internal, should never have been in the main init.
"add_start_docstrings", # Internal, should never have been in the main init.
"convert_tf_weight_name_to_pt_weight_name", # Internal used to convert model weights
"logger", # Internal logger
"logging", # External module
"requires_backends", # Internal function
]
# This list should be empty. Objects in it should get their own doc page.
SHOULD_HAVE_THEIR_OWN_PAGE = [
# Benchmarks
"PyTorchBenchmark",
"PyTorchBenchmarkArguments",
"TensorFlowBenchmark",
"TensorFlowBenchmarkArguments",
"BitBackbone",
"MaskFormerSwinBackbone",
"ResNetBackbone",
"AutoBackbone",
"MaskFormerSwinConfig",
"MaskFormerSwinModel",
]
def ignore_undocumented(name):
"""Rules to determine if `name` should be undocumented."""
# NOT DOCUMENTED ON PURPOSE.
# Constants uppercase are not documented.
if name.isupper():
return True
# PreTrainedModels / Encoders / Decoders / Layers / Embeddings / Attention are not documented.
if (
name.endswith("PreTrainedModel")
or name.endswith("Decoder")
or name.endswith("Encoder")
or name.endswith("Layer")
or name.endswith("Embeddings")
or name.endswith("Attention")
):
return True
# Submodules are not documented.
if os.path.isdir(os.path.join(PATH_TO_TRANSFORMERS, name)) or os.path.isfile(
os.path.join(PATH_TO_TRANSFORMERS, f"{name}.py")
):
return True
# All load functions are not documented.
if name.startswith("load_tf") or name.startswith("load_pytorch"):
return True
# is_xxx_available functions are not documented.
if name.startswith("is_") and name.endswith("_available"):
return True
# Deprecated objects are not documented.
if name in DEPRECATED_OBJECTS or name in UNDOCUMENTED_OBJECTS:
return True
# MMBT model does not really work.
if name.startswith("MMBT"):
return True
if name in SHOULD_HAVE_THEIR_OWN_PAGE:
return True
return False
def check_all_objects_are_documented():
"""Check all models are properly documented."""
documented_objs = find_all_documented_objects()
modules = transformers._modules
objects = [c for c in dir(transformers) if c not in modules and not c.startswith("_")]
undocumented_objs = [c for c in objects if c not in documented_objs and not ignore_undocumented(c)]
if len(undocumented_objs) > 0:
raise Exception(
"The following objects are in the public init so should be documented:\n - "
+ "\n - ".join(undocumented_objs)
)
check_docstrings_are_in_md()
check_model_type_doc_match()
def check_model_type_doc_match():
"""Check all doc pages have a corresponding model type."""
model_doc_folder = Path(PATH_TO_DOC) / "model_doc"
model_docs = [m.stem for m in model_doc_folder.glob("*.mdx")]
model_types = list(transformers.models.auto.configuration_auto.MODEL_NAMES_MAPPING.keys())
model_types = [MODEL_TYPE_TO_DOC_MAPPING[m] if m in MODEL_TYPE_TO_DOC_MAPPING else m for m in model_types]
errors = []
for m in model_docs:
if m not in model_types and m != "auto":
close_matches = get_close_matches(m, model_types)
error_message = f"{m} is not a proper model identifier."
if len(close_matches) > 0:
close_matches = "/".join(close_matches)
error_message += f" Did you mean {close_matches}?"
errors.append(error_message)
if len(errors) > 0:
raise ValueError(
"Some model doc pages do not match any existing model type:\n"
+ "\n".join(errors)
+ "\nYou can add any missing model type to the `MODEL_NAMES_MAPPING` constant in "
"models/auto/configuration_auto.py."
)
# Re pattern to catch :obj:`xx`, :class:`xx`, :func:`xx` or :meth:`xx`.
_re_rst_special_words = re.compile(r":(?:obj|func|class|meth):`([^`]+)`")
# Re pattern to catch things between double backquotes.
_re_double_backquotes = re.compile(r"(^|[^`])``([^`]+)``([^`]|$)")
# Re pattern to catch example introduction.
_re_rst_example = re.compile(r"^\s*Example.*::\s*$", flags=re.MULTILINE)
def is_rst_docstring(docstring):
"""
Returns `True` if `docstring` is written in rst.
"""
if _re_rst_special_words.search(docstring) is not None:
return True
if _re_double_backquotes.search(docstring) is not None:
return True
if _re_rst_example.search(docstring) is not None:
return True
return False
def check_docstrings_are_in_md():
"""Check all docstrings are in md"""
files_with_rst = []
for file in Path(PATH_TO_TRANSFORMERS).glob("**/*.py"):
with open(file, encoding="utf-8") as f:
code = f.read()
docstrings = code.split('"""')
for idx, docstring in enumerate(docstrings):
if idx % 2 == 0 or not is_rst_docstring(docstring):
continue
files_with_rst.append(file)
break
if len(files_with_rst) > 0:
raise ValueError(
"The following files have docstrings written in rst:\n"
+ "\n".join([f"- {f}" for f in files_with_rst])
+ "\nTo fix this run `doc-builder convert path_to_py_file` after installing `doc-builder`\n"
"(`pip install git+https://github.com/huggingface/doc-builder`)"
)
def check_repo_quality():
"""Check all models are properly tested and documented."""
print("Checking all models are included.")
check_model_list()
print("Checking all models are public.")
check_models_are_in_init()
print("Checking all models are properly tested.")
check_all_decorator_order()
check_all_models_are_tested()
print("Checking all objects are properly documented.")
check_all_objects_are_documented()
print("Checking all models are in at least one auto class.")
check_all_models_are_auto_configured()
if __name__ == "__main__":
check_repo_quality()
| 1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/layoutlmv2/image_processing_layoutlmv2.py | # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Image processor class for LayoutLMv2."""
from typing import Dict, Optional, Union
import numpy as np
from transformers.utils import is_vision_available
from transformers.utils.generic import TensorType
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
ChannelDimension,
ImageInput,
PILImageResampling,
infer_channel_dimension_format,
is_batched,
to_numpy_array,
valid_images,
)
from ...utils import is_pytesseract_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
logger = logging.get_logger(__name__)
def normalize_box(box, width, height):
return [
int(1000 * (box[0] / width)),
int(1000 * (box[1] / height)),
int(1000 * (box[2] / width)),
int(1000 * (box[3] / height)),
]
def apply_tesseract(image: np.ndarray, lang: Optional[str], tesseract_config: Optional[str] = None):
"""Applies Tesseract OCR on a document image, and returns recognized words + normalized bounding boxes."""
tesseract_config = tesseract_config if tesseract_config is not None else ""
# apply OCR
pil_image = to_pil_image(image)
image_width, image_height = pil_image.size
data = pytesseract.image_to_data(pil_image, lang=lang, output_type="dict", config=tesseract_config)
words, left, top, width, height = data["text"], data["left"], data["top"], data["width"], data["height"]
# filter empty words and corresponding coordinates
irrelevant_indices = [idx for idx, word in enumerate(words) if not word.strip()]
words = [word for idx, word in enumerate(words) if idx not in irrelevant_indices]
left = [coord for idx, coord in enumerate(left) if idx not in irrelevant_indices]
top = [coord for idx, coord in enumerate(top) if idx not in irrelevant_indices]
width = [coord for idx, coord in enumerate(width) if idx not in irrelevant_indices]
height = [coord for idx, coord in enumerate(height) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
actual_boxes = []
for x, y, w, h in zip(left, top, width, height):
actual_box = [x, y, x + w, y + h]
actual_boxes.append(actual_box)
# finally, normalize the bounding boxes
normalized_boxes = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(box, image_width, image_height))
assert len(words) == len(normalized_boxes), "Not as many words as there are bounding boxes"
return words, normalized_boxes
def flip_channel_order(image: np.ndarray, data_format: Optional[ChannelDimension] = None) -> np.ndarray:
input_data_format = infer_channel_dimension_format(image)
if input_data_format == ChannelDimension.LAST:
image = image[..., ::-1]
elif input_data_format == ChannelDimension.FIRST:
image = image[:, ::-1, ...]
else:
raise ValueError(f"Unsupported channel dimension: {input_data_format}")
if data_format is not None:
image = to_channel_dimension_format(image, data_format)
return image
class LayoutLMv2ImageProcessor(BaseImageProcessor):
r"""
Constructs a LayoutLMv2 image processor.
Args:
do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the image's (height, width) dimensions to `(size["height"], size["width"])`. Can be
overridden by `do_resize` in `preprocess`.
size (`Dict[str, int]` *optional*, defaults to `{"height": 224, "width": 224}`):
Size of the image after resizing. Can be overridden by `size` in `preprocess`.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BILINEAR`):
Resampling filter to use if resizing the image. Can be overridden by the `resample` parameter in the
`preprocess` method.
apply_ocr (`bool`, *optional*, defaults to `True`):
Whether to apply the Tesseract OCR engine to get words + normalized bounding boxes. Can be overridden by
`apply_ocr` in `preprocess`.
ocr_lang (`str`, *optional*):
The language, specified by its ISO code, to be used by the Tesseract OCR engine. By default, English is
used. Can be overridden by `ocr_lang` in `preprocess`.
tesseract_config (`str`, *optional*):
Any additional custom configuration flags that are forwarded to the `config` parameter when calling
Tesseract. For example: '--psm 6'. Can be overridden by `tesseract_config` in `preprocess`.
"""
model_input_names = ["pixel_values"]
def __init__(
self,
do_resize: bool = True,
size: Dict[str, int] = None,
resample: PILImageResampling = PILImageResampling.BILINEAR,
apply_ocr: bool = True,
ocr_lang: Optional[str] = None,
tesseract_config: Optional[str] = "",
**kwargs
) -> None:
super().__init__(**kwargs)
size = size if size is not None else {"height": 224, "width": 224}
size = get_size_dict(size)
self.do_resize = do_resize
self.size = size
self.resample = resample
self.apply_ocr = apply_ocr
self.ocr_lang = ocr_lang
self.tesseract_config = tesseract_config
def resize(
self,
image: np.ndarray,
size: Dict[str, int],
resample: PILImageResampling = PILImageResampling.BILINEAR,
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs
) -> np.ndarray:
"""
Resize an image to `(size["height"], size["width"])`.
Args:
image (`np.ndarray`):
Image to resize.
size (`Dict[str, int]`):
Size of the output image.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BILINEAR`):
Resampling filter to use when resizing the image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
size = get_size_dict(size)
if "height" not in size or "width" not in size:
raise ValueError(f"The size dictionary must contain the keys 'height' and 'width'. Got {size.keys()}")
output_size = (size["height"], size["width"])
return resize(image, size=output_size, resample=resample, data_format=data_format, **kwargs)
def preprocess(
self,
images: ImageInput,
do_resize: bool = None,
size: Dict[str, int] = None,
resample: PILImageResampling = None,
apply_ocr: bool = None,
ocr_lang: Optional[str] = None,
tesseract_config: Optional[str] = None,
return_tensors: Optional[Union[str, TensorType]] = None,
data_format: ChannelDimension = ChannelDimension.FIRST,
**kwargs,
) -> PIL.Image.Image:
"""
Preprocess an image or batch of images.
Args:
images (`ImageInput`):
Image to preprocess.
do_resize (`bool`, *optional*, defaults to `self.do_resize`):
Whether to resize the image.
size (`Dict[str, int]`, *optional*, defaults to `self.size`):
Desired size of the output image after resizing.
resample (`PILImageResampling`, *optional*, defaults to `self.resample`):
Resampling filter to use if resizing the image. This can be one of the enum `PIL.Image` resampling
filter. Only has an effect if `do_resize` is set to `True`.
apply_ocr (`bool`, *optional*, defaults to `self.apply_ocr`):
Whether to apply the Tesseract OCR engine to get words + normalized bounding boxes.
ocr_lang (`str`, *optional*, defaults to `self.ocr_lang`):
The language, specified by its ISO code, to be used by the Tesseract OCR engine. By default, English is
used.
tesseract_config (`str`, *optional*, defaults to `self.tesseract_config`):
Any additional custom configuration flags that are forwarded to the `config` parameter when calling
Tesseract.
return_tensors (`str` or `TensorType`, *optional*):
The type of tensors to return. Can be one of:
- Unset: Return a list of `np.ndarray`.
- `TensorType.TENSORFLOW` or `'tf'`: Return a batch of type `tf.Tensor`.
- `TensorType.PYTORCH` or `'pt'`: Return a batch of type `torch.Tensor`.
- `TensorType.NUMPY` or `'np'`: Return a batch of type `np.ndarray`.
- `TensorType.JAX` or `'jax'`: Return a batch of type `jax.numpy.ndarray`.
data_format (`ChannelDimension` or `str`, *optional*, defaults to `ChannelDimension.FIRST`):
The channel dimension format for the output image. Can be one of:
- `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
- `ChannelDimension.LAST`: image in (height, width, num_channels) format.
"""
do_resize = do_resize if do_resize is not None else self.do_resize
size = size if size is not None else self.size
size = get_size_dict(size)
resample = resample if resample is not None else self.resample
apply_ocr = apply_ocr if apply_ocr is not None else self.apply_ocr
ocr_lang = ocr_lang if ocr_lang is not None else self.ocr_lang
tesseract_config = tesseract_config if tesseract_config is not None else self.tesseract_config
if not is_batched(images):
images = [images]
if not valid_images(images):
raise ValueError(
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
"torch.Tensor, tf.Tensor or jax.ndarray."
)
if do_resize and size is None:
raise ValueError("Size must be specified if do_resize is True.")
# All transformations expect numpy arrays.
images = [to_numpy_array(image) for image in images]
if apply_ocr:
requires_backends(self, "pytesseract")
words_batch = []
boxes_batch = []
for image in images:
words, boxes = apply_tesseract(image, ocr_lang, tesseract_config)
words_batch.append(words)
boxes_batch.append(boxes)
if do_resize:
images = [self.resize(image=image, size=size, resample=resample) for image in images]
# flip color channels from RGB to BGR (as Detectron2 requires this)
images = [flip_channel_order(image) for image in images]
images = [to_channel_dimension_format(image, data_format) for image in images]
data = BatchFeature(data={"pixel_values": images}, tensor_type=return_tensors)
if apply_ocr:
data["words"] = words_batch
data["boxes"] = boxes_batch
return data
| # coding=utf-8
# Copyright 2022 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Image processor class for LayoutLMv2."""
from typing import Dict, Optional, Union
import numpy as np
from transformers.utils import is_vision_available
from transformers.utils.generic import TensorType
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
ChannelDimension,
ImageInput,
PILImageResampling,
infer_channel_dimension_format,
is_batched,
to_numpy_array,
valid_images,
)
from ...utils import is_pytesseract_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
logger = logging.get_logger(__name__)
def normalize_box(box, width, height):
return [
int(1000 * (box[0] / width)),
int(1000 * (box[1] / height)),
int(1000 * (box[2] / width)),
int(1000 * (box[3] / height)),
]
def apply_tesseract(image: np.ndarray, lang: Optional[str], tesseract_config: Optional[str] = None):
"""Applies Tesseract OCR on a document image, and returns recognized words + normalized bounding boxes."""
tesseract_config = tesseract_config if tesseract_config is not None else ""
# apply OCR
pil_image = to_pil_image(image)
image_width, image_height = pil_image.size
data = pytesseract.image_to_data(pil_image, lang=lang, output_type="dict", config=tesseract_config)
words, left, top, width, height = data["text"], data["left"], data["top"], data["width"], data["height"]
# filter empty words and corresponding coordinates
irrelevant_indices = [idx for idx, word in enumerate(words) if not word.strip()]
words = [word for idx, word in enumerate(words) if idx not in irrelevant_indices]
left = [coord for idx, coord in enumerate(left) if idx not in irrelevant_indices]
top = [coord for idx, coord in enumerate(top) if idx not in irrelevant_indices]
width = [coord for idx, coord in enumerate(width) if idx not in irrelevant_indices]
height = [coord for idx, coord in enumerate(height) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
actual_boxes = []
for x, y, w, h in zip(left, top, width, height):
actual_box = [x, y, x + w, y + h]
actual_boxes.append(actual_box)
# finally, normalize the bounding boxes
normalized_boxes = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(box, image_width, image_height))
assert len(words) == len(normalized_boxes), "Not as many words as there are bounding boxes"
return words, normalized_boxes
def flip_channel_order(image: np.ndarray, data_format: Optional[ChannelDimension] = None) -> np.ndarray:
input_data_format = infer_channel_dimension_format(image)
if input_data_format == ChannelDimension.LAST:
image = image[..., ::-1]
elif input_data_format == ChannelDimension.FIRST:
image = image[:, ::-1, ...]
else:
raise ValueError(f"Unsupported channel dimension: {input_data_format}")
if data_format is not None:
image = to_channel_dimension_format(image, data_format)
return image
class LayoutLMv2ImageProcessor(BaseImageProcessor):
r"""
Constructs a LayoutLMv2 image processor.
Args:
do_resize (`bool`, *optional*, defaults to `True`):
Whether to resize the image's (height, width) dimensions to `(size["height"], size["width"])`. Can be
overridden by `do_resize` in `preprocess`.
size (`Dict[str, int]` *optional*, defaults to `{"height": 224, "width": 224}`):
Size of the image after resizing. Can be overridden by `size` in `preprocess`.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BILINEAR`):
Resampling filter to use if resizing the image. Can be overridden by the `resample` parameter in the
`preprocess` method.
apply_ocr (`bool`, *optional*, defaults to `True`):
Whether to apply the Tesseract OCR engine to get words + normalized bounding boxes. Can be overridden by
`apply_ocr` in `preprocess`.
ocr_lang (`str`, *optional*):
The language, specified by its ISO code, to be used by the Tesseract OCR engine. By default, English is
used. Can be overridden by `ocr_lang` in `preprocess`.
tesseract_config (`str`, *optional*):
Any additional custom configuration flags that are forwarded to the `config` parameter when calling
Tesseract. For example: '--psm 6'. Can be overridden by `tesseract_config` in `preprocess`.
"""
model_input_names = ["pixel_values"]
def __init__(
self,
do_resize: bool = True,
size: Dict[str, int] = None,
resample: PILImageResampling = PILImageResampling.BILINEAR,
apply_ocr: bool = True,
ocr_lang: Optional[str] = None,
tesseract_config: Optional[str] = "",
**kwargs
) -> None:
super().__init__(**kwargs)
size = size if size is not None else {"height": 224, "width": 224}
size = get_size_dict(size)
self.do_resize = do_resize
self.size = size
self.resample = resample
self.apply_ocr = apply_ocr
self.ocr_lang = ocr_lang
self.tesseract_config = tesseract_config
def resize(
self,
image: np.ndarray,
size: Dict[str, int],
resample: PILImageResampling = PILImageResampling.BILINEAR,
data_format: Optional[Union[str, ChannelDimension]] = None,
**kwargs
) -> np.ndarray:
"""
Resize an image to `(size["height"], size["width"])`.
Args:
image (`np.ndarray`):
Image to resize.
size (`Dict[str, int]`):
Size of the output image.
resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BILINEAR`):
Resampling filter to use when resizing the image.
data_format (`str` or `ChannelDimension`, *optional*):
The channel dimension format of the image. If not provided, it will be the same as the input image.
"""
size = get_size_dict(size)
if "height" not in size or "width" not in size:
raise ValueError(f"The size dictionary must contain the keys 'height' and 'width'. Got {size.keys()}")
output_size = (size["height"], size["width"])
return resize(image, size=output_size, resample=resample, data_format=data_format, **kwargs)
def preprocess(
self,
images: ImageInput,
do_resize: bool = None,
size: Dict[str, int] = None,
resample: PILImageResampling = None,
apply_ocr: bool = None,
ocr_lang: Optional[str] = None,
tesseract_config: Optional[str] = None,
return_tensors: Optional[Union[str, TensorType]] = None,
data_format: ChannelDimension = ChannelDimension.FIRST,
**kwargs,
) -> PIL.Image.Image:
"""
Preprocess an image or batch of images.
Args:
images (`ImageInput`):
Image to preprocess.
do_resize (`bool`, *optional*, defaults to `self.do_resize`):
Whether to resize the image.
size (`Dict[str, int]`, *optional*, defaults to `self.size`):
Desired size of the output image after resizing.
resample (`PILImageResampling`, *optional*, defaults to `self.resample`):
Resampling filter to use if resizing the image. This can be one of the enum `PIL.Image` resampling
filter. Only has an effect if `do_resize` is set to `True`.
apply_ocr (`bool`, *optional*, defaults to `self.apply_ocr`):
Whether to apply the Tesseract OCR engine to get words + normalized bounding boxes.
ocr_lang (`str`, *optional*, defaults to `self.ocr_lang`):
The language, specified by its ISO code, to be used by the Tesseract OCR engine. By default, English is
used.
tesseract_config (`str`, *optional*, defaults to `self.tesseract_config`):
Any additional custom configuration flags that are forwarded to the `config` parameter when calling
Tesseract.
return_tensors (`str` or `TensorType`, *optional*):
The type of tensors to return. Can be one of:
- Unset: Return a list of `np.ndarray`.
- `TensorType.TENSORFLOW` or `'tf'`: Return a batch of type `tf.Tensor`.
- `TensorType.PYTORCH` or `'pt'`: Return a batch of type `torch.Tensor`.
- `TensorType.NUMPY` or `'np'`: Return a batch of type `np.ndarray`.
- `TensorType.JAX` or `'jax'`: Return a batch of type `jax.numpy.ndarray`.
data_format (`ChannelDimension` or `str`, *optional*, defaults to `ChannelDimension.FIRST`):
The channel dimension format for the output image. Can be one of:
- `ChannelDimension.FIRST`: image in (num_channels, height, width) format.
- `ChannelDimension.LAST`: image in (height, width, num_channels) format.
"""
do_resize = do_resize if do_resize is not None else self.do_resize
size = size if size is not None else self.size
size = get_size_dict(size)
resample = resample if resample is not None else self.resample
apply_ocr = apply_ocr if apply_ocr is not None else self.apply_ocr
ocr_lang = ocr_lang if ocr_lang is not None else self.ocr_lang
tesseract_config = tesseract_config if tesseract_config is not None else self.tesseract_config
if not is_batched(images):
images = [images]
if not valid_images(images):
raise ValueError(
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
"torch.Tensor, tf.Tensor or jax.ndarray."
)
if do_resize and size is None:
raise ValueError("Size must be specified if do_resize is True.")
# All transformations expect numpy arrays.
images = [to_numpy_array(image) for image in images]
if apply_ocr:
requires_backends(self, "pytesseract")
words_batch = []
boxes_batch = []
for image in images:
words, boxes = apply_tesseract(image, ocr_lang, tesseract_config)
words_batch.append(words)
boxes_batch.append(boxes)
if do_resize:
images = [self.resize(image=image, size=size, resample=resample) for image in images]
# flip color channels from RGB to BGR (as Detectron2 requires this)
images = [flip_channel_order(image) for image in images]
images = [to_channel_dimension_format(image, data_format) for image in images]
data = BatchFeature(data={"pixel_values": images}, tensor_type=return_tensors)
if apply_ocr:
data["words"] = words_batch
data["boxes"] = boxes_batch
return data
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/vit/convert_vit_timm_to_pytorch.py | # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Convert ViT and non-distilled DeiT checkpoints from the timm library."""
import argparse
import json
from pathlib import Path
import torch
from PIL import Image
import requests
import timm
from huggingface_hub import hf_hub_download
from transformers import DeiTFeatureExtractor, ViTConfig, ViTFeatureExtractor, ViTForImageClassification, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
logger = logging.get_logger(__name__)
# here we list all keys to be renamed (original name on the left, our name on the right)
def create_rename_keys(config, base_model=False):
rename_keys = []
for i in range(config.num_hidden_layers):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((f"blocks.{i}.norm1.weight", f"vit.encoder.layer.{i}.layernorm_before.weight"))
rename_keys.append((f"blocks.{i}.norm1.bias", f"vit.encoder.layer.{i}.layernorm_before.bias"))
rename_keys.append((f"blocks.{i}.attn.proj.weight", f"vit.encoder.layer.{i}.attention.output.dense.weight"))
rename_keys.append((f"blocks.{i}.attn.proj.bias", f"vit.encoder.layer.{i}.attention.output.dense.bias"))
rename_keys.append((f"blocks.{i}.norm2.weight", f"vit.encoder.layer.{i}.layernorm_after.weight"))
rename_keys.append((f"blocks.{i}.norm2.bias", f"vit.encoder.layer.{i}.layernorm_after.bias"))
rename_keys.append((f"blocks.{i}.mlp.fc1.weight", f"vit.encoder.layer.{i}.intermediate.dense.weight"))
rename_keys.append((f"blocks.{i}.mlp.fc1.bias", f"vit.encoder.layer.{i}.intermediate.dense.bias"))
rename_keys.append((f"blocks.{i}.mlp.fc2.weight", f"vit.encoder.layer.{i}.output.dense.weight"))
rename_keys.append((f"blocks.{i}.mlp.fc2.bias", f"vit.encoder.layer.{i}.output.dense.bias"))
# projection layer + position embeddings
rename_keys.extend(
[
("cls_token", "vit.embeddings.cls_token"),
("patch_embed.proj.weight", "vit.embeddings.patch_embeddings.projection.weight"),
("patch_embed.proj.bias", "vit.embeddings.patch_embeddings.projection.bias"),
("pos_embed", "vit.embeddings.position_embeddings"),
]
)
if base_model:
# layernorm + pooler
rename_keys.extend(
[
("norm.weight", "layernorm.weight"),
("norm.bias", "layernorm.bias"),
("pre_logits.fc.weight", "pooler.dense.weight"),
("pre_logits.fc.bias", "pooler.dense.bias"),
]
)
# if just the base model, we should remove "vit" from all keys that start with "vit"
rename_keys = [(pair[0], pair[1][4:]) if pair[1].startswith("vit") else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
("norm.weight", "vit.layernorm.weight"),
("norm.bias", "vit.layernorm.bias"),
("head.weight", "classifier.weight"),
("head.bias", "classifier.bias"),
]
)
return rename_keys
# we split up the matrix of each encoder layer into queries, keys and values
def read_in_q_k_v(state_dict, config, base_model=False):
for i in range(config.num_hidden_layers):
if base_model:
prefix = ""
else:
prefix = "vit."
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
in_proj_weight = state_dict.pop(f"blocks.{i}.attn.qkv.weight")
in_proj_bias = state_dict.pop(f"blocks.{i}.attn.qkv.bias")
# next, add query, keys and values (in that order) to the state dict
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.query.weight"] = in_proj_weight[
: config.hidden_size, :
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.query.bias"] = in_proj_bias[: config.hidden_size]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.key.weight"] = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.key.bias"] = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.value.weight"] = in_proj_weight[
-config.hidden_size :, :
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.value.bias"] = in_proj_bias[-config.hidden_size :]
def remove_classification_head_(state_dict):
ignore_keys = ["head.weight", "head.bias"]
for k in ignore_keys:
state_dict.pop(k, None)
def rename_key(dct, old, new):
val = dct.pop(old)
dct[new] = val
# We will verify our results on an image of cute cats
def prepare_img():
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
im = Image.open(requests.get(url, stream=True).raw)
return im
@torch.no_grad()
def convert_vit_checkpoint(vit_name, pytorch_dump_folder_path):
"""
Copy/paste/tweak model's weights to our ViT structure.
"""
# define default ViT configuration
config = ViTConfig()
base_model = False
# dataset (ImageNet-21k only or also fine-tuned on ImageNet 2012), patch_size and image_size
if vit_name[-5:] == "in21k":
base_model = True
config.patch_size = int(vit_name[-12:-10])
config.image_size = int(vit_name[-9:-6])
else:
config.num_labels = 1000
repo_id = "huggingface/label-files"
filename = "imagenet-1k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
config.patch_size = int(vit_name[-6:-4])
config.image_size = int(vit_name[-3:])
# size of the architecture
if "deit" in vit_name:
if vit_name[9:].startswith("tiny"):
config.hidden_size = 192
config.intermediate_size = 768
config.num_hidden_layers = 12
config.num_attention_heads = 3
elif vit_name[9:].startswith("small"):
config.hidden_size = 384
config.intermediate_size = 1536
config.num_hidden_layers = 12
config.num_attention_heads = 6
else:
pass
else:
if vit_name[4:].startswith("small"):
config.hidden_size = 768
config.intermediate_size = 2304
config.num_hidden_layers = 8
config.num_attention_heads = 8
elif vit_name[4:].startswith("base"):
pass
elif vit_name[4:].startswith("large"):
config.hidden_size = 1024
config.intermediate_size = 4096
config.num_hidden_layers = 24
config.num_attention_heads = 16
elif vit_name[4:].startswith("huge"):
config.hidden_size = 1280
config.intermediate_size = 5120
config.num_hidden_layers = 32
config.num_attention_heads = 16
# load original model from timm
timm_model = timm.create_model(vit_name, pretrained=True)
timm_model.eval()
# load state_dict of original model, remove and rename some keys
state_dict = timm_model.state_dict()
if base_model:
remove_classification_head_(state_dict)
rename_keys = create_rename_keys(config, base_model)
for src, dest in rename_keys:
rename_key(state_dict, src, dest)
read_in_q_k_v(state_dict, config, base_model)
# load HuggingFace model
if vit_name[-5:] == "in21k":
model = ViTModel(config).eval()
else:
model = ViTForImageClassification(config).eval()
model.load_state_dict(state_dict)
# Check outputs on an image, prepared by ViTFeatureExtractor/DeiTFeatureExtractor
if "deit" in vit_name:
feature_extractor = DeiTFeatureExtractor(size=config.image_size)
else:
feature_extractor = ViTFeatureExtractor(size=config.image_size)
encoding = feature_extractor(images=prepare_img(), return_tensors="pt")
pixel_values = encoding["pixel_values"]
outputs = model(pixel_values)
if base_model:
timm_pooled_output = timm_model.forward_features(pixel_values)
assert timm_pooled_output.shape == outputs.pooler_output.shape
assert torch.allclose(timm_pooled_output, outputs.pooler_output, atol=1e-3)
else:
timm_logits = timm_model(pixel_values)
assert timm_logits.shape == outputs.logits.shape
assert torch.allclose(timm_logits, outputs.logits, atol=1e-3)
Path(pytorch_dump_folder_path).mkdir(exist_ok=True)
print(f"Saving model {vit_name} to {pytorch_dump_folder_path}")
model.save_pretrained(pytorch_dump_folder_path)
print(f"Saving feature extractor to {pytorch_dump_folder_path}")
feature_extractor.save_pretrained(pytorch_dump_folder_path)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--vit_name",
default="vit_base_patch16_224",
type=str,
help="Name of the ViT timm model you'd like to convert.",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory."
)
args = parser.parse_args()
convert_vit_checkpoint(args.vit_name, args.pytorch_dump_folder_path)
| # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Convert ViT and non-distilled DeiT checkpoints from the timm library."""
import argparse
import json
from pathlib import Path
import torch
from PIL import Image
import requests
import timm
from huggingface_hub import hf_hub_download
from transformers import DeiTFeatureExtractor, ViTConfig, ViTFeatureExtractor, ViTForImageClassification, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
logger = logging.get_logger(__name__)
# here we list all keys to be renamed (original name on the left, our name on the right)
def create_rename_keys(config, base_model=False):
rename_keys = []
for i in range(config.num_hidden_layers):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((f"blocks.{i}.norm1.weight", f"vit.encoder.layer.{i}.layernorm_before.weight"))
rename_keys.append((f"blocks.{i}.norm1.bias", f"vit.encoder.layer.{i}.layernorm_before.bias"))
rename_keys.append((f"blocks.{i}.attn.proj.weight", f"vit.encoder.layer.{i}.attention.output.dense.weight"))
rename_keys.append((f"blocks.{i}.attn.proj.bias", f"vit.encoder.layer.{i}.attention.output.dense.bias"))
rename_keys.append((f"blocks.{i}.norm2.weight", f"vit.encoder.layer.{i}.layernorm_after.weight"))
rename_keys.append((f"blocks.{i}.norm2.bias", f"vit.encoder.layer.{i}.layernorm_after.bias"))
rename_keys.append((f"blocks.{i}.mlp.fc1.weight", f"vit.encoder.layer.{i}.intermediate.dense.weight"))
rename_keys.append((f"blocks.{i}.mlp.fc1.bias", f"vit.encoder.layer.{i}.intermediate.dense.bias"))
rename_keys.append((f"blocks.{i}.mlp.fc2.weight", f"vit.encoder.layer.{i}.output.dense.weight"))
rename_keys.append((f"blocks.{i}.mlp.fc2.bias", f"vit.encoder.layer.{i}.output.dense.bias"))
# projection layer + position embeddings
rename_keys.extend(
[
("cls_token", "vit.embeddings.cls_token"),
("patch_embed.proj.weight", "vit.embeddings.patch_embeddings.projection.weight"),
("patch_embed.proj.bias", "vit.embeddings.patch_embeddings.projection.bias"),
("pos_embed", "vit.embeddings.position_embeddings"),
]
)
if base_model:
# layernorm + pooler
rename_keys.extend(
[
("norm.weight", "layernorm.weight"),
("norm.bias", "layernorm.bias"),
("pre_logits.fc.weight", "pooler.dense.weight"),
("pre_logits.fc.bias", "pooler.dense.bias"),
]
)
# if just the base model, we should remove "vit" from all keys that start with "vit"
rename_keys = [(pair[0], pair[1][4:]) if pair[1].startswith("vit") else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
("norm.weight", "vit.layernorm.weight"),
("norm.bias", "vit.layernorm.bias"),
("head.weight", "classifier.weight"),
("head.bias", "classifier.bias"),
]
)
return rename_keys
# we split up the matrix of each encoder layer into queries, keys and values
def read_in_q_k_v(state_dict, config, base_model=False):
for i in range(config.num_hidden_layers):
if base_model:
prefix = ""
else:
prefix = "vit."
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
in_proj_weight = state_dict.pop(f"blocks.{i}.attn.qkv.weight")
in_proj_bias = state_dict.pop(f"blocks.{i}.attn.qkv.bias")
# next, add query, keys and values (in that order) to the state dict
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.query.weight"] = in_proj_weight[
: config.hidden_size, :
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.query.bias"] = in_proj_bias[: config.hidden_size]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.key.weight"] = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.key.bias"] = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.value.weight"] = in_proj_weight[
-config.hidden_size :, :
]
state_dict[f"{prefix}encoder.layer.{i}.attention.attention.value.bias"] = in_proj_bias[-config.hidden_size :]
def remove_classification_head_(state_dict):
ignore_keys = ["head.weight", "head.bias"]
for k in ignore_keys:
state_dict.pop(k, None)
def rename_key(dct, old, new):
val = dct.pop(old)
dct[new] = val
# We will verify our results on an image of cute cats
def prepare_img():
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
im = Image.open(requests.get(url, stream=True).raw)
return im
@torch.no_grad()
def convert_vit_checkpoint(vit_name, pytorch_dump_folder_path):
"""
Copy/paste/tweak model's weights to our ViT structure.
"""
# define default ViT configuration
config = ViTConfig()
base_model = False
# dataset (ImageNet-21k only or also fine-tuned on ImageNet 2012), patch_size and image_size
if vit_name[-5:] == "in21k":
base_model = True
config.patch_size = int(vit_name[-12:-10])
config.image_size = int(vit_name[-9:-6])
else:
config.num_labels = 1000
repo_id = "huggingface/label-files"
filename = "imagenet-1k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
config.patch_size = int(vit_name[-6:-4])
config.image_size = int(vit_name[-3:])
# size of the architecture
if "deit" in vit_name:
if vit_name[9:].startswith("tiny"):
config.hidden_size = 192
config.intermediate_size = 768
config.num_hidden_layers = 12
config.num_attention_heads = 3
elif vit_name[9:].startswith("small"):
config.hidden_size = 384
config.intermediate_size = 1536
config.num_hidden_layers = 12
config.num_attention_heads = 6
else:
pass
else:
if vit_name[4:].startswith("small"):
config.hidden_size = 768
config.intermediate_size = 2304
config.num_hidden_layers = 8
config.num_attention_heads = 8
elif vit_name[4:].startswith("base"):
pass
elif vit_name[4:].startswith("large"):
config.hidden_size = 1024
config.intermediate_size = 4096
config.num_hidden_layers = 24
config.num_attention_heads = 16
elif vit_name[4:].startswith("huge"):
config.hidden_size = 1280
config.intermediate_size = 5120
config.num_hidden_layers = 32
config.num_attention_heads = 16
# load original model from timm
timm_model = timm.create_model(vit_name, pretrained=True)
timm_model.eval()
# load state_dict of original model, remove and rename some keys
state_dict = timm_model.state_dict()
if base_model:
remove_classification_head_(state_dict)
rename_keys = create_rename_keys(config, base_model)
for src, dest in rename_keys:
rename_key(state_dict, src, dest)
read_in_q_k_v(state_dict, config, base_model)
# load HuggingFace model
if vit_name[-5:] == "in21k":
model = ViTModel(config).eval()
else:
model = ViTForImageClassification(config).eval()
model.load_state_dict(state_dict)
# Check outputs on an image, prepared by ViTFeatureExtractor/DeiTFeatureExtractor
if "deit" in vit_name:
feature_extractor = DeiTFeatureExtractor(size=config.image_size)
else:
feature_extractor = ViTFeatureExtractor(size=config.image_size)
encoding = feature_extractor(images=prepare_img(), return_tensors="pt")
pixel_values = encoding["pixel_values"]
outputs = model(pixel_values)
if base_model:
timm_pooled_output = timm_model.forward_features(pixel_values)
assert timm_pooled_output.shape == outputs.pooler_output.shape
assert torch.allclose(timm_pooled_output, outputs.pooler_output, atol=1e-3)
else:
timm_logits = timm_model(pixel_values)
assert timm_logits.shape == outputs.logits.shape
assert torch.allclose(timm_logits, outputs.logits, atol=1e-3)
Path(pytorch_dump_folder_path).mkdir(exist_ok=True)
print(f"Saving model {vit_name} to {pytorch_dump_folder_path}")
model.save_pretrained(pytorch_dump_folder_path)
print(f"Saving feature extractor to {pytorch_dump_folder_path}")
feature_extractor.save_pretrained(pytorch_dump_folder_path)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--vit_name",
default="vit_base_patch16_224",
type=str,
help="Name of the ViT timm model you'd like to convert.",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory."
)
args = parser.parse_args()
convert_vit_checkpoint(args.vit_name, args.pytorch_dump_folder_path)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./docs/source/en/model_doc/pegasus_x.mdx | <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# PEGASUS-X
## Overview
The PEGASUS-X model was proposed in [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao and Peter J. Liu.
PEGASUS-X (PEGASUS eXtended) extends the PEGASUS models for long input summarization through additional long input pretraining and using staggered block-local attention with global tokens in the encoder.
The abstract from the paper is the following:
*While large pretrained Transformer models have proven highly capable at tackling natural language tasks, handling long sequence inputs continues to be a significant challenge. One such task is long input summarization, where inputs are longer than the maximum input context of most pretrained models. Through an extensive set of experiments, we investigate what model architectural changes and pretraining paradigms can most efficiently adapt a pretrained Transformer for long input summarization. We find that a staggered, block-local Transformer with global encoder tokens strikes a good balance of performance and efficiency, and that an additional pretraining phase on long sequences meaningfully improves downstream summarization performance. Based on our findings, we introduce PEGASUS-X, an extension of the PEGASUS model with additional long input pretraining to handle inputs of up to 16K tokens. PEGASUS-X achieves strong performance on long input summarization tasks comparable with much larger models while adding few additional parameters and not requiring model parallelism to train.*
Tips:
* PEGASUS-X uses the same tokenizer as PEGASUS.
This model was contributed by [zphang](<https://huggingface.co/zphang). The original code can be found [here](https://github.com/google-research/pegasus).
## PegasusXConfig
[[autodoc]] PegasusXConfig
## PegasusXModel
[[autodoc]] PegasusXModel
- forward
## PegasusXForConditionalGeneration
[[autodoc]] PegasusXForConditionalGeneration
- forward
| <!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# PEGASUS-X
## Overview
The PEGASUS-X model was proposed in [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao and Peter J. Liu.
PEGASUS-X (PEGASUS eXtended) extends the PEGASUS models for long input summarization through additional long input pretraining and using staggered block-local attention with global tokens in the encoder.
The abstract from the paper is the following:
*While large pretrained Transformer models have proven highly capable at tackling natural language tasks, handling long sequence inputs continues to be a significant challenge. One such task is long input summarization, where inputs are longer than the maximum input context of most pretrained models. Through an extensive set of experiments, we investigate what model architectural changes and pretraining paradigms can most efficiently adapt a pretrained Transformer for long input summarization. We find that a staggered, block-local Transformer with global encoder tokens strikes a good balance of performance and efficiency, and that an additional pretraining phase on long sequences meaningfully improves downstream summarization performance. Based on our findings, we introduce PEGASUS-X, an extension of the PEGASUS model with additional long input pretraining to handle inputs of up to 16K tokens. PEGASUS-X achieves strong performance on long input summarization tasks comparable with much larger models while adding few additional parameters and not requiring model parallelism to train.*
Tips:
* PEGASUS-X uses the same tokenizer as PEGASUS.
This model was contributed by [zphang](<https://huggingface.co/zphang). The original code can be found [here](https://github.com/google-research/pegasus).
## PegasusXConfig
[[autodoc]] PegasusXConfig
## PegasusXModel
[[autodoc]] PegasusXModel
- forward
## PegasusXForConditionalGeneration
[[autodoc]] PegasusXForConditionalGeneration
- forward
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/lxmert/tokenization_lxmert.py | # coding=utf-8
# Copyright 2020 The Google AI Team, Stanford University and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import collections
import os
import unicodedata
from typing import List, Optional, Tuple
from ...tokenization_utils import PreTrainedTokenizer, _is_control, _is_punctuation, _is_whitespace
from ...utils import logging
logger = logging.get_logger(__name__)
VOCAB_FILES_NAMES = {"vocab_file": "vocab.txt"}
PRETRAINED_VOCAB_FILES_MAP = {
"vocab_file": {
"unc-nlp/lxmert-base-uncased": "https://huggingface.co/unc-nlp/lxmert-base-uncased/resolve/main/vocab.txt",
}
}
PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
"unc-nlp/lxmert-base-uncased": 512,
}
PRETRAINED_INIT_CONFIGURATION = {
"unc-nlp/lxmert-base-uncased": {"do_lower_case": True},
}
# Copied from transformers.models.bert.tokenization_bert.load_vocab
def load_vocab(vocab_file):
"""Loads a vocabulary file into a dictionary."""
vocab = collections.OrderedDict()
with open(vocab_file, "r", encoding="utf-8") as reader:
tokens = reader.readlines()
for index, token in enumerate(tokens):
token = token.rstrip("\n")
vocab[token] = index
return vocab
# Copied from transformers.models.bert.tokenization_bert.whitespace_tokenize
def whitespace_tokenize(text):
"""Runs basic whitespace cleaning and splitting on a piece of text."""
text = text.strip()
if not text:
return []
tokens = text.split()
return tokens
# Copied from transformers.models.bert.tokenization_bert.BertTokenizer with bert-base-cased->unc-nlp/lxmert-base-uncased, BERT->Lxmert, BertTokenizer->LxmertTokenizer
class LxmertTokenizer(PreTrainedTokenizer):
r"""
Construct a Lxmert tokenizer. Based on WordPiece.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to
this superclass for more information regarding those methods.
Args:
vocab_file (`str`):
File containing the vocabulary.
do_lower_case (`bool`, *optional*, defaults to `True`):
Whether or not to lowercase the input when tokenizing.
do_basic_tokenize (`bool`, *optional*, defaults to `True`):
Whether or not to do basic tokenization before WordPiece.
never_split (`Iterable`, *optional*):
Collection of tokens which will never be split during tokenization. Only has an effect when
`do_basic_tokenize=True`
unk_token (`str`, *optional*, defaults to `"[UNK]"`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
sep_token (`str`, *optional*, defaults to `"[SEP]"`):
The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
sequence classification or for a text and a question for question answering. It is also used as the last
token of a sequence built with special tokens.
pad_token (`str`, *optional*, defaults to `"[PAD]"`):
The token used for padding, for example when batching sequences of different lengths.
cls_token (`str`, *optional*, defaults to `"[CLS]"`):
The classifier token which is used when doing sequence classification (classification of the whole sequence
instead of per-token classification). It is the first token of the sequence when built with special tokens.
mask_token (`str`, *optional*, defaults to `"[MASK]"`):
The token used for masking values. This is the token used when training this model with masked language
modeling. This is the token which the model will try to predict.
tokenize_chinese_chars (`bool`, *optional*, defaults to `True`):
Whether or not to tokenize Chinese characters.
This should likely be deactivated for Japanese (see this
[issue](https://github.com/huggingface/transformers/issues/328)).
strip_accents (`bool`, *optional*):
Whether or not to strip all accents. If this option is not specified, then it will be determined by the
value for `lowercase` (as in the original Lxmert).
"""
vocab_files_names = VOCAB_FILES_NAMES
pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION
max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
def __init__(
self,
vocab_file,
do_lower_case=True,
do_basic_tokenize=True,
never_split=None,
unk_token="[UNK]",
sep_token="[SEP]",
pad_token="[PAD]",
cls_token="[CLS]",
mask_token="[MASK]",
tokenize_chinese_chars=True,
strip_accents=None,
**kwargs
):
super().__init__(
do_lower_case=do_lower_case,
do_basic_tokenize=do_basic_tokenize,
never_split=never_split,
unk_token=unk_token,
sep_token=sep_token,
pad_token=pad_token,
cls_token=cls_token,
mask_token=mask_token,
tokenize_chinese_chars=tokenize_chinese_chars,
strip_accents=strip_accents,
**kwargs,
)
if not os.path.isfile(vocab_file):
raise ValueError(
f"Can't find a vocabulary file at path '{vocab_file}'. To load the vocabulary from a Google pretrained"
" model use `tokenizer = LxmertTokenizer.from_pretrained(PRETRAINED_MODEL_NAME)`"
)
self.vocab = load_vocab(vocab_file)
self.ids_to_tokens = collections.OrderedDict([(ids, tok) for tok, ids in self.vocab.items()])
self.do_basic_tokenize = do_basic_tokenize
if do_basic_tokenize:
self.basic_tokenizer = BasicTokenizer(
do_lower_case=do_lower_case,
never_split=never_split,
tokenize_chinese_chars=tokenize_chinese_chars,
strip_accents=strip_accents,
)
self.wordpiece_tokenizer = WordpieceTokenizer(vocab=self.vocab, unk_token=self.unk_token)
@property
def do_lower_case(self):
return self.basic_tokenizer.do_lower_case
@property
def vocab_size(self):
return len(self.vocab)
def get_vocab(self):
return dict(self.vocab, **self.added_tokens_encoder)
def _tokenize(self, text):
split_tokens = []
if self.do_basic_tokenize:
for token in self.basic_tokenizer.tokenize(text, never_split=self.all_special_tokens):
# If the token is part of the never_split set
if token in self.basic_tokenizer.never_split:
split_tokens.append(token)
else:
split_tokens += self.wordpiece_tokenizer.tokenize(token)
else:
split_tokens = self.wordpiece_tokenizer.tokenize(text)
return split_tokens
def _convert_token_to_id(self, token):
"""Converts a token (str) in an id using the vocab."""
return self.vocab.get(token, self.vocab.get(self.unk_token))
def _convert_id_to_token(self, index):
"""Converts an index (integer) in a token (str) using the vocab."""
return self.ids_to_tokens.get(index, self.unk_token)
def convert_tokens_to_string(self, tokens):
"""Converts a sequence of tokens (string) in a single string."""
out_string = " ".join(tokens).replace(" ##", "").strip()
return out_string
def build_inputs_with_special_tokens(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
) -> List[int]:
"""
Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
adding special tokens. A Lxmert sequence has the following format:
- single sequence: `[CLS] X [SEP]`
- pair of sequences: `[CLS] A [SEP] B [SEP]`
Args:
token_ids_0 (`List[int]`):
List of IDs to which the special tokens will be added.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
`List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
"""
if token_ids_1 is None:
return [self.cls_token_id] + token_ids_0 + [self.sep_token_id]
cls = [self.cls_token_id]
sep = [self.sep_token_id]
return cls + token_ids_0 + sep + token_ids_1 + sep
def get_special_tokens_mask(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
) -> List[int]:
"""
Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer `prepare_for_model` method.
Args:
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
already_has_special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the token list is already formatted with special tokens for the model.
Returns:
`List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
"""
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
)
if token_ids_1 is not None:
return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1)) + [1]
return [1] + ([0] * len(token_ids_0)) + [1]
def create_token_type_ids_from_sequences(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
) -> List[int]:
"""
Create a mask from the two sequences passed to be used in a sequence-pair classification task. A Lxmert
sequence pair mask has the following format:
```
0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence | second sequence |
```
If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s).
Args:
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
`List[int]`: List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s).
"""
sep = [self.sep_token_id]
cls = [self.cls_token_id]
if token_ids_1 is None:
return len(cls + token_ids_0 + sep) * [0]
return len(cls + token_ids_0 + sep) * [0] + len(token_ids_1 + sep) * [1]
def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
index = 0
if os.path.isdir(save_directory):
vocab_file = os.path.join(
save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
)
else:
vocab_file = (filename_prefix + "-" if filename_prefix else "") + save_directory
with open(vocab_file, "w", encoding="utf-8") as writer:
for token, token_index in sorted(self.vocab.items(), key=lambda kv: kv[1]):
if index != token_index:
logger.warning(
f"Saving vocabulary to {vocab_file}: vocabulary indices are not consecutive."
" Please check that the vocabulary is not corrupted!"
)
index = token_index
writer.write(token + "\n")
index += 1
return (vocab_file,)
# Copied from transformers.models.bert.tokenization_bert.BasicTokenizer
class BasicTokenizer(object):
"""
Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
Args:
do_lower_case (`bool`, *optional*, defaults to `True`):
Whether or not to lowercase the input when tokenizing.
never_split (`Iterable`, *optional*):
Collection of tokens which will never be split during tokenization. Only has an effect when
`do_basic_tokenize=True`
tokenize_chinese_chars (`bool`, *optional*, defaults to `True`):
Whether or not to tokenize Chinese characters.
This should likely be deactivated for Japanese (see this
[issue](https://github.com/huggingface/transformers/issues/328)).
strip_accents (`bool`, *optional*):
Whether or not to strip all accents. If this option is not specified, then it will be determined by the
value for `lowercase` (as in the original BERT).
"""
def __init__(self, do_lower_case=True, never_split=None, tokenize_chinese_chars=True, strip_accents=None):
if never_split is None:
never_split = []
self.do_lower_case = do_lower_case
self.never_split = set(never_split)
self.tokenize_chinese_chars = tokenize_chinese_chars
self.strip_accents = strip_accents
def tokenize(self, text, never_split=None):
"""
Basic Tokenization of a piece of text. Split on "white spaces" only, for sub-word tokenization, see
WordPieceTokenizer.
Args:
never_split (`List[str]`, *optional*)
Kept for backward compatibility purposes. Now implemented directly at the base class level (see
[`PreTrainedTokenizer.tokenize`]) List of token not to split.
"""
# union() returns a new set by concatenating the two sets.
never_split = self.never_split.union(set(never_split)) if never_split else self.never_split
text = self._clean_text(text)
# This was added on November 1st, 2018 for the multilingual and Chinese
# models. This is also applied to the English models now, but it doesn't
# matter since the English models were not trained on any Chinese data
# and generally don't have any Chinese data in them (there are Chinese
# characters in the vocabulary because Wikipedia does have some Chinese
# words in the English Wikipedia.).
if self.tokenize_chinese_chars:
text = self._tokenize_chinese_chars(text)
orig_tokens = whitespace_tokenize(text)
split_tokens = []
for token in orig_tokens:
if token not in never_split:
if self.do_lower_case:
token = token.lower()
if self.strip_accents is not False:
token = self._run_strip_accents(token)
elif self.strip_accents:
token = self._run_strip_accents(token)
split_tokens.extend(self._run_split_on_punc(token, never_split))
output_tokens = whitespace_tokenize(" ".join(split_tokens))
return output_tokens
def _run_strip_accents(self, text):
"""Strips accents from a piece of text."""
text = unicodedata.normalize("NFD", text)
output = []
for char in text:
cat = unicodedata.category(char)
if cat == "Mn":
continue
output.append(char)
return "".join(output)
def _run_split_on_punc(self, text, never_split=None):
"""Splits punctuation on a piece of text."""
if never_split is not None and text in never_split:
return [text]
chars = list(text)
i = 0
start_new_word = True
output = []
while i < len(chars):
char = chars[i]
if _is_punctuation(char):
output.append([char])
start_new_word = True
else:
if start_new_word:
output.append([])
start_new_word = False
output[-1].append(char)
i += 1
return ["".join(x) for x in output]
def _tokenize_chinese_chars(self, text):
"""Adds whitespace around any CJK character."""
output = []
for char in text:
cp = ord(char)
if self._is_chinese_char(cp):
output.append(" ")
output.append(char)
output.append(" ")
else:
output.append(char)
return "".join(output)
def _is_chinese_char(self, cp):
"""Checks whether CP is the codepoint of a CJK character."""
# This defines a "chinese character" as anything in the CJK Unicode block:
# https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_(Unicode_block)
#
# Note that the CJK Unicode block is NOT all Japanese and Korean characters,
# despite its name. The modern Korean Hangul alphabet is a different block,
# as is Japanese Hiragana and Katakana. Those alphabets are used to write
# space-separated words, so they are not treated specially and handled
# like the all of the other languages.
if (
(cp >= 0x4E00 and cp <= 0x9FFF)
or (cp >= 0x3400 and cp <= 0x4DBF) #
or (cp >= 0x20000 and cp <= 0x2A6DF) #
or (cp >= 0x2A700 and cp <= 0x2B73F) #
or (cp >= 0x2B740 and cp <= 0x2B81F) #
or (cp >= 0x2B820 and cp <= 0x2CEAF) #
or (cp >= 0xF900 and cp <= 0xFAFF)
or (cp >= 0x2F800 and cp <= 0x2FA1F) #
): #
return True
return False
def _clean_text(self, text):
"""Performs invalid character removal and whitespace cleanup on text."""
output = []
for char in text:
cp = ord(char)
if cp == 0 or cp == 0xFFFD or _is_control(char):
continue
if _is_whitespace(char):
output.append(" ")
else:
output.append(char)
return "".join(output)
# Copied from transformers.models.bert.tokenization_bert.WordpieceTokenizer
class WordpieceTokenizer(object):
"""Runs WordPiece tokenization."""
def __init__(self, vocab, unk_token, max_input_chars_per_word=100):
self.vocab = vocab
self.unk_token = unk_token
self.max_input_chars_per_word = max_input_chars_per_word
def tokenize(self, text):
"""
Tokenizes a piece of text into its word pieces. This uses a greedy longest-match-first algorithm to perform
tokenization using the given vocabulary.
For example, `input = "unaffable"` wil return as output `["un", "##aff", "##able"]`.
Args:
text: A single token or whitespace separated tokens. This should have
already been passed through *BasicTokenizer*.
Returns:
A list of wordpiece tokens.
"""
output_tokens = []
for token in whitespace_tokenize(text):
chars = list(token)
if len(chars) > self.max_input_chars_per_word:
output_tokens.append(self.unk_token)
continue
is_bad = False
start = 0
sub_tokens = []
while start < len(chars):
end = len(chars)
cur_substr = None
while start < end:
substr = "".join(chars[start:end])
if start > 0:
substr = "##" + substr
if substr in self.vocab:
cur_substr = substr
break
end -= 1
if cur_substr is None:
is_bad = True
break
sub_tokens.append(cur_substr)
start = end
if is_bad:
output_tokens.append(self.unk_token)
else:
output_tokens.extend(sub_tokens)
return output_tokens
| # coding=utf-8
# Copyright 2020 The Google AI Team, Stanford University and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import collections
import os
import unicodedata
from typing import List, Optional, Tuple
from ...tokenization_utils import PreTrainedTokenizer, _is_control, _is_punctuation, _is_whitespace
from ...utils import logging
logger = logging.get_logger(__name__)
VOCAB_FILES_NAMES = {"vocab_file": "vocab.txt"}
PRETRAINED_VOCAB_FILES_MAP = {
"vocab_file": {
"unc-nlp/lxmert-base-uncased": "https://huggingface.co/unc-nlp/lxmert-base-uncased/resolve/main/vocab.txt",
}
}
PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {
"unc-nlp/lxmert-base-uncased": 512,
}
PRETRAINED_INIT_CONFIGURATION = {
"unc-nlp/lxmert-base-uncased": {"do_lower_case": True},
}
# Copied from transformers.models.bert.tokenization_bert.load_vocab
def load_vocab(vocab_file):
"""Loads a vocabulary file into a dictionary."""
vocab = collections.OrderedDict()
with open(vocab_file, "r", encoding="utf-8") as reader:
tokens = reader.readlines()
for index, token in enumerate(tokens):
token = token.rstrip("\n")
vocab[token] = index
return vocab
# Copied from transformers.models.bert.tokenization_bert.whitespace_tokenize
def whitespace_tokenize(text):
"""Runs basic whitespace cleaning and splitting on a piece of text."""
text = text.strip()
if not text:
return []
tokens = text.split()
return tokens
# Copied from transformers.models.bert.tokenization_bert.BertTokenizer with bert-base-cased->unc-nlp/lxmert-base-uncased, BERT->Lxmert, BertTokenizer->LxmertTokenizer
class LxmertTokenizer(PreTrainedTokenizer):
r"""
Construct a Lxmert tokenizer. Based on WordPiece.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to
this superclass for more information regarding those methods.
Args:
vocab_file (`str`):
File containing the vocabulary.
do_lower_case (`bool`, *optional*, defaults to `True`):
Whether or not to lowercase the input when tokenizing.
do_basic_tokenize (`bool`, *optional*, defaults to `True`):
Whether or not to do basic tokenization before WordPiece.
never_split (`Iterable`, *optional*):
Collection of tokens which will never be split during tokenization. Only has an effect when
`do_basic_tokenize=True`
unk_token (`str`, *optional*, defaults to `"[UNK]"`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
sep_token (`str`, *optional*, defaults to `"[SEP]"`):
The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
sequence classification or for a text and a question for question answering. It is also used as the last
token of a sequence built with special tokens.
pad_token (`str`, *optional*, defaults to `"[PAD]"`):
The token used for padding, for example when batching sequences of different lengths.
cls_token (`str`, *optional*, defaults to `"[CLS]"`):
The classifier token which is used when doing sequence classification (classification of the whole sequence
instead of per-token classification). It is the first token of the sequence when built with special tokens.
mask_token (`str`, *optional*, defaults to `"[MASK]"`):
The token used for masking values. This is the token used when training this model with masked language
modeling. This is the token which the model will try to predict.
tokenize_chinese_chars (`bool`, *optional*, defaults to `True`):
Whether or not to tokenize Chinese characters.
This should likely be deactivated for Japanese (see this
[issue](https://github.com/huggingface/transformers/issues/328)).
strip_accents (`bool`, *optional*):
Whether or not to strip all accents. If this option is not specified, then it will be determined by the
value for `lowercase` (as in the original Lxmert).
"""
vocab_files_names = VOCAB_FILES_NAMES
pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
pretrained_init_configuration = PRETRAINED_INIT_CONFIGURATION
max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
def __init__(
self,
vocab_file,
do_lower_case=True,
do_basic_tokenize=True,
never_split=None,
unk_token="[UNK]",
sep_token="[SEP]",
pad_token="[PAD]",
cls_token="[CLS]",
mask_token="[MASK]",
tokenize_chinese_chars=True,
strip_accents=None,
**kwargs
):
super().__init__(
do_lower_case=do_lower_case,
do_basic_tokenize=do_basic_tokenize,
never_split=never_split,
unk_token=unk_token,
sep_token=sep_token,
pad_token=pad_token,
cls_token=cls_token,
mask_token=mask_token,
tokenize_chinese_chars=tokenize_chinese_chars,
strip_accents=strip_accents,
**kwargs,
)
if not os.path.isfile(vocab_file):
raise ValueError(
f"Can't find a vocabulary file at path '{vocab_file}'. To load the vocabulary from a Google pretrained"
" model use `tokenizer = LxmertTokenizer.from_pretrained(PRETRAINED_MODEL_NAME)`"
)
self.vocab = load_vocab(vocab_file)
self.ids_to_tokens = collections.OrderedDict([(ids, tok) for tok, ids in self.vocab.items()])
self.do_basic_tokenize = do_basic_tokenize
if do_basic_tokenize:
self.basic_tokenizer = BasicTokenizer(
do_lower_case=do_lower_case,
never_split=never_split,
tokenize_chinese_chars=tokenize_chinese_chars,
strip_accents=strip_accents,
)
self.wordpiece_tokenizer = WordpieceTokenizer(vocab=self.vocab, unk_token=self.unk_token)
@property
def do_lower_case(self):
return self.basic_tokenizer.do_lower_case
@property
def vocab_size(self):
return len(self.vocab)
def get_vocab(self):
return dict(self.vocab, **self.added_tokens_encoder)
def _tokenize(self, text):
split_tokens = []
if self.do_basic_tokenize:
for token in self.basic_tokenizer.tokenize(text, never_split=self.all_special_tokens):
# If the token is part of the never_split set
if token in self.basic_tokenizer.never_split:
split_tokens.append(token)
else:
split_tokens += self.wordpiece_tokenizer.tokenize(token)
else:
split_tokens = self.wordpiece_tokenizer.tokenize(text)
return split_tokens
def _convert_token_to_id(self, token):
"""Converts a token (str) in an id using the vocab."""
return self.vocab.get(token, self.vocab.get(self.unk_token))
def _convert_id_to_token(self, index):
"""Converts an index (integer) in a token (str) using the vocab."""
return self.ids_to_tokens.get(index, self.unk_token)
def convert_tokens_to_string(self, tokens):
"""Converts a sequence of tokens (string) in a single string."""
out_string = " ".join(tokens).replace(" ##", "").strip()
return out_string
def build_inputs_with_special_tokens(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
) -> List[int]:
"""
Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
adding special tokens. A Lxmert sequence has the following format:
- single sequence: `[CLS] X [SEP]`
- pair of sequences: `[CLS] A [SEP] B [SEP]`
Args:
token_ids_0 (`List[int]`):
List of IDs to which the special tokens will be added.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
`List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
"""
if token_ids_1 is None:
return [self.cls_token_id] + token_ids_0 + [self.sep_token_id]
cls = [self.cls_token_id]
sep = [self.sep_token_id]
return cls + token_ids_0 + sep + token_ids_1 + sep
def get_special_tokens_mask(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
) -> List[int]:
"""
Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer `prepare_for_model` method.
Args:
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
already_has_special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the token list is already formatted with special tokens for the model.
Returns:
`List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
"""
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
)
if token_ids_1 is not None:
return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1)) + [1]
return [1] + ([0] * len(token_ids_0)) + [1]
def create_token_type_ids_from_sequences(
self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
) -> List[int]:
"""
Create a mask from the two sequences passed to be used in a sequence-pair classification task. A Lxmert
sequence pair mask has the following format:
```
0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence | second sequence |
```
If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s).
Args:
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
`List[int]`: List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s).
"""
sep = [self.sep_token_id]
cls = [self.cls_token_id]
if token_ids_1 is None:
return len(cls + token_ids_0 + sep) * [0]
return len(cls + token_ids_0 + sep) * [0] + len(token_ids_1 + sep) * [1]
def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
index = 0
if os.path.isdir(save_directory):
vocab_file = os.path.join(
save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
)
else:
vocab_file = (filename_prefix + "-" if filename_prefix else "") + save_directory
with open(vocab_file, "w", encoding="utf-8") as writer:
for token, token_index in sorted(self.vocab.items(), key=lambda kv: kv[1]):
if index != token_index:
logger.warning(
f"Saving vocabulary to {vocab_file}: vocabulary indices are not consecutive."
" Please check that the vocabulary is not corrupted!"
)
index = token_index
writer.write(token + "\n")
index += 1
return (vocab_file,)
# Copied from transformers.models.bert.tokenization_bert.BasicTokenizer
class BasicTokenizer(object):
"""
Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
Args:
do_lower_case (`bool`, *optional*, defaults to `True`):
Whether or not to lowercase the input when tokenizing.
never_split (`Iterable`, *optional*):
Collection of tokens which will never be split during tokenization. Only has an effect when
`do_basic_tokenize=True`
tokenize_chinese_chars (`bool`, *optional*, defaults to `True`):
Whether or not to tokenize Chinese characters.
This should likely be deactivated for Japanese (see this
[issue](https://github.com/huggingface/transformers/issues/328)).
strip_accents (`bool`, *optional*):
Whether or not to strip all accents. If this option is not specified, then it will be determined by the
value for `lowercase` (as in the original BERT).
"""
def __init__(self, do_lower_case=True, never_split=None, tokenize_chinese_chars=True, strip_accents=None):
if never_split is None:
never_split = []
self.do_lower_case = do_lower_case
self.never_split = set(never_split)
self.tokenize_chinese_chars = tokenize_chinese_chars
self.strip_accents = strip_accents
def tokenize(self, text, never_split=None):
"""
Basic Tokenization of a piece of text. Split on "white spaces" only, for sub-word tokenization, see
WordPieceTokenizer.
Args:
never_split (`List[str]`, *optional*)
Kept for backward compatibility purposes. Now implemented directly at the base class level (see
[`PreTrainedTokenizer.tokenize`]) List of token not to split.
"""
# union() returns a new set by concatenating the two sets.
never_split = self.never_split.union(set(never_split)) if never_split else self.never_split
text = self._clean_text(text)
# This was added on November 1st, 2018 for the multilingual and Chinese
# models. This is also applied to the English models now, but it doesn't
# matter since the English models were not trained on any Chinese data
# and generally don't have any Chinese data in them (there are Chinese
# characters in the vocabulary because Wikipedia does have some Chinese
# words in the English Wikipedia.).
if self.tokenize_chinese_chars:
text = self._tokenize_chinese_chars(text)
orig_tokens = whitespace_tokenize(text)
split_tokens = []
for token in orig_tokens:
if token not in never_split:
if self.do_lower_case:
token = token.lower()
if self.strip_accents is not False:
token = self._run_strip_accents(token)
elif self.strip_accents:
token = self._run_strip_accents(token)
split_tokens.extend(self._run_split_on_punc(token, never_split))
output_tokens = whitespace_tokenize(" ".join(split_tokens))
return output_tokens
def _run_strip_accents(self, text):
"""Strips accents from a piece of text."""
text = unicodedata.normalize("NFD", text)
output = []
for char in text:
cat = unicodedata.category(char)
if cat == "Mn":
continue
output.append(char)
return "".join(output)
def _run_split_on_punc(self, text, never_split=None):
"""Splits punctuation on a piece of text."""
if never_split is not None and text in never_split:
return [text]
chars = list(text)
i = 0
start_new_word = True
output = []
while i < len(chars):
char = chars[i]
if _is_punctuation(char):
output.append([char])
start_new_word = True
else:
if start_new_word:
output.append([])
start_new_word = False
output[-1].append(char)
i += 1
return ["".join(x) for x in output]
def _tokenize_chinese_chars(self, text):
"""Adds whitespace around any CJK character."""
output = []
for char in text:
cp = ord(char)
if self._is_chinese_char(cp):
output.append(" ")
output.append(char)
output.append(" ")
else:
output.append(char)
return "".join(output)
def _is_chinese_char(self, cp):
"""Checks whether CP is the codepoint of a CJK character."""
# This defines a "chinese character" as anything in the CJK Unicode block:
# https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_(Unicode_block)
#
# Note that the CJK Unicode block is NOT all Japanese and Korean characters,
# despite its name. The modern Korean Hangul alphabet is a different block,
# as is Japanese Hiragana and Katakana. Those alphabets are used to write
# space-separated words, so they are not treated specially and handled
# like the all of the other languages.
if (
(cp >= 0x4E00 and cp <= 0x9FFF)
or (cp >= 0x3400 and cp <= 0x4DBF) #
or (cp >= 0x20000 and cp <= 0x2A6DF) #
or (cp >= 0x2A700 and cp <= 0x2B73F) #
or (cp >= 0x2B740 and cp <= 0x2B81F) #
or (cp >= 0x2B820 and cp <= 0x2CEAF) #
or (cp >= 0xF900 and cp <= 0xFAFF)
or (cp >= 0x2F800 and cp <= 0x2FA1F) #
): #
return True
return False
def _clean_text(self, text):
"""Performs invalid character removal and whitespace cleanup on text."""
output = []
for char in text:
cp = ord(char)
if cp == 0 or cp == 0xFFFD or _is_control(char):
continue
if _is_whitespace(char):
output.append(" ")
else:
output.append(char)
return "".join(output)
# Copied from transformers.models.bert.tokenization_bert.WordpieceTokenizer
class WordpieceTokenizer(object):
"""Runs WordPiece tokenization."""
def __init__(self, vocab, unk_token, max_input_chars_per_word=100):
self.vocab = vocab
self.unk_token = unk_token
self.max_input_chars_per_word = max_input_chars_per_word
def tokenize(self, text):
"""
Tokenizes a piece of text into its word pieces. This uses a greedy longest-match-first algorithm to perform
tokenization using the given vocabulary.
For example, `input = "unaffable"` wil return as output `["un", "##aff", "##able"]`.
Args:
text: A single token or whitespace separated tokens. This should have
already been passed through *BasicTokenizer*.
Returns:
A list of wordpiece tokens.
"""
output_tokens = []
for token in whitespace_tokenize(text):
chars = list(token)
if len(chars) > self.max_input_chars_per_word:
output_tokens.append(self.unk_token)
continue
is_bad = False
start = 0
sub_tokens = []
while start < len(chars):
end = len(chars)
cur_substr = None
while start < end:
substr = "".join(chars[start:end])
if start > 0:
substr = "##" + substr
if substr in self.vocab:
cur_substr = substr
break
end -= 1
if cur_substr is None:
is_bad = True
break
sub_tokens.append(cur_substr)
start = end
if is_bad:
output_tokens.append(self.unk_token)
else:
output_tokens.extend(sub_tokens)
return output_tokens
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/beit/convert_beit_unilm_to_pytorch.py | # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Convert BEiT checkpoints from the unilm repository."""
import argparse
import json
from pathlib import Path
import torch
from datasets import load_dataset
from PIL import Image
import requests
from huggingface_hub import hf_hub_download
from transformers import (
BeitConfig,
BeitFeatureExtractor,
BeitForImageClassification,
BeitForMaskedImageModeling,
BeitForSemanticSegmentation,
)
from transformers.image_utils import PILImageResampling
from transformers.utils import logging
logging.set_verbosity_info()
logger = logging.get_logger(__name__)
# here we list all keys to be renamed (original name on the left, our name on the right)
def create_rename_keys(config, has_lm_head=False, is_semantic=False):
prefix = "backbone." if is_semantic else ""
rename_keys = []
for i in range(config.num_hidden_layers):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((f"{prefix}blocks.{i}.norm1.weight", f"beit.encoder.layer.{i}.layernorm_before.weight"))
rename_keys.append((f"{prefix}blocks.{i}.norm1.bias", f"beit.encoder.layer.{i}.layernorm_before.bias"))
rename_keys.append(
(f"{prefix}blocks.{i}.attn.proj.weight", f"beit.encoder.layer.{i}.attention.output.dense.weight")
)
rename_keys.append(
(f"{prefix}blocks.{i}.attn.proj.bias", f"beit.encoder.layer.{i}.attention.output.dense.bias")
)
rename_keys.append((f"{prefix}blocks.{i}.norm2.weight", f"beit.encoder.layer.{i}.layernorm_after.weight"))
rename_keys.append((f"{prefix}blocks.{i}.norm2.bias", f"beit.encoder.layer.{i}.layernorm_after.bias"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc1.weight", f"beit.encoder.layer.{i}.intermediate.dense.weight"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc1.bias", f"beit.encoder.layer.{i}.intermediate.dense.bias"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc2.weight", f"beit.encoder.layer.{i}.output.dense.weight"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc2.bias", f"beit.encoder.layer.{i}.output.dense.bias"))
# projection layer + position embeddings
rename_keys.extend(
[
(f"{prefix}cls_token", "beit.embeddings.cls_token"),
(f"{prefix}patch_embed.proj.weight", "beit.embeddings.patch_embeddings.projection.weight"),
(f"{prefix}patch_embed.proj.bias", "beit.embeddings.patch_embeddings.projection.bias"),
]
)
if has_lm_head:
# mask token + shared relative position bias + layernorm
rename_keys.extend(
[
("mask_token", "beit.embeddings.mask_token"),
(
"rel_pos_bias.relative_position_bias_table",
"beit.encoder.relative_position_bias.relative_position_bias_table",
),
(
"rel_pos_bias.relative_position_index",
"beit.encoder.relative_position_bias.relative_position_index",
),
("norm.weight", "layernorm.weight"),
("norm.bias", "layernorm.bias"),
]
)
elif is_semantic:
# semantic segmentation classification heads
rename_keys.extend(
[
("decode_head.conv_seg.weight", "decode_head.classifier.weight"),
("decode_head.conv_seg.bias", "decode_head.classifier.bias"),
("auxiliary_head.conv_seg.weight", "auxiliary_head.classifier.weight"),
("auxiliary_head.conv_seg.bias", "auxiliary_head.classifier.bias"),
]
)
else:
# layernorm + classification head
rename_keys.extend(
[
("fc_norm.weight", "beit.pooler.layernorm.weight"),
("fc_norm.bias", "beit.pooler.layernorm.bias"),
("head.weight", "classifier.weight"),
("head.bias", "classifier.bias"),
]
)
return rename_keys
# we split up the matrix of each encoder layer into queries, keys and values
def read_in_q_k_v(state_dict, config, has_lm_head=False, is_semantic=False):
for i in range(config.num_hidden_layers):
prefix = "backbone." if is_semantic else ""
# queries, keys and values
in_proj_weight = state_dict.pop(f"{prefix}blocks.{i}.attn.qkv.weight")
q_bias = state_dict.pop(f"{prefix}blocks.{i}.attn.q_bias")
v_bias = state_dict.pop(f"{prefix}blocks.{i}.attn.v_bias")
state_dict[f"beit.encoder.layer.{i}.attention.attention.query.weight"] = in_proj_weight[
: config.hidden_size, :
]
state_dict[f"beit.encoder.layer.{i}.attention.attention.query.bias"] = q_bias
state_dict[f"beit.encoder.layer.{i}.attention.attention.key.weight"] = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
state_dict[f"beit.encoder.layer.{i}.attention.attention.value.weight"] = in_proj_weight[
-config.hidden_size :, :
]
state_dict[f"beit.encoder.layer.{i}.attention.attention.value.bias"] = v_bias
# gamma_1 and gamma_2
# we call them lambda because otherwise they are renamed when using .from_pretrained
gamma_1 = state_dict.pop(f"{prefix}blocks.{i}.gamma_1")
gamma_2 = state_dict.pop(f"{prefix}blocks.{i}.gamma_2")
state_dict[f"beit.encoder.layer.{i}.lambda_1"] = gamma_1
state_dict[f"beit.encoder.layer.{i}.lambda_2"] = gamma_2
# relative_position bias table + index
if not has_lm_head:
# each layer has its own relative position bias
table = state_dict.pop(f"{prefix}blocks.{i}.attn.relative_position_bias_table")
index = state_dict.pop(f"{prefix}blocks.{i}.attn.relative_position_index")
state_dict[
f"beit.encoder.layer.{i}.attention.attention.relative_position_bias.relative_position_bias_table"
] = table
state_dict[
f"beit.encoder.layer.{i}.attention.attention.relative_position_bias.relative_position_index"
] = index
def rename_key(dct, old, new):
val = dct.pop(old)
dct[new] = val
# We will verify our results on an image of cute cats
def prepare_img():
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
im = Image.open(requests.get(url, stream=True).raw)
return im
@torch.no_grad()
def convert_beit_checkpoint(checkpoint_url, pytorch_dump_folder_path):
"""
Copy/paste/tweak model's weights to our BEiT structure.
"""
# define default BEiT configuration
config = BeitConfig()
has_lm_head = False
is_semantic = False
repo_id = "huggingface/label-files"
# set config parameters based on URL
if checkpoint_url[-9:-4] == "pt22k":
# masked image modeling
config.use_shared_relative_position_bias = True
config.use_mask_token = True
has_lm_head = True
elif checkpoint_url[-9:-4] == "ft22k":
# intermediate fine-tuning on ImageNet-22k
config.use_relative_position_bias = True
config.num_labels = 21841
filename = "imagenet-22k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
# this dataset contains 21843 labels but the model only has 21841
# we delete the classes as mentioned in https://github.com/google-research/big_transfer/issues/18
del id2label[9205]
del id2label[15027]
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
elif checkpoint_url[-8:-4] == "to1k":
# fine-tuning on ImageNet-1k
config.use_relative_position_bias = True
config.num_labels = 1000
filename = "imagenet-1k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
if "384" in checkpoint_url:
config.image_size = 384
if "512" in checkpoint_url:
config.image_size = 512
elif "ade20k" in checkpoint_url:
# fine-tuning
config.use_relative_position_bias = True
config.num_labels = 150
filename = "ade20k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
config.image_size = 640
is_semantic = True
else:
raise ValueError("Checkpoint not supported, URL should either end with 'pt22k', 'ft22k', 'to1k' or 'ade20k'")
# size of the architecture
if "base" in checkpoint_url:
pass
elif "large" in checkpoint_url:
config.hidden_size = 1024
config.intermediate_size = 4096
config.num_hidden_layers = 24
config.num_attention_heads = 16
if "ade20k" in checkpoint_url:
config.image_size = 640
config.out_indices = [7, 11, 15, 23]
else:
raise ValueError("Should either find 'base' or 'large' in checkpoint URL")
# load state_dict of original model, remove and rename some keys
state_dict = torch.hub.load_state_dict_from_url(checkpoint_url, map_location="cpu", check_hash=True)
state_dict = state_dict["model"] if "ade20k" not in checkpoint_url else state_dict["state_dict"]
rename_keys = create_rename_keys(config, has_lm_head=has_lm_head, is_semantic=is_semantic)
for src, dest in rename_keys:
rename_key(state_dict, src, dest)
read_in_q_k_v(state_dict, config, has_lm_head=has_lm_head, is_semantic=is_semantic)
if is_semantic:
# add prefix to decoder keys
for key, val in state_dict.copy().items():
val = state_dict.pop(key)
if key.startswith("backbone.fpn"):
key = key.replace("backbone.fpn", "fpn")
state_dict[key] = val
# load HuggingFace model
if checkpoint_url[-9:-4] == "pt22k":
model = BeitForMaskedImageModeling(config)
elif "ade20k" in checkpoint_url:
model = BeitForSemanticSegmentation(config)
else:
model = BeitForImageClassification(config)
model.eval()
model.load_state_dict(state_dict)
# Check outputs on an image
if is_semantic:
feature_extractor = BeitFeatureExtractor(size=config.image_size, do_center_crop=False)
ds = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
image = Image.open(ds[0]["file"])
else:
feature_extractor = BeitFeatureExtractor(
size=config.image_size, resample=PILImageResampling.BILINEAR, do_center_crop=False
)
image = prepare_img()
encoding = feature_extractor(images=image, return_tensors="pt")
pixel_values = encoding["pixel_values"]
outputs = model(pixel_values)
logits = outputs.logits
# verify logits
expected_shape = torch.Size([1, 1000])
if checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k"):
expected_shape = torch.Size([1, 196, 8192])
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k"):
expected_shape = torch.Size([1, 196, 8192])
elif checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k_ft22k"):
expected_shape = torch.Size([1, 21841])
expected_logits = torch.tensor([2.2288, 2.4671, 0.7395])
expected_class_idx = 2397
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k_ft22k"):
expected_shape = torch.Size([1, 21841])
expected_logits = torch.tensor([1.6881, -0.2787, 0.5901])
expected_class_idx = 2396
elif checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k_ft1k"):
expected_logits = torch.tensor([0.1241, 0.0798, -0.6569])
expected_class_idx = 285
elif checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-1.2385, -1.0987, -1.0108])
expected_class_idx = 281
elif checkpoint_url[:-4].endswith("beit_base_patch16_384_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-1.5303, -0.9484, -0.3147])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k_ft1k"):
expected_logits = torch.tensor([0.4610, -0.0928, 0.2086])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-0.4804, 0.6257, -0.1837])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_384_pt22k_ft22kto1k"):
expected_logits = torch.tensor([[-0.5122, 0.5117, -0.2113]])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_512_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-0.3062, 0.7261, 0.4852])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_base_patch16_640_pt22k_ft22ktoade20k"):
expected_shape = (1, 150, 160, 160)
expected_logits = torch.tensor(
[
[[-4.9225, -2.3954, -3.0522], [-2.8822, -1.0046, -1.7561], [-2.9549, -1.3228, -2.1347]],
[[-5.8168, -3.4129, -4.0778], [-3.8651, -2.2214, -3.0277], [-3.8356, -2.4643, -3.3535]],
[[-0.0078, 3.9952, 4.0754], [2.9856, 4.6944, 5.0035], [3.2413, 4.7813, 4.9969]],
]
)
elif checkpoint_url[:-4].endswith("beit_large_patch16_640_pt22k_ft22ktoade20k"):
expected_shape = (1, 150, 160, 160)
expected_logits = torch.tensor(
[
[[-4.3305, -2.3049, -3.0161], [-2.9591, -1.5305, -2.2251], [-3.4198, -1.8004, -2.9062]],
[[-5.8922, -3.7435, -4.3978], [-4.2063, -2.7872, -3.4755], [-4.2791, -3.1874, -4.1681]],
[[0.9895, 4.3467, 4.7663], [4.2476, 5.6830, 6.1518], [4.5550, 6.2495, 6.5154]],
]
)
else:
raise ValueError("Can't verify logits as model is not supported")
assert logits.shape == expected_shape, "Shape of logits not as expected"
if not has_lm_head:
if is_semantic:
assert torch.allclose(
logits[0, :3, :3, :3], expected_logits, atol=1e-3
), "First elements of logits not as expected"
else:
print("Predicted class idx:", logits.argmax(-1).item())
assert torch.allclose(
logits[0, :3], expected_logits, atol=1e-3
), "First elements of logits not as expected"
assert logits.argmax(-1).item() == expected_class_idx, "Predicted class index not as expected"
Path(pytorch_dump_folder_path).mkdir(exist_ok=True)
print(f"Saving model to {pytorch_dump_folder_path}")
model.save_pretrained(pytorch_dump_folder_path)
print(f"Saving feature extractor to {pytorch_dump_folder_path}")
feature_extractor.save_pretrained(pytorch_dump_folder_path)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--checkpoint_url",
default="https://conversationhub.blob.core.windows.net/beit-share-public/beit/beit_base_patch16_224_pt22k_ft22kto1k.pth",
type=str,
help="URL to the original PyTorch checkpoint (.pth file).",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the folder to output PyTorch model."
)
args = parser.parse_args()
convert_beit_checkpoint(args.checkpoint_url, args.pytorch_dump_folder_path)
| # coding=utf-8
# Copyright 2021 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Convert BEiT checkpoints from the unilm repository."""
import argparse
import json
from pathlib import Path
import torch
from datasets import load_dataset
from PIL import Image
import requests
from huggingface_hub import hf_hub_download
from transformers import (
BeitConfig,
BeitFeatureExtractor,
BeitForImageClassification,
BeitForMaskedImageModeling,
BeitForSemanticSegmentation,
)
from transformers.image_utils import PILImageResampling
from transformers.utils import logging
logging.set_verbosity_info()
logger = logging.get_logger(__name__)
# here we list all keys to be renamed (original name on the left, our name on the right)
def create_rename_keys(config, has_lm_head=False, is_semantic=False):
prefix = "backbone." if is_semantic else ""
rename_keys = []
for i in range(config.num_hidden_layers):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((f"{prefix}blocks.{i}.norm1.weight", f"beit.encoder.layer.{i}.layernorm_before.weight"))
rename_keys.append((f"{prefix}blocks.{i}.norm1.bias", f"beit.encoder.layer.{i}.layernorm_before.bias"))
rename_keys.append(
(f"{prefix}blocks.{i}.attn.proj.weight", f"beit.encoder.layer.{i}.attention.output.dense.weight")
)
rename_keys.append(
(f"{prefix}blocks.{i}.attn.proj.bias", f"beit.encoder.layer.{i}.attention.output.dense.bias")
)
rename_keys.append((f"{prefix}blocks.{i}.norm2.weight", f"beit.encoder.layer.{i}.layernorm_after.weight"))
rename_keys.append((f"{prefix}blocks.{i}.norm2.bias", f"beit.encoder.layer.{i}.layernorm_after.bias"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc1.weight", f"beit.encoder.layer.{i}.intermediate.dense.weight"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc1.bias", f"beit.encoder.layer.{i}.intermediate.dense.bias"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc2.weight", f"beit.encoder.layer.{i}.output.dense.weight"))
rename_keys.append((f"{prefix}blocks.{i}.mlp.fc2.bias", f"beit.encoder.layer.{i}.output.dense.bias"))
# projection layer + position embeddings
rename_keys.extend(
[
(f"{prefix}cls_token", "beit.embeddings.cls_token"),
(f"{prefix}patch_embed.proj.weight", "beit.embeddings.patch_embeddings.projection.weight"),
(f"{prefix}patch_embed.proj.bias", "beit.embeddings.patch_embeddings.projection.bias"),
]
)
if has_lm_head:
# mask token + shared relative position bias + layernorm
rename_keys.extend(
[
("mask_token", "beit.embeddings.mask_token"),
(
"rel_pos_bias.relative_position_bias_table",
"beit.encoder.relative_position_bias.relative_position_bias_table",
),
(
"rel_pos_bias.relative_position_index",
"beit.encoder.relative_position_bias.relative_position_index",
),
("norm.weight", "layernorm.weight"),
("norm.bias", "layernorm.bias"),
]
)
elif is_semantic:
# semantic segmentation classification heads
rename_keys.extend(
[
("decode_head.conv_seg.weight", "decode_head.classifier.weight"),
("decode_head.conv_seg.bias", "decode_head.classifier.bias"),
("auxiliary_head.conv_seg.weight", "auxiliary_head.classifier.weight"),
("auxiliary_head.conv_seg.bias", "auxiliary_head.classifier.bias"),
]
)
else:
# layernorm + classification head
rename_keys.extend(
[
("fc_norm.weight", "beit.pooler.layernorm.weight"),
("fc_norm.bias", "beit.pooler.layernorm.bias"),
("head.weight", "classifier.weight"),
("head.bias", "classifier.bias"),
]
)
return rename_keys
# we split up the matrix of each encoder layer into queries, keys and values
def read_in_q_k_v(state_dict, config, has_lm_head=False, is_semantic=False):
for i in range(config.num_hidden_layers):
prefix = "backbone." if is_semantic else ""
# queries, keys and values
in_proj_weight = state_dict.pop(f"{prefix}blocks.{i}.attn.qkv.weight")
q_bias = state_dict.pop(f"{prefix}blocks.{i}.attn.q_bias")
v_bias = state_dict.pop(f"{prefix}blocks.{i}.attn.v_bias")
state_dict[f"beit.encoder.layer.{i}.attention.attention.query.weight"] = in_proj_weight[
: config.hidden_size, :
]
state_dict[f"beit.encoder.layer.{i}.attention.attention.query.bias"] = q_bias
state_dict[f"beit.encoder.layer.{i}.attention.attention.key.weight"] = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
state_dict[f"beit.encoder.layer.{i}.attention.attention.value.weight"] = in_proj_weight[
-config.hidden_size :, :
]
state_dict[f"beit.encoder.layer.{i}.attention.attention.value.bias"] = v_bias
# gamma_1 and gamma_2
# we call them lambda because otherwise they are renamed when using .from_pretrained
gamma_1 = state_dict.pop(f"{prefix}blocks.{i}.gamma_1")
gamma_2 = state_dict.pop(f"{prefix}blocks.{i}.gamma_2")
state_dict[f"beit.encoder.layer.{i}.lambda_1"] = gamma_1
state_dict[f"beit.encoder.layer.{i}.lambda_2"] = gamma_2
# relative_position bias table + index
if not has_lm_head:
# each layer has its own relative position bias
table = state_dict.pop(f"{prefix}blocks.{i}.attn.relative_position_bias_table")
index = state_dict.pop(f"{prefix}blocks.{i}.attn.relative_position_index")
state_dict[
f"beit.encoder.layer.{i}.attention.attention.relative_position_bias.relative_position_bias_table"
] = table
state_dict[
f"beit.encoder.layer.{i}.attention.attention.relative_position_bias.relative_position_index"
] = index
def rename_key(dct, old, new):
val = dct.pop(old)
dct[new] = val
# We will verify our results on an image of cute cats
def prepare_img():
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
im = Image.open(requests.get(url, stream=True).raw)
return im
@torch.no_grad()
def convert_beit_checkpoint(checkpoint_url, pytorch_dump_folder_path):
"""
Copy/paste/tweak model's weights to our BEiT structure.
"""
# define default BEiT configuration
config = BeitConfig()
has_lm_head = False
is_semantic = False
repo_id = "huggingface/label-files"
# set config parameters based on URL
if checkpoint_url[-9:-4] == "pt22k":
# masked image modeling
config.use_shared_relative_position_bias = True
config.use_mask_token = True
has_lm_head = True
elif checkpoint_url[-9:-4] == "ft22k":
# intermediate fine-tuning on ImageNet-22k
config.use_relative_position_bias = True
config.num_labels = 21841
filename = "imagenet-22k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
# this dataset contains 21843 labels but the model only has 21841
# we delete the classes as mentioned in https://github.com/google-research/big_transfer/issues/18
del id2label[9205]
del id2label[15027]
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
elif checkpoint_url[-8:-4] == "to1k":
# fine-tuning on ImageNet-1k
config.use_relative_position_bias = True
config.num_labels = 1000
filename = "imagenet-1k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
if "384" in checkpoint_url:
config.image_size = 384
if "512" in checkpoint_url:
config.image_size = 512
elif "ade20k" in checkpoint_url:
# fine-tuning
config.use_relative_position_bias = True
config.num_labels = 150
filename = "ade20k-id2label.json"
id2label = json.load(open(hf_hub_download(repo_id, filename, repo_type="dataset"), "r"))
id2label = {int(k): v for k, v in id2label.items()}
config.id2label = id2label
config.label2id = {v: k for k, v in id2label.items()}
config.image_size = 640
is_semantic = True
else:
raise ValueError("Checkpoint not supported, URL should either end with 'pt22k', 'ft22k', 'to1k' or 'ade20k'")
# size of the architecture
if "base" in checkpoint_url:
pass
elif "large" in checkpoint_url:
config.hidden_size = 1024
config.intermediate_size = 4096
config.num_hidden_layers = 24
config.num_attention_heads = 16
if "ade20k" in checkpoint_url:
config.image_size = 640
config.out_indices = [7, 11, 15, 23]
else:
raise ValueError("Should either find 'base' or 'large' in checkpoint URL")
# load state_dict of original model, remove and rename some keys
state_dict = torch.hub.load_state_dict_from_url(checkpoint_url, map_location="cpu", check_hash=True)
state_dict = state_dict["model"] if "ade20k" not in checkpoint_url else state_dict["state_dict"]
rename_keys = create_rename_keys(config, has_lm_head=has_lm_head, is_semantic=is_semantic)
for src, dest in rename_keys:
rename_key(state_dict, src, dest)
read_in_q_k_v(state_dict, config, has_lm_head=has_lm_head, is_semantic=is_semantic)
if is_semantic:
# add prefix to decoder keys
for key, val in state_dict.copy().items():
val = state_dict.pop(key)
if key.startswith("backbone.fpn"):
key = key.replace("backbone.fpn", "fpn")
state_dict[key] = val
# load HuggingFace model
if checkpoint_url[-9:-4] == "pt22k":
model = BeitForMaskedImageModeling(config)
elif "ade20k" in checkpoint_url:
model = BeitForSemanticSegmentation(config)
else:
model = BeitForImageClassification(config)
model.eval()
model.load_state_dict(state_dict)
# Check outputs on an image
if is_semantic:
feature_extractor = BeitFeatureExtractor(size=config.image_size, do_center_crop=False)
ds = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
image = Image.open(ds[0]["file"])
else:
feature_extractor = BeitFeatureExtractor(
size=config.image_size, resample=PILImageResampling.BILINEAR, do_center_crop=False
)
image = prepare_img()
encoding = feature_extractor(images=image, return_tensors="pt")
pixel_values = encoding["pixel_values"]
outputs = model(pixel_values)
logits = outputs.logits
# verify logits
expected_shape = torch.Size([1, 1000])
if checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k"):
expected_shape = torch.Size([1, 196, 8192])
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k"):
expected_shape = torch.Size([1, 196, 8192])
elif checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k_ft22k"):
expected_shape = torch.Size([1, 21841])
expected_logits = torch.tensor([2.2288, 2.4671, 0.7395])
expected_class_idx = 2397
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k_ft22k"):
expected_shape = torch.Size([1, 21841])
expected_logits = torch.tensor([1.6881, -0.2787, 0.5901])
expected_class_idx = 2396
elif checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k_ft1k"):
expected_logits = torch.tensor([0.1241, 0.0798, -0.6569])
expected_class_idx = 285
elif checkpoint_url[:-4].endswith("beit_base_patch16_224_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-1.2385, -1.0987, -1.0108])
expected_class_idx = 281
elif checkpoint_url[:-4].endswith("beit_base_patch16_384_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-1.5303, -0.9484, -0.3147])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k_ft1k"):
expected_logits = torch.tensor([0.4610, -0.0928, 0.2086])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_224_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-0.4804, 0.6257, -0.1837])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_384_pt22k_ft22kto1k"):
expected_logits = torch.tensor([[-0.5122, 0.5117, -0.2113]])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_large_patch16_512_pt22k_ft22kto1k"):
expected_logits = torch.tensor([-0.3062, 0.7261, 0.4852])
expected_class_idx = 761
elif checkpoint_url[:-4].endswith("beit_base_patch16_640_pt22k_ft22ktoade20k"):
expected_shape = (1, 150, 160, 160)
expected_logits = torch.tensor(
[
[[-4.9225, -2.3954, -3.0522], [-2.8822, -1.0046, -1.7561], [-2.9549, -1.3228, -2.1347]],
[[-5.8168, -3.4129, -4.0778], [-3.8651, -2.2214, -3.0277], [-3.8356, -2.4643, -3.3535]],
[[-0.0078, 3.9952, 4.0754], [2.9856, 4.6944, 5.0035], [3.2413, 4.7813, 4.9969]],
]
)
elif checkpoint_url[:-4].endswith("beit_large_patch16_640_pt22k_ft22ktoade20k"):
expected_shape = (1, 150, 160, 160)
expected_logits = torch.tensor(
[
[[-4.3305, -2.3049, -3.0161], [-2.9591, -1.5305, -2.2251], [-3.4198, -1.8004, -2.9062]],
[[-5.8922, -3.7435, -4.3978], [-4.2063, -2.7872, -3.4755], [-4.2791, -3.1874, -4.1681]],
[[0.9895, 4.3467, 4.7663], [4.2476, 5.6830, 6.1518], [4.5550, 6.2495, 6.5154]],
]
)
else:
raise ValueError("Can't verify logits as model is not supported")
assert logits.shape == expected_shape, "Shape of logits not as expected"
if not has_lm_head:
if is_semantic:
assert torch.allclose(
logits[0, :3, :3, :3], expected_logits, atol=1e-3
), "First elements of logits not as expected"
else:
print("Predicted class idx:", logits.argmax(-1).item())
assert torch.allclose(
logits[0, :3], expected_logits, atol=1e-3
), "First elements of logits not as expected"
assert logits.argmax(-1).item() == expected_class_idx, "Predicted class index not as expected"
Path(pytorch_dump_folder_path).mkdir(exist_ok=True)
print(f"Saving model to {pytorch_dump_folder_path}")
model.save_pretrained(pytorch_dump_folder_path)
print(f"Saving feature extractor to {pytorch_dump_folder_path}")
feature_extractor.save_pretrained(pytorch_dump_folder_path)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--checkpoint_url",
default="https://conversationhub.blob.core.windows.net/beit-share-public/beit/beit_base_patch16_224_pt22k_ft22kto1k.pth",
type=str,
help="URL to the original PyTorch checkpoint (.pth file).",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the folder to output PyTorch model."
)
args = parser.parse_args()
convert_beit_checkpoint(args.checkpoint_url, args.pytorch_dump_folder_path)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./tests/models/mbart50/__init__.py | -1 |
||
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/roberta/modeling_flax_roberta.py | # coding=utf-8
# Copyright 2021 The Google Flax Team Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Callable, Optional, Tuple
import numpy as np
import flax.linen as nn
import jax
import jax.numpy as jnp
from flax.core.frozen_dict import FrozenDict, freeze, unfreeze
from flax.linen import combine_masks, make_causal_mask
from flax.linen import partitioning as nn_partitioning
from flax.linen.attention import dot_product_attention_weights
from flax.traverse_util import flatten_dict, unflatten_dict
from jax import lax
from ...modeling_flax_outputs import (
FlaxBaseModelOutputWithPastAndCrossAttentions,
FlaxBaseModelOutputWithPooling,
FlaxBaseModelOutputWithPoolingAndCrossAttentions,
FlaxCausalLMOutputWithCrossAttentions,
FlaxMaskedLMOutput,
FlaxMultipleChoiceModelOutput,
FlaxQuestionAnsweringModelOutput,
FlaxSequenceClassifierOutput,
FlaxTokenClassifierOutput,
)
from ...modeling_flax_utils import ACT2FN, FlaxPreTrainedModel, append_call_sample_docstring, overwrite_call_docstring
from ...utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_roberta import RobertaConfig
logger = logging.get_logger(__name__)
_CHECKPOINT_FOR_DOC = "roberta-base"
_CONFIG_FOR_DOC = "RobertaConfig"
_TOKENIZER_FOR_DOC = "RobertaTokenizer"
remat = nn_partitioning.remat
def create_position_ids_from_input_ids(input_ids, padding_idx):
"""
Replace non-padding symbols with their position numbers. Position numbers begin at padding_idx+1. Padding symbols
are ignored. This is modified from fairseq's `utils.make_positions`.
Args:
input_ids: jnp.ndarray
padding_idx: int
Returns: jnp.ndarray
"""
# The series of casts and type-conversions here are carefully balanced to both work with ONNX export and XLA.
mask = (input_ids != padding_idx).astype("i4")
if mask.ndim > 2:
mask = mask.reshape((-1, mask.shape[-1]))
incremental_indices = jnp.cumsum(mask, axis=1).astype("i4") * mask
incremental_indices = incremental_indices.reshape(input_ids.shape)
else:
incremental_indices = jnp.cumsum(mask, axis=1).astype("i4") * mask
return incremental_indices.astype("i4") + padding_idx
ROBERTA_START_DOCSTRING = r"""
This model inherits from [`FlaxPreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading, saving and converting weights from PyTorch models)
This model is also a Flax Linen [flax.linen.Module](https://flax.readthedocs.io/en/latest/flax.linen.html#module)
subclass. Use it as a regular Flax linen Module and refer to the Flax documentation for all matter related to
general usage and behavior.
Finally, this model supports inherent JAX features such as:
- [Just-In-Time (JIT) compilation](https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit)
- [Automatic Differentiation](https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation)
- [Vectorization](https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap)
- [Parallelization](https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap)
Parameters:
config ([`RobertaConfig`]): Model configuration class with all the parameters of the
model. Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~FlaxPreTrainedModel.from_pretrained`] method to load the model weights.
"""
ROBERTA_INPUTS_DOCSTRING = r"""
Args:
input_ids (`numpy.ndarray` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using [`BertTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`numpy.ndarray` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
token_type_ids (`numpy.ndarray` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0,
1]`:
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`numpy.ndarray` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0,
config.max_position_embeddings - 1]`.
head_mask (`numpy.ndarray` of shape `({0})`, `optional):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertEmbeddings with Bert->Roberta
class FlaxRobertaEmbeddings(nn.Module):
"""Construct the embeddings from word, position and token_type embeddings."""
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.word_embeddings = nn.Embed(
self.config.vocab_size,
self.config.hidden_size,
embedding_init=jax.nn.initializers.normal(stddev=self.config.initializer_range),
dtype=self.dtype,
)
self.position_embeddings = nn.Embed(
self.config.max_position_embeddings,
self.config.hidden_size,
embedding_init=jax.nn.initializers.normal(stddev=self.config.initializer_range),
dtype=self.dtype,
)
self.token_type_embeddings = nn.Embed(
self.config.type_vocab_size,
self.config.hidden_size,
embedding_init=jax.nn.initializers.normal(stddev=self.config.initializer_range),
dtype=self.dtype,
)
self.LayerNorm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
def __call__(self, input_ids, token_type_ids, position_ids, attention_mask, deterministic: bool = True):
# Embed
inputs_embeds = self.word_embeddings(input_ids.astype("i4"))
position_embeds = self.position_embeddings(position_ids.astype("i4"))
token_type_embeddings = self.token_type_embeddings(token_type_ids.astype("i4"))
# Sum all embeddings
hidden_states = inputs_embeds + token_type_embeddings + position_embeds
# Layer Norm
hidden_states = self.LayerNorm(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertSelfAttention with Bert->Roberta
class FlaxRobertaSelfAttention(nn.Module):
config: RobertaConfig
causal: bool = False
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.head_dim = self.config.hidden_size // self.config.num_attention_heads
if self.config.hidden_size % self.config.num_attention_heads != 0:
raise ValueError(
"`config.hidden_size`: {self.config.hidden_size} has to be a multiple of `config.num_attention_heads` "
" : {self.config.num_attention_heads}"
)
self.query = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.key = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.value = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
if self.causal:
self.causal_mask = make_causal_mask(
jnp.ones((1, self.config.max_position_embeddings), dtype="bool"), dtype="bool"
)
def _split_heads(self, hidden_states):
return hidden_states.reshape(hidden_states.shape[:2] + (self.config.num_attention_heads, self.head_dim))
def _merge_heads(self, hidden_states):
return hidden_states.reshape(hidden_states.shape[:2] + (self.config.hidden_size,))
@nn.compact
# Copied from transformers.models.bart.modeling_flax_bart.FlaxBartAttention._concatenate_to_cache
def _concatenate_to_cache(self, key, value, query, attention_mask):
"""
This function takes projected key, value states from a single input token and concatenates the states to cached
states from previous steps. This function is slighly adapted from the official Flax repository:
https://github.com/google/flax/blob/491ce18759622506588784b4fca0e4bf05f8c8cd/flax/linen/attention.py#L252
"""
# detect if we're initializing by absence of existing cache data.
is_initialized = self.has_variable("cache", "cached_key")
cached_key = self.variable("cache", "cached_key", jnp.zeros, key.shape, key.dtype)
cached_value = self.variable("cache", "cached_value", jnp.zeros, value.shape, value.dtype)
cache_index = self.variable("cache", "cache_index", lambda: jnp.array(0, dtype=jnp.int32))
if is_initialized:
*batch_dims, max_length, num_heads, depth_per_head = cached_key.value.shape
# update key, value caches with our new 1d spatial slices
cur_index = cache_index.value
indices = (0,) * len(batch_dims) + (cur_index, 0, 0)
key = lax.dynamic_update_slice(cached_key.value, key, indices)
value = lax.dynamic_update_slice(cached_value.value, value, indices)
cached_key.value = key
cached_value.value = value
num_updated_cache_vectors = query.shape[1]
cache_index.value = cache_index.value + num_updated_cache_vectors
# causal mask for cached decoder self-attention: our single query position should only attend to those key positions that have already been generated and cached, not the remaining zero elements.
pad_mask = jnp.broadcast_to(
jnp.arange(max_length) < cur_index + num_updated_cache_vectors,
tuple(batch_dims) + (1, num_updated_cache_vectors, max_length),
)
attention_mask = combine_masks(pad_mask, attention_mask)
return key, value, attention_mask
def __call__(
self,
hidden_states,
attention_mask,
layer_head_mask,
key_value_states: Optional[jnp.array] = None,
init_cache: bool = False,
deterministic=True,
output_attentions: bool = False,
):
# if key_value_states are provided this layer is used as a cross-attention layer
# for the decoder
is_cross_attention = key_value_states is not None
batch_size = hidden_states.shape[0]
# get query proj
query_states = self.query(hidden_states)
# get key, value proj
if is_cross_attention:
# cross_attentions
key_states = self.key(key_value_states)
value_states = self.value(key_value_states)
else:
# self_attention
key_states = self.key(hidden_states)
value_states = self.value(hidden_states)
query_states = self._split_heads(query_states)
key_states = self._split_heads(key_states)
value_states = self._split_heads(value_states)
# handle cache prepare causal attention mask
if self.causal:
query_length, key_length = query_states.shape[1], key_states.shape[1]
if self.has_variable("cache", "cached_key"):
mask_shift = self.variables["cache"]["cache_index"]
max_decoder_length = self.variables["cache"]["cached_key"].shape[1]
causal_mask = lax.dynamic_slice(
self.causal_mask, (0, 0, mask_shift, 0), (1, 1, query_length, max_decoder_length)
)
else:
causal_mask = self.causal_mask[:, :, :query_length, :key_length]
causal_mask = jnp.broadcast_to(causal_mask, (batch_size,) + causal_mask.shape[1:])
# combine masks if needed
if attention_mask is not None and self.causal:
attention_mask = jnp.broadcast_to(jnp.expand_dims(attention_mask, axis=(-3, -2)), causal_mask.shape)
attention_mask = combine_masks(attention_mask, causal_mask)
elif self.causal:
attention_mask = causal_mask
elif attention_mask is not None:
attention_mask = jnp.expand_dims(attention_mask, axis=(-3, -2))
# During fast autoregressive decoding, we feed one position at a time,
# and cache the keys and values step by step.
if self.causal and (self.has_variable("cache", "cached_key") or init_cache):
key_states, value_states, attention_mask = self._concatenate_to_cache(
key_states, value_states, query_states, attention_mask
)
# Convert the boolean attention mask to an attention bias.
if attention_mask is not None:
# attention mask in the form of attention bias
attention_bias = lax.select(
attention_mask > 0,
jnp.full(attention_mask.shape, 0.0).astype(self.dtype),
jnp.full(attention_mask.shape, -1e10).astype(self.dtype),
)
else:
attention_bias = None
dropout_rng = None
if not deterministic and self.config.attention_probs_dropout_prob > 0.0:
dropout_rng = self.make_rng("dropout")
attn_weights = dot_product_attention_weights(
query_states,
key_states,
bias=attention_bias,
dropout_rng=dropout_rng,
dropout_rate=self.config.attention_probs_dropout_prob,
broadcast_dropout=True,
deterministic=deterministic,
dtype=self.dtype,
precision=None,
)
# Mask heads if we want to
if layer_head_mask is not None:
attn_weights = jnp.einsum("...hqk,h->...hqk", attn_weights, layer_head_mask)
attn_output = jnp.einsum("...hqk,...khd->...qhd", attn_weights, value_states)
attn_output = attn_output.reshape(attn_output.shape[:2] + (-1,))
outputs = (attn_output, attn_weights) if output_attentions else (attn_output,)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertSelfOutput with Bert->Roberta
class FlaxRobertaSelfOutput(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
self.LayerNorm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
def __call__(self, hidden_states, input_tensor, deterministic: bool = True):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.LayerNorm(hidden_states + input_tensor)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertAttention with Bert->Roberta
class FlaxRobertaAttention(nn.Module):
config: RobertaConfig
causal: bool = False
dtype: jnp.dtype = jnp.float32
def setup(self):
self.self = FlaxRobertaSelfAttention(self.config, causal=self.causal, dtype=self.dtype)
self.output = FlaxRobertaSelfOutput(self.config, dtype=self.dtype)
def __call__(
self,
hidden_states,
attention_mask,
layer_head_mask,
key_value_states=None,
init_cache=False,
deterministic=True,
output_attentions: bool = False,
):
# Attention mask comes in as attention_mask.shape == (*batch_sizes, kv_length)
# FLAX expects: attention_mask.shape == (*batch_sizes, 1, 1, kv_length) such that it is broadcastable
# with attn_weights.shape == (*batch_sizes, num_heads, q_length, kv_length)
attn_outputs = self.self(
hidden_states,
attention_mask,
layer_head_mask=layer_head_mask,
key_value_states=key_value_states,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
)
attn_output = attn_outputs[0]
hidden_states = self.output(attn_output, hidden_states, deterministic=deterministic)
outputs = (hidden_states,)
if output_attentions:
outputs += (attn_outputs[1],)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertIntermediate with Bert->Roberta
class FlaxRobertaIntermediate(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.intermediate_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
self.activation = ACT2FN[self.config.hidden_act]
def __call__(self, hidden_states):
hidden_states = self.dense(hidden_states)
hidden_states = self.activation(hidden_states)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertOutput with Bert->Roberta
class FlaxRobertaOutput(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
self.LayerNorm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
def __call__(self, hidden_states, attention_output, deterministic: bool = True):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.LayerNorm(hidden_states + attention_output)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertLayer with Bert->Roberta
class FlaxRobertaLayer(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.attention = FlaxRobertaAttention(self.config, causal=self.config.is_decoder, dtype=self.dtype)
self.intermediate = FlaxRobertaIntermediate(self.config, dtype=self.dtype)
self.output = FlaxRobertaOutput(self.config, dtype=self.dtype)
if self.config.add_cross_attention:
self.crossattention = FlaxRobertaAttention(self.config, causal=False, dtype=self.dtype)
def __call__(
self,
hidden_states,
attention_mask,
layer_head_mask,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
):
# Self Attention
attention_outputs = self.attention(
hidden_states,
attention_mask,
layer_head_mask=layer_head_mask,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
)
attention_output = attention_outputs[0]
# Cross-Attention Block
if encoder_hidden_states is not None:
cross_attention_outputs = self.crossattention(
attention_output,
attention_mask=encoder_attention_mask,
layer_head_mask=layer_head_mask,
key_value_states=encoder_hidden_states,
deterministic=deterministic,
output_attentions=output_attentions,
)
attention_output = cross_attention_outputs[0]
hidden_states = self.intermediate(attention_output)
hidden_states = self.output(hidden_states, attention_output, deterministic=deterministic)
outputs = (hidden_states,)
if output_attentions:
outputs += (attention_outputs[1],)
if encoder_hidden_states is not None:
outputs += (cross_attention_outputs[1],)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertLayerCollection with Bert->Roberta
class FlaxRobertaLayerCollection(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
gradient_checkpointing: bool = False
def setup(self):
if self.gradient_checkpointing:
FlaxRobertaCheckpointLayer = remat(FlaxRobertaLayer, static_argnums=(5, 6, 7))
self.layers = [
FlaxRobertaCheckpointLayer(self.config, name=str(i), dtype=self.dtype)
for i in range(self.config.num_hidden_layers)
]
else:
self.layers = [
FlaxRobertaLayer(self.config, name=str(i), dtype=self.dtype)
for i in range(self.config.num_hidden_layers)
]
def __call__(
self,
hidden_states,
attention_mask,
head_mask,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
all_attentions = () if output_attentions else None
all_hidden_states = () if output_hidden_states else None
all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None
# Check if head_mask has a correct number of layers specified if desired
if head_mask is not None:
if head_mask.shape[0] != (len(self.layers)):
raise ValueError(
f"The head_mask should be specified for {len(self.layers)} layers, but it is for "
f" {head_mask.shape[0]}."
)
for i, layer in enumerate(self.layers):
if output_hidden_states:
all_hidden_states += (hidden_states,)
layer_outputs = layer(
hidden_states,
attention_mask,
head_mask[i] if head_mask is not None else None,
encoder_hidden_states,
encoder_attention_mask,
init_cache,
deterministic,
output_attentions,
)
hidden_states = layer_outputs[0]
if output_attentions:
all_attentions += (layer_outputs[1],)
if encoder_hidden_states is not None:
all_cross_attentions += (layer_outputs[2],)
if output_hidden_states:
all_hidden_states += (hidden_states,)
outputs = (hidden_states, all_hidden_states, all_attentions, all_cross_attentions)
if not return_dict:
return tuple(v for v in outputs if v is not None)
return FlaxBaseModelOutputWithPastAndCrossAttentions(
last_hidden_state=hidden_states,
hidden_states=all_hidden_states,
attentions=all_attentions,
cross_attentions=all_cross_attentions,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertEncoder with Bert->Roberta
class FlaxRobertaEncoder(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
gradient_checkpointing: bool = False
def setup(self):
self.layer = FlaxRobertaLayerCollection(
self.config,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
def __call__(
self,
hidden_states,
attention_mask,
head_mask,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
return self.layer(
hidden_states,
attention_mask,
head_mask=head_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertPooler with Bert->Roberta
class FlaxRobertaPooler(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
def __call__(self, hidden_states):
cls_hidden_state = hidden_states[:, 0]
cls_hidden_state = self.dense(cls_hidden_state)
return nn.tanh(cls_hidden_state)
class FlaxRobertaLMHead(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
bias_init: Callable[..., np.ndarray] = jax.nn.initializers.zeros
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.layer_norm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
self.decoder = nn.Dense(
self.config.vocab_size,
dtype=self.dtype,
use_bias=False,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.bias = self.param("bias", self.bias_init, (self.config.vocab_size,))
def __call__(self, hidden_states, shared_embedding=None):
hidden_states = self.dense(hidden_states)
hidden_states = ACT2FN["gelu"](hidden_states)
hidden_states = self.layer_norm(hidden_states)
if shared_embedding is not None:
hidden_states = self.decoder.apply({"params": {"kernel": shared_embedding.T}}, hidden_states)
else:
hidden_states = self.decoder(hidden_states)
bias = jnp.asarray(self.bias, self.dtype)
hidden_states += bias
return hidden_states
class FlaxRobertaClassificationHead(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
classifier_dropout = (
self.config.classifier_dropout
if self.config.classifier_dropout is not None
else self.config.hidden_dropout_prob
)
self.dropout = nn.Dropout(rate=classifier_dropout)
self.out_proj = nn.Dense(
self.config.num_labels,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
def __call__(self, hidden_states, deterministic=True):
hidden_states = hidden_states[:, 0, :] # take <s> token (equiv. to [CLS])
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.dense(hidden_states)
hidden_states = nn.tanh(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.out_proj(hidden_states)
return hidden_states
class FlaxRobertaPreTrainedModel(FlaxPreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = RobertaConfig
base_model_prefix = "roberta"
module_class: nn.Module = None
def __init__(
self,
config: RobertaConfig,
input_shape: Tuple = (1, 1),
seed: int = 0,
dtype: jnp.dtype = jnp.float32,
_do_init: bool = True,
gradient_checkpointing: bool = False,
**kwargs
):
module = self.module_class(config=config, dtype=dtype, gradient_checkpointing=gradient_checkpointing, **kwargs)
super().__init__(config, module, input_shape=input_shape, seed=seed, dtype=dtype, _do_init=_do_init)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertPreTrainedModel.enable_gradient_checkpointing
def enable_gradient_checkpointing(self):
self._module = self.module_class(
config=self.config,
dtype=self.dtype,
gradient_checkpointing=True,
)
def init_weights(self, rng: jax.random.PRNGKey, input_shape: Tuple, params: FrozenDict = None) -> FrozenDict:
# init input tensors
input_ids = jnp.zeros(input_shape, dtype="i4")
token_type_ids = jnp.ones_like(input_ids)
position_ids = create_position_ids_from_input_ids(input_ids, self.config.pad_token_id)
attention_mask = jnp.ones_like(input_ids)
head_mask = jnp.ones((self.config.num_hidden_layers, self.config.num_attention_heads))
params_rng, dropout_rng = jax.random.split(rng)
rngs = {"params": params_rng, "dropout": dropout_rng}
if self.config.add_cross_attention:
encoder_hidden_states = jnp.zeros(input_shape + (self.config.hidden_size,))
encoder_attention_mask = attention_mask
module_init_outputs = self.module.init(
rngs,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
encoder_hidden_states,
encoder_attention_mask,
return_dict=False,
)
else:
module_init_outputs = self.module.init(
rngs, input_ids, attention_mask, token_type_ids, position_ids, head_mask, return_dict=False
)
random_params = module_init_outputs["params"]
if params is not None:
random_params = flatten_dict(unfreeze(random_params))
params = flatten_dict(unfreeze(params))
for missing_key in self._missing_keys:
params[missing_key] = random_params[missing_key]
self._missing_keys = set()
return freeze(unflatten_dict(params))
else:
return random_params
# Copied from transformers.models.bart.modeling_flax_bart.FlaxBartDecoderPreTrainedModel.init_cache
def init_cache(self, batch_size, max_length):
r"""
Args:
batch_size (`int`):
batch_size used for fast auto-regressive decoding. Defines the batch size of the initialized cache.
max_length (`int`):
maximum possible length for auto-regressive decoding. Defines the sequence length of the initialized
cache.
"""
# init input variables to retrieve cache
input_ids = jnp.ones((batch_size, max_length), dtype="i4")
attention_mask = jnp.ones_like(input_ids, dtype="i4")
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
init_variables = self.module.init(
jax.random.PRNGKey(0), input_ids, attention_mask, position_ids, return_dict=False, init_cache=True
)
return unfreeze(init_variables["cache"])
@add_start_docstrings_to_model_forward(ROBERTA_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
def __call__(
self,
input_ids,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
encoder_hidden_states=None,
encoder_attention_mask=None,
params: dict = None,
dropout_rng: jax.random.PRNGKey = None,
train: bool = False,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
past_key_values: dict = None,
):
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.return_dict
# init input tensors if not passed
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
if position_ids is None:
position_ids = create_position_ids_from_input_ids(input_ids, self.config.pad_token_id)
if attention_mask is None:
attention_mask = jnp.ones_like(input_ids)
if head_mask is None:
head_mask = jnp.ones((self.config.num_hidden_layers, self.config.num_attention_heads))
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
inputs = {"params": params or self.params}
if self.config.add_cross_attention:
# if past_key_values are passed then cache is already initialized a private flag init_cache has to be passed
# down to ensure cache is used. It has to be made sure that cache is marked as mutable so that it can be
# changed by FlaxRobertaAttention module
if past_key_values:
inputs["cache"] = past_key_values
mutable = ["cache"]
else:
mutable = False
outputs = self.module.apply(
inputs,
jnp.array(input_ids, dtype="i4"),
jnp.array(attention_mask, dtype="i4"),
token_type_ids=jnp.array(token_type_ids, dtype="i4"),
position_ids=jnp.array(position_ids, dtype="i4"),
head_mask=jnp.array(head_mask, dtype="i4"),
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
deterministic=not train,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
rngs=rngs,
mutable=mutable,
)
# add updated cache to model output
if past_key_values is not None and return_dict:
outputs, past_key_values = outputs
outputs["past_key_values"] = unfreeze(past_key_values["cache"])
return outputs
elif past_key_values is not None and not return_dict:
outputs, past_key_values = outputs
outputs = outputs[:1] + (unfreeze(past_key_values["cache"]),) + outputs[1:]
else:
outputs = self.module.apply(
inputs,
jnp.array(input_ids, dtype="i4"),
jnp.array(attention_mask, dtype="i4"),
token_type_ids=jnp.array(token_type_ids, dtype="i4"),
position_ids=jnp.array(position_ids, dtype="i4"),
head_mask=jnp.array(head_mask, dtype="i4"),
deterministic=not train,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
rngs=rngs,
)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertModule with Bert->Roberta
class FlaxRobertaModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
add_pooling_layer: bool = True
gradient_checkpointing: bool = False
def setup(self):
self.embeddings = FlaxRobertaEmbeddings(self.config, dtype=self.dtype)
self.encoder = FlaxRobertaEncoder(
self.config,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.pooler = FlaxRobertaPooler(self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids: Optional[jnp.ndarray] = None,
position_ids: Optional[jnp.ndarray] = None,
head_mask: Optional[jnp.ndarray] = None,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# make sure `token_type_ids` is correctly initialized when not passed
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
# make sure `position_ids` is correctly initialized when not passed
if position_ids is None:
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
hidden_states = self.embeddings(
input_ids, token_type_ids, position_ids, attention_mask, deterministic=deterministic
)
outputs = self.encoder(
hidden_states,
attention_mask,
head_mask=head_mask,
deterministic=deterministic,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
init_cache=init_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
pooled = self.pooler(hidden_states) if self.add_pooling_layer else None
if not return_dict:
# if pooled is None, don't return it
if pooled is None:
return (hidden_states,) + outputs[1:]
return (hidden_states, pooled) + outputs[1:]
return FlaxBaseModelOutputWithPoolingAndCrossAttentions(
last_hidden_state=hidden_states,
pooler_output=pooled,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)
@add_start_docstrings(
"The bare RoBERTa Model transformer outputting raw hidden-states without any specific head on top.",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaModel(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaModule
append_call_sample_docstring(
FlaxRobertaModel, _TOKENIZER_FOR_DOC, _CHECKPOINT_FOR_DOC, FlaxBaseModelOutputWithPooling, _CONFIG_FOR_DOC
)
class FlaxRobertaForMaskedLMModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
add_pooling_layer=False,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.lm_head = FlaxRobertaLMHead(config=self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
if self.config.tie_word_embeddings:
shared_embedding = self.roberta.variables["params"]["embeddings"]["word_embeddings"]["embedding"]
else:
shared_embedding = None
# Compute the prediction scores
logits = self.lm_head(hidden_states, shared_embedding=shared_embedding)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxMaskedLMOutput(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings("""RoBERTa Model with a `language modeling` head on top.""", ROBERTA_START_DOCSTRING)
class FlaxRobertaForMaskedLM(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForMaskedLMModule
append_call_sample_docstring(
FlaxRobertaForMaskedLM,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxBaseModelOutputWithPooling,
_CONFIG_FOR_DOC,
mask="<mask>",
)
class FlaxRobertaForSequenceClassificationModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
add_pooling_layer=False,
gradient_checkpointing=self.gradient_checkpointing,
)
self.classifier = FlaxRobertaClassificationHead(config=self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
logits = self.classifier(sequence_output, deterministic=deterministic)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxSequenceClassifierOutput(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model transformer with a sequence classification/regression head on top (a linear layer on top of the
pooled output) e.g. for GLUE tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForSequenceClassification(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForSequenceClassificationModule
append_call_sample_docstring(
FlaxRobertaForSequenceClassification,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxSequenceClassifierOutput,
_CONFIG_FOR_DOC,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertForMultipleChoiceModule with Bert->Roberta, with self.bert->self.roberta
class FlaxRobertaForMultipleChoiceModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
self.classifier = nn.Dense(1, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
num_choices = input_ids.shape[1]
input_ids = input_ids.reshape(-1, input_ids.shape[-1]) if input_ids is not None else None
attention_mask = attention_mask.reshape(-1, attention_mask.shape[-1]) if attention_mask is not None else None
token_type_ids = token_type_ids.reshape(-1, token_type_ids.shape[-1]) if token_type_ids is not None else None
position_ids = position_ids.reshape(-1, position_ids.shape[-1]) if position_ids is not None else None
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
pooled_output = outputs[1]
pooled_output = self.dropout(pooled_output, deterministic=deterministic)
logits = self.classifier(pooled_output)
reshaped_logits = logits.reshape(-1, num_choices)
if not return_dict:
return (reshaped_logits,) + outputs[2:]
return FlaxMultipleChoiceModelOutput(
logits=reshaped_logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model with a multiple choice classification head on top (a linear layer on top of the pooled output and a
softmax) e.g. for RocStories/SWAG tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForMultipleChoice(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForMultipleChoiceModule
overwrite_call_docstring(
FlaxRobertaForMultipleChoice, ROBERTA_INPUTS_DOCSTRING.format("batch_size, num_choices, sequence_length")
)
append_call_sample_docstring(
FlaxRobertaForMultipleChoice,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxMultipleChoiceModelOutput,
_CONFIG_FOR_DOC,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertForTokenClassificationModule with Bert->Roberta, with self.bert->self.roberta
class FlaxRobertaForTokenClassificationModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
add_pooling_layer=False,
gradient_checkpointing=self.gradient_checkpointing,
)
classifier_dropout = (
self.config.classifier_dropout
if self.config.classifier_dropout is not None
else self.config.hidden_dropout_prob
)
self.dropout = nn.Dropout(rate=classifier_dropout)
self.classifier = nn.Dense(self.config.num_labels, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
logits = self.classifier(hidden_states)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxTokenClassifierOutput(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for
Named-Entity-Recognition (NER) tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForTokenClassification(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForTokenClassificationModule
append_call_sample_docstring(
FlaxRobertaForTokenClassification,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxTokenClassifierOutput,
_CONFIG_FOR_DOC,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertForQuestionAnsweringModule with Bert->Roberta, with self.bert->self.roberta
class FlaxRobertaForQuestionAnsweringModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
add_pooling_layer=False,
gradient_checkpointing=self.gradient_checkpointing,
)
self.qa_outputs = nn.Dense(self.config.num_labels, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
logits = self.qa_outputs(hidden_states)
start_logits, end_logits = logits.split(self.config.num_labels, axis=-1)
start_logits = start_logits.squeeze(-1)
end_logits = end_logits.squeeze(-1)
if not return_dict:
return (start_logits, end_logits) + outputs[1:]
return FlaxQuestionAnsweringModelOutput(
start_logits=start_logits,
end_logits=end_logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear
layers on top of the hidden-states output to compute `span start logits` and `span end logits`).
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForQuestionAnswering(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForQuestionAnsweringModule
append_call_sample_docstring(
FlaxRobertaForQuestionAnswering,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxQuestionAnsweringModelOutput,
_CONFIG_FOR_DOC,
)
class FlaxRobertaForCausalLMModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
add_pooling_layer=False,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.lm_head = FlaxRobertaLMHead(config=self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
position_ids,
token_type_ids: Optional[jnp.ndarray] = None,
head_mask: Optional[jnp.ndarray] = None,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
if self.config.tie_word_embeddings:
shared_embedding = self.roberta.variables["params"]["embeddings"]["word_embeddings"]["embedding"]
else:
shared_embedding = None
# Compute the prediction scores
logits = self.lm_head(hidden_states, shared_embedding=shared_embedding)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxCausalLMOutputWithCrossAttentions(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)
@add_start_docstrings(
"""
Roberta Model with a language modeling head on top (a linear layer on top of the hidden-states output) e.g for
autoregressive tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForCausalLM(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForCausalLMModule
def prepare_inputs_for_generation(self, input_ids, max_length, attention_mask: Optional[jnp.DeviceArray] = None):
# initializing the cache
batch_size, seq_length = input_ids.shape
past_key_values = self.init_cache(batch_size, max_length)
# Note that usually one would have to put 0's in the attention_mask for x > input_ids.shape[-1] and x < cache_length.
# But since the decoder uses a causal mask, those positions are masked anyway.
# Thus, we can create a single static attention_mask here, which is more efficient for compilation
extended_attention_mask = jnp.ones((batch_size, max_length), dtype="i4")
if attention_mask is not None:
position_ids = attention_mask.cumsum(axis=-1) - 1
extended_attention_mask = lax.dynamic_update_slice(extended_attention_mask, attention_mask, (0, 0))
else:
position_ids = jnp.broadcast_to(jnp.arange(seq_length, dtype="i4")[None, :], (batch_size, seq_length))
return {
"past_key_values": past_key_values,
"attention_mask": extended_attention_mask,
"position_ids": position_ids,
}
def update_inputs_for_generation(self, model_outputs, model_kwargs):
model_kwargs["past_key_values"] = model_outputs.past_key_values
model_kwargs["position_ids"] = model_kwargs["position_ids"][:, -1:] + 1
return model_kwargs
append_call_sample_docstring(
FlaxRobertaForCausalLM,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxCausalLMOutputWithCrossAttentions,
_CONFIG_FOR_DOC,
)
| # coding=utf-8
# Copyright 2021 The Google Flax Team Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import Callable, Optional, Tuple
import numpy as np
import flax.linen as nn
import jax
import jax.numpy as jnp
from flax.core.frozen_dict import FrozenDict, freeze, unfreeze
from flax.linen import combine_masks, make_causal_mask
from flax.linen import partitioning as nn_partitioning
from flax.linen.attention import dot_product_attention_weights
from flax.traverse_util import flatten_dict, unflatten_dict
from jax import lax
from ...modeling_flax_outputs import (
FlaxBaseModelOutputWithPastAndCrossAttentions,
FlaxBaseModelOutputWithPooling,
FlaxBaseModelOutputWithPoolingAndCrossAttentions,
FlaxCausalLMOutputWithCrossAttentions,
FlaxMaskedLMOutput,
FlaxMultipleChoiceModelOutput,
FlaxQuestionAnsweringModelOutput,
FlaxSequenceClassifierOutput,
FlaxTokenClassifierOutput,
)
from ...modeling_flax_utils import ACT2FN, FlaxPreTrainedModel, append_call_sample_docstring, overwrite_call_docstring
from ...utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_roberta import RobertaConfig
logger = logging.get_logger(__name__)
_CHECKPOINT_FOR_DOC = "roberta-base"
_CONFIG_FOR_DOC = "RobertaConfig"
_TOKENIZER_FOR_DOC = "RobertaTokenizer"
remat = nn_partitioning.remat
def create_position_ids_from_input_ids(input_ids, padding_idx):
"""
Replace non-padding symbols with their position numbers. Position numbers begin at padding_idx+1. Padding symbols
are ignored. This is modified from fairseq's `utils.make_positions`.
Args:
input_ids: jnp.ndarray
padding_idx: int
Returns: jnp.ndarray
"""
# The series of casts and type-conversions here are carefully balanced to both work with ONNX export and XLA.
mask = (input_ids != padding_idx).astype("i4")
if mask.ndim > 2:
mask = mask.reshape((-1, mask.shape[-1]))
incremental_indices = jnp.cumsum(mask, axis=1).astype("i4") * mask
incremental_indices = incremental_indices.reshape(input_ids.shape)
else:
incremental_indices = jnp.cumsum(mask, axis=1).astype("i4") * mask
return incremental_indices.astype("i4") + padding_idx
ROBERTA_START_DOCSTRING = r"""
This model inherits from [`FlaxPreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading, saving and converting weights from PyTorch models)
This model is also a Flax Linen [flax.linen.Module](https://flax.readthedocs.io/en/latest/flax.linen.html#module)
subclass. Use it as a regular Flax linen Module and refer to the Flax documentation for all matter related to
general usage and behavior.
Finally, this model supports inherent JAX features such as:
- [Just-In-Time (JIT) compilation](https://jax.readthedocs.io/en/latest/jax.html#just-in-time-compilation-jit)
- [Automatic Differentiation](https://jax.readthedocs.io/en/latest/jax.html#automatic-differentiation)
- [Vectorization](https://jax.readthedocs.io/en/latest/jax.html#vectorization-vmap)
- [Parallelization](https://jax.readthedocs.io/en/latest/jax.html#parallelization-pmap)
Parameters:
config ([`RobertaConfig`]): Model configuration class with all the parameters of the
model. Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~FlaxPreTrainedModel.from_pretrained`] method to load the model weights.
"""
ROBERTA_INPUTS_DOCSTRING = r"""
Args:
input_ids (`numpy.ndarray` of shape `({0})`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using [`BertTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`numpy.ndarray` of shape `({0})`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
token_type_ids (`numpy.ndarray` of shape `({0})`, *optional*):
Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0,
1]`:
- 0 corresponds to a *sentence A* token,
- 1 corresponds to a *sentence B* token.
[What are token type IDs?](../glossary#token-type-ids)
position_ids (`numpy.ndarray` of shape `({0})`, *optional*):
Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0,
config.max_position_embeddings - 1]`.
head_mask (`numpy.ndarray` of shape `({0})`, `optional):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertEmbeddings with Bert->Roberta
class FlaxRobertaEmbeddings(nn.Module):
"""Construct the embeddings from word, position and token_type embeddings."""
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.word_embeddings = nn.Embed(
self.config.vocab_size,
self.config.hidden_size,
embedding_init=jax.nn.initializers.normal(stddev=self.config.initializer_range),
dtype=self.dtype,
)
self.position_embeddings = nn.Embed(
self.config.max_position_embeddings,
self.config.hidden_size,
embedding_init=jax.nn.initializers.normal(stddev=self.config.initializer_range),
dtype=self.dtype,
)
self.token_type_embeddings = nn.Embed(
self.config.type_vocab_size,
self.config.hidden_size,
embedding_init=jax.nn.initializers.normal(stddev=self.config.initializer_range),
dtype=self.dtype,
)
self.LayerNorm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
def __call__(self, input_ids, token_type_ids, position_ids, attention_mask, deterministic: bool = True):
# Embed
inputs_embeds = self.word_embeddings(input_ids.astype("i4"))
position_embeds = self.position_embeddings(position_ids.astype("i4"))
token_type_embeddings = self.token_type_embeddings(token_type_ids.astype("i4"))
# Sum all embeddings
hidden_states = inputs_embeds + token_type_embeddings + position_embeds
# Layer Norm
hidden_states = self.LayerNorm(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertSelfAttention with Bert->Roberta
class FlaxRobertaSelfAttention(nn.Module):
config: RobertaConfig
causal: bool = False
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.head_dim = self.config.hidden_size // self.config.num_attention_heads
if self.config.hidden_size % self.config.num_attention_heads != 0:
raise ValueError(
"`config.hidden_size`: {self.config.hidden_size} has to be a multiple of `config.num_attention_heads` "
" : {self.config.num_attention_heads}"
)
self.query = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.key = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.value = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
if self.causal:
self.causal_mask = make_causal_mask(
jnp.ones((1, self.config.max_position_embeddings), dtype="bool"), dtype="bool"
)
def _split_heads(self, hidden_states):
return hidden_states.reshape(hidden_states.shape[:2] + (self.config.num_attention_heads, self.head_dim))
def _merge_heads(self, hidden_states):
return hidden_states.reshape(hidden_states.shape[:2] + (self.config.hidden_size,))
@nn.compact
# Copied from transformers.models.bart.modeling_flax_bart.FlaxBartAttention._concatenate_to_cache
def _concatenate_to_cache(self, key, value, query, attention_mask):
"""
This function takes projected key, value states from a single input token and concatenates the states to cached
states from previous steps. This function is slighly adapted from the official Flax repository:
https://github.com/google/flax/blob/491ce18759622506588784b4fca0e4bf05f8c8cd/flax/linen/attention.py#L252
"""
# detect if we're initializing by absence of existing cache data.
is_initialized = self.has_variable("cache", "cached_key")
cached_key = self.variable("cache", "cached_key", jnp.zeros, key.shape, key.dtype)
cached_value = self.variable("cache", "cached_value", jnp.zeros, value.shape, value.dtype)
cache_index = self.variable("cache", "cache_index", lambda: jnp.array(0, dtype=jnp.int32))
if is_initialized:
*batch_dims, max_length, num_heads, depth_per_head = cached_key.value.shape
# update key, value caches with our new 1d spatial slices
cur_index = cache_index.value
indices = (0,) * len(batch_dims) + (cur_index, 0, 0)
key = lax.dynamic_update_slice(cached_key.value, key, indices)
value = lax.dynamic_update_slice(cached_value.value, value, indices)
cached_key.value = key
cached_value.value = value
num_updated_cache_vectors = query.shape[1]
cache_index.value = cache_index.value + num_updated_cache_vectors
# causal mask for cached decoder self-attention: our single query position should only attend to those key positions that have already been generated and cached, not the remaining zero elements.
pad_mask = jnp.broadcast_to(
jnp.arange(max_length) < cur_index + num_updated_cache_vectors,
tuple(batch_dims) + (1, num_updated_cache_vectors, max_length),
)
attention_mask = combine_masks(pad_mask, attention_mask)
return key, value, attention_mask
def __call__(
self,
hidden_states,
attention_mask,
layer_head_mask,
key_value_states: Optional[jnp.array] = None,
init_cache: bool = False,
deterministic=True,
output_attentions: bool = False,
):
# if key_value_states are provided this layer is used as a cross-attention layer
# for the decoder
is_cross_attention = key_value_states is not None
batch_size = hidden_states.shape[0]
# get query proj
query_states = self.query(hidden_states)
# get key, value proj
if is_cross_attention:
# cross_attentions
key_states = self.key(key_value_states)
value_states = self.value(key_value_states)
else:
# self_attention
key_states = self.key(hidden_states)
value_states = self.value(hidden_states)
query_states = self._split_heads(query_states)
key_states = self._split_heads(key_states)
value_states = self._split_heads(value_states)
# handle cache prepare causal attention mask
if self.causal:
query_length, key_length = query_states.shape[1], key_states.shape[1]
if self.has_variable("cache", "cached_key"):
mask_shift = self.variables["cache"]["cache_index"]
max_decoder_length = self.variables["cache"]["cached_key"].shape[1]
causal_mask = lax.dynamic_slice(
self.causal_mask, (0, 0, mask_shift, 0), (1, 1, query_length, max_decoder_length)
)
else:
causal_mask = self.causal_mask[:, :, :query_length, :key_length]
causal_mask = jnp.broadcast_to(causal_mask, (batch_size,) + causal_mask.shape[1:])
# combine masks if needed
if attention_mask is not None and self.causal:
attention_mask = jnp.broadcast_to(jnp.expand_dims(attention_mask, axis=(-3, -2)), causal_mask.shape)
attention_mask = combine_masks(attention_mask, causal_mask)
elif self.causal:
attention_mask = causal_mask
elif attention_mask is not None:
attention_mask = jnp.expand_dims(attention_mask, axis=(-3, -2))
# During fast autoregressive decoding, we feed one position at a time,
# and cache the keys and values step by step.
if self.causal and (self.has_variable("cache", "cached_key") or init_cache):
key_states, value_states, attention_mask = self._concatenate_to_cache(
key_states, value_states, query_states, attention_mask
)
# Convert the boolean attention mask to an attention bias.
if attention_mask is not None:
# attention mask in the form of attention bias
attention_bias = lax.select(
attention_mask > 0,
jnp.full(attention_mask.shape, 0.0).astype(self.dtype),
jnp.full(attention_mask.shape, -1e10).astype(self.dtype),
)
else:
attention_bias = None
dropout_rng = None
if not deterministic and self.config.attention_probs_dropout_prob > 0.0:
dropout_rng = self.make_rng("dropout")
attn_weights = dot_product_attention_weights(
query_states,
key_states,
bias=attention_bias,
dropout_rng=dropout_rng,
dropout_rate=self.config.attention_probs_dropout_prob,
broadcast_dropout=True,
deterministic=deterministic,
dtype=self.dtype,
precision=None,
)
# Mask heads if we want to
if layer_head_mask is not None:
attn_weights = jnp.einsum("...hqk,h->...hqk", attn_weights, layer_head_mask)
attn_output = jnp.einsum("...hqk,...khd->...qhd", attn_weights, value_states)
attn_output = attn_output.reshape(attn_output.shape[:2] + (-1,))
outputs = (attn_output, attn_weights) if output_attentions else (attn_output,)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertSelfOutput with Bert->Roberta
class FlaxRobertaSelfOutput(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
self.LayerNorm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
def __call__(self, hidden_states, input_tensor, deterministic: bool = True):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.LayerNorm(hidden_states + input_tensor)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertAttention with Bert->Roberta
class FlaxRobertaAttention(nn.Module):
config: RobertaConfig
causal: bool = False
dtype: jnp.dtype = jnp.float32
def setup(self):
self.self = FlaxRobertaSelfAttention(self.config, causal=self.causal, dtype=self.dtype)
self.output = FlaxRobertaSelfOutput(self.config, dtype=self.dtype)
def __call__(
self,
hidden_states,
attention_mask,
layer_head_mask,
key_value_states=None,
init_cache=False,
deterministic=True,
output_attentions: bool = False,
):
# Attention mask comes in as attention_mask.shape == (*batch_sizes, kv_length)
# FLAX expects: attention_mask.shape == (*batch_sizes, 1, 1, kv_length) such that it is broadcastable
# with attn_weights.shape == (*batch_sizes, num_heads, q_length, kv_length)
attn_outputs = self.self(
hidden_states,
attention_mask,
layer_head_mask=layer_head_mask,
key_value_states=key_value_states,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
)
attn_output = attn_outputs[0]
hidden_states = self.output(attn_output, hidden_states, deterministic=deterministic)
outputs = (hidden_states,)
if output_attentions:
outputs += (attn_outputs[1],)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertIntermediate with Bert->Roberta
class FlaxRobertaIntermediate(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.intermediate_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
self.activation = ACT2FN[self.config.hidden_act]
def __call__(self, hidden_states):
hidden_states = self.dense(hidden_states)
hidden_states = self.activation(hidden_states)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertOutput with Bert->Roberta
class FlaxRobertaOutput(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
self.LayerNorm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
def __call__(self, hidden_states, attention_output, deterministic: bool = True):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.LayerNorm(hidden_states + attention_output)
return hidden_states
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertLayer with Bert->Roberta
class FlaxRobertaLayer(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.attention = FlaxRobertaAttention(self.config, causal=self.config.is_decoder, dtype=self.dtype)
self.intermediate = FlaxRobertaIntermediate(self.config, dtype=self.dtype)
self.output = FlaxRobertaOutput(self.config, dtype=self.dtype)
if self.config.add_cross_attention:
self.crossattention = FlaxRobertaAttention(self.config, causal=False, dtype=self.dtype)
def __call__(
self,
hidden_states,
attention_mask,
layer_head_mask,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
):
# Self Attention
attention_outputs = self.attention(
hidden_states,
attention_mask,
layer_head_mask=layer_head_mask,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
)
attention_output = attention_outputs[0]
# Cross-Attention Block
if encoder_hidden_states is not None:
cross_attention_outputs = self.crossattention(
attention_output,
attention_mask=encoder_attention_mask,
layer_head_mask=layer_head_mask,
key_value_states=encoder_hidden_states,
deterministic=deterministic,
output_attentions=output_attentions,
)
attention_output = cross_attention_outputs[0]
hidden_states = self.intermediate(attention_output)
hidden_states = self.output(hidden_states, attention_output, deterministic=deterministic)
outputs = (hidden_states,)
if output_attentions:
outputs += (attention_outputs[1],)
if encoder_hidden_states is not None:
outputs += (cross_attention_outputs[1],)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertLayerCollection with Bert->Roberta
class FlaxRobertaLayerCollection(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
gradient_checkpointing: bool = False
def setup(self):
if self.gradient_checkpointing:
FlaxRobertaCheckpointLayer = remat(FlaxRobertaLayer, static_argnums=(5, 6, 7))
self.layers = [
FlaxRobertaCheckpointLayer(self.config, name=str(i), dtype=self.dtype)
for i in range(self.config.num_hidden_layers)
]
else:
self.layers = [
FlaxRobertaLayer(self.config, name=str(i), dtype=self.dtype)
for i in range(self.config.num_hidden_layers)
]
def __call__(
self,
hidden_states,
attention_mask,
head_mask,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
all_attentions = () if output_attentions else None
all_hidden_states = () if output_hidden_states else None
all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None
# Check if head_mask has a correct number of layers specified if desired
if head_mask is not None:
if head_mask.shape[0] != (len(self.layers)):
raise ValueError(
f"The head_mask should be specified for {len(self.layers)} layers, but it is for "
f" {head_mask.shape[0]}."
)
for i, layer in enumerate(self.layers):
if output_hidden_states:
all_hidden_states += (hidden_states,)
layer_outputs = layer(
hidden_states,
attention_mask,
head_mask[i] if head_mask is not None else None,
encoder_hidden_states,
encoder_attention_mask,
init_cache,
deterministic,
output_attentions,
)
hidden_states = layer_outputs[0]
if output_attentions:
all_attentions += (layer_outputs[1],)
if encoder_hidden_states is not None:
all_cross_attentions += (layer_outputs[2],)
if output_hidden_states:
all_hidden_states += (hidden_states,)
outputs = (hidden_states, all_hidden_states, all_attentions, all_cross_attentions)
if not return_dict:
return tuple(v for v in outputs if v is not None)
return FlaxBaseModelOutputWithPastAndCrossAttentions(
last_hidden_state=hidden_states,
hidden_states=all_hidden_states,
attentions=all_attentions,
cross_attentions=all_cross_attentions,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertEncoder with Bert->Roberta
class FlaxRobertaEncoder(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
gradient_checkpointing: bool = False
def setup(self):
self.layer = FlaxRobertaLayerCollection(
self.config,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
def __call__(
self,
hidden_states,
attention_mask,
head_mask,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
return self.layer(
hidden_states,
attention_mask,
head_mask=head_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertPooler with Bert->Roberta
class FlaxRobertaPooler(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
dtype=self.dtype,
)
def __call__(self, hidden_states):
cls_hidden_state = hidden_states[:, 0]
cls_hidden_state = self.dense(cls_hidden_state)
return nn.tanh(cls_hidden_state)
class FlaxRobertaLMHead(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
bias_init: Callable[..., np.ndarray] = jax.nn.initializers.zeros
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.layer_norm = nn.LayerNorm(epsilon=self.config.layer_norm_eps, dtype=self.dtype)
self.decoder = nn.Dense(
self.config.vocab_size,
dtype=self.dtype,
use_bias=False,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
self.bias = self.param("bias", self.bias_init, (self.config.vocab_size,))
def __call__(self, hidden_states, shared_embedding=None):
hidden_states = self.dense(hidden_states)
hidden_states = ACT2FN["gelu"](hidden_states)
hidden_states = self.layer_norm(hidden_states)
if shared_embedding is not None:
hidden_states = self.decoder.apply({"params": {"kernel": shared_embedding.T}}, hidden_states)
else:
hidden_states = self.decoder(hidden_states)
bias = jnp.asarray(self.bias, self.dtype)
hidden_states += bias
return hidden_states
class FlaxRobertaClassificationHead(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
def setup(self):
self.dense = nn.Dense(
self.config.hidden_size,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
classifier_dropout = (
self.config.classifier_dropout
if self.config.classifier_dropout is not None
else self.config.hidden_dropout_prob
)
self.dropout = nn.Dropout(rate=classifier_dropout)
self.out_proj = nn.Dense(
self.config.num_labels,
dtype=self.dtype,
kernel_init=jax.nn.initializers.normal(self.config.initializer_range),
)
def __call__(self, hidden_states, deterministic=True):
hidden_states = hidden_states[:, 0, :] # take <s> token (equiv. to [CLS])
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.dense(hidden_states)
hidden_states = nn.tanh(hidden_states)
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
hidden_states = self.out_proj(hidden_states)
return hidden_states
class FlaxRobertaPreTrainedModel(FlaxPreTrainedModel):
"""
An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
models.
"""
config_class = RobertaConfig
base_model_prefix = "roberta"
module_class: nn.Module = None
def __init__(
self,
config: RobertaConfig,
input_shape: Tuple = (1, 1),
seed: int = 0,
dtype: jnp.dtype = jnp.float32,
_do_init: bool = True,
gradient_checkpointing: bool = False,
**kwargs
):
module = self.module_class(config=config, dtype=dtype, gradient_checkpointing=gradient_checkpointing, **kwargs)
super().__init__(config, module, input_shape=input_shape, seed=seed, dtype=dtype, _do_init=_do_init)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertPreTrainedModel.enable_gradient_checkpointing
def enable_gradient_checkpointing(self):
self._module = self.module_class(
config=self.config,
dtype=self.dtype,
gradient_checkpointing=True,
)
def init_weights(self, rng: jax.random.PRNGKey, input_shape: Tuple, params: FrozenDict = None) -> FrozenDict:
# init input tensors
input_ids = jnp.zeros(input_shape, dtype="i4")
token_type_ids = jnp.ones_like(input_ids)
position_ids = create_position_ids_from_input_ids(input_ids, self.config.pad_token_id)
attention_mask = jnp.ones_like(input_ids)
head_mask = jnp.ones((self.config.num_hidden_layers, self.config.num_attention_heads))
params_rng, dropout_rng = jax.random.split(rng)
rngs = {"params": params_rng, "dropout": dropout_rng}
if self.config.add_cross_attention:
encoder_hidden_states = jnp.zeros(input_shape + (self.config.hidden_size,))
encoder_attention_mask = attention_mask
module_init_outputs = self.module.init(
rngs,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
encoder_hidden_states,
encoder_attention_mask,
return_dict=False,
)
else:
module_init_outputs = self.module.init(
rngs, input_ids, attention_mask, token_type_ids, position_ids, head_mask, return_dict=False
)
random_params = module_init_outputs["params"]
if params is not None:
random_params = flatten_dict(unfreeze(random_params))
params = flatten_dict(unfreeze(params))
for missing_key in self._missing_keys:
params[missing_key] = random_params[missing_key]
self._missing_keys = set()
return freeze(unflatten_dict(params))
else:
return random_params
# Copied from transformers.models.bart.modeling_flax_bart.FlaxBartDecoderPreTrainedModel.init_cache
def init_cache(self, batch_size, max_length):
r"""
Args:
batch_size (`int`):
batch_size used for fast auto-regressive decoding. Defines the batch size of the initialized cache.
max_length (`int`):
maximum possible length for auto-regressive decoding. Defines the sequence length of the initialized
cache.
"""
# init input variables to retrieve cache
input_ids = jnp.ones((batch_size, max_length), dtype="i4")
attention_mask = jnp.ones_like(input_ids, dtype="i4")
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
init_variables = self.module.init(
jax.random.PRNGKey(0), input_ids, attention_mask, position_ids, return_dict=False, init_cache=True
)
return unfreeze(init_variables["cache"])
@add_start_docstrings_to_model_forward(ROBERTA_INPUTS_DOCSTRING.format("batch_size, sequence_length"))
def __call__(
self,
input_ids,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
encoder_hidden_states=None,
encoder_attention_mask=None,
params: dict = None,
dropout_rng: jax.random.PRNGKey = None,
train: bool = False,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
past_key_values: dict = None,
):
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.return_dict
# init input tensors if not passed
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
if position_ids is None:
position_ids = create_position_ids_from_input_ids(input_ids, self.config.pad_token_id)
if attention_mask is None:
attention_mask = jnp.ones_like(input_ids)
if head_mask is None:
head_mask = jnp.ones((self.config.num_hidden_layers, self.config.num_attention_heads))
# Handle any PRNG if needed
rngs = {}
if dropout_rng is not None:
rngs["dropout"] = dropout_rng
inputs = {"params": params or self.params}
if self.config.add_cross_attention:
# if past_key_values are passed then cache is already initialized a private flag init_cache has to be passed
# down to ensure cache is used. It has to be made sure that cache is marked as mutable so that it can be
# changed by FlaxRobertaAttention module
if past_key_values:
inputs["cache"] = past_key_values
mutable = ["cache"]
else:
mutable = False
outputs = self.module.apply(
inputs,
jnp.array(input_ids, dtype="i4"),
jnp.array(attention_mask, dtype="i4"),
token_type_ids=jnp.array(token_type_ids, dtype="i4"),
position_ids=jnp.array(position_ids, dtype="i4"),
head_mask=jnp.array(head_mask, dtype="i4"),
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
deterministic=not train,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
rngs=rngs,
mutable=mutable,
)
# add updated cache to model output
if past_key_values is not None and return_dict:
outputs, past_key_values = outputs
outputs["past_key_values"] = unfreeze(past_key_values["cache"])
return outputs
elif past_key_values is not None and not return_dict:
outputs, past_key_values = outputs
outputs = outputs[:1] + (unfreeze(past_key_values["cache"]),) + outputs[1:]
else:
outputs = self.module.apply(
inputs,
jnp.array(input_ids, dtype="i4"),
jnp.array(attention_mask, dtype="i4"),
token_type_ids=jnp.array(token_type_ids, dtype="i4"),
position_ids=jnp.array(position_ids, dtype="i4"),
head_mask=jnp.array(head_mask, dtype="i4"),
deterministic=not train,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
rngs=rngs,
)
return outputs
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertModule with Bert->Roberta
class FlaxRobertaModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32 # the dtype of the computation
add_pooling_layer: bool = True
gradient_checkpointing: bool = False
def setup(self):
self.embeddings = FlaxRobertaEmbeddings(self.config, dtype=self.dtype)
self.encoder = FlaxRobertaEncoder(
self.config,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.pooler = FlaxRobertaPooler(self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids: Optional[jnp.ndarray] = None,
position_ids: Optional[jnp.ndarray] = None,
head_mask: Optional[jnp.ndarray] = None,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# make sure `token_type_ids` is correctly initialized when not passed
if token_type_ids is None:
token_type_ids = jnp.zeros_like(input_ids)
# make sure `position_ids` is correctly initialized when not passed
if position_ids is None:
position_ids = jnp.broadcast_to(jnp.arange(jnp.atleast_2d(input_ids).shape[-1]), input_ids.shape)
hidden_states = self.embeddings(
input_ids, token_type_ids, position_ids, attention_mask, deterministic=deterministic
)
outputs = self.encoder(
hidden_states,
attention_mask,
head_mask=head_mask,
deterministic=deterministic,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
init_cache=init_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
pooled = self.pooler(hidden_states) if self.add_pooling_layer else None
if not return_dict:
# if pooled is None, don't return it
if pooled is None:
return (hidden_states,) + outputs[1:]
return (hidden_states, pooled) + outputs[1:]
return FlaxBaseModelOutputWithPoolingAndCrossAttentions(
last_hidden_state=hidden_states,
pooler_output=pooled,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)
@add_start_docstrings(
"The bare RoBERTa Model transformer outputting raw hidden-states without any specific head on top.",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaModel(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaModule
append_call_sample_docstring(
FlaxRobertaModel, _TOKENIZER_FOR_DOC, _CHECKPOINT_FOR_DOC, FlaxBaseModelOutputWithPooling, _CONFIG_FOR_DOC
)
class FlaxRobertaForMaskedLMModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
add_pooling_layer=False,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.lm_head = FlaxRobertaLMHead(config=self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
if self.config.tie_word_embeddings:
shared_embedding = self.roberta.variables["params"]["embeddings"]["word_embeddings"]["embedding"]
else:
shared_embedding = None
# Compute the prediction scores
logits = self.lm_head(hidden_states, shared_embedding=shared_embedding)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxMaskedLMOutput(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings("""RoBERTa Model with a `language modeling` head on top.""", ROBERTA_START_DOCSTRING)
class FlaxRobertaForMaskedLM(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForMaskedLMModule
append_call_sample_docstring(
FlaxRobertaForMaskedLM,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxBaseModelOutputWithPooling,
_CONFIG_FOR_DOC,
mask="<mask>",
)
class FlaxRobertaForSequenceClassificationModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
add_pooling_layer=False,
gradient_checkpointing=self.gradient_checkpointing,
)
self.classifier = FlaxRobertaClassificationHead(config=self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
sequence_output = outputs[0]
logits = self.classifier(sequence_output, deterministic=deterministic)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxSequenceClassifierOutput(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model transformer with a sequence classification/regression head on top (a linear layer on top of the
pooled output) e.g. for GLUE tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForSequenceClassification(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForSequenceClassificationModule
append_call_sample_docstring(
FlaxRobertaForSequenceClassification,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxSequenceClassifierOutput,
_CONFIG_FOR_DOC,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertForMultipleChoiceModule with Bert->Roberta, with self.bert->self.roberta
class FlaxRobertaForMultipleChoiceModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.dropout = nn.Dropout(rate=self.config.hidden_dropout_prob)
self.classifier = nn.Dense(1, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
num_choices = input_ids.shape[1]
input_ids = input_ids.reshape(-1, input_ids.shape[-1]) if input_ids is not None else None
attention_mask = attention_mask.reshape(-1, attention_mask.shape[-1]) if attention_mask is not None else None
token_type_ids = token_type_ids.reshape(-1, token_type_ids.shape[-1]) if token_type_ids is not None else None
position_ids = position_ids.reshape(-1, position_ids.shape[-1]) if position_ids is not None else None
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
pooled_output = outputs[1]
pooled_output = self.dropout(pooled_output, deterministic=deterministic)
logits = self.classifier(pooled_output)
reshaped_logits = logits.reshape(-1, num_choices)
if not return_dict:
return (reshaped_logits,) + outputs[2:]
return FlaxMultipleChoiceModelOutput(
logits=reshaped_logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model with a multiple choice classification head on top (a linear layer on top of the pooled output and a
softmax) e.g. for RocStories/SWAG tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForMultipleChoice(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForMultipleChoiceModule
overwrite_call_docstring(
FlaxRobertaForMultipleChoice, ROBERTA_INPUTS_DOCSTRING.format("batch_size, num_choices, sequence_length")
)
append_call_sample_docstring(
FlaxRobertaForMultipleChoice,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxMultipleChoiceModelOutput,
_CONFIG_FOR_DOC,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertForTokenClassificationModule with Bert->Roberta, with self.bert->self.roberta
class FlaxRobertaForTokenClassificationModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
add_pooling_layer=False,
gradient_checkpointing=self.gradient_checkpointing,
)
classifier_dropout = (
self.config.classifier_dropout
if self.config.classifier_dropout is not None
else self.config.hidden_dropout_prob
)
self.dropout = nn.Dropout(rate=classifier_dropout)
self.classifier = nn.Dense(self.config.num_labels, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
hidden_states = self.dropout(hidden_states, deterministic=deterministic)
logits = self.classifier(hidden_states)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxTokenClassifierOutput(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for
Named-Entity-Recognition (NER) tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForTokenClassification(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForTokenClassificationModule
append_call_sample_docstring(
FlaxRobertaForTokenClassification,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxTokenClassifierOutput,
_CONFIG_FOR_DOC,
)
# Copied from transformers.models.bert.modeling_flax_bert.FlaxBertForQuestionAnsweringModule with Bert->Roberta, with self.bert->self.roberta
class FlaxRobertaForQuestionAnsweringModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
dtype=self.dtype,
add_pooling_layer=False,
gradient_checkpointing=self.gradient_checkpointing,
)
self.qa_outputs = nn.Dense(self.config.num_labels, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
logits = self.qa_outputs(hidden_states)
start_logits, end_logits = logits.split(self.config.num_labels, axis=-1)
start_logits = start_logits.squeeze(-1)
end_logits = end_logits.squeeze(-1)
if not return_dict:
return (start_logits, end_logits) + outputs[1:]
return FlaxQuestionAnsweringModelOutput(
start_logits=start_logits,
end_logits=end_logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
@add_start_docstrings(
"""
Roberta Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear
layers on top of the hidden-states output to compute `span start logits` and `span end logits`).
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForQuestionAnswering(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForQuestionAnsweringModule
append_call_sample_docstring(
FlaxRobertaForQuestionAnswering,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxQuestionAnsweringModelOutput,
_CONFIG_FOR_DOC,
)
class FlaxRobertaForCausalLMModule(nn.Module):
config: RobertaConfig
dtype: jnp.dtype = jnp.float32
gradient_checkpointing: bool = False
def setup(self):
self.roberta = FlaxRobertaModule(
config=self.config,
add_pooling_layer=False,
dtype=self.dtype,
gradient_checkpointing=self.gradient_checkpointing,
)
self.lm_head = FlaxRobertaLMHead(config=self.config, dtype=self.dtype)
def __call__(
self,
input_ids,
attention_mask,
position_ids,
token_type_ids: Optional[jnp.ndarray] = None,
head_mask: Optional[jnp.ndarray] = None,
encoder_hidden_states: Optional[jnp.ndarray] = None,
encoder_attention_mask: Optional[jnp.ndarray] = None,
init_cache: bool = False,
deterministic: bool = True,
output_attentions: bool = False,
output_hidden_states: bool = False,
return_dict: bool = True,
):
# Model
outputs = self.roberta(
input_ids,
attention_mask,
token_type_ids,
position_ids,
head_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
init_cache=init_cache,
deterministic=deterministic,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = outputs[0]
if self.config.tie_word_embeddings:
shared_embedding = self.roberta.variables["params"]["embeddings"]["word_embeddings"]["embedding"]
else:
shared_embedding = None
# Compute the prediction scores
logits = self.lm_head(hidden_states, shared_embedding=shared_embedding)
if not return_dict:
return (logits,) + outputs[1:]
return FlaxCausalLMOutputWithCrossAttentions(
logits=logits,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)
@add_start_docstrings(
"""
Roberta Model with a language modeling head on top (a linear layer on top of the hidden-states output) e.g for
autoregressive tasks.
""",
ROBERTA_START_DOCSTRING,
)
class FlaxRobertaForCausalLM(FlaxRobertaPreTrainedModel):
module_class = FlaxRobertaForCausalLMModule
def prepare_inputs_for_generation(self, input_ids, max_length, attention_mask: Optional[jnp.DeviceArray] = None):
# initializing the cache
batch_size, seq_length = input_ids.shape
past_key_values = self.init_cache(batch_size, max_length)
# Note that usually one would have to put 0's in the attention_mask for x > input_ids.shape[-1] and x < cache_length.
# But since the decoder uses a causal mask, those positions are masked anyway.
# Thus, we can create a single static attention_mask here, which is more efficient for compilation
extended_attention_mask = jnp.ones((batch_size, max_length), dtype="i4")
if attention_mask is not None:
position_ids = attention_mask.cumsum(axis=-1) - 1
extended_attention_mask = lax.dynamic_update_slice(extended_attention_mask, attention_mask, (0, 0))
else:
position_ids = jnp.broadcast_to(jnp.arange(seq_length, dtype="i4")[None, :], (batch_size, seq_length))
return {
"past_key_values": past_key_values,
"attention_mask": extended_attention_mask,
"position_ids": position_ids,
}
def update_inputs_for_generation(self, model_outputs, model_kwargs):
model_kwargs["past_key_values"] = model_outputs.past_key_values
model_kwargs["position_ids"] = model_kwargs["position_ids"][:, -1:] + 1
return model_kwargs
append_call_sample_docstring(
FlaxRobertaForCausalLM,
_TOKENIZER_FOR_DOC,
_CHECKPOINT_FOR_DOC,
FlaxCausalLMOutputWithCrossAttentions,
_CONFIG_FOR_DOC,
)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./tests/utils/test_hf_argparser.py | # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import json
import os
import tempfile
import unittest
from argparse import Namespace
from dataclasses import dataclass, field
from enum import Enum
from pathlib import Path
from typing import List, Optional
import yaml
from transformers import HfArgumentParser, TrainingArguments
from transformers.hf_argparser import make_choice_type_function, string_to_bool
try:
# For Python versions <3.8, Literal is not in typing: https://peps.python.org/pep-0586/
from typing import Literal
except ImportError:
# For Python 3.7
from typing_extensions import Literal
def list_field(default=None, metadata=None):
return field(default_factory=lambda: default, metadata=metadata)
@dataclass
class BasicExample:
foo: int
bar: float
baz: str
flag: bool
@dataclass
class WithDefaultExample:
foo: int = 42
baz: str = field(default="toto", metadata={"help": "help message"})
@dataclass
class WithDefaultBoolExample:
foo: bool = False
baz: bool = True
opt: Optional[bool] = None
class BasicEnum(Enum):
titi = "titi"
toto = "toto"
class MixedTypeEnum(Enum):
titi = "titi"
toto = "toto"
fourtytwo = 42
@dataclass
class EnumExample:
foo: BasicEnum = "toto"
def __post_init__(self):
self.foo = BasicEnum(self.foo)
@dataclass
class MixedTypeEnumExample:
foo: MixedTypeEnum = "toto"
def __post_init__(self):
self.foo = MixedTypeEnum(self.foo)
@dataclass
class OptionalExample:
foo: Optional[int] = None
bar: Optional[float] = field(default=None, metadata={"help": "help message"})
baz: Optional[str] = None
ces: Optional[List[str]] = list_field(default=[])
des: Optional[List[int]] = list_field(default=[])
@dataclass
class ListExample:
foo_int: List[int] = list_field(default=[])
bar_int: List[int] = list_field(default=[1, 2, 3])
foo_str: List[str] = list_field(default=["Hallo", "Bonjour", "Hello"])
foo_float: List[float] = list_field(default=[0.1, 0.2, 0.3])
@dataclass
class RequiredExample:
required_list: List[int] = field()
required_str: str = field()
required_enum: BasicEnum = field()
def __post_init__(self):
self.required_enum = BasicEnum(self.required_enum)
@dataclass
class StringLiteralAnnotationExample:
foo: int
required_enum: "BasicEnum" = field()
opt: "Optional[bool]" = None
baz: "str" = field(default="toto", metadata={"help": "help message"})
foo_str: "List[str]" = list_field(default=["Hallo", "Bonjour", "Hello"])
class HfArgumentParserTest(unittest.TestCase):
def argparsersEqual(self, a: argparse.ArgumentParser, b: argparse.ArgumentParser):
"""
Small helper to check pseudo-equality of parsed arguments on `ArgumentParser` instances.
"""
self.assertEqual(len(a._actions), len(b._actions))
for x, y in zip(a._actions, b._actions):
xx = {k: v for k, v in vars(x).items() if k != "container"}
yy = {k: v for k, v in vars(y).items() if k != "container"}
# Choices with mixed type have custom function as "type"
# So we need to compare results directly for equality
if xx.get("choices", None) and yy.get("choices", None):
for expected_choice in yy["choices"] + xx["choices"]:
self.assertEqual(xx["type"](expected_choice), yy["type"](expected_choice))
del xx["type"], yy["type"]
self.assertEqual(xx, yy)
def test_basic(self):
parser = HfArgumentParser(BasicExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", type=int, required=True)
expected.add_argument("--bar", type=float, required=True)
expected.add_argument("--baz", type=str, required=True)
expected.add_argument("--flag", type=string_to_bool, default=False, const=True, nargs="?")
self.argparsersEqual(parser, expected)
args = ["--foo", "1", "--baz", "quux", "--bar", "0.5"]
(example,) = parser.parse_args_into_dataclasses(args, look_for_args_file=False)
self.assertFalse(example.flag)
def test_with_default(self):
parser = HfArgumentParser(WithDefaultExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", default=42, type=int)
expected.add_argument("--baz", default="toto", type=str, help="help message")
self.argparsersEqual(parser, expected)
def test_with_default_bool(self):
parser = HfArgumentParser(WithDefaultBoolExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", type=string_to_bool, default=False, const=True, nargs="?")
expected.add_argument("--baz", type=string_to_bool, default=True, const=True, nargs="?")
# A boolean no_* argument always has to come after its "default: True" regular counter-part
# and its default must be set to False
expected.add_argument("--no_baz", action="store_false", default=False, dest="baz")
expected.add_argument("--opt", type=string_to_bool, default=None)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args, Namespace(foo=False, baz=True, opt=None))
args = parser.parse_args(["--foo", "--no_baz"])
self.assertEqual(args, Namespace(foo=True, baz=False, opt=None))
args = parser.parse_args(["--foo", "--baz"])
self.assertEqual(args, Namespace(foo=True, baz=True, opt=None))
args = parser.parse_args(["--foo", "True", "--baz", "True", "--opt", "True"])
self.assertEqual(args, Namespace(foo=True, baz=True, opt=True))
args = parser.parse_args(["--foo", "False", "--baz", "False", "--opt", "False"])
self.assertEqual(args, Namespace(foo=False, baz=False, opt=False))
def test_with_enum(self):
parser = HfArgumentParser(MixedTypeEnumExample)
expected = argparse.ArgumentParser()
expected.add_argument(
"--foo",
default="toto",
choices=["titi", "toto", 42],
type=make_choice_type_function(["titi", "toto", 42]),
)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args.foo, "toto")
enum_ex = parser.parse_args_into_dataclasses([])[0]
self.assertEqual(enum_ex.foo, MixedTypeEnum.toto)
args = parser.parse_args(["--foo", "titi"])
self.assertEqual(args.foo, "titi")
enum_ex = parser.parse_args_into_dataclasses(["--foo", "titi"])[0]
self.assertEqual(enum_ex.foo, MixedTypeEnum.titi)
args = parser.parse_args(["--foo", "42"])
self.assertEqual(args.foo, 42)
enum_ex = parser.parse_args_into_dataclasses(["--foo", "42"])[0]
self.assertEqual(enum_ex.foo, MixedTypeEnum.fourtytwo)
def test_with_literal(self):
@dataclass
class LiteralExample:
foo: Literal["titi", "toto", 42] = "toto"
parser = HfArgumentParser(LiteralExample)
expected = argparse.ArgumentParser()
expected.add_argument(
"--foo",
default="toto",
choices=("titi", "toto", 42),
type=make_choice_type_function(["titi", "toto", 42]),
)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args.foo, "toto")
args = parser.parse_args(["--foo", "titi"])
self.assertEqual(args.foo, "titi")
args = parser.parse_args(["--foo", "42"])
self.assertEqual(args.foo, 42)
def test_with_list(self):
parser = HfArgumentParser(ListExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo_int", nargs="+", default=[], type=int)
expected.add_argument("--bar_int", nargs="+", default=[1, 2, 3], type=int)
expected.add_argument("--foo_str", nargs="+", default=["Hallo", "Bonjour", "Hello"], type=str)
expected.add_argument("--foo_float", nargs="+", default=[0.1, 0.2, 0.3], type=float)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(
args,
Namespace(foo_int=[], bar_int=[1, 2, 3], foo_str=["Hallo", "Bonjour", "Hello"], foo_float=[0.1, 0.2, 0.3]),
)
args = parser.parse_args("--foo_int 1 --bar_int 2 3 --foo_str a b c --foo_float 0.1 0.7".split())
self.assertEqual(args, Namespace(foo_int=[1], bar_int=[2, 3], foo_str=["a", "b", "c"], foo_float=[0.1, 0.7]))
def test_with_optional(self):
parser = HfArgumentParser(OptionalExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", default=None, type=int)
expected.add_argument("--bar", default=None, type=float, help="help message")
expected.add_argument("--baz", default=None, type=str)
expected.add_argument("--ces", nargs="+", default=[], type=str)
expected.add_argument("--des", nargs="+", default=[], type=int)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args, Namespace(foo=None, bar=None, baz=None, ces=[], des=[]))
args = parser.parse_args("--foo 12 --bar 3.14 --baz 42 --ces a b c --des 1 2 3".split())
self.assertEqual(args, Namespace(foo=12, bar=3.14, baz="42", ces=["a", "b", "c"], des=[1, 2, 3]))
def test_with_required(self):
parser = HfArgumentParser(RequiredExample)
expected = argparse.ArgumentParser()
expected.add_argument("--required_list", nargs="+", type=int, required=True)
expected.add_argument("--required_str", type=str, required=True)
expected.add_argument(
"--required_enum",
type=make_choice_type_function(["titi", "toto"]),
choices=["titi", "toto"],
required=True,
)
self.argparsersEqual(parser, expected)
def test_with_string_literal_annotation(self):
parser = HfArgumentParser(StringLiteralAnnotationExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", type=int, required=True)
expected.add_argument(
"--required_enum",
type=make_choice_type_function(["titi", "toto"]),
choices=["titi", "toto"],
required=True,
)
expected.add_argument("--opt", type=string_to_bool, default=None)
expected.add_argument("--baz", default="toto", type=str, help="help message")
expected.add_argument("--foo_str", nargs="+", default=["Hallo", "Bonjour", "Hello"], type=str)
self.argparsersEqual(parser, expected)
def test_parse_dict(self):
parser = HfArgumentParser(BasicExample)
args_dict = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
}
parsed_args = parser.parse_dict(args_dict)[0]
args = BasicExample(**args_dict)
self.assertEqual(parsed_args, args)
def test_parse_dict_extra_key(self):
parser = HfArgumentParser(BasicExample)
args_dict = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
"extra": 42,
}
self.assertRaises(ValueError, parser.parse_dict, args_dict, allow_extra_keys=False)
def test_parse_json(self):
parser = HfArgumentParser(BasicExample)
args_dict_for_json = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
}
with tempfile.TemporaryDirectory() as tmp_dir:
temp_local_path = os.path.join(tmp_dir, "temp_json")
os.mkdir(temp_local_path)
with open(temp_local_path + ".json", "w+") as f:
json.dump(args_dict_for_json, f)
parsed_args = parser.parse_yaml_file(Path(temp_local_path + ".json"))[0]
args = BasicExample(**args_dict_for_json)
self.assertEqual(parsed_args, args)
def test_parse_yaml(self):
parser = HfArgumentParser(BasicExample)
args_dict_for_yaml = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
}
with tempfile.TemporaryDirectory() as tmp_dir:
temp_local_path = os.path.join(tmp_dir, "temp_yaml")
os.mkdir(temp_local_path)
with open(temp_local_path + ".yaml", "w+") as f:
yaml.dump(args_dict_for_yaml, f)
parsed_args = parser.parse_yaml_file(Path(temp_local_path + ".yaml"))[0]
args = BasicExample(**args_dict_for_yaml)
self.assertEqual(parsed_args, args)
def test_integration_training_args(self):
parser = HfArgumentParser(TrainingArguments)
self.assertIsNotNone(parser)
| # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import json
import os
import tempfile
import unittest
from argparse import Namespace
from dataclasses import dataclass, field
from enum import Enum
from pathlib import Path
from typing import List, Optional
import yaml
from transformers import HfArgumentParser, TrainingArguments
from transformers.hf_argparser import make_choice_type_function, string_to_bool
try:
# For Python versions <3.8, Literal is not in typing: https://peps.python.org/pep-0586/
from typing import Literal
except ImportError:
# For Python 3.7
from typing_extensions import Literal
def list_field(default=None, metadata=None):
return field(default_factory=lambda: default, metadata=metadata)
@dataclass
class BasicExample:
foo: int
bar: float
baz: str
flag: bool
@dataclass
class WithDefaultExample:
foo: int = 42
baz: str = field(default="toto", metadata={"help": "help message"})
@dataclass
class WithDefaultBoolExample:
foo: bool = False
baz: bool = True
opt: Optional[bool] = None
class BasicEnum(Enum):
titi = "titi"
toto = "toto"
class MixedTypeEnum(Enum):
titi = "titi"
toto = "toto"
fourtytwo = 42
@dataclass
class EnumExample:
foo: BasicEnum = "toto"
def __post_init__(self):
self.foo = BasicEnum(self.foo)
@dataclass
class MixedTypeEnumExample:
foo: MixedTypeEnum = "toto"
def __post_init__(self):
self.foo = MixedTypeEnum(self.foo)
@dataclass
class OptionalExample:
foo: Optional[int] = None
bar: Optional[float] = field(default=None, metadata={"help": "help message"})
baz: Optional[str] = None
ces: Optional[List[str]] = list_field(default=[])
des: Optional[List[int]] = list_field(default=[])
@dataclass
class ListExample:
foo_int: List[int] = list_field(default=[])
bar_int: List[int] = list_field(default=[1, 2, 3])
foo_str: List[str] = list_field(default=["Hallo", "Bonjour", "Hello"])
foo_float: List[float] = list_field(default=[0.1, 0.2, 0.3])
@dataclass
class RequiredExample:
required_list: List[int] = field()
required_str: str = field()
required_enum: BasicEnum = field()
def __post_init__(self):
self.required_enum = BasicEnum(self.required_enum)
@dataclass
class StringLiteralAnnotationExample:
foo: int
required_enum: "BasicEnum" = field()
opt: "Optional[bool]" = None
baz: "str" = field(default="toto", metadata={"help": "help message"})
foo_str: "List[str]" = list_field(default=["Hallo", "Bonjour", "Hello"])
class HfArgumentParserTest(unittest.TestCase):
def argparsersEqual(self, a: argparse.ArgumentParser, b: argparse.ArgumentParser):
"""
Small helper to check pseudo-equality of parsed arguments on `ArgumentParser` instances.
"""
self.assertEqual(len(a._actions), len(b._actions))
for x, y in zip(a._actions, b._actions):
xx = {k: v for k, v in vars(x).items() if k != "container"}
yy = {k: v for k, v in vars(y).items() if k != "container"}
# Choices with mixed type have custom function as "type"
# So we need to compare results directly for equality
if xx.get("choices", None) and yy.get("choices", None):
for expected_choice in yy["choices"] + xx["choices"]:
self.assertEqual(xx["type"](expected_choice), yy["type"](expected_choice))
del xx["type"], yy["type"]
self.assertEqual(xx, yy)
def test_basic(self):
parser = HfArgumentParser(BasicExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", type=int, required=True)
expected.add_argument("--bar", type=float, required=True)
expected.add_argument("--baz", type=str, required=True)
expected.add_argument("--flag", type=string_to_bool, default=False, const=True, nargs="?")
self.argparsersEqual(parser, expected)
args = ["--foo", "1", "--baz", "quux", "--bar", "0.5"]
(example,) = parser.parse_args_into_dataclasses(args, look_for_args_file=False)
self.assertFalse(example.flag)
def test_with_default(self):
parser = HfArgumentParser(WithDefaultExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", default=42, type=int)
expected.add_argument("--baz", default="toto", type=str, help="help message")
self.argparsersEqual(parser, expected)
def test_with_default_bool(self):
parser = HfArgumentParser(WithDefaultBoolExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", type=string_to_bool, default=False, const=True, nargs="?")
expected.add_argument("--baz", type=string_to_bool, default=True, const=True, nargs="?")
# A boolean no_* argument always has to come after its "default: True" regular counter-part
# and its default must be set to False
expected.add_argument("--no_baz", action="store_false", default=False, dest="baz")
expected.add_argument("--opt", type=string_to_bool, default=None)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args, Namespace(foo=False, baz=True, opt=None))
args = parser.parse_args(["--foo", "--no_baz"])
self.assertEqual(args, Namespace(foo=True, baz=False, opt=None))
args = parser.parse_args(["--foo", "--baz"])
self.assertEqual(args, Namespace(foo=True, baz=True, opt=None))
args = parser.parse_args(["--foo", "True", "--baz", "True", "--opt", "True"])
self.assertEqual(args, Namespace(foo=True, baz=True, opt=True))
args = parser.parse_args(["--foo", "False", "--baz", "False", "--opt", "False"])
self.assertEqual(args, Namespace(foo=False, baz=False, opt=False))
def test_with_enum(self):
parser = HfArgumentParser(MixedTypeEnumExample)
expected = argparse.ArgumentParser()
expected.add_argument(
"--foo",
default="toto",
choices=["titi", "toto", 42],
type=make_choice_type_function(["titi", "toto", 42]),
)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args.foo, "toto")
enum_ex = parser.parse_args_into_dataclasses([])[0]
self.assertEqual(enum_ex.foo, MixedTypeEnum.toto)
args = parser.parse_args(["--foo", "titi"])
self.assertEqual(args.foo, "titi")
enum_ex = parser.parse_args_into_dataclasses(["--foo", "titi"])[0]
self.assertEqual(enum_ex.foo, MixedTypeEnum.titi)
args = parser.parse_args(["--foo", "42"])
self.assertEqual(args.foo, 42)
enum_ex = parser.parse_args_into_dataclasses(["--foo", "42"])[0]
self.assertEqual(enum_ex.foo, MixedTypeEnum.fourtytwo)
def test_with_literal(self):
@dataclass
class LiteralExample:
foo: Literal["titi", "toto", 42] = "toto"
parser = HfArgumentParser(LiteralExample)
expected = argparse.ArgumentParser()
expected.add_argument(
"--foo",
default="toto",
choices=("titi", "toto", 42),
type=make_choice_type_function(["titi", "toto", 42]),
)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args.foo, "toto")
args = parser.parse_args(["--foo", "titi"])
self.assertEqual(args.foo, "titi")
args = parser.parse_args(["--foo", "42"])
self.assertEqual(args.foo, 42)
def test_with_list(self):
parser = HfArgumentParser(ListExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo_int", nargs="+", default=[], type=int)
expected.add_argument("--bar_int", nargs="+", default=[1, 2, 3], type=int)
expected.add_argument("--foo_str", nargs="+", default=["Hallo", "Bonjour", "Hello"], type=str)
expected.add_argument("--foo_float", nargs="+", default=[0.1, 0.2, 0.3], type=float)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(
args,
Namespace(foo_int=[], bar_int=[1, 2, 3], foo_str=["Hallo", "Bonjour", "Hello"], foo_float=[0.1, 0.2, 0.3]),
)
args = parser.parse_args("--foo_int 1 --bar_int 2 3 --foo_str a b c --foo_float 0.1 0.7".split())
self.assertEqual(args, Namespace(foo_int=[1], bar_int=[2, 3], foo_str=["a", "b", "c"], foo_float=[0.1, 0.7]))
def test_with_optional(self):
parser = HfArgumentParser(OptionalExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", default=None, type=int)
expected.add_argument("--bar", default=None, type=float, help="help message")
expected.add_argument("--baz", default=None, type=str)
expected.add_argument("--ces", nargs="+", default=[], type=str)
expected.add_argument("--des", nargs="+", default=[], type=int)
self.argparsersEqual(parser, expected)
args = parser.parse_args([])
self.assertEqual(args, Namespace(foo=None, bar=None, baz=None, ces=[], des=[]))
args = parser.parse_args("--foo 12 --bar 3.14 --baz 42 --ces a b c --des 1 2 3".split())
self.assertEqual(args, Namespace(foo=12, bar=3.14, baz="42", ces=["a", "b", "c"], des=[1, 2, 3]))
def test_with_required(self):
parser = HfArgumentParser(RequiredExample)
expected = argparse.ArgumentParser()
expected.add_argument("--required_list", nargs="+", type=int, required=True)
expected.add_argument("--required_str", type=str, required=True)
expected.add_argument(
"--required_enum",
type=make_choice_type_function(["titi", "toto"]),
choices=["titi", "toto"],
required=True,
)
self.argparsersEqual(parser, expected)
def test_with_string_literal_annotation(self):
parser = HfArgumentParser(StringLiteralAnnotationExample)
expected = argparse.ArgumentParser()
expected.add_argument("--foo", type=int, required=True)
expected.add_argument(
"--required_enum",
type=make_choice_type_function(["titi", "toto"]),
choices=["titi", "toto"],
required=True,
)
expected.add_argument("--opt", type=string_to_bool, default=None)
expected.add_argument("--baz", default="toto", type=str, help="help message")
expected.add_argument("--foo_str", nargs="+", default=["Hallo", "Bonjour", "Hello"], type=str)
self.argparsersEqual(parser, expected)
def test_parse_dict(self):
parser = HfArgumentParser(BasicExample)
args_dict = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
}
parsed_args = parser.parse_dict(args_dict)[0]
args = BasicExample(**args_dict)
self.assertEqual(parsed_args, args)
def test_parse_dict_extra_key(self):
parser = HfArgumentParser(BasicExample)
args_dict = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
"extra": 42,
}
self.assertRaises(ValueError, parser.parse_dict, args_dict, allow_extra_keys=False)
def test_parse_json(self):
parser = HfArgumentParser(BasicExample)
args_dict_for_json = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
}
with tempfile.TemporaryDirectory() as tmp_dir:
temp_local_path = os.path.join(tmp_dir, "temp_json")
os.mkdir(temp_local_path)
with open(temp_local_path + ".json", "w+") as f:
json.dump(args_dict_for_json, f)
parsed_args = parser.parse_yaml_file(Path(temp_local_path + ".json"))[0]
args = BasicExample(**args_dict_for_json)
self.assertEqual(parsed_args, args)
def test_parse_yaml(self):
parser = HfArgumentParser(BasicExample)
args_dict_for_yaml = {
"foo": 12,
"bar": 3.14,
"baz": "42",
"flag": True,
}
with tempfile.TemporaryDirectory() as tmp_dir:
temp_local_path = os.path.join(tmp_dir, "temp_yaml")
os.mkdir(temp_local_path)
with open(temp_local_path + ".yaml", "w+") as f:
yaml.dump(args_dict_for_yaml, f)
parsed_args = parser.parse_yaml_file(Path(temp_local_path + ".yaml"))[0]
args = BasicExample(**args_dict_for_yaml)
self.assertEqual(parsed_args, args)
def test_integration_training_args(self):
parser = HfArgumentParser(TrainingArguments)
self.assertIsNotNone(parser)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/blenderbot/modeling_blenderbot.py | # coding=utf-8
# Copyright 2021 The Facebook, Inc. and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PyTorch Blenderbot model."""
import copy
import math
import os
import random
import warnings
from typing import List, Optional, Tuple, Union
import torch
import torch.utils.checkpoint
from torch import nn
from torch.nn import CrossEntropyLoss
from ...activations import ACT2FN
from ...modeling_outputs import (
BaseModelOutput,
BaseModelOutputWithPastAndCrossAttentions,
CausalLMOutputWithCrossAttentions,
Seq2SeqLMOutput,
Seq2SeqModelOutput,
)
from ...modeling_utils import PreTrainedModel
from ...utils import (
add_end_docstrings,
add_start_docstrings,
add_start_docstrings_to_model_forward,
logging,
replace_return_docstrings,
)
from ..blenderbot_small import BlenderbotSmallForConditionalGeneration, BlenderbotSmallModel
from .configuration_blenderbot import BlenderbotConfig
logger = logging.get_logger(__name__)
_CONFIG_FOR_DOC = "BlenderbotConfig"
_TOKENIZER_FOR_DOC = "BlenderbotTokenizer"
_CHECKPOINT_FOR_DOC = "facebook/blenderbot-400M-distill"
BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST = [
"facebook/blenderbot-3B",
# See all Blenderbot models at https://huggingface.co/models?filter=blenderbot
]
# Copied from transformers.models.bart.modeling_bart.shift_tokens_right
def shift_tokens_right(input_ids: torch.Tensor, pad_token_id: int, decoder_start_token_id: int):
"""
Shift input ids one token to the right.
"""
shifted_input_ids = input_ids.new_zeros(input_ids.shape)
shifted_input_ids[:, 1:] = input_ids[:, :-1].clone()
shifted_input_ids[:, 0] = decoder_start_token_id
if pad_token_id is None:
raise ValueError("self.model.config.pad_token_id has to be defined.")
# replace possible -100 values in labels by `pad_token_id`
shifted_input_ids.masked_fill_(shifted_input_ids == -100, pad_token_id)
return shifted_input_ids
# Copied from transformers.models.bart.modeling_bart._make_causal_mask
def _make_causal_mask(input_ids_shape: torch.Size, dtype: torch.dtype, past_key_values_length: int = 0):
"""
Make causal mask used for bi-directional self-attention.
"""
bsz, tgt_len = input_ids_shape
mask = torch.full((tgt_len, tgt_len), torch.tensor(torch.finfo(dtype).min))
mask_cond = torch.arange(mask.size(-1))
mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
mask = mask.to(dtype)
if past_key_values_length > 0:
mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype), mask], dim=-1)
return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)
# Copied from transformers.models.bart.modeling_bart._expand_mask
def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
"""
Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
"""
bsz, src_len = mask.size()
tgt_len = tgt_len if tgt_len is not None else src_len
expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)
inverted_mask = 1.0 - expanded_mask
return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
class BlenderbotLearnedPositionalEmbedding(nn.Embedding):
"""
This module learns positional embeddings up to a fixed maximum size.
"""
def __init__(self, num_embeddings: int, embedding_dim: int):
super().__init__(num_embeddings, embedding_dim)
def forward(self, input_ids_shape: torch.Size, past_key_values_length: int = 0):
"""`input_ids_shape` is expected to be [bsz x seqlen]."""
bsz, seq_len = input_ids_shape[:2]
positions = torch.arange(
past_key_values_length, past_key_values_length + seq_len, dtype=torch.long, device=self.weight.device
)
return super().forward(positions)
# Copied from transformers.models.bart.modeling_bart.BartAttention with Bart->Blenderbot
class BlenderbotAttention(nn.Module):
"""Multi-headed attention from 'Attention Is All You Need' paper"""
def __init__(
self,
embed_dim: int,
num_heads: int,
dropout: float = 0.0,
is_decoder: bool = False,
bias: bool = True,
):
super().__init__()
self.embed_dim = embed_dim
self.num_heads = num_heads
self.dropout = dropout
self.head_dim = embed_dim // num_heads
if (self.head_dim * num_heads) != self.embed_dim:
raise ValueError(
f"embed_dim must be divisible by num_heads (got `embed_dim`: {self.embed_dim}"
f" and `num_heads`: {num_heads})."
)
self.scaling = self.head_dim**-0.5
self.is_decoder = is_decoder
self.k_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
self.v_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
def forward(
self,
hidden_states: torch.Tensor,
key_value_states: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
output_attentions: bool = False,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
"""Input shape: Batch x Time x Channel"""
# if key_value_states are provided this layer is used as a cross-attention layer
# for the decoder
is_cross_attention = key_value_states is not None
bsz, tgt_len, _ = hidden_states.size()
# get query proj
query_states = self.q_proj(hidden_states) * self.scaling
# get key, value proj
# `past_key_value[0].shape[2] == key_value_states.shape[1]`
# is checking that the `sequence_length` of the `past_key_value` is the same as
# the provided `key_value_states` to support prefix tuning
if (
is_cross_attention
and past_key_value is not None
and past_key_value[0].shape[2] == key_value_states.shape[1]
):
# reuse k,v, cross_attentions
key_states = past_key_value[0]
value_states = past_key_value[1]
elif is_cross_attention:
# cross_attentions
key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
elif past_key_value is not None:
# reuse k, v, self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
key_states = torch.cat([past_key_value[0], key_states], dim=2)
value_states = torch.cat([past_key_value[1], value_states], dim=2)
else:
# self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
if self.is_decoder:
# if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states.
# Further calls to cross_attention layer can then reuse all cross-attention
# key/value_states (first "if" case)
# if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of
# all previous decoder key/value_states. Further calls to uni-directional self-attention
# can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
# if encoder bi-directional self-attention `past_key_value` is always `None`
past_key_value = (key_states, value_states)
proj_shape = (bsz * self.num_heads, -1, self.head_dim)
query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
key_states = key_states.view(*proj_shape)
value_states = value_states.view(*proj_shape)
src_len = key_states.size(1)
attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))
if attn_weights.size() != (bsz * self.num_heads, tgt_len, src_len):
raise ValueError(
f"Attention weights should be of size {(bsz * self.num_heads, tgt_len, src_len)}, but is"
f" {attn_weights.size()}"
)
if attention_mask is not None:
if attention_mask.size() != (bsz, 1, tgt_len, src_len):
raise ValueError(
f"Attention mask should be of size {(bsz, 1, tgt_len, src_len)}, but is {attention_mask.size()}"
)
attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + attention_mask
attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
if layer_head_mask is not None:
if layer_head_mask.size() != (self.num_heads,):
raise ValueError(
f"Head mask for a single layer should be of size {(self.num_heads,)}, but is"
f" {layer_head_mask.size()}"
)
attn_weights = layer_head_mask.view(1, -1, 1, 1) * attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)
if output_attentions:
# this operation is a bit awkward, but it's required to
# make sure that attn_weights keeps its gradient.
# In order to do so, attn_weights have to be reshaped
# twice and have to be reused in the following
attn_weights_reshaped = attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
attn_weights = attn_weights_reshaped.view(bsz * self.num_heads, tgt_len, src_len)
else:
attn_weights_reshaped = None
attn_probs = nn.functional.dropout(attn_weights, p=self.dropout, training=self.training)
attn_output = torch.bmm(attn_probs, value_states)
if attn_output.size() != (bsz * self.num_heads, tgt_len, self.head_dim):
raise ValueError(
f"`attn_output` should be of size {(bsz, self.num_heads, tgt_len, self.head_dim)}, but is"
f" {attn_output.size()}"
)
attn_output = attn_output.view(bsz, self.num_heads, tgt_len, self.head_dim)
attn_output = attn_output.transpose(1, 2)
# Use the `embed_dim` from the config (stored in the class) rather than `hidden_state` because `attn_output` can be
# partitioned aross GPUs when using tensor-parallelism.
attn_output = attn_output.reshape(bsz, tgt_len, self.embed_dim)
attn_output = self.out_proj(attn_output)
return attn_output, attn_weights_reshaped, past_key_value
# Copied from transformers.models.mbart.modeling_mbart.MBartEncoderLayer with MBart->Blenderbot
class BlenderbotEncoderLayer(nn.Module):
def __init__(self, config: BlenderbotConfig):
super().__init__()
self.embed_dim = config.d_model
self.self_attn = BlenderbotAttention(
embed_dim=self.embed_dim,
num_heads=config.encoder_attention_heads,
dropout=config.attention_dropout,
)
self.self_attn_layer_norm = nn.LayerNorm(self.embed_dim)
self.dropout = config.dropout
self.activation_fn = ACT2FN[config.activation_function]
self.activation_dropout = config.activation_dropout
self.fc1 = nn.Linear(self.embed_dim, config.encoder_ffn_dim)
self.fc2 = nn.Linear(config.encoder_ffn_dim, self.embed_dim)
self.final_layer_norm = nn.LayerNorm(self.embed_dim)
def forward(
self,
hidden_states: torch.Tensor,
attention_mask: torch.Tensor,
layer_head_mask: torch.Tensor,
output_attentions: bool = False,
) -> torch.Tensor:
"""
Args:
hidden_states (`torch.FloatTensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`torch.FloatTensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
*(encoder_attention_heads,)*.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
"""
residual = hidden_states
hidden_states = self.self_attn_layer_norm(hidden_states)
hidden_states, attn_weights, _ = self.self_attn(
hidden_states=hidden_states,
attention_mask=attention_mask,
layer_head_mask=layer_head_mask,
output_attentions=output_attentions,
)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
residual = hidden_states
hidden_states = self.final_layer_norm(hidden_states)
hidden_states = self.activation_fn(self.fc1(hidden_states))
hidden_states = nn.functional.dropout(hidden_states, p=self.activation_dropout, training=self.training)
hidden_states = self.fc2(hidden_states)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
if hidden_states.dtype == torch.float16 and (
torch.isinf(hidden_states).any() or torch.isnan(hidden_states).any()
):
clamp_value = torch.finfo(hidden_states.dtype).max - 1000
hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value)
outputs = (hidden_states,)
if output_attentions:
outputs += (attn_weights,)
return outputs
# Copied from transformers.models.mbart.modeling_mbart.MBartDecoderLayer with MBart->Blenderbot
class BlenderbotDecoderLayer(nn.Module):
def __init__(self, config: BlenderbotConfig):
super().__init__()
self.embed_dim = config.d_model
self.self_attn = BlenderbotAttention(
embed_dim=self.embed_dim,
num_heads=config.decoder_attention_heads,
dropout=config.attention_dropout,
is_decoder=True,
)
self.dropout = config.dropout
self.activation_fn = ACT2FN[config.activation_function]
self.activation_dropout = config.activation_dropout
self.self_attn_layer_norm = nn.LayerNorm(self.embed_dim)
self.encoder_attn = BlenderbotAttention(
self.embed_dim,
config.decoder_attention_heads,
dropout=config.attention_dropout,
is_decoder=True,
)
self.encoder_attn_layer_norm = nn.LayerNorm(self.embed_dim)
self.fc1 = nn.Linear(self.embed_dim, config.decoder_ffn_dim)
self.fc2 = nn.Linear(config.decoder_ffn_dim, self.embed_dim)
self.final_layer_norm = nn.LayerNorm(self.embed_dim)
def forward(
self,
hidden_states: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
encoder_hidden_states: Optional[torch.Tensor] = None,
encoder_attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
cross_attn_layer_head_mask: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
output_attentions: Optional[bool] = False,
use_cache: Optional[bool] = True,
) -> torch.Tensor:
"""
Args:
hidden_states (`torch.FloatTensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`torch.FloatTensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
encoder_hidden_states (`torch.FloatTensor`):
cross attention input to the layer of shape *(seq_len, batch, embed_dim)*
encoder_attention_mask (`torch.FloatTensor`): encoder attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
*(encoder_attention_heads,)*.
cross_attn_layer_head_mask (`torch.FloatTensor`): mask for cross-attention heads in a given layer of
size *(decoder_attention_heads,)*.
past_key_value (`Tuple(torch.FloatTensor)`): cached past key and value projection states
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
"""
residual = hidden_states
hidden_states = self.self_attn_layer_norm(hidden_states)
# Self Attention
# decoder uni-directional self-attention cached key/values tuple is at positions 1,2
self_attn_past_key_value = past_key_value[:2] if past_key_value is not None else None
# add present self-attn cache to positions 1,2 of present_key_value tuple
hidden_states, self_attn_weights, present_key_value = self.self_attn(
hidden_states=hidden_states,
past_key_value=self_attn_past_key_value,
attention_mask=attention_mask,
layer_head_mask=layer_head_mask,
output_attentions=output_attentions,
)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
# Cross-Attention Block
cross_attn_present_key_value = None
cross_attn_weights = None
if encoder_hidden_states is not None:
residual = hidden_states
hidden_states = self.encoder_attn_layer_norm(hidden_states)
# cross_attn cached key/values tuple is at positions 3,4 of present_key_value tuple
cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None
hidden_states, cross_attn_weights, cross_attn_present_key_value = self.encoder_attn(
hidden_states=hidden_states,
key_value_states=encoder_hidden_states,
attention_mask=encoder_attention_mask,
layer_head_mask=cross_attn_layer_head_mask,
past_key_value=cross_attn_past_key_value,
output_attentions=output_attentions,
)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
# add cross-attn to positions 3,4 of present_key_value tuple
present_key_value = present_key_value + cross_attn_present_key_value
# Fully Connected
residual = hidden_states
hidden_states = self.final_layer_norm(hidden_states)
hidden_states = self.activation_fn(self.fc1(hidden_states))
hidden_states = nn.functional.dropout(hidden_states, p=self.activation_dropout, training=self.training)
hidden_states = self.fc2(hidden_states)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
outputs = (hidden_states,)
if output_attentions:
outputs += (self_attn_weights, cross_attn_weights)
if use_cache:
outputs += (present_key_value,)
return outputs
class BlenderbotPreTrainedModel(PreTrainedModel):
config_class = BlenderbotConfig
base_model_prefix = "model"
supports_gradient_checkpointing = True
def _init_weights(self, module):
std = self.config.init_std
if isinstance(module, nn.Linear):
module.weight.data.normal_(mean=0.0, std=std)
if module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.Embedding):
module.weight.data.normal_(mean=0.0, std=std)
if module.padding_idx is not None:
module.weight.data[module.padding_idx].zero_()
def _set_gradient_checkpointing(self, module, value=False):
if isinstance(module, (BlenderbotDecoder, BlenderbotEncoder)):
module.gradient_checkpointing = value
@property
def dummy_inputs(self):
pad_token = self.config.pad_token_id
input_ids = torch.tensor([[0, 6, 10, 4, 2], [0, 8, 12, 2, pad_token]], device=self.device)
dummy_inputs = {
"attention_mask": input_ids.ne(pad_token),
"input_ids": input_ids,
"decoder_input_ids": input_ids,
}
return dummy_inputs
BLENDERBOT_START_DOCSTRING = r"""
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
and behavior.
Parameters:
config ([`BlenderbotConfig`]):
Model configuration class with all the parameters of the model. Initializing with a config file does not
load the weights associated with the model, only the configuration. Check out the
[`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""
BLENDERBOT_GENERATION_EXAMPLE = r"""
Conversation example:
```python
>>> from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration
>>> mname = "facebook/blenderbot-400M-distill"
>>> model = BlenderbotForConditionalGeneration.from_pretrained(mname)
>>> tokenizer = BlenderbotTokenizer.from_pretrained(mname)
>>> UTTERANCE = "My friends are cool but they eat too many carbs."
>>> print("Human: ", UTTERANCE)
Human: My friends are cool but they eat too many carbs.
>>> inputs = tokenizer([UTTERANCE], return_tensors="pt")
>>> reply_ids = model.generate(**inputs)
>>> print("Bot: ", tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0])
Bot: That's unfortunate. Are they trying to lose weight or are they just trying to be healthier?
>>> REPLY = "I'm not sure"
>>> print("Human: ", REPLY)
Human: I'm not sure
>>> NEXT_UTTERANCE = (
... "My friends are cool but they eat too many carbs.</s> <s>That's unfortunate. "
... "Are they trying to lose weight or are they just trying to be healthier?</s> "
... "<s> I'm not sure."
... )
>>> inputs = tokenizer([NEXT_UTTERANCE], return_tensors="pt")
>>> next_reply_ids = model.generate(**inputs)
>>> print("Bot: ", tokenizer.batch_decode(next_reply_ids, skip_special_tokens=True)[0])
Bot: That's too bad. Have you tried encouraging them to change their eating habits?
```
"""
BLENDERBOT_INPUTS_DOCSTRING = r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Indices of decoder input sequence tokens in the vocabulary.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are decoder input IDs?](../glossary#decoder-input-ids)
Blenderbot uses the `bos_token_id` as the starting token for `decoder_input_ids` generation. If
`past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
`past_key_values`).
decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
be used by default.
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
decoder_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules in the decoder. Mask values selected in `[0,
1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
encoder_outputs (`tuple(tuple(torch.FloatTensor)`, *optional*):
Tuple consists of (`last_hidden_state`, *optional*: `hidden_states`, *optional*: `attentions`)
`last_hidden_state` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) is a sequence of
hidden-states at the output of the last layer of the encoder. Used in the cross-attention of the decoder.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape
`(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of shape
`(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention
blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that
don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all
`decoder_input_ids` of shape `(batch_size, sequence_length)`. inputs_embeds (`torch.FloatTensor` of shape
`(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you
can choose to directly pass an embedded representation. This is useful if you want more control over how to
convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
decoder_inputs_embeds (`torch.FloatTensor` of shape `(batch_size, target_sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `decoder_input_ids` you can choose to directly pass an embedded
representation. If `past_key_values` is used, optionally only the last `decoder_inputs_embeds` have to be
input (see `past_key_values`). This is useful if you want more control over how to convert
`decoder_input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
If `decoder_input_ids` and `decoder_inputs_embeds` are both unset, `decoder_inputs_embeds` takes the value
of `inputs_embeds`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
`past_key_values`).
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
class BlenderbotEncoder(BlenderbotPreTrainedModel):
"""
Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer is a
[`BlenderbotEncoderLayer`].
Args:
config: BlenderbotConfig
embed_tokens (nn.Embedding): output embedding
"""
def __init__(self, config: BlenderbotConfig, embed_tokens: Optional[nn.Embedding] = None):
super().__init__(config)
self.dropout = config.dropout
self.layerdrop = config.encoder_layerdrop
embed_dim = config.d_model
self.padding_idx = config.pad_token_id
self.max_source_positions = config.max_position_embeddings
self.embed_scale = math.sqrt(embed_dim) if config.scale_embedding else 1.0
if embed_tokens is not None:
self.embed_tokens = embed_tokens
else:
self.embed_tokens = nn.Embedding(config.vocab_size, embed_dim, self.padding_idx)
self.embed_positions = BlenderbotLearnedPositionalEmbedding(
config.max_position_embeddings,
embed_dim,
)
self.layers = nn.ModuleList([BlenderbotEncoderLayer(config) for _ in range(config.encoder_layers)])
self.layer_norm = nn.LayerNorm(config.d_model)
self.gradient_checkpointing = False
# Initialize weights and apply final processing
self.post_init()
def forward(
self,
input_ids=None,
attention_mask=None,
head_mask=None,
inputs_embeds=None,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated vectors
than the model's internal embedding lookup matrix.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# retrieve input_ids and inputs_embeds
if input_ids is not None and inputs_embeds is not None:
raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
elif input_ids is not None:
input_shape = input_ids.size()
input_ids = input_ids.view(-1, input_shape[-1])
elif inputs_embeds is not None:
input_shape = inputs_embeds.size()[:-1]
else:
raise ValueError("You have to specify either input_ids or inputs_embeds")
if inputs_embeds is None:
inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale
embed_pos = self.embed_positions(input_shape)
hidden_states = inputs_embeds + embed_pos
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
# expand attention_mask
if attention_mask is not None:
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
attention_mask = _expand_mask(attention_mask, inputs_embeds.dtype)
encoder_states = () if output_hidden_states else None
all_attentions = () if output_attentions else None
# check if head_mask has a correct number of layers specified if desired
if head_mask is not None:
if head_mask.size()[0] != len(self.layers):
raise ValueError(
f"The head_mask should be specified for {len(self.layers)} layers, but it is for"
f" {head_mask.size()[0]}."
)
for idx, encoder_layer in enumerate(self.layers):
if output_hidden_states:
encoder_states = encoder_states + (hidden_states,)
# add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
dropout_probability = random.uniform(0, 1)
if self.training and (dropout_probability < self.layerdrop): # skip the layer
layer_outputs = (None, None)
else:
if self.gradient_checkpointing and self.training:
def create_custom_forward(module):
def custom_forward(*inputs):
return module(*inputs, output_attentions)
return custom_forward
layer_outputs = torch.utils.checkpoint.checkpoint(
create_custom_forward(encoder_layer),
hidden_states,
attention_mask,
(head_mask[idx] if head_mask is not None else None),
)
else:
layer_outputs = encoder_layer(
hidden_states,
attention_mask,
layer_head_mask=(head_mask[idx] if head_mask is not None else None),
output_attentions=output_attentions,
)
hidden_states = layer_outputs[0]
if output_attentions:
all_attentions = all_attentions + (layer_outputs[1],)
# add final layer norm
hidden_states = self.layer_norm(hidden_states)
if output_hidden_states:
encoder_states = encoder_states + (hidden_states,)
if not return_dict:
return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None)
return BaseModelOutput(
last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions
)
class BlenderbotDecoder(BlenderbotPreTrainedModel):
"""
Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a [`BlenderbotDecoderLayer`]
Args:
config: BlenderbotConfig
embed_tokens (nn.Embedding): output embedding
"""
def __init__(self, config: BlenderbotConfig, embed_tokens: Optional[nn.Embedding] = None):
super().__init__(config)
self.dropout = config.dropout
self.layerdrop = config.decoder_layerdrop
self.padding_idx = config.pad_token_id
self.max_target_positions = config.max_position_embeddings
self.embed_scale = math.sqrt(config.d_model) if config.scale_embedding else 1.0
if embed_tokens is not None:
self.embed_tokens = embed_tokens
else:
self.embed_tokens = nn.Embedding(config.vocab_size, config.d_model, self.padding_idx)
self.embed_positions = BlenderbotLearnedPositionalEmbedding(
config.max_position_embeddings,
config.d_model,
)
self.layers = nn.ModuleList([BlenderbotDecoderLayer(config) for _ in range(config.decoder_layers)])
self.layer_norm = nn.LayerNorm(config.d_model)
self.gradient_checkpointing = False
# Initialize weights and apply final processing
self.post_init()
def get_input_embeddings(self):
return self.embed_tokens
def set_input_embeddings(self, value):
self.embed_tokens = value
# Copied from transformers.models.bart.modeling_bart.BartDecoder._prepare_decoder_attention_mask
def _prepare_decoder_attention_mask(self, attention_mask, input_shape, inputs_embeds, past_key_values_length):
# create causal mask
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
combined_attention_mask = None
if input_shape[-1] > 1:
combined_attention_mask = _make_causal_mask(
input_shape, inputs_embeds.dtype, past_key_values_length=past_key_values_length
).to(inputs_embeds.device)
if attention_mask is not None:
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
expanded_attn_mask = _expand_mask(attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]).to(
inputs_embeds.device
)
combined_attention_mask = (
expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask + combined_attention_mask
)
return combined_attention_mask
def forward(
self,
input_ids=None,
attention_mask=None,
encoder_hidden_states=None,
encoder_attention_mask=None,
head_mask=None,
cross_attn_head_mask=None,
past_key_values=None,
inputs_embeds=None,
use_cache=None,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
of the decoder.
encoder_attention_mask (`torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional*):
Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values
selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in `[0,
1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules in the decoder to avoid performing
cross-attention on hidden heads. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those
that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of
all `decoder_input_ids` of shape `(batch_size, sequence_length)`. inputs_embeds (`torch.FloatTensor` of
shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing
`input_ids` you can choose to directly pass an embedded representation. This is useful if you want more
control over how to convert `input_ids` indices into associated vectors than the model's internal
embedding lookup matrix.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
use_cache = use_cache if use_cache is not None else self.config.use_cache
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# retrieve input_ids and inputs_embeds
if input_ids is not None and inputs_embeds is not None:
raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
elif input_ids is not None:
input_shape = input_ids.size()
input_ids = input_ids.view(-1, input_shape[-1])
elif inputs_embeds is not None:
input_shape = inputs_embeds.size()[:-1]
else:
raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")
# past_key_values_length
past_key_values_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0
if inputs_embeds is None:
inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale
attention_mask = self._prepare_decoder_attention_mask(
attention_mask, input_shape, inputs_embeds, past_key_values_length
)
# expand encoder attention mask
if encoder_hidden_states is not None and encoder_attention_mask is not None:
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
encoder_attention_mask = _expand_mask(encoder_attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1])
# embed positions
positions = self.embed_positions(input_shape, past_key_values_length)
hidden_states = inputs_embeds + positions
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
# decoder layers
all_hidden_states = () if output_hidden_states else None
all_self_attns = () if output_attentions else None
all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None
next_decoder_cache = () if use_cache else None
# check if head_mask/cross_attn_head_mask has a correct number of layers specified if desired
for attn_mask, mask_name in zip([head_mask, cross_attn_head_mask], ["head_mask", "cross_attn_head_mask"]):
if attn_mask is not None:
if attn_mask.size()[0] != len(self.layers):
raise ValueError(
f"The `{mask_name}` should be specified for {len(self.layers)} layers, but it is for"
f" {head_mask.size()[0]}."
)
for idx, decoder_layer in enumerate(self.layers):
# add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
if output_hidden_states:
all_hidden_states += (hidden_states,)
dropout_probability = random.uniform(0, 1)
if self.training and (dropout_probability < self.layerdrop):
continue
past_key_value = past_key_values[idx] if past_key_values is not None else None
if self.gradient_checkpointing and self.training:
if use_cache:
logger.warning(
"`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
)
use_cache = False
def create_custom_forward(module):
def custom_forward(*inputs):
# None for past_key_value
return module(*inputs, output_attentions, use_cache)
return custom_forward
layer_outputs = torch.utils.checkpoint.checkpoint(
create_custom_forward(decoder_layer),
hidden_states,
attention_mask,
encoder_hidden_states,
encoder_attention_mask,
head_mask[idx] if head_mask is not None else None,
cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None,
None,
)
else:
layer_outputs = decoder_layer(
hidden_states,
attention_mask=attention_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
layer_head_mask=(head_mask[idx] if head_mask is not None else None),
cross_attn_layer_head_mask=(
cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None
),
past_key_value=past_key_value,
output_attentions=output_attentions,
use_cache=use_cache,
)
hidden_states = layer_outputs[0]
if use_cache:
next_decoder_cache += (layer_outputs[3 if output_attentions else 1],)
if output_attentions:
all_self_attns += (layer_outputs[1],)
if encoder_hidden_states is not None:
all_cross_attentions += (layer_outputs[2],)
# add final layer norm
hidden_states = self.layer_norm(hidden_states)
# add hidden states from the last decoder layer
if output_hidden_states:
all_hidden_states += (hidden_states,)
next_cache = next_decoder_cache if use_cache else None
if not return_dict:
return tuple(
v
for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, all_cross_attentions]
if v is not None
)
return BaseModelOutputWithPastAndCrossAttentions(
last_hidden_state=hidden_states,
past_key_values=next_cache,
hidden_states=all_hidden_states,
attentions=all_self_attns,
cross_attentions=all_cross_attentions,
)
@add_start_docstrings(
"The bare Blenderbot Model outputting raw hidden-states without any specific head on top.",
BLENDERBOT_START_DOCSTRING,
)
class BlenderbotModel(BlenderbotPreTrainedModel):
_keys_to_ignore_on_load_missing = ["decoder.embed_tokens.weight", "encoder.embed_tokens.weight"]
def __init__(self, config: BlenderbotConfig):
super().__init__(config)
padding_idx, vocab_size = config.pad_token_id, config.vocab_size
self.shared = nn.Embedding(vocab_size, config.d_model, padding_idx)
self.encoder = BlenderbotEncoder(config, self.shared)
self.decoder = BlenderbotDecoder(config, self.shared)
# Initialize weights and apply final processing
self.post_init()
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
if pretrained_model_name_or_path == "facebook/blenderbot-90M":
warnings.warn(
"The checkpoint `facebook/blenderbot-90M` is deprecated. In the future, please use the identical"
" checkpoint `facebook/small_blenderbot-90M` with"
" `BlenderbotSmallModel.from_pretrained('facebook/small_blenderbot-90M')` instead.",
FutureWarning,
)
return BlenderbotSmallModel.from_pretrained(pretrained_model_name_or_path)
return super(BlenderbotModel, cls).from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
def get_input_embeddings(self):
return self.shared
def set_input_embeddings(self, value):
self.shared = value
self.encoder.embed_tokens = self.shared
self.decoder.embed_tokens = self.shared
def get_encoder(self):
return self.encoder
def get_decoder(self):
return self.decoder
@add_start_docstrings_to_model_forward(BLENDERBOT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=Seq2SeqModelOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.LongTensor] = None,
head_mask: Optional[torch.Tensor] = None,
decoder_head_mask: Optional[torch.Tensor] = None,
cross_attn_head_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Union[Tuple, BaseModelOutput]] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.Tensor] = None,
decoder_inputs_embeds: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple[torch.FloatTensor], Seq2SeqModelOutput]:
r"""
Returns:
Example:
```python
>>> from transformers import BlenderbotTokenizer, BlenderbotModel
>>> model = BlenderbotModel.from_pretrained("facebook/blenderbot-400M-distill")
>>> tokenizer = BlenderbotTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
>>> inputs = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt")
>>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids # Batch size 1
>>> outputs = model(input_ids=inputs.input_ids, decoder_input_ids=decoder_input_ids)
>>> last_hidden_states = outputs.last_hidden_state
>>> list(last_hidden_states.shape)
[1, 6, 1280]
```"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
use_cache = use_cache if use_cache is not None else self.config.use_cache
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if encoder_outputs is None:
encoder_outputs = self.encoder(
input_ids=input_ids,
attention_mask=attention_mask,
head_mask=head_mask,
inputs_embeds=inputs_embeds,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
# If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOutput when return_dict=True
elif return_dict and not isinstance(encoder_outputs, BaseModelOutput):
encoder_outputs = BaseModelOutput(
last_hidden_state=encoder_outputs[0],
hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None,
attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None,
)
# decoder outputs consists of (dec_features, past_key_value, dec_hidden, dec_attn)
decoder_outputs = self.decoder(
input_ids=decoder_input_ids,
attention_mask=decoder_attention_mask,
encoder_hidden_states=encoder_outputs[0],
encoder_attention_mask=attention_mask,
head_mask=decoder_head_mask,
cross_attn_head_mask=cross_attn_head_mask,
past_key_values=past_key_values,
inputs_embeds=decoder_inputs_embeds,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
if not return_dict:
return decoder_outputs + encoder_outputs
return Seq2SeqModelOutput(
last_hidden_state=decoder_outputs.last_hidden_state,
past_key_values=decoder_outputs.past_key_values,
decoder_hidden_states=decoder_outputs.hidden_states,
decoder_attentions=decoder_outputs.attentions,
cross_attentions=decoder_outputs.cross_attentions,
encoder_last_hidden_state=encoder_outputs.last_hidden_state,
encoder_hidden_states=encoder_outputs.hidden_states,
encoder_attentions=encoder_outputs.attentions,
)
@add_start_docstrings(
"The Blenderbot Model with a language modeling head. Can be used for summarization.", BLENDERBOT_START_DOCSTRING
)
class BlenderbotForConditionalGeneration(BlenderbotPreTrainedModel):
base_model_prefix = "model"
_keys_to_ignore_on_load_missing = [
r"final_logits_bias",
r"encoder.version",
r"decoder.version",
r"lm_head.weight",
"decoder.embed_tokens.weight",
"encoder.embed_tokens.weight",
]
def __init__(self, config: BlenderbotConfig):
super().__init__(config)
self.model = BlenderbotModel(config)
self.register_buffer("final_logits_bias", torch.zeros((1, self.model.shared.num_embeddings)))
self.lm_head = nn.Linear(config.d_model, self.model.shared.num_embeddings, bias=False)
# Initialize weights and apply final processing
self.post_init()
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
if pretrained_model_name_or_path == "facebook/blenderbot-90M":
warnings.warn(
"The checkpoint `facebook/blenderbot-90M` is deprecated. In the future, please use the identical"
" checkpoint `facebook/small_blenderbot-90M` with"
" `BlenderbotSmallForConditionalGeneration.from_pretrained('facebook/small_blenderbot-90M')` instead.",
FutureWarning,
)
return BlenderbotSmallForConditionalGeneration.from_pretrained(pretrained_model_name_or_path)
return super(BlenderbotForConditionalGeneration, cls).from_pretrained(
pretrained_model_name_or_path, *model_args, **kwargs
)
def get_encoder(self):
return self.model.get_encoder()
def get_decoder(self):
return self.model.get_decoder()
def resize_token_embeddings(self, new_num_tokens: int) -> nn.Embedding:
new_embeddings = super().resize_token_embeddings(new_num_tokens)
self._resize_final_logits_bias(new_num_tokens)
return new_embeddings
def _resize_final_logits_bias(self, new_num_tokens: int) -> None:
old_num_tokens = self.final_logits_bias.shape[-1]
if new_num_tokens <= old_num_tokens:
new_bias = self.final_logits_bias[:, :new_num_tokens]
else:
extra_bias = torch.zeros((1, new_num_tokens - old_num_tokens), device=self.final_logits_bias.device)
new_bias = torch.cat([self.final_logits_bias, extra_bias], dim=1)
self.register_buffer("final_logits_bias", new_bias)
def get_output_embeddings(self):
return self.lm_head
def set_output_embeddings(self, new_embeddings):
self.lm_head = new_embeddings
@add_start_docstrings_to_model_forward(BLENDERBOT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=Seq2SeqLMOutput, config_class=_CONFIG_FOR_DOC)
@add_end_docstrings(BLENDERBOT_GENERATION_EXAMPLE)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.LongTensor] = None,
head_mask: Optional[torch.Tensor] = None,
decoder_head_mask: Optional[torch.Tensor] = None,
cross_attn_head_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Union[Tuple, BaseModelOutput]] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.Tensor] = None,
decoder_inputs_embeds: Optional[torch.FloatTensor] = None,
labels: Optional[torch.LongTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple[torch.FloatTensor], Seq2SeqLMOutput]:
r"""
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
Returns:
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if labels is not None:
if use_cache:
logger.warning("The `use_cache` argument is changed to `False` since `labels` is provided.")
use_cache = False
if decoder_input_ids is None and decoder_inputs_embeds is None:
decoder_input_ids = shift_tokens_right(
labels, self.config.pad_token_id, self.config.decoder_start_token_id
)
outputs = self.model(
input_ids,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
encoder_outputs=encoder_outputs,
decoder_attention_mask=decoder_attention_mask,
head_mask=head_mask,
decoder_head_mask=decoder_head_mask,
cross_attn_head_mask=cross_attn_head_mask,
past_key_values=past_key_values,
inputs_embeds=inputs_embeds,
decoder_inputs_embeds=decoder_inputs_embeds,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
lm_logits = self.lm_head(outputs[0]) + self.final_logits_bias
masked_lm_loss = None
if labels is not None:
loss_fct = CrossEntropyLoss()
masked_lm_loss = loss_fct(lm_logits.view(-1, self.config.vocab_size), labels.view(-1))
if not return_dict:
output = (lm_logits,) + outputs[1:]
return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output
return Seq2SeqLMOutput(
loss=masked_lm_loss,
logits=lm_logits,
past_key_values=outputs.past_key_values,
decoder_hidden_states=outputs.decoder_hidden_states,
decoder_attentions=outputs.decoder_attentions,
cross_attentions=outputs.cross_attentions,
encoder_last_hidden_state=outputs.encoder_last_hidden_state,
encoder_hidden_states=outputs.encoder_hidden_states,
encoder_attentions=outputs.encoder_attentions,
)
def prepare_inputs_for_generation(
self,
decoder_input_ids,
past=None,
attention_mask=None,
head_mask=None,
decoder_head_mask=None,
cross_attn_head_mask=None,
use_cache=None,
encoder_outputs=None,
**kwargs
):
# cut decoder_input_ids if past is used
if past is not None:
decoder_input_ids = decoder_input_ids[:, -1:]
return {
"input_ids": None, # encoder_outputs is defined. input_ids not needed
"encoder_outputs": encoder_outputs,
"past_key_values": past,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"head_mask": head_mask,
"decoder_head_mask": decoder_head_mask,
"cross_attn_head_mask": cross_attn_head_mask,
"use_cache": use_cache, # change this to avoid caching (presumably for debugging)
}
@staticmethod
def _reorder_cache(past, beam_idx):
reordered_past = ()
for layer_past in past:
# cached cross_attention states don't have to be reordered -> they are always the same
reordered_past += (
tuple(past_state.index_select(0, beam_idx) for past_state in layer_past[:2]) + layer_past[2:],
)
return reordered_past
# Copied from transformers.models.bart.modeling_bart.BartDecoderWrapper with Bart->Blenderbot
class BlenderbotDecoderWrapper(BlenderbotPreTrainedModel):
"""
This wrapper class is a helper class to correctly load pretrained checkpoints when the causal language model is
used in combination with the [`EncoderDecoderModel`] framework.
"""
def __init__(self, config):
super().__init__(config)
self.decoder = BlenderbotDecoder(config)
def forward(self, *args, **kwargs):
return self.decoder(*args, **kwargs)
# Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->Blenderbot, facebook/bart-base->facebook/blenderbot-400M-distill
class BlenderbotForCausalLM(BlenderbotPreTrainedModel):
_keys_to_ignore_on_load_missing = ["lm_head.weight"]
def __init__(self, config):
config = copy.deepcopy(config)
config.is_decoder = True
config.is_encoder_decoder = False
super().__init__(config)
self.model = BlenderbotDecoderWrapper(config)
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
# Initialize weights and apply final processing
self.post_init()
def get_input_embeddings(self):
return self.model.decoder.embed_tokens
def set_input_embeddings(self, value):
self.model.decoder.embed_tokens = value
def get_output_embeddings(self):
return self.lm_head
def set_output_embeddings(self, new_embeddings):
self.lm_head = new_embeddings
def set_decoder(self, decoder):
self.model.decoder = decoder
def get_decoder(self):
return self.model.decoder
@replace_return_docstrings(output_type=CausalLMOutputWithCrossAttentions, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: torch.LongTensor = None,
attention_mask: Optional[torch.Tensor] = None,
encoder_hidden_states: Optional[torch.FloatTensor] = None,
encoder_attention_mask: Optional[torch.FloatTensor] = None,
head_mask: Optional[torch.Tensor] = None,
cross_attn_head_mask: Optional[torch.Tensor] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.FloatTensor] = None,
labels: Optional[torch.LongTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple, CausalLMOutputWithCrossAttentions]:
r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
if the model is configured as a decoder.
encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used
in the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. The two additional
tensors are only required when the model is used as a decoder in a Sequence to Sequence model.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those
that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of
all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
(see `past_key_values`).
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
Returns:
Example:
```python
>>> from transformers import BlenderbotTokenizer, BlenderbotForCausalLM
>>> tokenizer = BlenderbotTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
>>> model = BlenderbotForCausalLM.from_pretrained(
... "facebook/blenderbot-400M-distill", add_cross_attention=False
... )
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> expected_shape = [1, inputs.input_ids.shape[-1], model.config.vocab_size]
>>> list(logits.shape) == expected_shape
True
```"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
outputs = self.model.decoder(
input_ids=input_ids,
attention_mask=attention_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
head_mask=head_mask,
cross_attn_head_mask=cross_attn_head_mask,
past_key_values=past_key_values,
inputs_embeds=inputs_embeds,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
logits = self.lm_head(outputs[0])
loss = None
if labels is not None:
loss_fct = CrossEntropyLoss()
loss = loss_fct(logits.view(-1, self.config.vocab_size), labels.view(-1))
if not return_dict:
output = (logits,) + outputs[1:]
return (loss,) + output if loss is not None else output
return CausalLMOutputWithCrossAttentions(
loss=loss,
logits=logits,
past_key_values=outputs.past_key_values,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)
def prepare_inputs_for_generation(self, input_ids, past=None, attention_mask=None, use_cache=None, **kwargs):
# if model is used as a decoder in encoder-decoder model, the decoder attention mask is created on the fly
if attention_mask is None:
attention_mask = input_ids.new_ones(input_ids.shape)
if past:
input_ids = input_ids[:, -1:]
# first step, decoder_cached_states are empty
return {
"input_ids": input_ids, # encoder_outputs is defined. input_ids not needed
"attention_mask": attention_mask,
"past_key_values": past,
"use_cache": use_cache,
}
@staticmethod
def _reorder_cache(past, beam_idx):
reordered_past = ()
for layer_past in past:
reordered_past += (tuple(past_state.index_select(0, beam_idx) for past_state in layer_past),)
return reordered_past
| # coding=utf-8
# Copyright 2021 The Facebook, Inc. and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" PyTorch Blenderbot model."""
import copy
import math
import os
import random
import warnings
from typing import List, Optional, Tuple, Union
import torch
import torch.utils.checkpoint
from torch import nn
from torch.nn import CrossEntropyLoss
from ...activations import ACT2FN
from ...modeling_outputs import (
BaseModelOutput,
BaseModelOutputWithPastAndCrossAttentions,
CausalLMOutputWithCrossAttentions,
Seq2SeqLMOutput,
Seq2SeqModelOutput,
)
from ...modeling_utils import PreTrainedModel
from ...utils import (
add_end_docstrings,
add_start_docstrings,
add_start_docstrings_to_model_forward,
logging,
replace_return_docstrings,
)
from ..blenderbot_small import BlenderbotSmallForConditionalGeneration, BlenderbotSmallModel
from .configuration_blenderbot import BlenderbotConfig
logger = logging.get_logger(__name__)
_CONFIG_FOR_DOC = "BlenderbotConfig"
_TOKENIZER_FOR_DOC = "BlenderbotTokenizer"
_CHECKPOINT_FOR_DOC = "facebook/blenderbot-400M-distill"
BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST = [
"facebook/blenderbot-3B",
# See all Blenderbot models at https://huggingface.co/models?filter=blenderbot
]
# Copied from transformers.models.bart.modeling_bart.shift_tokens_right
def shift_tokens_right(input_ids: torch.Tensor, pad_token_id: int, decoder_start_token_id: int):
"""
Shift input ids one token to the right.
"""
shifted_input_ids = input_ids.new_zeros(input_ids.shape)
shifted_input_ids[:, 1:] = input_ids[:, :-1].clone()
shifted_input_ids[:, 0] = decoder_start_token_id
if pad_token_id is None:
raise ValueError("self.model.config.pad_token_id has to be defined.")
# replace possible -100 values in labels by `pad_token_id`
shifted_input_ids.masked_fill_(shifted_input_ids == -100, pad_token_id)
return shifted_input_ids
# Copied from transformers.models.bart.modeling_bart._make_causal_mask
def _make_causal_mask(input_ids_shape: torch.Size, dtype: torch.dtype, past_key_values_length: int = 0):
"""
Make causal mask used for bi-directional self-attention.
"""
bsz, tgt_len = input_ids_shape
mask = torch.full((tgt_len, tgt_len), torch.tensor(torch.finfo(dtype).min))
mask_cond = torch.arange(mask.size(-1))
mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
mask = mask.to(dtype)
if past_key_values_length > 0:
mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype), mask], dim=-1)
return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)
# Copied from transformers.models.bart.modeling_bart._expand_mask
def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
"""
Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
"""
bsz, src_len = mask.size()
tgt_len = tgt_len if tgt_len is not None else src_len
expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)
inverted_mask = 1.0 - expanded_mask
return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
class BlenderbotLearnedPositionalEmbedding(nn.Embedding):
"""
This module learns positional embeddings up to a fixed maximum size.
"""
def __init__(self, num_embeddings: int, embedding_dim: int):
super().__init__(num_embeddings, embedding_dim)
def forward(self, input_ids_shape: torch.Size, past_key_values_length: int = 0):
"""`input_ids_shape` is expected to be [bsz x seqlen]."""
bsz, seq_len = input_ids_shape[:2]
positions = torch.arange(
past_key_values_length, past_key_values_length + seq_len, dtype=torch.long, device=self.weight.device
)
return super().forward(positions)
# Copied from transformers.models.bart.modeling_bart.BartAttention with Bart->Blenderbot
class BlenderbotAttention(nn.Module):
"""Multi-headed attention from 'Attention Is All You Need' paper"""
def __init__(
self,
embed_dim: int,
num_heads: int,
dropout: float = 0.0,
is_decoder: bool = False,
bias: bool = True,
):
super().__init__()
self.embed_dim = embed_dim
self.num_heads = num_heads
self.dropout = dropout
self.head_dim = embed_dim // num_heads
if (self.head_dim * num_heads) != self.embed_dim:
raise ValueError(
f"embed_dim must be divisible by num_heads (got `embed_dim`: {self.embed_dim}"
f" and `num_heads`: {num_heads})."
)
self.scaling = self.head_dim**-0.5
self.is_decoder = is_decoder
self.k_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
self.v_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
self.q_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
self.out_proj = nn.Linear(embed_dim, embed_dim, bias=bias)
def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
def forward(
self,
hidden_states: torch.Tensor,
key_value_states: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
output_attentions: bool = False,
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
"""Input shape: Batch x Time x Channel"""
# if key_value_states are provided this layer is used as a cross-attention layer
# for the decoder
is_cross_attention = key_value_states is not None
bsz, tgt_len, _ = hidden_states.size()
# get query proj
query_states = self.q_proj(hidden_states) * self.scaling
# get key, value proj
# `past_key_value[0].shape[2] == key_value_states.shape[1]`
# is checking that the `sequence_length` of the `past_key_value` is the same as
# the provided `key_value_states` to support prefix tuning
if (
is_cross_attention
and past_key_value is not None
and past_key_value[0].shape[2] == key_value_states.shape[1]
):
# reuse k,v, cross_attentions
key_states = past_key_value[0]
value_states = past_key_value[1]
elif is_cross_attention:
# cross_attentions
key_states = self._shape(self.k_proj(key_value_states), -1, bsz)
value_states = self._shape(self.v_proj(key_value_states), -1, bsz)
elif past_key_value is not None:
# reuse k, v, self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
key_states = torch.cat([past_key_value[0], key_states], dim=2)
value_states = torch.cat([past_key_value[1], value_states], dim=2)
else:
# self_attention
key_states = self._shape(self.k_proj(hidden_states), -1, bsz)
value_states = self._shape(self.v_proj(hidden_states), -1, bsz)
if self.is_decoder:
# if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states.
# Further calls to cross_attention layer can then reuse all cross-attention
# key/value_states (first "if" case)
# if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of
# all previous decoder key/value_states. Further calls to uni-directional self-attention
# can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
# if encoder bi-directional self-attention `past_key_value` is always `None`
past_key_value = (key_states, value_states)
proj_shape = (bsz * self.num_heads, -1, self.head_dim)
query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
key_states = key_states.view(*proj_shape)
value_states = value_states.view(*proj_shape)
src_len = key_states.size(1)
attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))
if attn_weights.size() != (bsz * self.num_heads, tgt_len, src_len):
raise ValueError(
f"Attention weights should be of size {(bsz * self.num_heads, tgt_len, src_len)}, but is"
f" {attn_weights.size()}"
)
if attention_mask is not None:
if attention_mask.size() != (bsz, 1, tgt_len, src_len):
raise ValueError(
f"Attention mask should be of size {(bsz, 1, tgt_len, src_len)}, but is {attention_mask.size()}"
)
attn_weights = attn_weights.view(bsz, self.num_heads, tgt_len, src_len) + attention_mask
attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
if layer_head_mask is not None:
if layer_head_mask.size() != (self.num_heads,):
raise ValueError(
f"Head mask for a single layer should be of size {(self.num_heads,)}, but is"
f" {layer_head_mask.size()}"
)
attn_weights = layer_head_mask.view(1, -1, 1, 1) * attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
attn_weights = attn_weights.view(bsz * self.num_heads, tgt_len, src_len)
if output_attentions:
# this operation is a bit awkward, but it's required to
# make sure that attn_weights keeps its gradient.
# In order to do so, attn_weights have to be reshaped
# twice and have to be reused in the following
attn_weights_reshaped = attn_weights.view(bsz, self.num_heads, tgt_len, src_len)
attn_weights = attn_weights_reshaped.view(bsz * self.num_heads, tgt_len, src_len)
else:
attn_weights_reshaped = None
attn_probs = nn.functional.dropout(attn_weights, p=self.dropout, training=self.training)
attn_output = torch.bmm(attn_probs, value_states)
if attn_output.size() != (bsz * self.num_heads, tgt_len, self.head_dim):
raise ValueError(
f"`attn_output` should be of size {(bsz, self.num_heads, tgt_len, self.head_dim)}, but is"
f" {attn_output.size()}"
)
attn_output = attn_output.view(bsz, self.num_heads, tgt_len, self.head_dim)
attn_output = attn_output.transpose(1, 2)
# Use the `embed_dim` from the config (stored in the class) rather than `hidden_state` because `attn_output` can be
# partitioned aross GPUs when using tensor-parallelism.
attn_output = attn_output.reshape(bsz, tgt_len, self.embed_dim)
attn_output = self.out_proj(attn_output)
return attn_output, attn_weights_reshaped, past_key_value
# Copied from transformers.models.mbart.modeling_mbart.MBartEncoderLayer with MBart->Blenderbot
class BlenderbotEncoderLayer(nn.Module):
def __init__(self, config: BlenderbotConfig):
super().__init__()
self.embed_dim = config.d_model
self.self_attn = BlenderbotAttention(
embed_dim=self.embed_dim,
num_heads=config.encoder_attention_heads,
dropout=config.attention_dropout,
)
self.self_attn_layer_norm = nn.LayerNorm(self.embed_dim)
self.dropout = config.dropout
self.activation_fn = ACT2FN[config.activation_function]
self.activation_dropout = config.activation_dropout
self.fc1 = nn.Linear(self.embed_dim, config.encoder_ffn_dim)
self.fc2 = nn.Linear(config.encoder_ffn_dim, self.embed_dim)
self.final_layer_norm = nn.LayerNorm(self.embed_dim)
def forward(
self,
hidden_states: torch.Tensor,
attention_mask: torch.Tensor,
layer_head_mask: torch.Tensor,
output_attentions: bool = False,
) -> torch.Tensor:
"""
Args:
hidden_states (`torch.FloatTensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`torch.FloatTensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
*(encoder_attention_heads,)*.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
"""
residual = hidden_states
hidden_states = self.self_attn_layer_norm(hidden_states)
hidden_states, attn_weights, _ = self.self_attn(
hidden_states=hidden_states,
attention_mask=attention_mask,
layer_head_mask=layer_head_mask,
output_attentions=output_attentions,
)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
residual = hidden_states
hidden_states = self.final_layer_norm(hidden_states)
hidden_states = self.activation_fn(self.fc1(hidden_states))
hidden_states = nn.functional.dropout(hidden_states, p=self.activation_dropout, training=self.training)
hidden_states = self.fc2(hidden_states)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
if hidden_states.dtype == torch.float16 and (
torch.isinf(hidden_states).any() or torch.isnan(hidden_states).any()
):
clamp_value = torch.finfo(hidden_states.dtype).max - 1000
hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value)
outputs = (hidden_states,)
if output_attentions:
outputs += (attn_weights,)
return outputs
# Copied from transformers.models.mbart.modeling_mbart.MBartDecoderLayer with MBart->Blenderbot
class BlenderbotDecoderLayer(nn.Module):
def __init__(self, config: BlenderbotConfig):
super().__init__()
self.embed_dim = config.d_model
self.self_attn = BlenderbotAttention(
embed_dim=self.embed_dim,
num_heads=config.decoder_attention_heads,
dropout=config.attention_dropout,
is_decoder=True,
)
self.dropout = config.dropout
self.activation_fn = ACT2FN[config.activation_function]
self.activation_dropout = config.activation_dropout
self.self_attn_layer_norm = nn.LayerNorm(self.embed_dim)
self.encoder_attn = BlenderbotAttention(
self.embed_dim,
config.decoder_attention_heads,
dropout=config.attention_dropout,
is_decoder=True,
)
self.encoder_attn_layer_norm = nn.LayerNorm(self.embed_dim)
self.fc1 = nn.Linear(self.embed_dim, config.decoder_ffn_dim)
self.fc2 = nn.Linear(config.decoder_ffn_dim, self.embed_dim)
self.final_layer_norm = nn.LayerNorm(self.embed_dim)
def forward(
self,
hidden_states: torch.Tensor,
attention_mask: Optional[torch.Tensor] = None,
encoder_hidden_states: Optional[torch.Tensor] = None,
encoder_attention_mask: Optional[torch.Tensor] = None,
layer_head_mask: Optional[torch.Tensor] = None,
cross_attn_layer_head_mask: Optional[torch.Tensor] = None,
past_key_value: Optional[Tuple[torch.Tensor]] = None,
output_attentions: Optional[bool] = False,
use_cache: Optional[bool] = True,
) -> torch.Tensor:
"""
Args:
hidden_states (`torch.FloatTensor`): input to the layer of shape *(seq_len, batch, embed_dim)*
attention_mask (`torch.FloatTensor`): attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
encoder_hidden_states (`torch.FloatTensor`):
cross attention input to the layer of shape *(seq_len, batch, embed_dim)*
encoder_attention_mask (`torch.FloatTensor`): encoder attention mask of size
*(batch, 1, tgt_len, src_len)* where padding elements are indicated by very large negative values.
layer_head_mask (`torch.FloatTensor`): mask for attention heads in a given layer of size
*(encoder_attention_heads,)*.
cross_attn_layer_head_mask (`torch.FloatTensor`): mask for cross-attention heads in a given layer of
size *(decoder_attention_heads,)*.
past_key_value (`Tuple(torch.FloatTensor)`): cached past key and value projection states
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
"""
residual = hidden_states
hidden_states = self.self_attn_layer_norm(hidden_states)
# Self Attention
# decoder uni-directional self-attention cached key/values tuple is at positions 1,2
self_attn_past_key_value = past_key_value[:2] if past_key_value is not None else None
# add present self-attn cache to positions 1,2 of present_key_value tuple
hidden_states, self_attn_weights, present_key_value = self.self_attn(
hidden_states=hidden_states,
past_key_value=self_attn_past_key_value,
attention_mask=attention_mask,
layer_head_mask=layer_head_mask,
output_attentions=output_attentions,
)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
# Cross-Attention Block
cross_attn_present_key_value = None
cross_attn_weights = None
if encoder_hidden_states is not None:
residual = hidden_states
hidden_states = self.encoder_attn_layer_norm(hidden_states)
# cross_attn cached key/values tuple is at positions 3,4 of present_key_value tuple
cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None
hidden_states, cross_attn_weights, cross_attn_present_key_value = self.encoder_attn(
hidden_states=hidden_states,
key_value_states=encoder_hidden_states,
attention_mask=encoder_attention_mask,
layer_head_mask=cross_attn_layer_head_mask,
past_key_value=cross_attn_past_key_value,
output_attentions=output_attentions,
)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
# add cross-attn to positions 3,4 of present_key_value tuple
present_key_value = present_key_value + cross_attn_present_key_value
# Fully Connected
residual = hidden_states
hidden_states = self.final_layer_norm(hidden_states)
hidden_states = self.activation_fn(self.fc1(hidden_states))
hidden_states = nn.functional.dropout(hidden_states, p=self.activation_dropout, training=self.training)
hidden_states = self.fc2(hidden_states)
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
hidden_states = residual + hidden_states
outputs = (hidden_states,)
if output_attentions:
outputs += (self_attn_weights, cross_attn_weights)
if use_cache:
outputs += (present_key_value,)
return outputs
class BlenderbotPreTrainedModel(PreTrainedModel):
config_class = BlenderbotConfig
base_model_prefix = "model"
supports_gradient_checkpointing = True
def _init_weights(self, module):
std = self.config.init_std
if isinstance(module, nn.Linear):
module.weight.data.normal_(mean=0.0, std=std)
if module.bias is not None:
module.bias.data.zero_()
elif isinstance(module, nn.Embedding):
module.weight.data.normal_(mean=0.0, std=std)
if module.padding_idx is not None:
module.weight.data[module.padding_idx].zero_()
def _set_gradient_checkpointing(self, module, value=False):
if isinstance(module, (BlenderbotDecoder, BlenderbotEncoder)):
module.gradient_checkpointing = value
@property
def dummy_inputs(self):
pad_token = self.config.pad_token_id
input_ids = torch.tensor([[0, 6, 10, 4, 2], [0, 8, 12, 2, pad_token]], device=self.device)
dummy_inputs = {
"attention_mask": input_ids.ne(pad_token),
"input_ids": input_ids,
"decoder_input_ids": input_ids,
}
return dummy_inputs
BLENDERBOT_START_DOCSTRING = r"""
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
and behavior.
Parameters:
config ([`BlenderbotConfig`]):
Model configuration class with all the parameters of the model. Initializing with a config file does not
load the weights associated with the model, only the configuration. Check out the
[`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""
BLENDERBOT_GENERATION_EXAMPLE = r"""
Conversation example:
```python
>>> from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration
>>> mname = "facebook/blenderbot-400M-distill"
>>> model = BlenderbotForConditionalGeneration.from_pretrained(mname)
>>> tokenizer = BlenderbotTokenizer.from_pretrained(mname)
>>> UTTERANCE = "My friends are cool but they eat too many carbs."
>>> print("Human: ", UTTERANCE)
Human: My friends are cool but they eat too many carbs.
>>> inputs = tokenizer([UTTERANCE], return_tensors="pt")
>>> reply_ids = model.generate(**inputs)
>>> print("Bot: ", tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0])
Bot: That's unfortunate. Are they trying to lose weight or are they just trying to be healthier?
>>> REPLY = "I'm not sure"
>>> print("Human: ", REPLY)
Human: I'm not sure
>>> NEXT_UTTERANCE = (
... "My friends are cool but they eat too many carbs.</s> <s>That's unfortunate. "
... "Are they trying to lose weight or are they just trying to be healthier?</s> "
... "<s> I'm not sure."
... )
>>> inputs = tokenizer([NEXT_UTTERANCE], return_tensors="pt")
>>> next_reply_ids = model.generate(**inputs)
>>> print("Bot: ", tokenizer.batch_decode(next_reply_ids, skip_special_tokens=True)[0])
Bot: That's too bad. Have you tried encouraging them to change their eating habits?
```
"""
BLENDERBOT_INPUTS_DOCSTRING = r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Indices of decoder input sequence tokens in the vocabulary.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are decoder input IDs?](../glossary#decoder-input-ids)
Blenderbot uses the `bos_token_id` as the starting token for `decoder_input_ids` generation. If
`past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see
`past_key_values`).
decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
be used by default.
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
decoder_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the decoder. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules in the decoder. Mask values selected in `[0,
1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
encoder_outputs (`tuple(tuple(torch.FloatTensor)`, *optional*):
Tuple consists of (`last_hidden_state`, *optional*: `hidden_states`, *optional*: `attentions`)
`last_hidden_state` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) is a sequence of
hidden-states at the output of the last layer of the encoder. Used in the cross-attention of the decoder.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape
`(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of shape
`(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention
blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that
don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all
`decoder_input_ids` of shape `(batch_size, sequence_length)`. inputs_embeds (`torch.FloatTensor` of shape
`(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you
can choose to directly pass an embedded representation. This is useful if you want more control over how to
convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
decoder_inputs_embeds (`torch.FloatTensor` of shape `(batch_size, target_sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `decoder_input_ids` you can choose to directly pass an embedded
representation. If `past_key_values` is used, optionally only the last `decoder_inputs_embeds` have to be
input (see `past_key_values`). This is useful if you want more control over how to convert
`decoder_input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
If `decoder_input_ids` and `decoder_inputs_embeds` are both unset, `decoder_inputs_embeds` takes the value
of `inputs_embeds`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
`past_key_values`).
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
class BlenderbotEncoder(BlenderbotPreTrainedModel):
"""
Transformer encoder consisting of *config.encoder_layers* self attention layers. Each layer is a
[`BlenderbotEncoderLayer`].
Args:
config: BlenderbotConfig
embed_tokens (nn.Embedding): output embedding
"""
def __init__(self, config: BlenderbotConfig, embed_tokens: Optional[nn.Embedding] = None):
super().__init__(config)
self.dropout = config.dropout
self.layerdrop = config.encoder_layerdrop
embed_dim = config.d_model
self.padding_idx = config.pad_token_id
self.max_source_positions = config.max_position_embeddings
self.embed_scale = math.sqrt(embed_dim) if config.scale_embedding else 1.0
if embed_tokens is not None:
self.embed_tokens = embed_tokens
else:
self.embed_tokens = nn.Embedding(config.vocab_size, embed_dim, self.padding_idx)
self.embed_positions = BlenderbotLearnedPositionalEmbedding(
config.max_position_embeddings,
embed_dim,
)
self.layers = nn.ModuleList([BlenderbotEncoderLayer(config) for _ in range(config.encoder_layers)])
self.layer_norm = nn.LayerNorm(config.d_model)
self.gradient_checkpointing = False
# Initialize weights and apply final processing
self.post_init()
def forward(
self,
input_ids=None,
attention_mask=None,
head_mask=None,
inputs_embeds=None,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation.
This is useful if you want more control over how to convert `input_ids` indices into associated vectors
than the model's internal embedding lookup matrix.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# retrieve input_ids and inputs_embeds
if input_ids is not None and inputs_embeds is not None:
raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
elif input_ids is not None:
input_shape = input_ids.size()
input_ids = input_ids.view(-1, input_shape[-1])
elif inputs_embeds is not None:
input_shape = inputs_embeds.size()[:-1]
else:
raise ValueError("You have to specify either input_ids or inputs_embeds")
if inputs_embeds is None:
inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale
embed_pos = self.embed_positions(input_shape)
hidden_states = inputs_embeds + embed_pos
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
# expand attention_mask
if attention_mask is not None:
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
attention_mask = _expand_mask(attention_mask, inputs_embeds.dtype)
encoder_states = () if output_hidden_states else None
all_attentions = () if output_attentions else None
# check if head_mask has a correct number of layers specified if desired
if head_mask is not None:
if head_mask.size()[0] != len(self.layers):
raise ValueError(
f"The head_mask should be specified for {len(self.layers)} layers, but it is for"
f" {head_mask.size()[0]}."
)
for idx, encoder_layer in enumerate(self.layers):
if output_hidden_states:
encoder_states = encoder_states + (hidden_states,)
# add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
dropout_probability = random.uniform(0, 1)
if self.training and (dropout_probability < self.layerdrop): # skip the layer
layer_outputs = (None, None)
else:
if self.gradient_checkpointing and self.training:
def create_custom_forward(module):
def custom_forward(*inputs):
return module(*inputs, output_attentions)
return custom_forward
layer_outputs = torch.utils.checkpoint.checkpoint(
create_custom_forward(encoder_layer),
hidden_states,
attention_mask,
(head_mask[idx] if head_mask is not None else None),
)
else:
layer_outputs = encoder_layer(
hidden_states,
attention_mask,
layer_head_mask=(head_mask[idx] if head_mask is not None else None),
output_attentions=output_attentions,
)
hidden_states = layer_outputs[0]
if output_attentions:
all_attentions = all_attentions + (layer_outputs[1],)
# add final layer norm
hidden_states = self.layer_norm(hidden_states)
if output_hidden_states:
encoder_states = encoder_states + (hidden_states,)
if not return_dict:
return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None)
return BaseModelOutput(
last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions
)
class BlenderbotDecoder(BlenderbotPreTrainedModel):
"""
Transformer decoder consisting of *config.decoder_layers* layers. Each layer is a [`BlenderbotDecoderLayer`]
Args:
config: BlenderbotConfig
embed_tokens (nn.Embedding): output embedding
"""
def __init__(self, config: BlenderbotConfig, embed_tokens: Optional[nn.Embedding] = None):
super().__init__(config)
self.dropout = config.dropout
self.layerdrop = config.decoder_layerdrop
self.padding_idx = config.pad_token_id
self.max_target_positions = config.max_position_embeddings
self.embed_scale = math.sqrt(config.d_model) if config.scale_embedding else 1.0
if embed_tokens is not None:
self.embed_tokens = embed_tokens
else:
self.embed_tokens = nn.Embedding(config.vocab_size, config.d_model, self.padding_idx)
self.embed_positions = BlenderbotLearnedPositionalEmbedding(
config.max_position_embeddings,
config.d_model,
)
self.layers = nn.ModuleList([BlenderbotDecoderLayer(config) for _ in range(config.decoder_layers)])
self.layer_norm = nn.LayerNorm(config.d_model)
self.gradient_checkpointing = False
# Initialize weights and apply final processing
self.post_init()
def get_input_embeddings(self):
return self.embed_tokens
def set_input_embeddings(self, value):
self.embed_tokens = value
# Copied from transformers.models.bart.modeling_bart.BartDecoder._prepare_decoder_attention_mask
def _prepare_decoder_attention_mask(self, attention_mask, input_shape, inputs_embeds, past_key_values_length):
# create causal mask
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
combined_attention_mask = None
if input_shape[-1] > 1:
combined_attention_mask = _make_causal_mask(
input_shape, inputs_embeds.dtype, past_key_values_length=past_key_values_length
).to(inputs_embeds.device)
if attention_mask is not None:
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
expanded_attn_mask = _expand_mask(attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]).to(
inputs_embeds.device
)
combined_attention_mask = (
expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask + combined_attention_mask
)
return combined_attention_mask
def forward(
self,
input_ids=None,
attention_mask=None,
encoder_hidden_states=None,
encoder_attention_mask=None,
head_mask=None,
cross_attn_head_mask=None,
past_key_values=None,
inputs_embeds=None,
use_cache=None,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
of the decoder.
encoder_attention_mask (`torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional*):
Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values
selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
head_mask (`torch.Tensor` of shape `(encoder_layers, encoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules in the encoder. Mask values selected in `[0,
1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules in the decoder to avoid performing
cross-attention on hidden heads. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those
that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of
all `decoder_input_ids` of shape `(batch_size, sequence_length)`. inputs_embeds (`torch.FloatTensor` of
shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing
`input_ids` you can choose to directly pass an embedded representation. This is useful if you want more
control over how to convert `input_ids` indices into associated vectors than the model's internal
embedding lookup matrix.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
use_cache = use_cache if use_cache is not None else self.config.use_cache
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# retrieve input_ids and inputs_embeds
if input_ids is not None and inputs_embeds is not None:
raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
elif input_ids is not None:
input_shape = input_ids.size()
input_ids = input_ids.view(-1, input_shape[-1])
elif inputs_embeds is not None:
input_shape = inputs_embeds.size()[:-1]
else:
raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")
# past_key_values_length
past_key_values_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0
if inputs_embeds is None:
inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale
attention_mask = self._prepare_decoder_attention_mask(
attention_mask, input_shape, inputs_embeds, past_key_values_length
)
# expand encoder attention mask
if encoder_hidden_states is not None and encoder_attention_mask is not None:
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
encoder_attention_mask = _expand_mask(encoder_attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1])
# embed positions
positions = self.embed_positions(input_shape, past_key_values_length)
hidden_states = inputs_embeds + positions
hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)
# decoder layers
all_hidden_states = () if output_hidden_states else None
all_self_attns = () if output_attentions else None
all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None
next_decoder_cache = () if use_cache else None
# check if head_mask/cross_attn_head_mask has a correct number of layers specified if desired
for attn_mask, mask_name in zip([head_mask, cross_attn_head_mask], ["head_mask", "cross_attn_head_mask"]):
if attn_mask is not None:
if attn_mask.size()[0] != len(self.layers):
raise ValueError(
f"The `{mask_name}` should be specified for {len(self.layers)} layers, but it is for"
f" {head_mask.size()[0]}."
)
for idx, decoder_layer in enumerate(self.layers):
# add LayerDrop (see https://arxiv.org/abs/1909.11556 for description)
if output_hidden_states:
all_hidden_states += (hidden_states,)
dropout_probability = random.uniform(0, 1)
if self.training and (dropout_probability < self.layerdrop):
continue
past_key_value = past_key_values[idx] if past_key_values is not None else None
if self.gradient_checkpointing and self.training:
if use_cache:
logger.warning(
"`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
)
use_cache = False
def create_custom_forward(module):
def custom_forward(*inputs):
# None for past_key_value
return module(*inputs, output_attentions, use_cache)
return custom_forward
layer_outputs = torch.utils.checkpoint.checkpoint(
create_custom_forward(decoder_layer),
hidden_states,
attention_mask,
encoder_hidden_states,
encoder_attention_mask,
head_mask[idx] if head_mask is not None else None,
cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None,
None,
)
else:
layer_outputs = decoder_layer(
hidden_states,
attention_mask=attention_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
layer_head_mask=(head_mask[idx] if head_mask is not None else None),
cross_attn_layer_head_mask=(
cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None
),
past_key_value=past_key_value,
output_attentions=output_attentions,
use_cache=use_cache,
)
hidden_states = layer_outputs[0]
if use_cache:
next_decoder_cache += (layer_outputs[3 if output_attentions else 1],)
if output_attentions:
all_self_attns += (layer_outputs[1],)
if encoder_hidden_states is not None:
all_cross_attentions += (layer_outputs[2],)
# add final layer norm
hidden_states = self.layer_norm(hidden_states)
# add hidden states from the last decoder layer
if output_hidden_states:
all_hidden_states += (hidden_states,)
next_cache = next_decoder_cache if use_cache else None
if not return_dict:
return tuple(
v
for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, all_cross_attentions]
if v is not None
)
return BaseModelOutputWithPastAndCrossAttentions(
last_hidden_state=hidden_states,
past_key_values=next_cache,
hidden_states=all_hidden_states,
attentions=all_self_attns,
cross_attentions=all_cross_attentions,
)
@add_start_docstrings(
"The bare Blenderbot Model outputting raw hidden-states without any specific head on top.",
BLENDERBOT_START_DOCSTRING,
)
class BlenderbotModel(BlenderbotPreTrainedModel):
_keys_to_ignore_on_load_missing = ["decoder.embed_tokens.weight", "encoder.embed_tokens.weight"]
def __init__(self, config: BlenderbotConfig):
super().__init__(config)
padding_idx, vocab_size = config.pad_token_id, config.vocab_size
self.shared = nn.Embedding(vocab_size, config.d_model, padding_idx)
self.encoder = BlenderbotEncoder(config, self.shared)
self.decoder = BlenderbotDecoder(config, self.shared)
# Initialize weights and apply final processing
self.post_init()
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
if pretrained_model_name_or_path == "facebook/blenderbot-90M":
warnings.warn(
"The checkpoint `facebook/blenderbot-90M` is deprecated. In the future, please use the identical"
" checkpoint `facebook/small_blenderbot-90M` with"
" `BlenderbotSmallModel.from_pretrained('facebook/small_blenderbot-90M')` instead.",
FutureWarning,
)
return BlenderbotSmallModel.from_pretrained(pretrained_model_name_or_path)
return super(BlenderbotModel, cls).from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
def get_input_embeddings(self):
return self.shared
def set_input_embeddings(self, value):
self.shared = value
self.encoder.embed_tokens = self.shared
self.decoder.embed_tokens = self.shared
def get_encoder(self):
return self.encoder
def get_decoder(self):
return self.decoder
@add_start_docstrings_to_model_forward(BLENDERBOT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=Seq2SeqModelOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.LongTensor] = None,
head_mask: Optional[torch.Tensor] = None,
decoder_head_mask: Optional[torch.Tensor] = None,
cross_attn_head_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Union[Tuple, BaseModelOutput]] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.Tensor] = None,
decoder_inputs_embeds: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple[torch.FloatTensor], Seq2SeqModelOutput]:
r"""
Returns:
Example:
```python
>>> from transformers import BlenderbotTokenizer, BlenderbotModel
>>> model = BlenderbotModel.from_pretrained("facebook/blenderbot-400M-distill")
>>> tokenizer = BlenderbotTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
>>> inputs = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt")
>>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids # Batch size 1
>>> outputs = model(input_ids=inputs.input_ids, decoder_input_ids=decoder_input_ids)
>>> last_hidden_states = outputs.last_hidden_state
>>> list(last_hidden_states.shape)
[1, 6, 1280]
```"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
use_cache = use_cache if use_cache is not None else self.config.use_cache
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if encoder_outputs is None:
encoder_outputs = self.encoder(
input_ids=input_ids,
attention_mask=attention_mask,
head_mask=head_mask,
inputs_embeds=inputs_embeds,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
# If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOutput when return_dict=True
elif return_dict and not isinstance(encoder_outputs, BaseModelOutput):
encoder_outputs = BaseModelOutput(
last_hidden_state=encoder_outputs[0],
hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None,
attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None,
)
# decoder outputs consists of (dec_features, past_key_value, dec_hidden, dec_attn)
decoder_outputs = self.decoder(
input_ids=decoder_input_ids,
attention_mask=decoder_attention_mask,
encoder_hidden_states=encoder_outputs[0],
encoder_attention_mask=attention_mask,
head_mask=decoder_head_mask,
cross_attn_head_mask=cross_attn_head_mask,
past_key_values=past_key_values,
inputs_embeds=decoder_inputs_embeds,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
if not return_dict:
return decoder_outputs + encoder_outputs
return Seq2SeqModelOutput(
last_hidden_state=decoder_outputs.last_hidden_state,
past_key_values=decoder_outputs.past_key_values,
decoder_hidden_states=decoder_outputs.hidden_states,
decoder_attentions=decoder_outputs.attentions,
cross_attentions=decoder_outputs.cross_attentions,
encoder_last_hidden_state=encoder_outputs.last_hidden_state,
encoder_hidden_states=encoder_outputs.hidden_states,
encoder_attentions=encoder_outputs.attentions,
)
@add_start_docstrings(
"The Blenderbot Model with a language modeling head. Can be used for summarization.", BLENDERBOT_START_DOCSTRING
)
class BlenderbotForConditionalGeneration(BlenderbotPreTrainedModel):
base_model_prefix = "model"
_keys_to_ignore_on_load_missing = [
r"final_logits_bias",
r"encoder.version",
r"decoder.version",
r"lm_head.weight",
"decoder.embed_tokens.weight",
"encoder.embed_tokens.weight",
]
def __init__(self, config: BlenderbotConfig):
super().__init__(config)
self.model = BlenderbotModel(config)
self.register_buffer("final_logits_bias", torch.zeros((1, self.model.shared.num_embeddings)))
self.lm_head = nn.Linear(config.d_model, self.model.shared.num_embeddings, bias=False)
# Initialize weights and apply final processing
self.post_init()
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
if pretrained_model_name_or_path == "facebook/blenderbot-90M":
warnings.warn(
"The checkpoint `facebook/blenderbot-90M` is deprecated. In the future, please use the identical"
" checkpoint `facebook/small_blenderbot-90M` with"
" `BlenderbotSmallForConditionalGeneration.from_pretrained('facebook/small_blenderbot-90M')` instead.",
FutureWarning,
)
return BlenderbotSmallForConditionalGeneration.from_pretrained(pretrained_model_name_or_path)
return super(BlenderbotForConditionalGeneration, cls).from_pretrained(
pretrained_model_name_or_path, *model_args, **kwargs
)
def get_encoder(self):
return self.model.get_encoder()
def get_decoder(self):
return self.model.get_decoder()
def resize_token_embeddings(self, new_num_tokens: int) -> nn.Embedding:
new_embeddings = super().resize_token_embeddings(new_num_tokens)
self._resize_final_logits_bias(new_num_tokens)
return new_embeddings
def _resize_final_logits_bias(self, new_num_tokens: int) -> None:
old_num_tokens = self.final_logits_bias.shape[-1]
if new_num_tokens <= old_num_tokens:
new_bias = self.final_logits_bias[:, :new_num_tokens]
else:
extra_bias = torch.zeros((1, new_num_tokens - old_num_tokens), device=self.final_logits_bias.device)
new_bias = torch.cat([self.final_logits_bias, extra_bias], dim=1)
self.register_buffer("final_logits_bias", new_bias)
def get_output_embeddings(self):
return self.lm_head
def set_output_embeddings(self, new_embeddings):
self.lm_head = new_embeddings
@add_start_docstrings_to_model_forward(BLENDERBOT_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=Seq2SeqLMOutput, config_class=_CONFIG_FOR_DOC)
@add_end_docstrings(BLENDERBOT_GENERATION_EXAMPLE)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.LongTensor] = None,
head_mask: Optional[torch.Tensor] = None,
decoder_head_mask: Optional[torch.Tensor] = None,
cross_attn_head_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Union[Tuple, BaseModelOutput]] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.Tensor] = None,
decoder_inputs_embeds: Optional[torch.FloatTensor] = None,
labels: Optional[torch.LongTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple[torch.FloatTensor], Seq2SeqLMOutput]:
r"""
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
Returns:
"""
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
if labels is not None:
if use_cache:
logger.warning("The `use_cache` argument is changed to `False` since `labels` is provided.")
use_cache = False
if decoder_input_ids is None and decoder_inputs_embeds is None:
decoder_input_ids = shift_tokens_right(
labels, self.config.pad_token_id, self.config.decoder_start_token_id
)
outputs = self.model(
input_ids,
attention_mask=attention_mask,
decoder_input_ids=decoder_input_ids,
encoder_outputs=encoder_outputs,
decoder_attention_mask=decoder_attention_mask,
head_mask=head_mask,
decoder_head_mask=decoder_head_mask,
cross_attn_head_mask=cross_attn_head_mask,
past_key_values=past_key_values,
inputs_embeds=inputs_embeds,
decoder_inputs_embeds=decoder_inputs_embeds,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
lm_logits = self.lm_head(outputs[0]) + self.final_logits_bias
masked_lm_loss = None
if labels is not None:
loss_fct = CrossEntropyLoss()
masked_lm_loss = loss_fct(lm_logits.view(-1, self.config.vocab_size), labels.view(-1))
if not return_dict:
output = (lm_logits,) + outputs[1:]
return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output
return Seq2SeqLMOutput(
loss=masked_lm_loss,
logits=lm_logits,
past_key_values=outputs.past_key_values,
decoder_hidden_states=outputs.decoder_hidden_states,
decoder_attentions=outputs.decoder_attentions,
cross_attentions=outputs.cross_attentions,
encoder_last_hidden_state=outputs.encoder_last_hidden_state,
encoder_hidden_states=outputs.encoder_hidden_states,
encoder_attentions=outputs.encoder_attentions,
)
def prepare_inputs_for_generation(
self,
decoder_input_ids,
past=None,
attention_mask=None,
head_mask=None,
decoder_head_mask=None,
cross_attn_head_mask=None,
use_cache=None,
encoder_outputs=None,
**kwargs
):
# cut decoder_input_ids if past is used
if past is not None:
decoder_input_ids = decoder_input_ids[:, -1:]
return {
"input_ids": None, # encoder_outputs is defined. input_ids not needed
"encoder_outputs": encoder_outputs,
"past_key_values": past,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"head_mask": head_mask,
"decoder_head_mask": decoder_head_mask,
"cross_attn_head_mask": cross_attn_head_mask,
"use_cache": use_cache, # change this to avoid caching (presumably for debugging)
}
@staticmethod
def _reorder_cache(past, beam_idx):
reordered_past = ()
for layer_past in past:
# cached cross_attention states don't have to be reordered -> they are always the same
reordered_past += (
tuple(past_state.index_select(0, beam_idx) for past_state in layer_past[:2]) + layer_past[2:],
)
return reordered_past
# Copied from transformers.models.bart.modeling_bart.BartDecoderWrapper with Bart->Blenderbot
class BlenderbotDecoderWrapper(BlenderbotPreTrainedModel):
"""
This wrapper class is a helper class to correctly load pretrained checkpoints when the causal language model is
used in combination with the [`EncoderDecoderModel`] framework.
"""
def __init__(self, config):
super().__init__(config)
self.decoder = BlenderbotDecoder(config)
def forward(self, *args, **kwargs):
return self.decoder(*args, **kwargs)
# Copied from transformers.models.bart.modeling_bart.BartForCausalLM with Bart->Blenderbot, facebook/bart-base->facebook/blenderbot-400M-distill
class BlenderbotForCausalLM(BlenderbotPreTrainedModel):
_keys_to_ignore_on_load_missing = ["lm_head.weight"]
def __init__(self, config):
config = copy.deepcopy(config)
config.is_decoder = True
config.is_encoder_decoder = False
super().__init__(config)
self.model = BlenderbotDecoderWrapper(config)
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
# Initialize weights and apply final processing
self.post_init()
def get_input_embeddings(self):
return self.model.decoder.embed_tokens
def set_input_embeddings(self, value):
self.model.decoder.embed_tokens = value
def get_output_embeddings(self):
return self.lm_head
def set_output_embeddings(self, new_embeddings):
self.lm_head = new_embeddings
def set_decoder(self, decoder):
self.model.decoder = decoder
def get_decoder(self):
return self.model.decoder
@replace_return_docstrings(output_type=CausalLMOutputWithCrossAttentions, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: torch.LongTensor = None,
attention_mask: Optional[torch.Tensor] = None,
encoder_hidden_states: Optional[torch.FloatTensor] = None,
encoder_attention_mask: Optional[torch.FloatTensor] = None,
head_mask: Optional[torch.Tensor] = None,
cross_attn_head_mask: Optional[torch.Tensor] = None,
past_key_values: Optional[List[torch.FloatTensor]] = None,
inputs_embeds: Optional[torch.FloatTensor] = None,
labels: Optional[torch.LongTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
return_dict: Optional[bool] = None,
) -> Union[Tuple, CausalLMOutputWithCrossAttentions]:
r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you
provide it.
Indices can be obtained using [`BlenderbotTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
if the model is configured as a decoder.
encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used
in the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*):
Mask to nullify selected heads of the cross-attention modules. Mask values selected in `[0, 1]`:
- 1 indicates the head is **not masked**,
- 0 indicates the head is **masked**.
past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of
shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of
shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. The two additional
tensors are only required when the model is used as a decoder in a Sequence to Sequence model.
Contains pre-computed hidden-states (key and values in the self-attention blocks and in the
cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding.
If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those
that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of
all `decoder_input_ids` of shape `(batch_size, sequence_length)`.
labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
(masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
use_cache (`bool`, *optional*):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
(see `past_key_values`).
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under
returned tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
for more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
Returns:
Example:
```python
>>> from transformers import BlenderbotTokenizer, BlenderbotForCausalLM
>>> tokenizer = BlenderbotTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
>>> model = BlenderbotForCausalLM.from_pretrained(
... "facebook/blenderbot-400M-distill", add_cross_attention=False
... )
>>> assert model.config.is_decoder, f"{model.__class__} has to be configured as a decoder."
>>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt")
>>> outputs = model(**inputs)
>>> logits = outputs.logits
>>> expected_shape = [1, inputs.input_ids.shape[-1], model.config.vocab_size]
>>> list(logits.shape) == expected_shape
True
```"""
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
outputs = self.model.decoder(
input_ids=input_ids,
attention_mask=attention_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
head_mask=head_mask,
cross_attn_head_mask=cross_attn_head_mask,
past_key_values=past_key_values,
inputs_embeds=inputs_embeds,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
logits = self.lm_head(outputs[0])
loss = None
if labels is not None:
loss_fct = CrossEntropyLoss()
loss = loss_fct(logits.view(-1, self.config.vocab_size), labels.view(-1))
if not return_dict:
output = (logits,) + outputs[1:]
return (loss,) + output if loss is not None else output
return CausalLMOutputWithCrossAttentions(
loss=loss,
logits=logits,
past_key_values=outputs.past_key_values,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
cross_attentions=outputs.cross_attentions,
)
def prepare_inputs_for_generation(self, input_ids, past=None, attention_mask=None, use_cache=None, **kwargs):
# if model is used as a decoder in encoder-decoder model, the decoder attention mask is created on the fly
if attention_mask is None:
attention_mask = input_ids.new_ones(input_ids.shape)
if past:
input_ids = input_ids[:, -1:]
# first step, decoder_cached_states are empty
return {
"input_ids": input_ids, # encoder_outputs is defined. input_ids not needed
"attention_mask": attention_mask,
"past_key_values": past,
"use_cache": use_cache,
}
@staticmethod
def _reorder_cache(past, beam_idx):
reordered_past = ()
for layer_past in past:
reordered_past += (tuple(past_state.index_select(0, beam_idx) for past_state in layer_past),)
return reordered_past
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/data/datasets/glue.py | # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
import time
import warnings
from dataclasses import dataclass, field
from enum import Enum
from typing import List, Optional, Union
import torch
from torch.utils.data import Dataset
from filelock import FileLock
from ...tokenization_utils_base import PreTrainedTokenizerBase
from ...utils import logging
from ..processors.glue import glue_convert_examples_to_features, glue_output_modes, glue_processors
from ..processors.utils import InputFeatures
logger = logging.get_logger(__name__)
@dataclass
class GlueDataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
Using `HfArgumentParser` we can turn this class into argparse arguments to be able to specify them on the command
line.
"""
task_name: str = field(metadata={"help": "The name of the task to train on: " + ", ".join(glue_processors.keys())})
data_dir: str = field(
metadata={"help": "The input data dir. Should contain the .tsv files (or other data files) for the task."}
)
max_seq_length: int = field(
default=128,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
def __post_init__(self):
self.task_name = self.task_name.lower()
class Split(Enum):
train = "train"
dev = "dev"
test = "test"
class GlueDataset(Dataset):
"""
This will be superseded by a framework-agnostic approach soon.
"""
args: GlueDataTrainingArguments
output_mode: str
features: List[InputFeatures]
def __init__(
self,
args: GlueDataTrainingArguments,
tokenizer: PreTrainedTokenizerBase,
limit_length: Optional[int] = None,
mode: Union[str, Split] = Split.train,
cache_dir: Optional[str] = None,
):
warnings.warn(
"This dataset will be removed from the library soon, preprocessing should be handled with the 🤗 Datasets "
"library. You can have a look at this example script for pointers: "
"https://github.com/huggingface/transformers/blob/main/examples/pytorch/text-classification/run_glue.py",
FutureWarning,
)
self.args = args
self.processor = glue_processors[args.task_name]()
self.output_mode = glue_output_modes[args.task_name]
if isinstance(mode, str):
try:
mode = Split[mode]
except KeyError:
raise KeyError("mode is not a valid split name")
# Load data features from cache or dataset file
cached_features_file = os.path.join(
cache_dir if cache_dir is not None else args.data_dir,
f"cached_{mode.value}_{tokenizer.__class__.__name__}_{args.max_seq_length}_{args.task_name}",
)
label_list = self.processor.get_labels()
if args.task_name in ["mnli", "mnli-mm"] and tokenizer.__class__.__name__ in (
"RobertaTokenizer",
"RobertaTokenizerFast",
"XLMRobertaTokenizer",
"BartTokenizer",
"BartTokenizerFast",
):
# HACK(label indices are swapped in RoBERTa pretrained model)
label_list[1], label_list[2] = label_list[2], label_list[1]
self.label_list = label_list
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
lock_path = cached_features_file + ".lock"
with FileLock(lock_path):
if os.path.exists(cached_features_file) and not args.overwrite_cache:
start = time.time()
self.features = torch.load(cached_features_file)
logger.info(
f"Loading features from cached file {cached_features_file} [took %.3f s]", time.time() - start
)
else:
logger.info(f"Creating features from dataset file at {args.data_dir}")
if mode == Split.dev:
examples = self.processor.get_dev_examples(args.data_dir)
elif mode == Split.test:
examples = self.processor.get_test_examples(args.data_dir)
else:
examples = self.processor.get_train_examples(args.data_dir)
if limit_length is not None:
examples = examples[:limit_length]
self.features = glue_convert_examples_to_features(
examples,
tokenizer,
max_length=args.max_seq_length,
label_list=label_list,
output_mode=self.output_mode,
)
start = time.time()
torch.save(self.features, cached_features_file)
# ^ This seems to take a lot of time so I want to investigate why and how we can improve.
logger.info(
f"Saving features into cached file {cached_features_file} [took {time.time() - start:.3f} s]"
)
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
def get_labels(self):
return self.label_list
| # Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
import time
import warnings
from dataclasses import dataclass, field
from enum import Enum
from typing import List, Optional, Union
import torch
from torch.utils.data import Dataset
from filelock import FileLock
from ...tokenization_utils_base import PreTrainedTokenizerBase
from ...utils import logging
from ..processors.glue import glue_convert_examples_to_features, glue_output_modes, glue_processors
from ..processors.utils import InputFeatures
logger = logging.get_logger(__name__)
@dataclass
class GlueDataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
Using `HfArgumentParser` we can turn this class into argparse arguments to be able to specify them on the command
line.
"""
task_name: str = field(metadata={"help": "The name of the task to train on: " + ", ".join(glue_processors.keys())})
data_dir: str = field(
metadata={"help": "The input data dir. Should contain the .tsv files (or other data files) for the task."}
)
max_seq_length: int = field(
default=128,
metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
)
def __post_init__(self):
self.task_name = self.task_name.lower()
class Split(Enum):
train = "train"
dev = "dev"
test = "test"
class GlueDataset(Dataset):
"""
This will be superseded by a framework-agnostic approach soon.
"""
args: GlueDataTrainingArguments
output_mode: str
features: List[InputFeatures]
def __init__(
self,
args: GlueDataTrainingArguments,
tokenizer: PreTrainedTokenizerBase,
limit_length: Optional[int] = None,
mode: Union[str, Split] = Split.train,
cache_dir: Optional[str] = None,
):
warnings.warn(
"This dataset will be removed from the library soon, preprocessing should be handled with the 🤗 Datasets "
"library. You can have a look at this example script for pointers: "
"https://github.com/huggingface/transformers/blob/main/examples/pytorch/text-classification/run_glue.py",
FutureWarning,
)
self.args = args
self.processor = glue_processors[args.task_name]()
self.output_mode = glue_output_modes[args.task_name]
if isinstance(mode, str):
try:
mode = Split[mode]
except KeyError:
raise KeyError("mode is not a valid split name")
# Load data features from cache or dataset file
cached_features_file = os.path.join(
cache_dir if cache_dir is not None else args.data_dir,
f"cached_{mode.value}_{tokenizer.__class__.__name__}_{args.max_seq_length}_{args.task_name}",
)
label_list = self.processor.get_labels()
if args.task_name in ["mnli", "mnli-mm"] and tokenizer.__class__.__name__ in (
"RobertaTokenizer",
"RobertaTokenizerFast",
"XLMRobertaTokenizer",
"BartTokenizer",
"BartTokenizerFast",
):
# HACK(label indices are swapped in RoBERTa pretrained model)
label_list[1], label_list[2] = label_list[2], label_list[1]
self.label_list = label_list
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
lock_path = cached_features_file + ".lock"
with FileLock(lock_path):
if os.path.exists(cached_features_file) and not args.overwrite_cache:
start = time.time()
self.features = torch.load(cached_features_file)
logger.info(
f"Loading features from cached file {cached_features_file} [took %.3f s]", time.time() - start
)
else:
logger.info(f"Creating features from dataset file at {args.data_dir}")
if mode == Split.dev:
examples = self.processor.get_dev_examples(args.data_dir)
elif mode == Split.test:
examples = self.processor.get_test_examples(args.data_dir)
else:
examples = self.processor.get_train_examples(args.data_dir)
if limit_length is not None:
examples = examples[:limit_length]
self.features = glue_convert_examples_to_features(
examples,
tokenizer,
max_length=args.max_seq_length,
label_list=label_list,
output_mode=self.output_mode,
)
start = time.time()
torch.save(self.features, cached_features_file)
# ^ This seems to take a lot of time so I want to investigate why and how we can improve.
logger.info(
f"Saving features into cached file {cached_features_file} [took {time.time() - start:.3f} s]"
)
def __len__(self):
return len(self.features)
def __getitem__(self, i) -> InputFeatures:
return self.features[i]
def get_labels(self):
return self.label_list
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/electra/__init__.py | # flake8: noqa
# There's no way to ignore "F401 '...' imported but unused" warnings in this
# module, but to preserve other warnings. So, don't check this module at all.
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
_import_structure = {
"configuration_electra": ["ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP", "ElectraConfig", "ElectraOnnxConfig"],
"tokenization_electra": ["ElectraTokenizer"],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["tokenization_electra_fast"] = ["ElectraTokenizerFast"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_electra"] = [
"ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"ElectraForCausalLM",
"ElectraForMaskedLM",
"ElectraForMultipleChoice",
"ElectraForPreTraining",
"ElectraForQuestionAnswering",
"ElectraForSequenceClassification",
"ElectraForTokenClassification",
"ElectraModel",
"ElectraPreTrainedModel",
"load_tf_weights_in_electra",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tf_electra"] = [
"TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFElectraForMaskedLM",
"TFElectraForMultipleChoice",
"TFElectraForPreTraining",
"TFElectraForQuestionAnswering",
"TFElectraForSequenceClassification",
"TFElectraForTokenClassification",
"TFElectraModel",
"TFElectraPreTrainedModel",
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_flax_electra"] = [
"FlaxElectraForCausalLM",
"FlaxElectraForMaskedLM",
"FlaxElectraForMultipleChoice",
"FlaxElectraForPreTraining",
"FlaxElectraForQuestionAnswering",
"FlaxElectraForSequenceClassification",
"FlaxElectraForTokenClassification",
"FlaxElectraModel",
"FlaxElectraPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_electra import ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, ElectraConfig, ElectraOnnxConfig
from .tokenization_electra import ElectraTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_electra_fast import ElectraTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_electra import (
ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
ElectraForCausalLM,
ElectraForMaskedLM,
ElectraForMultipleChoice,
ElectraForPreTraining,
ElectraForQuestionAnswering,
ElectraForSequenceClassification,
ElectraForTokenClassification,
ElectraModel,
ElectraPreTrainedModel,
load_tf_weights_in_electra,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_electra import (
TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFElectraForMaskedLM,
TFElectraForMultipleChoice,
TFElectraForPreTraining,
TFElectraForQuestionAnswering,
TFElectraForSequenceClassification,
TFElectraForTokenClassification,
TFElectraModel,
TFElectraPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_electra import (
FlaxElectraForCausalLM,
FlaxElectraForMaskedLM,
FlaxElectraForMultipleChoice,
FlaxElectraForPreTraining,
FlaxElectraForQuestionAnswering,
FlaxElectraForSequenceClassification,
FlaxElectraForTokenClassification,
FlaxElectraModel,
FlaxElectraPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| # flake8: noqa
# There's no way to ignore "F401 '...' imported but unused" warnings in this
# module, but to preserve other warnings. So, don't check this module at all.
# Copyright 2020 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
_import_structure = {
"configuration_electra": ["ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP", "ElectraConfig", "ElectraOnnxConfig"],
"tokenization_electra": ["ElectraTokenizer"],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["tokenization_electra_fast"] = ["ElectraTokenizerFast"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_electra"] = [
"ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"ElectraForCausalLM",
"ElectraForMaskedLM",
"ElectraForMultipleChoice",
"ElectraForPreTraining",
"ElectraForQuestionAnswering",
"ElectraForSequenceClassification",
"ElectraForTokenClassification",
"ElectraModel",
"ElectraPreTrainedModel",
"load_tf_weights_in_electra",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_tf_electra"] = [
"TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST",
"TFElectraForMaskedLM",
"TFElectraForMultipleChoice",
"TFElectraForPreTraining",
"TFElectraForQuestionAnswering",
"TFElectraForSequenceClassification",
"TFElectraForTokenClassification",
"TFElectraModel",
"TFElectraPreTrainedModel",
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_import_structure["modeling_flax_electra"] = [
"FlaxElectraForCausalLM",
"FlaxElectraForMaskedLM",
"FlaxElectraForMultipleChoice",
"FlaxElectraForPreTraining",
"FlaxElectraForQuestionAnswering",
"FlaxElectraForSequenceClassification",
"FlaxElectraForTokenClassification",
"FlaxElectraModel",
"FlaxElectraPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_electra import ELECTRA_PRETRAINED_CONFIG_ARCHIVE_MAP, ElectraConfig, ElectraOnnxConfig
from .tokenization_electra import ElectraTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_electra_fast import ElectraTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_electra import (
ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
ElectraForCausalLM,
ElectraForMaskedLM,
ElectraForMultipleChoice,
ElectraForPreTraining,
ElectraForQuestionAnswering,
ElectraForSequenceClassification,
ElectraForTokenClassification,
ElectraModel,
ElectraPreTrainedModel,
load_tf_weights_in_electra,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_electra import (
TF_ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFElectraForMaskedLM,
TFElectraForMultipleChoice,
TFElectraForPreTraining,
TFElectraForQuestionAnswering,
TFElectraForSequenceClassification,
TFElectraForTokenClassification,
TFElectraModel,
TFElectraPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_electra import (
FlaxElectraForCausalLM,
FlaxElectraForMaskedLM,
FlaxElectraForMultipleChoice,
FlaxElectraForPreTraining,
FlaxElectraForQuestionAnswering,
FlaxElectraForSequenceClassification,
FlaxElectraForTokenClassification,
FlaxElectraModel,
FlaxElectraPreTrainedModel,
)
else:
import sys
sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/xlnet/configuration_xlnet.py | # coding=utf-8
# Copyright 2018 Google AI, Google Brain and Carnegie Mellon University Authors and the HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" XLNet configuration"""
import warnings
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"xlnet-base-cased": "https://huggingface.co/xlnet-base-cased/resolve/main/config.json",
"xlnet-large-cased": "https://huggingface.co/xlnet-large-cased/resolve/main/config.json",
}
class XLNetConfig(PretrainedConfig):
"""
This is the configuration class to store the configuration of a [`XLNetModel`] or a [`TFXLNetModel`]. It is used to
instantiate a XLNet model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the
[xlnet-large-cased](https://huggingface.co/xlnet-large-cased) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 32000):
Vocabulary size of the XLNet model. Defines the number of different tokens that can be represented by the
`inputs_ids` passed when calling [`XLNetModel`] or [`TFXLNetModel`].
d_model (`int`, *optional*, defaults to 1024):
Dimensionality of the encoder layers and the pooler layer.
n_layer (`int`, *optional*, defaults to 24):
Number of hidden layers in the Transformer encoder.
n_head (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
d_inner (`int`, *optional*, defaults to 4096):
Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
ff_activation (`str` or `Callable`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the If string, `"gelu"`, `"relu"`, `"silu"` and
`"gelu_new"` are supported.
untie_r (`bool`, *optional*, defaults to `True`):
Whether or not to untie relative position biases
attn_type (`str`, *optional*, defaults to `"bi"`):
The attention type used by the model. Set `"bi"` for XLNet, `"uni"` for Transformer-XL.
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
mem_len (`int` or `None`, *optional*):
The number of tokens to cache. The key/value pairs that have already been pre-computed in a previous
forward pass won't be re-computed. See the
[quickstart](https://huggingface.co/transformers/quickstart.html#using-the-past) for more information.
reuse_len (`int`, *optional*):
The number of tokens in the current batch to be cached and reused in the future.
bi_data (`bool`, *optional*, defaults to `False`):
Whether or not to use bidirectional input pipeline. Usually set to `True` during pretraining and `False`
during finetuning.
clamp_len (`int`, *optional*, defaults to -1):
Clamp all relative distances larger than clamp_len. Setting this attribute to -1 means no clamping.
same_length (`bool`, *optional*, defaults to `False`):
Whether or not to use the same attention length for each token.
summary_type (`str`, *optional*, defaults to "last"):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Has to be one of the following options:
- `"last"`: Take the last token hidden state (like XLNet).
- `"first"`: Take the first token hidden state (like BERT).
- `"mean"`: Take the mean of all tokens hidden states.
- `"cls_index"`: Supply a Tensor of classification token position (like GPT/GPT-2).
- `"attn"`: Not implemented now, use multi-head attention.
summary_use_proj (`bool`, *optional*, defaults to `True`):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Whether or not to add a projection after the vector extraction.
summary_activation (`str`, *optional*):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Pass `"tanh"` for a tanh activation to the output, any other value will result in no activation.
summary_proj_to_labels (`boo`, *optional*, defaults to `True`):
Used in the sequence classification and multiple choice models.
Whether the projection outputs should have `config.num_labels` or `config.hidden_size` classes.
summary_last_dropout (`float`, *optional*, defaults to 0.1):
Used in the sequence classification and multiple choice models.
The dropout ratio to be used after the projection and activation.
start_n_top (`int`, *optional*, defaults to 5):
Used in the SQuAD evaluation script.
end_n_top (`int`, *optional*, defaults to 5):
Used in the SQuAD evaluation script.
use_mems_eval (`bool`, *optional*, defaults to `True`):
Whether or not the model should make use of the recurrent memory mechanism in evaluation mode.
use_mems_train (`bool`, *optional*, defaults to `False`):
Whether or not the model should make use of the recurrent memory mechanism in train mode.
<Tip>
For pretraining, it is recommended to set `use_mems_train` to `True`. For fine-tuning, it is recommended to
set `use_mems_train` to `False` as discussed
[here](https://github.com/zihangdai/xlnet/issues/41#issuecomment-505102587). If `use_mems_train` is set to
`True`, one has to make sure that the train batches are correctly pre-processed, *e.g.* `batch_1 = [[This
line is], [This is the]]` and `batch_2 = [[ the first line], [ second line]]` and that all batches are of
equal size.
</Tip>
Examples:
```python
>>> from transformers import XLNetConfig, XLNetModel
>>> # Initializing a XLNet configuration
>>> configuration = XLNetConfig()
>>> # Initializing a model (with random weights) from the configuration
>>> model = XLNetModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "xlnet"
keys_to_ignore_at_inference = ["mems"]
attribute_map = {
"n_token": "vocab_size", # Backward compatibility
"hidden_size": "d_model",
"num_attention_heads": "n_head",
"num_hidden_layers": "n_layer",
}
def __init__(
self,
vocab_size=32000,
d_model=1024,
n_layer=24,
n_head=16,
d_inner=4096,
ff_activation="gelu",
untie_r=True,
attn_type="bi",
initializer_range=0.02,
layer_norm_eps=1e-12,
dropout=0.1,
mem_len=512,
reuse_len=None,
use_mems_eval=True,
use_mems_train=False,
bi_data=False,
clamp_len=-1,
same_length=False,
summary_type="last",
summary_use_proj=True,
summary_activation="tanh",
summary_last_dropout=0.1,
start_n_top=5,
end_n_top=5,
pad_token_id=5,
bos_token_id=1,
eos_token_id=2,
**kwargs
):
"""Constructs XLNetConfig."""
self.vocab_size = vocab_size
self.d_model = d_model
self.n_layer = n_layer
self.n_head = n_head
if d_model % n_head != 0:
raise ValueError(f"'d_model % n_head' ({d_model % n_head}) should be equal to 0")
if "d_head" in kwargs:
if kwargs["d_head"] != d_model // n_head:
raise ValueError(
f"`d_head` ({kwargs['d_head']}) should be equal to `d_model // n_head` ({d_model // n_head})"
)
self.d_head = d_model // n_head
self.ff_activation = ff_activation
self.d_inner = d_inner
self.untie_r = untie_r
self.attn_type = attn_type
self.initializer_range = initializer_range
self.layer_norm_eps = layer_norm_eps
self.dropout = dropout
self.mem_len = mem_len
self.reuse_len = reuse_len
self.bi_data = bi_data
self.clamp_len = clamp_len
self.same_length = same_length
self.summary_type = summary_type
self.summary_use_proj = summary_use_proj
self.summary_activation = summary_activation
self.summary_last_dropout = summary_last_dropout
self.start_n_top = start_n_top
self.end_n_top = end_n_top
self.bos_token_id = bos_token_id
self.pad_token_id = pad_token_id
self.eos_token_id = eos_token_id
if "use_cache" in kwargs:
warnings.warn(
"The `use_cache` argument is deprecated and will be removed in a future version, use `use_mems_eval`"
" instead.",
FutureWarning,
)
use_mems_eval = kwargs["use_cache"]
self.use_mems_eval = use_mems_eval
self.use_mems_train = use_mems_train
super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
@property
def max_position_embeddings(self):
logger.info(f"The model {self.model_type} is one of the few models that has no sequence length limit.")
return -1
@max_position_embeddings.setter
def max_position_embeddings(self, value):
# Message copied from Transformer-XL documentation
raise NotImplementedError(
f"The model {self.model_type} is one of the few models that has no sequence length limit."
)
| # coding=utf-8
# Copyright 2018 Google AI, Google Brain and Carnegie Mellon University Authors and the HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" XLNet configuration"""
import warnings
from ...configuration_utils import PretrainedConfig
from ...utils import logging
logger = logging.get_logger(__name__)
XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP = {
"xlnet-base-cased": "https://huggingface.co/xlnet-base-cased/resolve/main/config.json",
"xlnet-large-cased": "https://huggingface.co/xlnet-large-cased/resolve/main/config.json",
}
class XLNetConfig(PretrainedConfig):
"""
This is the configuration class to store the configuration of a [`XLNetModel`] or a [`TFXLNetModel`]. It is used to
instantiate a XLNet model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the
[xlnet-large-cased](https://huggingface.co/xlnet-large-cased) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 32000):
Vocabulary size of the XLNet model. Defines the number of different tokens that can be represented by the
`inputs_ids` passed when calling [`XLNetModel`] or [`TFXLNetModel`].
d_model (`int`, *optional*, defaults to 1024):
Dimensionality of the encoder layers and the pooler layer.
n_layer (`int`, *optional*, defaults to 24):
Number of hidden layers in the Transformer encoder.
n_head (`int`, *optional*, defaults to 16):
Number of attention heads for each attention layer in the Transformer encoder.
d_inner (`int`, *optional*, defaults to 4096):
Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
ff_activation (`str` or `Callable`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the If string, `"gelu"`, `"relu"`, `"silu"` and
`"gelu_new"` are supported.
untie_r (`bool`, *optional*, defaults to `True`):
Whether or not to untie relative position biases
attn_type (`str`, *optional*, defaults to `"bi"`):
The attention type used by the model. Set `"bi"` for XLNet, `"uni"` for Transformer-XL.
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
layer_norm_eps (`float`, *optional*, defaults to 1e-12):
The epsilon used by the layer normalization layers.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
mem_len (`int` or `None`, *optional*):
The number of tokens to cache. The key/value pairs that have already been pre-computed in a previous
forward pass won't be re-computed. See the
[quickstart](https://huggingface.co/transformers/quickstart.html#using-the-past) for more information.
reuse_len (`int`, *optional*):
The number of tokens in the current batch to be cached and reused in the future.
bi_data (`bool`, *optional*, defaults to `False`):
Whether or not to use bidirectional input pipeline. Usually set to `True` during pretraining and `False`
during finetuning.
clamp_len (`int`, *optional*, defaults to -1):
Clamp all relative distances larger than clamp_len. Setting this attribute to -1 means no clamping.
same_length (`bool`, *optional*, defaults to `False`):
Whether or not to use the same attention length for each token.
summary_type (`str`, *optional*, defaults to "last"):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Has to be one of the following options:
- `"last"`: Take the last token hidden state (like XLNet).
- `"first"`: Take the first token hidden state (like BERT).
- `"mean"`: Take the mean of all tokens hidden states.
- `"cls_index"`: Supply a Tensor of classification token position (like GPT/GPT-2).
- `"attn"`: Not implemented now, use multi-head attention.
summary_use_proj (`bool`, *optional*, defaults to `True`):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Whether or not to add a projection after the vector extraction.
summary_activation (`str`, *optional*):
Argument used when doing sequence summary. Used in the sequence classification and multiple choice models.
Pass `"tanh"` for a tanh activation to the output, any other value will result in no activation.
summary_proj_to_labels (`boo`, *optional*, defaults to `True`):
Used in the sequence classification and multiple choice models.
Whether the projection outputs should have `config.num_labels` or `config.hidden_size` classes.
summary_last_dropout (`float`, *optional*, defaults to 0.1):
Used in the sequence classification and multiple choice models.
The dropout ratio to be used after the projection and activation.
start_n_top (`int`, *optional*, defaults to 5):
Used in the SQuAD evaluation script.
end_n_top (`int`, *optional*, defaults to 5):
Used in the SQuAD evaluation script.
use_mems_eval (`bool`, *optional*, defaults to `True`):
Whether or not the model should make use of the recurrent memory mechanism in evaluation mode.
use_mems_train (`bool`, *optional*, defaults to `False`):
Whether or not the model should make use of the recurrent memory mechanism in train mode.
<Tip>
For pretraining, it is recommended to set `use_mems_train` to `True`. For fine-tuning, it is recommended to
set `use_mems_train` to `False` as discussed
[here](https://github.com/zihangdai/xlnet/issues/41#issuecomment-505102587). If `use_mems_train` is set to
`True`, one has to make sure that the train batches are correctly pre-processed, *e.g.* `batch_1 = [[This
line is], [This is the]]` and `batch_2 = [[ the first line], [ second line]]` and that all batches are of
equal size.
</Tip>
Examples:
```python
>>> from transformers import XLNetConfig, XLNetModel
>>> # Initializing a XLNet configuration
>>> configuration = XLNetConfig()
>>> # Initializing a model (with random weights) from the configuration
>>> model = XLNetModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "xlnet"
keys_to_ignore_at_inference = ["mems"]
attribute_map = {
"n_token": "vocab_size", # Backward compatibility
"hidden_size": "d_model",
"num_attention_heads": "n_head",
"num_hidden_layers": "n_layer",
}
def __init__(
self,
vocab_size=32000,
d_model=1024,
n_layer=24,
n_head=16,
d_inner=4096,
ff_activation="gelu",
untie_r=True,
attn_type="bi",
initializer_range=0.02,
layer_norm_eps=1e-12,
dropout=0.1,
mem_len=512,
reuse_len=None,
use_mems_eval=True,
use_mems_train=False,
bi_data=False,
clamp_len=-1,
same_length=False,
summary_type="last",
summary_use_proj=True,
summary_activation="tanh",
summary_last_dropout=0.1,
start_n_top=5,
end_n_top=5,
pad_token_id=5,
bos_token_id=1,
eos_token_id=2,
**kwargs
):
"""Constructs XLNetConfig."""
self.vocab_size = vocab_size
self.d_model = d_model
self.n_layer = n_layer
self.n_head = n_head
if d_model % n_head != 0:
raise ValueError(f"'d_model % n_head' ({d_model % n_head}) should be equal to 0")
if "d_head" in kwargs:
if kwargs["d_head"] != d_model // n_head:
raise ValueError(
f"`d_head` ({kwargs['d_head']}) should be equal to `d_model // n_head` ({d_model // n_head})"
)
self.d_head = d_model // n_head
self.ff_activation = ff_activation
self.d_inner = d_inner
self.untie_r = untie_r
self.attn_type = attn_type
self.initializer_range = initializer_range
self.layer_norm_eps = layer_norm_eps
self.dropout = dropout
self.mem_len = mem_len
self.reuse_len = reuse_len
self.bi_data = bi_data
self.clamp_len = clamp_len
self.same_length = same_length
self.summary_type = summary_type
self.summary_use_proj = summary_use_proj
self.summary_activation = summary_activation
self.summary_last_dropout = summary_last_dropout
self.start_n_top = start_n_top
self.end_n_top = end_n_top
self.bos_token_id = bos_token_id
self.pad_token_id = pad_token_id
self.eos_token_id = eos_token_id
if "use_cache" in kwargs:
warnings.warn(
"The `use_cache` argument is deprecated and will be removed in a future version, use `use_mems_eval`"
" instead.",
FutureWarning,
)
use_mems_eval = kwargs["use_cache"]
self.use_mems_eval = use_mems_eval
self.use_mems_train = use_mems_train
super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
@property
def max_position_embeddings(self):
logger.info(f"The model {self.model_type} is one of the few models that has no sequence length limit.")
return -1
@max_position_embeddings.setter
def max_position_embeddings(self, value):
# Message copied from Transformer-XL documentation
raise NotImplementedError(
f"The model {self.model_type} is one of the few models that has no sequence length limit."
)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./tests/models/cpm/test_tokenization_cpm.py | # coding=utf-8
# Copyright 2018 HuggingFace Inc. team.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from transformers.models.cpm.tokenization_cpm import CpmTokenizer
from transformers.testing_utils import custom_tokenizers
from ..xlnet.test_modeling_xlnet import XLNetModelTest
@custom_tokenizers
class CpmTokenizationTest(XLNetModelTest):
def test_pre_tokenization(self):
tokenizer = CpmTokenizer.from_pretrained("TsinghuaAI/CPM-Generate")
text = "Hugging Face大法好，谁用谁知道。"
normalized_text = "Hugging Face大法好,谁用谁知道。<unk>"
bpe_tokens = "▁Hu gg ing ▁ ▂ ▁F ace ▁大法 ▁好 ▁ , ▁谁 ▁用 ▁谁 ▁知 道 ▁ 。".split()
tokens = tokenizer.tokenize(text)
self.assertListEqual(tokens, bpe_tokens)
input_tokens = tokens + [tokenizer.unk_token]
input_bpe_tokens = [13789, 13283, 1421, 8, 10, 1164, 13608, 16528, 63, 8, 9, 440, 108, 440, 121, 90, 8, 12, 0]
self.assertListEqual(tokenizer.convert_tokens_to_ids(input_tokens), input_bpe_tokens)
reconstructed_text = tokenizer.decode(input_bpe_tokens)
self.assertEqual(reconstructed_text, normalized_text)
| # coding=utf-8
# Copyright 2018 HuggingFace Inc. team.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from transformers.models.cpm.tokenization_cpm import CpmTokenizer
from transformers.testing_utils import custom_tokenizers
from ..xlnet.test_modeling_xlnet import XLNetModelTest
@custom_tokenizers
class CpmTokenizationTest(XLNetModelTest):
def test_pre_tokenization(self):
tokenizer = CpmTokenizer.from_pretrained("TsinghuaAI/CPM-Generate")
text = "Hugging Face大法好，谁用谁知道。"
normalized_text = "Hugging Face大法好,谁用谁知道。<unk>"
bpe_tokens = "▁Hu gg ing ▁ ▂ ▁F ace ▁大法 ▁好 ▁ , ▁谁 ▁用 ▁谁 ▁知 道 ▁ 。".split()
tokens = tokenizer.tokenize(text)
self.assertListEqual(tokens, bpe_tokens)
input_tokens = tokens + [tokenizer.unk_token]
input_bpe_tokens = [13789, 13283, 1421, 8, 10, 1164, 13608, 16528, 63, 8, 9, 440, 108, 440, 121, 90, 8, 12, 0]
self.assertListEqual(tokenizer.convert_tokens_to_ids(input_tokens), input_bpe_tokens)
reconstructed_text = tokenizer.decode(input_bpe_tokens)
self.assertEqual(reconstructed_text, normalized_text)
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./tests/models/__init__.py | -1 |
||
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./docs/source/en/model_doc/m2m_100.mdx | <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# M2M100
## Overview
The M2M100 model was proposed in [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky,
Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy
Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
The abstract from the paper is the following:
*Existing work in translation demonstrated the potential of massively multilingual machine translation by training a
single model able to translate between any pair of languages. However, much of this work is English-Centric by training
only on data which was translated from or to English. While this is supported by large sources of training data, it
does not reflect translation needs worldwide. In this work, we create a true Many-to-Many multilingual translation
model that can translate directly between any pair of 100 languages. We build and open source a training dataset that
covers thousands of language directions with supervised data, created through large-scale mining. Then, we explore how
to effectively increase model capacity through a combination of dense scaling and language-specific sparse parameters
to create high quality models. Our focus on non-English-Centric models brings gains of more than 10 BLEU when directly
translating between non-English directions while performing competitively to the best single systems of WMT. We
open-source our scripts so that others may reproduce the data, evaluation, and final M2M-100 model.*
This model was contributed by [valhalla](https://huggingface.co/valhalla).
### Training and Generation
M2M100 is a multilingual encoder-decoder (seq-to-seq) model primarily intended for translation tasks. As the model is
multilingual it expects the sequences in a certain format: A special language id token is used as prefix in both the
source and target text. The source text format is `[lang_code] X [eos]`, where `lang_code` is source language
id for source text and target language id for target text, with `X` being the source or target text.
The [`M2M100Tokenizer`] depends on `sentencepiece` so be sure to install it before running the
examples. To install `sentencepiece` run `pip install sentencepiece`.
- Supervised Training
```python
from transformers import M2M100Config, M2M100ForConditionalGeneration, M2M100Tokenizer
model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M")
tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M", src_lang="en", tgt_lang="fr")
src_text = "Life is like a box of chocolates."
tgt_text = "La vie est comme une boîte de chocolat."
model_inputs = tokenizer(src_text, text_target=tgt_text, return_tensors="pt")
loss = model(**model_inputs).loss # forward pass
```
- Generation
M2M100 uses the `eos_token_id` as the `decoder_start_token_id` for generation with the target language id
being forced as the first generated token. To force the target language id as the first generated token, pass the
*forced_bos_token_id* parameter to the *generate* method. The following example shows how to translate between
Hindi to French and Chinese to English using the *facebook/m2m100_418M* checkpoint.
```python
>>> from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
>>> hi_text = "जीवन एक चॉकलेट बॉक्स की तरह है।"
>>> chinese_text = "生活就像一盒巧克力。"
>>> model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M")
>>> tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M")
>>> # translate Hindi to French
>>> tokenizer.src_lang = "hi"
>>> encoded_hi = tokenizer(hi_text, return_tensors="pt")
>>> generated_tokens = model.generate(**encoded_hi, forced_bos_token_id=tokenizer.get_lang_id("fr"))
>>> tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
"La vie est comme une boîte de chocolat."
>>> # translate Chinese to English
>>> tokenizer.src_lang = "zh"
>>> encoded_zh = tokenizer(chinese_text, return_tensors="pt")
>>> generated_tokens = model.generate(**encoded_zh, forced_bos_token_id=tokenizer.get_lang_id("en"))
>>> tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
"Life is like a box of chocolate."
```
## M2M100Config
[[autodoc]] M2M100Config
## M2M100Tokenizer
[[autodoc]] M2M100Tokenizer
- build_inputs_with_special_tokens
- get_special_tokens_mask
- create_token_type_ids_from_sequences
- save_vocabulary
## M2M100Model
[[autodoc]] M2M100Model
- forward
## M2M100ForConditionalGeneration
[[autodoc]] M2M100ForConditionalGeneration
- forward
| <!--Copyright 2020 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
-->
# M2M100
## Overview
The M2M100 model was proposed in [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky,
Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy
Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
The abstract from the paper is the following:
*Existing work in translation demonstrated the potential of massively multilingual machine translation by training a
single model able to translate between any pair of languages. However, much of this work is English-Centric by training
only on data which was translated from or to English. While this is supported by large sources of training data, it
does not reflect translation needs worldwide. In this work, we create a true Many-to-Many multilingual translation
model that can translate directly between any pair of 100 languages. We build and open source a training dataset that
covers thousands of language directions with supervised data, created through large-scale mining. Then, we explore how
to effectively increase model capacity through a combination of dense scaling and language-specific sparse parameters
to create high quality models. Our focus on non-English-Centric models brings gains of more than 10 BLEU when directly
translating between non-English directions while performing competitively to the best single systems of WMT. We
open-source our scripts so that others may reproduce the data, evaluation, and final M2M-100 model.*
This model was contributed by [valhalla](https://huggingface.co/valhalla).
### Training and Generation
M2M100 is a multilingual encoder-decoder (seq-to-seq) model primarily intended for translation tasks. As the model is
multilingual it expects the sequences in a certain format: A special language id token is used as prefix in both the
source and target text. The source text format is `[lang_code] X [eos]`, where `lang_code` is source language
id for source text and target language id for target text, with `X` being the source or target text.
The [`M2M100Tokenizer`] depends on `sentencepiece` so be sure to install it before running the
examples. To install `sentencepiece` run `pip install sentencepiece`.
- Supervised Training
```python
from transformers import M2M100Config, M2M100ForConditionalGeneration, M2M100Tokenizer
model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M")
tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M", src_lang="en", tgt_lang="fr")
src_text = "Life is like a box of chocolates."
tgt_text = "La vie est comme une boîte de chocolat."
model_inputs = tokenizer(src_text, text_target=tgt_text, return_tensors="pt")
loss = model(**model_inputs).loss # forward pass
```
- Generation
M2M100 uses the `eos_token_id` as the `decoder_start_token_id` for generation with the target language id
being forced as the first generated token. To force the target language id as the first generated token, pass the
*forced_bos_token_id* parameter to the *generate* method. The following example shows how to translate between
Hindi to French and Chinese to English using the *facebook/m2m100_418M* checkpoint.
```python
>>> from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
>>> hi_text = "जीवन एक चॉकलेट बॉक्स की तरह है।"
>>> chinese_text = "生活就像一盒巧克力。"
>>> model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M")
>>> tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M")
>>> # translate Hindi to French
>>> tokenizer.src_lang = "hi"
>>> encoded_hi = tokenizer(hi_text, return_tensors="pt")
>>> generated_tokens = model.generate(**encoded_hi, forced_bos_token_id=tokenizer.get_lang_id("fr"))
>>> tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
"La vie est comme une boîte de chocolat."
>>> # translate Chinese to English
>>> tokenizer.src_lang = "zh"
>>> encoded_zh = tokenizer(chinese_text, return_tensors="pt")
>>> generated_tokens = model.generate(**encoded_zh, forced_bos_token_id=tokenizer.get_lang_id("en"))
>>> tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
"Life is like a box of chocolate."
```
## M2M100Config
[[autodoc]] M2M100Config
## M2M100Tokenizer
[[autodoc]] M2M100Tokenizer
- build_inputs_with_special_tokens
- get_special_tokens_mask
- create_token_type_ids_from_sequences
- save_vocabulary
## M2M100Model
[[autodoc]] M2M100Model
- forward
## M2M100ForConditionalGeneration
[[autodoc]] M2M100ForConditionalGeneration
- forward
| -1 |
huggingface/transformers | 20,209 | Add gpt-sw3 model to transformers | This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ekgren | "2022-11-14T14:04:00Z" | "2022-12-12T18:12:13Z" | b58beebe7286bf53a80f137e0e5cd100ccb77ae2 | 5f94855dc31242d15d755b0d97ec6a0479ee0ea9 | Add gpt-sw3 model to transformers. This adds the gpt-sw3 models and tokenizer to hf. The models are developed by AI Sweden and others. They are gpt models trained from scratch with the nemo-megatron framework and will initially range in sizes from 128m to 20B. The models are multilingual and the languages in the models are English, Swedish, Norwegian, Danish and Icelandic.
Fixes # (issue) https://github.com/huggingface/transformers/issues/20176
@ArthurZucker | ./src/transformers/models/rag/modeling_rag.py | # coding=utf-8
# Copyright 2020, The RAG Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""RAG model implementation."""
from dataclasses import dataclass
from typing import Callable, List, Optional, Tuple, Union
import torch
from torch import nn
from ...configuration_utils import PretrainedConfig
from ...generation import BeamSearchScorer, LogitsProcessorList, StoppingCriteriaList
from ...modeling_outputs import ModelOutput
from ...modeling_utils import PreTrainedModel
from ...utils import add_start_docstrings_to_model_forward, logging, replace_return_docstrings
from .configuration_rag import RagConfig
from .retrieval_rag import RagRetriever
logger = logging.get_logger(__name__)
_CONFIG_FOR_DOC = "RagConfig"
@dataclass
class RetrievAugLMMarginOutput(ModelOutput):
"""
Base class for retriever augmented marginalized models outputs.
Args:
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
Language modeling loss.
logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
Prediction scores of the language modeling head. The score is possibly marginalized over all documents for
each vocabulary token.
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
past_key_values (`List[torch.FloatTensor]`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
List of `torch.FloatTensor` of length `config.n_layers`, with each tensor of shape `(2, batch_size,
num_heads, sequence_length, embed_size_per_head)`).
Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used
(see `past_key_values` input) to speed up sequential decoding.
retrieved_doc_embeds (`torch.FloatTensor` of shape `(batch_size, config.n_docs, hidden_size)`, *optional*, returned when *output_retrieved=True*):
Embedded documents retrieved by the retriever. Is used with `question_encoder_last_hidden_state` to compute
the `doc_scores`.
retrieved_doc_ids (`torch.LongTensor` of shape `(batch_size, config.n_docs)`, *optional*, returned when *output_retrieved=True*):
The indexes of the embedded documents retrieved by the retriever.
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
question_encoder_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden states at the output of the last layer of the question encoder pooled output of the
model.
question_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.
question_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the question encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_enc_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the generator encoder of the model.
generator_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.
generator_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_dec_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.
generator_dec_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_cross_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the
weighted average in the cross-attention heads.
"""
loss: Optional[torch.FloatTensor] = None
logits: torch.FloatTensor = None
doc_scores: torch.FloatTensor = None
past_key_values: Optional[List[torch.FloatTensor]] = None
retrieved_doc_embeds: Optional[torch.FloatTensor] = None
retrieved_doc_ids: Optional[torch.LongTensor] = None
context_input_ids: Optional[torch.LongTensor] = None
context_attention_mask: Optional[torch.LongTensor] = None
question_encoder_last_hidden_state: Optional[torch.FloatTensor] = None
question_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
question_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_last_hidden_state: Optional[torch.FloatTensor] = None
generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_cross_attentions: Optional[Tuple[torch.FloatTensor]] = None
@dataclass
class RetrievAugLMOutput(ModelOutput):
"""
Args:
logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
Prediction scores of the language modeling head. The score is possibly marginalized over all documents for
each vocabulary token.
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
past_key_values (`List[torch.FloatTensor]`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
List of `torch.FloatTensor` of length `config.n_layers`, with each tensor of shape `(2, batch_size,
num_heads, sequence_length, embed_size_per_head)`).
Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used
(see `past_key_values` input) to speed up sequential decoding.
retrieved_doc_embeds (`torch.FloatTensor` of shape `(batch_size, config.n_docs, hidden_size)`, *optional*, returned when *output_retrieved=True*):
Embedded documents retrieved by the retriever. Is used with `question_encoder_last_hidden_state` to compute
the `doc_scores`.
retrieved_doc_ids (`torch.LongTensor` of shape `(batch_size, config.n_docs)`, *optional*, returned when *output_retrieved=True*):
The indexes of the embedded documents retrieved by the retriever.
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
question_encoder_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden states at the output of the last layer of the question encoder pooled output of the
model.
question_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.
question_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the question encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_enc_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the generator encoder of the model.
generator_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.
generator_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_dec_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.
generator_dec_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_cross_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the
weighted average in the cross-attention heads.
"""
logits: torch.FloatTensor = None
doc_scores: torch.FloatTensor = None
past_key_values: Optional[List[torch.FloatTensor]] = None
retrieved_doc_embeds: Optional[torch.FloatTensor] = None
retrieved_doc_ids: Optional[torch.LongTensor] = None
context_input_ids: Optional[torch.LongTensor] = None
context_attention_mask: Optional[torch.LongTensor] = None
question_encoder_last_hidden_state: Optional[torch.FloatTensor] = None
question_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
question_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_last_hidden_state: Optional[torch.FloatTensor] = None
generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_cross_attentions: Optional[Tuple[torch.FloatTensor]] = None
class RagPreTrainedModel(PreTrainedModel):
r"""
RAG models were released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP
Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandra Piktus et al.
RAG is a retriever augmented model and encapsulate three components: a question encoder, a dataset retriever and a
generator, the encoder and generator are trainable while the retriever is just an indexed dataset.
"""
config_class = RagConfig
base_model_prefix = "rag"
_keys_to_ignore_on_load_missing = [r"position_ids"]
@classmethod
def from_pretrained(cls, *args, **kwargs):
# At the moment fast initialization is not supported
# for composite models
kwargs["_fast_init"] = False
return super().from_pretrained(*args, **kwargs)
@classmethod
def from_pretrained_question_encoder_generator(
cls,
question_encoder_pretrained_model_name_or_path: str = None,
generator_pretrained_model_name_or_path: str = None,
retriever: RagRetriever = None,
**kwargs
) -> PreTrainedModel:
r"""
Instantiates an question encoder and a generator from one or two base classes of the library from pretrained
model checkpoints.
The model is set in evaluation mode by default using `model.eval()` (Dropout modules are deactivated). To train
the model, you need to first set it back in training mode with `model.train()`.
Params:
question_encoder_pretrained_model_name_or_path (`str`, *optional*, defaults to `None`):
Information necessary to initiate the question encoder. Can be either:
- A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing model weights saved using
[`~PreTrainedModel.save_pretrained`], e.g., `./my_model_directory/`.
- A path or url to a *tensorflow index checkpoint file* (e.g, `./tf_model/model.ckpt.index`). In
this case, `from_tf` should be set to `True` and a configuration object should be provided as
`config` argument. This loading path is slower than converting the TensorFlow checkpoint in a
PyTorch model using the provided conversion scripts and loading the PyTorch model afterwards.
generator_pretrained_model_name_or_path (`str`, *optional*, defaults to `None`):
Information necessary to initiate the generator. Can be either:
- A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing model weights saved using
[`~PreTrainedModel.save_pretrained`], e.g., `./my_model_directory/`.
- A path or url to a *tensorflow index checkpoint file* (e.g, `./tf_model/model.ckpt.index`). In
this case, `from_tf` should be set to `True` and a configuration object should be provided as
`config` argument. This loading path is slower than converting the TensorFlow checkpoint in a
PyTorch model using the provided conversion scripts and loading the PyTorch model afterwards.
model_args (remaining positional arguments, *optional*):
All remaining positional arguments will be passed to the underlying model's `__init__` method.
retriever ([`RagRetriever`], *optional*):
The retriever to use.
kwwargs (remaining dictionary of keyword arguments, *optional*):
Can be used to update the configuration object (after it being loaded) and initiate the model (e.g.,
`output_attentions=True`).
- To update the question_encoder configuration, use the prefix *question_encoder_* for each
configuration parameter.
- To update the generator configuration, use the prefix *generator_* for each configuration parameter.
- To update the parent model configuration, do not use a prefix for each configuration parameter.
Behaves differently depending on whether a `config` is provided or automatically loaded.
Example:
```python
>>> from transformers import RagModel
>>> # initialize a RAG from two pretrained models.
>>> model = RagModel.from_pretrained_question_encoder_generator(
... "facebook/dpr-question_encoder-single-nq-base", "t5-small"
... )
>>> # saving model after fine-tuning
>>> model.save_pretrained("./rag")
>>> # load fine-tuned model
>>> model = RagModel.from_pretrained("./rag")
```"""
kwargs_question_encoder = {
argument[len("question_encoder_") :]: value
for argument, value in kwargs.items()
if argument.startswith("question_encoder_")
}
kwargs_generator = {
argument[len("generator_") :]: value
for argument, value in kwargs.items()
if argument.startswith("generator_")
}
# remove question_encoder, generator kwargs from kwargs
for key in kwargs_question_encoder.keys():
del kwargs["question_encoder_" + key]
for key in kwargs_generator.keys():
del kwargs["generator_" + key]
# Load and initialize the question_encoder and generator
# The distinction between question_encoder and generator at the model level is made
# by the value of the flag `is_generator` that we need to set correctly.
question_encoder = kwargs_question_encoder.pop("model", None)
if question_encoder is None:
assert question_encoder_pretrained_model_name_or_path is not None, (
"If `model` is not defined as an argument, a `question_encoder_pretrained_model_name_or_path` has to"
" be defined"
)
from ..auto.modeling_auto import AutoModel
if "config" not in kwargs_question_encoder:
from ..auto.configuration_auto import AutoConfig
question_encoder_config, kwargs_question_encoder = AutoConfig.from_pretrained(
question_encoder_pretrained_model_name_or_path,
**kwargs_question_encoder,
return_unused_kwargs=True,
)
kwargs_question_encoder["config"] = question_encoder_config
question_encoder = AutoModel.from_pretrained(
question_encoder_pretrained_model_name_or_path, **kwargs_question_encoder
)
generator = kwargs_generator.pop("model", None)
if generator is None:
assert generator_pretrained_model_name_or_path is not None, (
"If `generator_model` is not defined as an argument, a `generator_pretrained_model_name_or_path` has"
" to be defined"
)
from ..auto.modeling_auto import AutoModelForSeq2SeqLM
if "config" not in kwargs_generator:
from ..auto.configuration_auto import AutoConfig
generator_config, kwargs_generator = AutoConfig.from_pretrained(
generator_pretrained_model_name_or_path, **kwargs_generator, return_unused_kwargs=True
)
kwargs_generator["config"] = generator_config
generator = AutoModelForSeq2SeqLM.from_pretrained(
generator_pretrained_model_name_or_path, **kwargs_generator
)
# instantiate config with corresponding kwargs
config = kwargs.get("config", None)
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
return cls(question_encoder=question_encoder, generator=generator, config=config, retriever=retriever)
RAG_START_DOCSTRING = r"""
RAG is a seq2seq model which encapsulates two core components: a question encoder and a generator. During a forward
pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context
documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.
The question encoder can be any *autoencoding* model, preferably [`DPRQuestionEncoder`], and the generator can be
any *seq2seq* model, preferably [`BartForConditionalGeneration`].
The model can be initialized with a [`RagRetriever`] for end-to-end generation or used in combination with the
outputs of a retriever in multiple steps---see examples for more details. The model is compatible any
*autoencoding* model as the `question_encoder` and any *seq2seq* model with language model head as the `generator`.
It has been tested with [`DPRQuestionEncoder`] as the `question_encoder` and [`BartForConditionalGeneration`] or
[`T5ForConditionalGeneration`] as the `generator`.
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
and behavior.
Args:
config ([`RagConfig`]):
Model configuration class with all the parameters of the model. Initializing with a config file does not
load the weights associated with the model, only the configuration. Check out the
[`~PreTrainedModel.from_pretrained`] method to load the model weights.
question_encoder ([`PreTrainedModel`]):
An encoder model compatible with the faiss index encapsulated by the `retriever`.
generator ([`PreTrainedModel`]):
A seq2seq model used as the generator in the RAG architecture.
retriever ([`RagRetriever`]):
A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.
"""
RAG_FORWARD_INPUTS_DOCSTRING = r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. [`RagConfig`], used to initialize the model, specifies
which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to
obtain the indices.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
encoder_outputs (`tuple(tuple(torch.FloatTensor)`, *optional*)
Tuple consists of (`generator_enc_last_hidden_state`, *optional*: `generator_enc_hidden_states`,
*optional*: `generator_enc_attentions`). `generator_enc_last_hidden_state` of shape `(batch_size, n_docs *
sequence_length, hidden_size)` is a sequence of hidden-states at the output of the last layer of the
generator's encoder.
Used by the ([`RagModel`]) model during decoding.
decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Provide for generation tasks. `None` by default, construct as per instructions for the generator model
you're using with your RAG instance.
decoder_attention_mask (`torch.BoolTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
be used by default.
past_key_values (`tuple(tuple(torch.FloatTensor))`):
Tuple consists of two elements: `encoder_outputs` of the RAG model (see `encoder_outputs`) and
`past_key_values` of the underlying generator. Can be used to speed up decoding. `past_key_values` are used
in the ([`RagTokenForGeneration`]) model during decoding.
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`. If the model has is not initialized with a `retriever` `doc_scores`
has to be provided to the forward pass. `doc_scores` can be computed via
`question_encoder_last_hidden_state` and `retrieved_doc_embeds`, see examples for more information.
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input IDs post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model has is not initialized with a `retriever` ``context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`]. context_attention_mask
(`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*,
returned when *output_retrieved=True*): Attention mask post-processed from the retrieved documents and the
question encoder `input_ids` by the retriever.
If the model has is not initialized with a `retriever` `context_attention_mask` has to be provided to the
forward pass. `context_attention_mask` are returned by [`~RagRetriever.__call__`].
use_cache (`bool`, *optional*, defaults to `True`):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
`past_key_values`).
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
output_retrieved(`bool`, *optional*):
Whether or not to return the `retrieved_doc_embeds`, `retrieved_doc_ids`, `context_input_ids` and
`context_attention_mask`. See returned tensors for more detail.
n_docs (`int`, *optional*, defaults to `config.n_docs``)
Number of documents to retrieve and/or number of documents for which to generate an answer.
"""
@add_start_docstrings_to_model_forward(RAG_START_DOCSTRING)
class RagModel(RagPreTrainedModel):
def __init__(
self,
config: Optional[PretrainedConfig] = None,
question_encoder: Optional[PreTrainedModel] = None,
generator: Optional[PreTrainedModel] = None,
retriever: Optional[RagRetriever] = None, # or maybe just use a `set_retriever(...)` method
**kwargs,
):
assert config is not None or (
question_encoder is not None and generator is not None
), "Either a configuration or an question_encoder and a generator has to be provided."
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
else:
assert isinstance(config, self.config_class), f"config: {config} has to be of type {self.config_class}"
super().__init__(config)
if question_encoder is None:
from ..auto.modeling_auto import AutoModel
question_encoder = AutoModel.from_config(config.question_encoder)
if generator is None:
from ..auto.modeling_auto import AutoModelForSeq2SeqLM
generator = AutoModelForSeq2SeqLM.from_config(config.generator)
self.retriever = retriever
if self.retriever is not None:
assert isinstance(
retriever, RagRetriever
), f"`self.retriever` is of type {type(self.retriever)}, but should be of type `RagRetriever`"
self.retriever = retriever
self.question_encoder = question_encoder
self.generator = generator
self.ctx_encoder = None
self.context_encoder_training = False
@add_start_docstrings_to_model_forward(RAG_FORWARD_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=RetrievAugLMOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.BoolTensor] = None,
past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
doc_scores: Optional[torch.FloatTensor] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask=None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
output_retrieved: Optional[bool] = None,
n_docs: Optional[int] = None,
) -> Union[Tuple[torch.Tensor], RetrievAugLMOutput]:
r"""
Returns:
Example:
```python
>>> from transformers import RagTokenizer, RagRetriever, RagModel
>>> import torch
>>> tokenizer = RagTokenizer.from_pretrained("facebook/rag-token-base")
>>> retriever = RagRetriever.from_pretrained(
... "facebook/rag-token-base", index_name="exact", use_dummy_dataset=True
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = RagModel.from_pretrained("facebook/rag-token-base", retriever=retriever)
>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> outputs = model(input_ids=inputs["input_ids"])
```"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
use_cache = use_cache if use_cache is not None else self.config.use_cache
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
output_retrieved = output_retrieved if output_retrieved is not None else self.config.output_retrieved
# whether retriever has to be used
has_to_retrieve = (
self.retriever is not None
and (context_input_ids is None or context_attention_mask is None or doc_scores is None)
and encoder_outputs is None
)
# encoder_outputs are pre-computed during RAG-token generation
if encoder_outputs is None:
if has_to_retrieve:
question_enc_outputs = self.question_encoder(
input_ids, attention_mask=attention_mask, return_dict=True
)
question_encoder_last_hidden_state = question_enc_outputs[0] # hidden states of question encoder
retriever_outputs = self.retriever(
input_ids,
question_encoder_last_hidden_state.cpu().detach().to(torch.float32).numpy(),
prefix=self.generator.config.prefix,
n_docs=n_docs,
return_tensors="pt",
)
if self.context_encoder_training:
(
context_input_ids,
context_attention_mask,
retrieved_doc_embeds,
retrived_doc_input_ids,
retrived_doc_attention_mask,
retrieved_doc_ids,
) = (
retriever_outputs["context_input_ids"],
retriever_outputs["context_attention_mask"],
retriever_outputs["retrieved_doc_embeds"],
retriever_outputs["tokenized_doc_ids"],
retriever_outputs["tokenized_doc_attention_mask"],
retriever_outputs["doc_ids"],
)
context_input_ids = context_input_ids.to(input_ids)
context_attention_mask = context_attention_mask.to(input_ids)
retrived_doc_input_ids = retrived_doc_input_ids.to(input_ids)
retrived_doc_attention_mask = retrived_doc_attention_mask.to(input_ids)
retrieved_doc_embeds = self.ctx_encoder(
retrived_doc_input_ids, attention_mask=retrived_doc_attention_mask, return_dict=True
).pooler_output
retrieved_doc_embeds = retrieved_doc_embeds.view(
-1, n_docs, question_encoder_last_hidden_state.shape[1]
) # reshaping
# compute doc_scores involving ctx_encoder
doc_scores = torch.bmm(
question_encoder_last_hidden_state.unsqueeze(1), retrieved_doc_embeds.transpose(1, 2)
).squeeze(1)
else:
context_input_ids, context_attention_mask, retrieved_doc_embeds, retrieved_doc_ids = (
retriever_outputs["context_input_ids"],
retriever_outputs["context_attention_mask"],
retriever_outputs["retrieved_doc_embeds"],
retriever_outputs["doc_ids"],
)
# set to correct device
retrieved_doc_embeds = retrieved_doc_embeds.to(question_encoder_last_hidden_state)
context_input_ids = context_input_ids.to(input_ids)
context_attention_mask = context_attention_mask.to(input_ids)
# compute doc_scores
doc_scores = torch.bmm(
question_encoder_last_hidden_state.unsqueeze(1), retrieved_doc_embeds.transpose(1, 2)
).squeeze(1)
else:
assert context_input_ids is not None, (
"Make sure that `context_input_ids` are passed, if no `retriever` is set. Alternatively, you can"
" set a retriever using the `set_retriever(...)` function."
)
assert context_attention_mask is not None, (
"Make sure that `context_attention_mask` are passed, if no `retriever` is set. Alternatively, you"
" can set a retriever using the `set_retriever(...)` function."
)
assert doc_scores is not None, (
"Make sure that `doc_scores` are passed, if no `retriever` is set. Alternatively, you can set a"
" retriever using the `set_retriever(...)` function."
)
assert (
doc_scores is not None
), "Make sure that `doc_scores` are passed when passing `encoder_outputs` to the forward function."
assert (doc_scores.shape[1] % n_docs) == 0, (
f" The first dimension of `context_input_ids` should be a multiple of `n_docs`={n_docs}, but is"
f" {context_input_ids.shape[0]}."
)
# Decoder input without context documents
if decoder_input_ids is not None:
decoder_input_ids = decoder_input_ids.repeat_interleave(n_docs, dim=0)
if decoder_attention_mask is not None:
decoder_attention_mask = decoder_attention_mask.repeat_interleave(n_docs, dim=0)
gen_outputs = self.generator(
input_ids=context_input_ids,
attention_mask=context_attention_mask,
encoder_outputs=encoder_outputs,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
past_key_values=past_key_values,
use_cache=use_cache,
output_attentions=output_attentions,
return_dict=True,
)
if not has_to_retrieve:
question_encoder_last_hidden_state = None
question_enc_hidden_states = None
question_enc_attentions = None
retrieved_doc_embeds = None
retrieved_doc_ids = None
else:
question_enc_hidden_states = question_enc_outputs.hidden_states
question_enc_attentions = question_enc_outputs.attentions
if not has_to_retrieve or not output_retrieved:
# don't output retrieved docs
context_input_ids = (None,)
context_attention_mask = None
retrieved_doc_embeds = None
retrieved_doc_ids = None
return RetrievAugLMOutput(
logits=gen_outputs.logits,
doc_scores=doc_scores,
past_key_values=gen_outputs.past_key_values,
context_input_ids=context_input_ids,
context_attention_mask=context_attention_mask,
retrieved_doc_embeds=retrieved_doc_embeds,
retrieved_doc_ids=retrieved_doc_ids,
question_encoder_last_hidden_state=question_encoder_last_hidden_state,
question_enc_hidden_states=question_enc_hidden_states,
question_enc_attentions=question_enc_attentions,
generator_enc_last_hidden_state=gen_outputs.encoder_last_hidden_state,
generator_enc_hidden_states=gen_outputs.encoder_hidden_states,
generator_enc_attentions=gen_outputs.encoder_attentions,
generator_dec_hidden_states=gen_outputs.decoder_hidden_states,
generator_dec_attentions=gen_outputs.decoder_attentions,
generator_cross_attentions=gen_outputs.cross_attentions,
)
@add_start_docstrings_to_model_forward(
"""
A RAG-sequence model implementation. It performs RAG-sequence specific marginalization in the forward pass.
""",
RAG_START_DOCSTRING,
)
class RagSequenceForGeneration(RagPreTrainedModel):
def __init__(
self,
config: Optional[PretrainedConfig] = None,
question_encoder: Optional[PreTrainedModel] = None,
generator: Optional[PreTrainedModel] = None,
retriever: Optional[RagRetriever] = None,
**kwargs,
):
assert config is not None or (
question_encoder is not None and generator is not None
), "Either a configuration or an encoder and a generator has to be provided."
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
super().__init__(config)
# instantiate model
self.rag = RagModel(config=config, question_encoder=question_encoder, generator=generator, retriever=retriever)
def set_retriever(self, retriever: RagRetriever):
self.rag.retriever = retriever
def set_context_encoder_for_training(self, ctx_encoder: PreTrainedModel):
self.rag.context_encoder_training = True
self.rag.ctx_encoder = ctx_encoder
@add_start_docstrings_to_model_forward(RAG_FORWARD_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=RetrievAugLMMarginOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Tuple[Tuple[torch.Tensor]]] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.BoolTensor] = None,
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
output_retrieved: Optional[bool] = None,
exclude_bos_score: Optional[bool] = None,
reduce_loss: Optional[bool] = None,
labels: Optional[torch.LongTensor] = None,
n_docs: Optional[int] = None,
**kwargs # needs kwargs for generation
) -> RetrievAugLMMarginOutput:
r"""
exclude_bos_score (`bool`, *optional*):
Only relevant if `labels` is passed. If `True`, the score of the BOS token is disregarded when computing
the loss.
reduce_loss (`bool`, *optional*):
Only relevant if `labels` is passed. If `True`, the NLL loss is reduced using the `torch.Tensor.sum`
operation.
kwargs (`Dict[str, any]`, optional, defaults to *{}*):
Legacy dictionary, which is required so that model can use *generate()* function.
Returns:
Example:
```python
>>> from transformers import RagTokenizer, RagRetriever, RagSequenceForGeneration
>>> import torch
>>> tokenizer = RagTokenizer.from_pretrained("facebook/rag-sequence-nq")
>>> retriever = RagRetriever.from_pretrained(
... "facebook/rag-sequence-nq", index_name="exact", use_dummy_dataset=True
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = RagSequenceForGeneration.from_pretrained("facebook/rag-token-nq", retriever=retriever)
>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> targets = tokenizer(text_target="In Paris, there are 10 million people.", return_tensors="pt")
>>> input_ids = inputs["input_ids"]
>>> labels = targets["input_ids"]
>>> outputs = model(input_ids=input_ids, labels=labels)
>>> # or use retriever separately
>>> model = RagSequenceForGeneration.from_pretrained("facebook/rag-sequence-nq", use_dummy_dataset=True)
>>> # 1. Encode
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.detach().numpy(), return_tensors="pt")
>>> doc_scores = torch.bmm(
... question_hidden_states.unsqueeze(1), docs_dict["retrieved_doc_embeds"].float().transpose(1, 2)
... ).squeeze(1)
>>> # 3. Forward to generator
>>> outputs = model(
... context_input_ids=docs_dict["context_input_ids"],
... context_attention_mask=docs_dict["context_attention_mask"],
... doc_scores=doc_scores,
... decoder_input_ids=labels,
... )
```"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
exclude_bos_score = exclude_bos_score if exclude_bos_score is not None else self.config.exclude_bos_score
reduce_loss = reduce_loss if reduce_loss is not None else self.config.reduce_loss
if labels is not None:
if decoder_input_ids is None:
decoder_input_ids = labels
use_cache = False
outputs = self.rag(
input_ids=input_ids,
attention_mask=attention_mask,
encoder_outputs=encoder_outputs,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
context_input_ids=context_input_ids,
context_attention_mask=context_attention_mask,
doc_scores=doc_scores,
past_key_values=past_key_values,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
output_retrieved=output_retrieved,
n_docs=n_docs,
)
loss = None
if labels is not None:
loss = self.get_nll(
outputs.logits,
outputs.doc_scores,
decoder_input_ids,
reduce_loss=reduce_loss,
epsilon=self.config.label_smoothing,
exclude_bos_score=exclude_bos_score,
n_docs=n_docs,
)
return RetrievAugLMMarginOutput(
loss=loss,
logits=outputs.logits,
doc_scores=outputs.doc_scores,
past_key_values=outputs.past_key_values,
context_input_ids=outputs.context_input_ids,
context_attention_mask=outputs.context_attention_mask,
retrieved_doc_embeds=outputs.retrieved_doc_embeds,
retrieved_doc_ids=outputs.retrieved_doc_ids,
question_encoder_last_hidden_state=outputs.question_encoder_last_hidden_state,
question_enc_hidden_states=outputs.question_enc_hidden_states,
question_enc_attentions=outputs.question_enc_attentions,
generator_enc_last_hidden_state=outputs.generator_enc_last_hidden_state,
generator_enc_hidden_states=outputs.generator_enc_hidden_states,
generator_enc_attentions=outputs.generator_enc_attentions,
generator_dec_hidden_states=outputs.generator_dec_hidden_states,
generator_dec_attentions=outputs.generator_dec_attentions,
generator_cross_attentions=outputs.generator_cross_attentions,
)
@property
def retriever(self):
return self.rag.retriever
@property
def generator(self):
return self.rag.generator
@property
def question_encoder(self):
return self.rag.question_encoder
@torch.no_grad()
def generate(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.LongTensor] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
do_deduplication: Optional[bool] = None, # defaults to True
num_return_sequences: Optional[int] = None, # defaults to 1
num_beams: Optional[int] = None, # defaults to 1
n_docs: Optional[int] = None,
**model_kwargs
) -> torch.LongTensor:
"""
Implements RAG sequence "thorough" decoding. Read the [`~generation.GenerationMixin.generate`]` documentation
for more information on how to set other generate input parameters.
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
The sequence used as a prompt for the generation. If `input_ids` is not passed, then
`context_input_ids` has to be provided.
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input IDs post-processed from the retrieved documents and the question encoder input_ids by the
retriever.
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model is not initialized with a `retriever` or `input_ids` is not given, `context_input_ids` and
`context_attention_mask` have to be provided to the forward pass. They are returned by
[`~RagRetriever.__call__`].
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
If the model is not initialized with a `retriever` or `input_ids` is not given, `doc_scores` has to be
provided to the forward pass. `doc_scores` are returned by [`~RagRetriever.__call__`].
do_deduplication (`bool`, *optional*):
Whether or not to deduplicate the generations from different context documents for a given input. Has
to be set to `False` if used while training with distributed backend.
num_return_sequences(`int`, *optional*, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the `generator`'s `[`~generation.GenerationMixin.generate`]` function,
where we set `num_return_sequences` to `num_beams`.
num_beams (`int`, *optional*, defaults to 1):
Number of beams for beam search. 1 means no beam search.
n_docs (`int`, *optional*, defaults to `config.n_docs`)
Number of documents to retrieve and/or number of documents for which to generate an answer.
kwargs:
Additional kwargs will be passed to [`~generation.GenerationMixin.generate`].
Return:
`torch.LongTensor` of shape `(batch_size * num_return_sequences, sequence_length)`: The generated
sequences. The second dimension (sequence length) is either equal to `max_length` or shorter if all batches
finished early due to the `eos_token_id`.
"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
do_deduplication = do_deduplication if do_deduplication is not None else self.config.do_deduplication
num_doc_return_sequences = (
num_return_sequences if num_return_sequences is not None else self.config.num_return_sequences
)
num_beams = num_beams if num_beams is not None else self.config.num_beams
assert (
input_ids is not None or context_input_ids is not None
), " At least one of input_ids or context_input_ids must be given"
if self.retriever is not None and context_input_ids is None:
question_hidden_states = self.question_encoder(input_ids, attention_mask=attention_mask)[0]
context_input_ids = self.retriever(
input_ids,
question_hidden_states.cpu().detach().to(torch.float32).numpy(),
prefix=self.generator.config.prefix,
n_docs=n_docs,
return_tensors="pt",
)["context_input_ids"]
# set to correct device
context_input_ids = context_input_ids.to(input_ids)
hypos = []
model_kwargs["num_beams"] = num_beams
model_kwargs["num_return_sequences"] = num_beams
model_kwargs["attention_mask"] = None
batch_size = input_ids.shape[0] if input_ids is not None else context_input_ids.shape[0] // n_docs
for index in range(batch_size):
# first, generate beams from documents:
generator_input_ids = context_input_ids[index * n_docs : (index + 1) * n_docs] # (n_docs, max_len)
output_sequences = self.generator.generate(
generator_input_ids,
**model_kwargs,
) # n_docs * n_beam, tgt_len
if do_deduplication:
# do_deduplication, max_output_len
output_sequences = torch.stack(list({str(k.tolist()): k for k in output_sequences}.values()))
num_candidates = output_sequences.shape[
0
] # after deduplication, this number can be less than n_docs*n_beam
# then, run model forwards to get nll scores:
if input_ids is not None:
new_input_ids = input_ids[index : index + 1].repeat(num_candidates, 1)
outputs = self(new_input_ids, labels=output_sequences, exclude_bos_score=True)
else: # input_ids is None, need context_input_ids/mask and doc_scores
assert context_attention_mask is not None, (
"Make sure that `context_attention_mask` are passed, if no `input_ids` is set. Alternatively, you"
" can set a retriever using the `set_retriever(...)` function."
)
assert doc_scores is not None, (
"Make sure that `doc_scores` are passed, if no `input_ids` is set. Alternatively, you can set a"
" retriever using the `set_retriever(...)` function."
)
individual_input_ids = generator_input_ids.repeat(
num_candidates, 1
) # (num_candidates*n_docs, max_len)
individual_attention_mask = context_attention_mask[index * n_docs : (index + 1) * n_docs]
individual_attention_mask = individual_attention_mask.repeat(num_candidates, 1)
individual_doc_scores = doc_scores[index : (index + 1), :] # doc_scores.shape = [batch, n_docs]
individual_doc_scores = individual_doc_scores.repeat(num_candidates, 1) # [num_candidates, n_docs]
outputs = self(
context_input_ids=individual_input_ids,
context_attention_mask=individual_attention_mask,
doc_scores=individual_doc_scores,
labels=output_sequences,
exclude_bos_score=True,
)
top_cand_inds = (-outputs["loss"]).topk(num_doc_return_sequences)[1]
# add hypothesis
hypos.append(output_sequences[top_cand_inds])
return self._cat_and_pad(hypos, pad_token_id=self.config.generator.pad_token_id)
def get_nll(
self, seq_logits, doc_scores, target, reduce_loss=False, epsilon=0.0, exclude_bos_score=False, n_docs=None
):
# shift tokens left
target = torch.cat(
[target[:, 1:], target.new(target.shape[0], 1).fill_(self.config.generator.pad_token_id)], 1
)
n_docs = n_docs if n_docs is not None else self.config.n_docs
# bos_token_id is None for T5
bos_token_id = self.config.bos_token_id or self.config.generator.bos_token_id
use_bos = bos_token_id is not None and target[:, 0].eq(bos_token_id).all()
def _mask_pads(ll, smooth_obj):
pad_mask = target.eq(self.config.generator.pad_token_id)
if pad_mask.any():
ll.masked_fill_(pad_mask, 0.0)
smooth_obj.masked_fill_(pad_mask, 0.0)
return ll.squeeze(-1), smooth_obj.squeeze(-1)
# seq_logits dim = (batch*n_docs, tgt_len , #vocabs)
seq_logprobs = nn.functional.log_softmax(seq_logits, dim=-1).view(
seq_logits.shape[0] // n_docs, n_docs, -1, seq_logits.size(-1)
) # batch_size x n_docs x tgt_len x #vocab_size
doc_logprobs = nn.functional.log_softmax(doc_scores, dim=1).unsqueeze(-1).unsqueeze(-1)
# RAG-sequence marginalization
first_token_scores = seq_logprobs[:, :, :1, :]
second_token_scores = seq_logprobs[:, :, 1:2, :]
remainder = seq_logprobs[:, :, 2:, :]
rag_logprobs = torch.cat([first_token_scores, second_token_scores + doc_logprobs, remainder], dim=2)
# calculate loss
target = target.unsqueeze(1).unsqueeze(-1).repeat(1, n_docs, 1, 1)
assert target.dim() == rag_logprobs.dim()
ll = rag_logprobs.gather(dim=-1, index=target)
smooth_obj = rag_logprobs.sum(dim=-1, keepdim=True) # total sum of all (normalised) logits
ll, smooth_obj = _mask_pads(ll, smooth_obj)
# sum over tokens, exclude bos while scoring
ll = ll[:, :, 1:].sum(2) if exclude_bos_score and use_bos else ll.sum(2)
smooth_obj = smooth_obj.sum(2)
ll = ll.logsumexp(1) # logsumexp over docs
smooth_obj = smooth_obj.logsumexp(1)
nll_loss = -ll
smooth_loss = -smooth_obj
if reduce_loss:
nll_loss = nll_loss.sum()
smooth_loss = smooth_loss.sum()
eps_i = epsilon / rag_logprobs.size(-1)
loss = (1.0 - epsilon) * nll_loss + eps_i * smooth_loss
return loss
@staticmethod
def _cat_and_pad(tensors, pad_token_id):
output = (
tensors[0].new(sum([t.shape[0] for t in tensors]), max([t.shape[1] for t in tensors])).fill_(pad_token_id)
)
ind = 0
for t in tensors:
output[ind : ind + t.shape[0], : t.shape[1]] = t
ind += t.shape[0]
return output
@add_start_docstrings_to_model_forward(
"""
A RAG-token model implementation. It performs RAG-token specific marginalization in the forward pass.
""",
RAG_START_DOCSTRING,
)
class RagTokenForGeneration(RagPreTrainedModel):
def __init__(
self,
config: Optional[PretrainedConfig] = None,
question_encoder: Optional[PreTrainedModel] = None,
generator: Optional[PreTrainedModel] = None,
retriever: Optional[RagRetriever] = None,
**kwargs,
):
assert config is not None or (
question_encoder is not None and generator is not None
), "Either a configuration or an encoder and a generator has to be provided."
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
super().__init__(config)
# instantiate model
self.rag = RagModel(config=config, question_encoder=question_encoder, generator=generator, retriever=retriever)
def set_retriever(self, retriever: RagRetriever):
self.rag.retriever = retriever
def set_context_encoder_for_training(self, ctx_encoder: PreTrainedModel):
self.rag.context_encoder_training = True
self.rag.ctx_encoder = ctx_encoder
def prepare_inputs_for_generation(
self,
decoder_input_ids,
past=None,
attention_mask=None,
use_cache=None,
encoder_outputs=None,
doc_scores=None,
n_docs=None,
**kwargs
):
if past is not None:
# if past is defined use only last decoder_input_ids
decoder_input_ids = decoder_input_ids[:, -1:]
return {
"input_ids": None,
"encoder_outputs": encoder_outputs,
"doc_scores": doc_scores,
"context_attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"past_key_values": past,
"use_cache": use_cache,
"do_marginalize": True,
"n_docs": n_docs,
}
@property
def retriever(self):
return self.rag.retriever
@property
def generator(self):
return self.rag.generator
@property
def question_encoder(self):
return self.rag.question_encoder
@staticmethod
def _reorder_cache(past, beam_idx):
"""Reorders cache for generation. BART-inspired but we need to take care of the extra dimension for docs"""
def _reorder_stacked(hidden_states, new_order):
n_docs = hidden_states.shape[0] // new_order.shape[0]
hidden_states = hidden_states.view(-1, n_docs, *hidden_states.shape[1:])
hidden_states = hidden_states.index_select(0, new_order)
result = hidden_states.view(-1, *hidden_states.shape[2:])
return result
reordered_past = ()
for layer_past in past:
# get the correct batch idx from decoder layer's batch dim for cross and self-attn
reordered_past += (tuple(_reorder_stacked(past_state, beam_idx) for past_state in layer_past),)
return reordered_past
def marginalize(self, seq_logits, doc_scores, n_docs=None):
n_docs = n_docs if n_docs is not None else self.config.n_docs
# RAG-token marginalization
seq_logprobs = nn.functional.log_softmax(seq_logits, dim=-1).view(
seq_logits.shape[0] // n_docs, n_docs, -1, seq_logits.size(-1)
)
doc_logprobs = torch.log_softmax(doc_scores, dim=1)
log_prob_sum = seq_logprobs + doc_logprobs.unsqueeze(-1).unsqueeze(-1)
return torch.logsumexp(log_prob_sum, dim=1)
@add_start_docstrings_to_model_forward(RAG_FORWARD_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=RetrievAugLMMarginOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.FloatTensor] = None,
encoder_outputs: Optional[Tuple[Tuple[torch.Tensor]]] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.BoolTensor] = None,
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
output_retrieved: Optional[bool] = None,
do_marginalize: Optional[bool] = None,
reduce_loss: Optional[bool] = None,
labels: Optional[torch.LongTensor] = None,
n_docs: Optional[int] = None,
**kwargs # needs kwargs for generation
) -> RetrievAugLMMarginOutput:
r"""
do_marginalize (`bool`, *optional*):
If `True`, the logits are marginalized over all documents by making use of
`torch.nn.functional.log_softmax`.
reduce_loss (`bool`, *optional*):
Only relevant if `labels` is passed. If `True`, the NLL loss is reduced using the `torch.Tensor.sum`
operation.
kwargs (`Dict[str, any]`, optional, defaults to *{}*):
Legacy dictionary, which is required so that model can use *generate()* function.
Returns:
Example:
```python
>>> from transformers import RagTokenizer, RagRetriever, RagTokenForGeneration
>>> import torch
>>> tokenizer = RagTokenizer.from_pretrained("facebook/rag-token-nq")
>>> retriever = RagRetriever.from_pretrained(
... "facebook/rag-token-nq", index_name="exact", use_dummy_dataset=True
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = RagTokenForGeneration.from_pretrained("facebook/rag-token-nq", retriever=retriever)
>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> targets = tokenizer(text_target="In Paris, there are 10 million people.", return_tensors="pt")
>>> input_ids = inputs["input_ids"]
>>> labels = targets["input_ids"]
>>> outputs = model(input_ids=input_ids, labels=labels)
>>> # or use retriever separately
>>> model = RagTokenForGeneration.from_pretrained("facebook/rag-token-nq", use_dummy_dataset=True)
>>> # 1. Encode
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.detach().numpy(), return_tensors="pt")
>>> doc_scores = torch.bmm(
... question_hidden_states.unsqueeze(1), docs_dict["retrieved_doc_embeds"].float().transpose(1, 2)
... ).squeeze(1)
>>> # 3. Forward to generator
>>> outputs = model(
... context_input_ids=docs_dict["context_input_ids"],
... context_attention_mask=docs_dict["context_attention_mask"],
... doc_scores=doc_scores,
... decoder_input_ids=labels,
... )
>>> # or directly generate
>>> generated = model.generate(
... context_input_ids=docs_dict["context_input_ids"],
... context_attention_mask=docs_dict["context_attention_mask"],
... doc_scores=doc_scores,
... )
>>> generated_string = tokenizer.batch_decode(generated, skip_special_tokens=True)
```"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
do_marginalize = do_marginalize if do_marginalize is not None else self.config.do_marginalize
reduce_loss = reduce_loss if reduce_loss is not None else self.config.reduce_loss
if labels is not None:
if decoder_input_ids is None:
decoder_input_ids = labels
use_cache = False
outputs = self.rag(
input_ids=input_ids,
attention_mask=attention_mask,
encoder_outputs=encoder_outputs,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
context_input_ids=context_input_ids,
context_attention_mask=context_attention_mask,
doc_scores=doc_scores,
past_key_values=past_key_values,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
output_retrieved=output_retrieved,
n_docs=n_docs,
)
loss = None
logits = outputs.logits
if labels is not None:
assert decoder_input_ids is not None
loss = self.get_nll(
outputs.logits,
outputs.doc_scores,
labels,
reduce_loss=reduce_loss,
epsilon=self.config.label_smoothing,
n_docs=n_docs,
)
if do_marginalize:
logits = self.marginalize(logits, outputs.doc_scores, n_docs)
return RetrievAugLMMarginOutput(
loss=loss,
logits=logits,
doc_scores=outputs.doc_scores,
past_key_values=outputs.past_key_values,
context_input_ids=outputs.context_input_ids,
context_attention_mask=outputs.context_attention_mask,
retrieved_doc_embeds=outputs.retrieved_doc_embeds,
retrieved_doc_ids=outputs.retrieved_doc_ids,
question_encoder_last_hidden_state=outputs.question_encoder_last_hidden_state,
question_enc_hidden_states=outputs.question_enc_hidden_states,
question_enc_attentions=outputs.question_enc_attentions,
generator_enc_last_hidden_state=outputs.generator_enc_last_hidden_state,
generator_enc_hidden_states=outputs.generator_enc_hidden_states,
generator_enc_attentions=outputs.generator_enc_attentions,
generator_dec_hidden_states=outputs.generator_dec_hidden_states,
generator_dec_attentions=outputs.generator_dec_attentions,
generator_cross_attentions=outputs.generator_cross_attentions,
)
@torch.no_grad()
def generate(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.LongTensor] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
max_length: Optional[int] = None,
min_length: Optional[int] = None,
early_stopping: Optional[bool] = None,
use_cache: Optional[bool] = None,
num_beams: Optional[int] = None,
num_beam_groups: Optional[int] = None,
diversity_penalty: Optional[float] = None,
bos_token_id: Optional[int] = None,
pad_token_id: Optional[int] = None,
eos_token_id: Optional[int] = None,
length_penalty: Optional[float] = None,
no_repeat_ngram_size: Optional[int] = None,
encoder_no_repeat_ngram_size: Optional[int] = None,
repetition_penalty: Optional[float] = None,
bad_words_ids: Optional[List[List[int]]] = None,
num_return_sequences: Optional[int] = None,
decoder_start_token_id: Optional[int] = None,
n_docs: Optional[int] = None,
prefix_allowed_tokens_fn: Callable[[int, torch.Tensor], List[int]] = None,
logits_processor: Optional[LogitsProcessorList] = LogitsProcessorList(),
renormalize_logits: Optional[bool] = None,
stopping_criteria: Optional[StoppingCriteriaList] = StoppingCriteriaList(),
forced_bos_token_id: Optional[int] = None,
forced_eos_token_id: Optional[int] = None,
remove_invalid_values: Optional[bool] = None,
exponential_decay_length_penalty: Optional[Tuple[Union[int, float]]] = None,
**model_kwargs
) -> torch.LongTensor:
"""
Implements RAG token decoding.
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
The sequence used as a prompt for the generation. If `input_ids` is not passed, then
`context_input_ids` has to be provided.
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input IDs post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model has is not initialized with a `retriever`, `context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`].
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model has is not initialized with a `retriever`, `context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`].
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
If the model has is not initialized with a `retriever`, `context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`].
max_length (`int`, *optional*, defaults to 20):
The maximum length of the sequence to be generated.
min_length (`int`, *optional*, defaults to 10):
The minimum length of the sequence to be generated.
early_stopping (`bool`, *optional*, defaults to `False`):
Whether or not to stop the beam search when at least `num_beams` sentences are finished per batch or
not.
use_cache: (`bool`, *optional*, defaults to `True`):
Whether or not the model should use the past last key/values attentions (if applicable to the model) to
speed up decoding.
pad_token_id (`int`, *optional*):
The id of the *padding* token.
bos_token_id (`int`, *optional*):
The id of the *beginning-of-sequence* token.
eos_token_id (`int`, *optional*):
The id of the *end-of-sequence* token.
length_penalty (`float`, *optional*, defaults to 1.0):
Exponential penalty to the length that is used with beam-based generation. It is applied as an exponent
to the sequence length, which in turn is used to divide the score of the sequence. Since the score is
the log likelihood of the sequence (i.e. negative), `length_penalty` > 0.0 promotes longer sequences,
while `length_penalty` < 0.0 encourages shorter sequences.
no_repeat_ngram_size (`int`, *optional*, defaults to 0):
If set to int > 0, all ngrams of that size can only occur once.
encoder_no_repeat_ngram_size (`int`, *optional*, defaults to 0):
If set to int > 0, all ngrams of that size that occur in the `encoder_input_ids` cannot occur in the
`decoder_input_ids`.
bad_words_ids(`List[int]`, *optional*):
List of token ids that are not allowed to be generated. In order to get the tokens of the words that
should not appear in the generated text, use `tokenizer.encode(bad_word, add_prefix_space=True)`.
num_beams (`int`, *optional*, defaults to 1):
Number of beams for beam search. 1 means no beam search.
num_beam_groups (`int`, *optional*, defaults to 1):
Number of groups to divide `num_beams` into in order to ensure diversity among different groups of
beams. [this paper](https://arxiv.org/pdf/1610.02424.pdf) for more details.
diversity_penalty (`float`, *optional*, defaults to 0.0):
This value is subtracted from a beam's score if it generates a token same as any beam from other group
at a particular time. Note that `diversity_penalty` is only effective if `group beam search` is
enabled.
num_return_sequences(`int`, *optional*, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the `generator`'s `[`~generation.GenerationMixin.generate`] function, where
we set `num_return_sequences` to `num_beams`. decoder_start_token_id (`int`, *optional*): If an
encoder-decoder model starts decoding with a different token than *bos*, the id of that token.
n_docs (`int`, *optional*, defaults to `config.n_docs`)
Number of documents to retrieve and/or number of documents for which to generate an answer.
prefix_allowed_tokens_fn: (`Callable[[int, torch.Tensor], List[int]]`, *optional*):
If provided, this function constraints the beam search to allowed tokens only at each step. If not
provided no constraint is applied. This function takes 2 arguments `inputs_ids` and the batch ID
`batch_id`. It has to return a list with the allowed tokens for the next generation step conditioned on
the previously generated tokens `inputs_ids` and the batch ID `batch_id`. This argument is useful for
constrained generation conditioned on the prefix, as described in [Autoregressive Entity
Retrieval](https://arxiv.org/abs/2010.00904).
logits_processor (`LogitsProcessorList`, *optional*):
Custom logits processors that complement the default logits processors built from arguments and a
model's config. If a logit processor is passed that is already created with the arguments or a model's
config an error is thrown.
stopping_criteria (`StoppingCriteriaList`, *optional*):
Custom stopping criteria that complement the default stopping criteria built from arguments and a
model's config. If a stopping criteria is passed that is already created with the arguments or a
model's config an error is thrown.
forced_bos_token_id (`int`, *optional*):
The id of the token to force as the first generated token after the `decoder_start_token_id`. Useful
for multilingual models like [mBART](../model_doc/mbart) where the first generated token needs to be
the target language token.
forced_eos_token_id (`int`, *optional*):
The id of the token to force as the last generated token when `max_length` is reached.
remove_invalid_values (`bool`, *optional*):
Whether to remove possible *nan* and *inf* outputs of the model to prevent the generation method to
crash. Note that using `remove_invalid_values` can slow down generation.
Return:
`torch.LongTensor` of shape `(batch_size * num_return_sequences, sequence_length)`: The generated
sequences. The second dimension (sequence_length) is either equal to `max_length` or shorter if all batches
finished early due to the `eos_token_id`.
"""
# set default parameters
n_docs = n_docs if n_docs is not None else self.config.n_docs
num_beams = num_beams if num_beams is not None else self.config.num_beams
num_beam_groups = num_beam_groups if num_beam_groups is not None else self.config.num_beam_groups
max_length = max_length if max_length is not None else self.config.max_length
num_return_sequences = (
num_return_sequences if num_return_sequences is not None else self.config.num_return_sequences
)
bos_token_id = bos_token_id if bos_token_id is not None else self.config.generator.bos_token_id
eos_token_id = eos_token_id if eos_token_id is not None else self.config.generator.eos_token_id
pad_token_id = pad_token_id if pad_token_id is not None else self.config.generator.pad_token_id
use_cache = use_cache if use_cache is not None else self.config.use_cache
decoder_start_token_id = (
decoder_start_token_id
if decoder_start_token_id is not None
else self.config.generator.decoder_start_token_id
)
remove_invalid_values = (
remove_invalid_values if remove_invalid_values is not None else self.config.remove_invalid_values
)
exponential_decay_length_penalty = (
exponential_decay_length_penalty
if exponential_decay_length_penalty is not None
else self.config.exponential_decay_length_penalty
)
# retrieve docs
if self.retriever is not None and context_input_ids is None:
question_hidden_states = self.question_encoder(input_ids, attention_mask=attention_mask)[0]
out = self.retriever(
input_ids,
question_hidden_states.cpu().detach().to(torch.float32).numpy(),
prefix=self.generator.config.prefix,
n_docs=n_docs,
return_tensors="pt",
)
context_input_ids, context_attention_mask, retrieved_doc_embeds = (
out["context_input_ids"],
out["context_attention_mask"],
out["retrieved_doc_embeds"],
)
# set to correct device
retrieved_doc_embeds = retrieved_doc_embeds.to(question_hidden_states)
context_input_ids = context_input_ids.to(input_ids)
context_attention_mask = context_attention_mask.to(input_ids)
# compute doc_scores
doc_scores = torch.bmm(question_hidden_states.unsqueeze(1), retrieved_doc_embeds.transpose(1, 2)).squeeze(
1
)
assert (context_input_ids.shape[0] % n_docs) == 0, (
f" The first dimension of `context_input_ids` should be a multiple of `n_docs`={n_docs}, but is"
f" {context_input_ids.shape[0]}."
)
# batch_size
batch_size = context_input_ids.shape[0] // n_docs
encoder = self.rag.generator.get_encoder()
encoder_outputs = encoder(input_ids=context_input_ids, attention_mask=context_attention_mask, return_dict=True)
input_ids = torch.full(
(batch_size * num_beams, 1),
decoder_start_token_id,
dtype=torch.long,
device=next(self.parameters()).device,
)
input_ids_seq_length = input_ids.shape[-1]
last_hidden_state = encoder_outputs["last_hidden_state"]
def extend_enc_output(tensor, num_beams=None):
# split into `batch_size`, `num_beams`, `num_docs`
tensor = tensor[None, None, :].reshape((batch_size, 1, n_docs) + tensor.shape[1:])
# repeat same last hidden states over `num_beams` dimension
tensor = tensor.expand((batch_size, num_beams, n_docs) + tensor.shape[3:])
# merge `batch_size`, `num_beams`, `num_docs` dims again
return tensor.reshape((batch_size * num_beams * n_docs,) + tensor.shape[3:])
# correctly extend last_hidden_state and attention mask
context_attention_mask = extend_enc_output(context_attention_mask, num_beams=num_beams)
encoder_outputs["last_hidden_state"] = extend_enc_output(last_hidden_state, num_beams=num_beams)
doc_scores = doc_scores.repeat_interleave(num_beams, dim=0)
# define start_len & additional parameters
model_kwargs["doc_scores"] = doc_scores
model_kwargs["encoder_outputs"] = encoder_outputs
model_kwargs["attention_mask"] = context_attention_mask
model_kwargs["n_docs"] = n_docs
pre_processor = self._get_logits_processor(
repetition_penalty=repetition_penalty,
no_repeat_ngram_size=no_repeat_ngram_size,
encoder_no_repeat_ngram_size=encoder_no_repeat_ngram_size,
input_ids_seq_length=input_ids_seq_length,
encoder_input_ids=context_input_ids,
bad_words_ids=bad_words_ids,
min_length=min_length,
max_length=max_length,
eos_token_id=eos_token_id,
forced_bos_token_id=forced_bos_token_id,
forced_eos_token_id=forced_eos_token_id,
prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
num_beams=num_beams,
num_beam_groups=num_beam_groups,
diversity_penalty=diversity_penalty,
remove_invalid_values=remove_invalid_values,
exponential_decay_length_penalty=exponential_decay_length_penalty,
logits_processor=logits_processor,
renormalize_logits=renormalize_logits,
)
if num_beams == 1:
if num_return_sequences > 1:
raise ValueError(
f"num_return_sequences has to be 1, but is {num_return_sequences} when doing greedy search."
)
return self.greedy_search(
input_ids,
logits_processor=pre_processor,
max_length=max_length,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
**model_kwargs,
)
elif num_beams > 1:
length_penalty = length_penalty if length_penalty is not None else self.config.length_penalty
early_stopping = early_stopping if early_stopping is not None else self.config.early_stopping
if num_return_sequences > num_beams:
raise ValueError("`num_return_sequences` has to be smaller or equal to `num_beams`.")
beam_scorer = BeamSearchScorer(
batch_size=batch_size,
num_beams=num_beams,
device=self.device,
length_penalty=length_penalty,
do_early_stopping=early_stopping,
num_beam_hyps_to_keep=num_return_sequences,
)
return self.beam_search(
input_ids,
beam_scorer,
logits_processor=pre_processor,
max_length=max_length,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
**model_kwargs,
)
else:
raise ValueError(f"`num_beams` has to be an integer strictly superior to 0 (≥ 1), but is {num_beams}")
def get_input_embeddings(self):
return self.rag.generator.get_input_embeddings()
def get_output_embeddings(self):
return self.rag.generator.get_output_embeddings()
def set_output_embeddings(self, new_embeddings):
return self.rag.generator.set_output_embeddings(new_embeddings)
def shift_tokens_right(self, input_ids, start_token_id=None):
"""Shift input ids one token to the right, and pad with start_token_id"""
if start_token_id is None:
start_token_id = self.config.decoder_start_token_id
shifted_input_ids = input_ids.new_zeros(input_ids.shape)
shifted_input_ids[:, 1:] = input_ids[:, :-1].clone()
shifted_input_ids[:, 0] = start_token_id
return shifted_input_ids
def get_nll(self, seq_logits, doc_scores, target, reduce_loss=False, epsilon=0.0, n_docs=None):
n_docs = n_docs if n_docs is not None else self.config.n_docs
# shift tokens left
target = torch.cat(
[target[:, 1:], target.new(target.shape[0], 1).fill_(self.config.generator.pad_token_id)], 1
)
def _mask_pads(ll, smooth_obj):
pad_mask = target.eq(self.config.generator.pad_token_id)
if pad_mask.any():
ll.masked_fill_(pad_mask, 0.0)
smooth_obj.masked_fill_(pad_mask, 0.0)
return ll.squeeze(-1), smooth_obj.squeeze(-1)
rag_logprobs = self.marginalize(seq_logits, doc_scores, n_docs)
target = target.unsqueeze(-1)
assert target.dim() == rag_logprobs.dim()
ll = rag_logprobs.gather(dim=-1, index=target)
smooth_obj = rag_logprobs.sum(dim=-1, keepdim=True) # total sum of all (normalised) logits
ll, smooth_obj = _mask_pads(ll, smooth_obj)
ll = ll.sum(1) # sum over tokens
smooth_obj = smooth_obj.sum(1)
nll_loss = -ll
smooth_loss = -smooth_obj
if reduce_loss:
nll_loss = nll_loss.sum()
smooth_loss = smooth_loss.sum()
eps_i = epsilon / rag_logprobs.size(-1)
loss = (1.0 - epsilon) * nll_loss + eps_i * smooth_loss
return loss
| # coding=utf-8
# Copyright 2020, The RAG Authors and The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""RAG model implementation."""
from dataclasses import dataclass
from typing import Callable, List, Optional, Tuple, Union
import torch
from torch import nn
from ...configuration_utils import PretrainedConfig
from ...generation import BeamSearchScorer, LogitsProcessorList, StoppingCriteriaList
from ...modeling_outputs import ModelOutput
from ...modeling_utils import PreTrainedModel
from ...utils import add_start_docstrings_to_model_forward, logging, replace_return_docstrings
from .configuration_rag import RagConfig
from .retrieval_rag import RagRetriever
logger = logging.get_logger(__name__)
_CONFIG_FOR_DOC = "RagConfig"
@dataclass
class RetrievAugLMMarginOutput(ModelOutput):
"""
Base class for retriever augmented marginalized models outputs.
Args:
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
Language modeling loss.
logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
Prediction scores of the language modeling head. The score is possibly marginalized over all documents for
each vocabulary token.
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
past_key_values (`List[torch.FloatTensor]`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
List of `torch.FloatTensor` of length `config.n_layers`, with each tensor of shape `(2, batch_size,
num_heads, sequence_length, embed_size_per_head)`).
Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used
(see `past_key_values` input) to speed up sequential decoding.
retrieved_doc_embeds (`torch.FloatTensor` of shape `(batch_size, config.n_docs, hidden_size)`, *optional*, returned when *output_retrieved=True*):
Embedded documents retrieved by the retriever. Is used with `question_encoder_last_hidden_state` to compute
the `doc_scores`.
retrieved_doc_ids (`torch.LongTensor` of shape `(batch_size, config.n_docs)`, *optional*, returned when *output_retrieved=True*):
The indexes of the embedded documents retrieved by the retriever.
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
question_encoder_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden states at the output of the last layer of the question encoder pooled output of the
model.
question_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.
question_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the question encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_enc_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the generator encoder of the model.
generator_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.
generator_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_dec_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.
generator_dec_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_cross_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the
weighted average in the cross-attention heads.
"""
loss: Optional[torch.FloatTensor] = None
logits: torch.FloatTensor = None
doc_scores: torch.FloatTensor = None
past_key_values: Optional[List[torch.FloatTensor]] = None
retrieved_doc_embeds: Optional[torch.FloatTensor] = None
retrieved_doc_ids: Optional[torch.LongTensor] = None
context_input_ids: Optional[torch.LongTensor] = None
context_attention_mask: Optional[torch.LongTensor] = None
question_encoder_last_hidden_state: Optional[torch.FloatTensor] = None
question_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
question_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_last_hidden_state: Optional[torch.FloatTensor] = None
generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_cross_attentions: Optional[Tuple[torch.FloatTensor]] = None
@dataclass
class RetrievAugLMOutput(ModelOutput):
"""
Args:
logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
Prediction scores of the language modeling head. The score is possibly marginalized over all documents for
each vocabulary token.
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
past_key_values (`List[torch.FloatTensor]`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
List of `torch.FloatTensor` of length `config.n_layers`, with each tensor of shape `(2, batch_size,
num_heads, sequence_length, embed_size_per_head)`).
Contains precomputed hidden-states (key and values in the attention blocks) of the decoder that can be used
(see `past_key_values` input) to speed up sequential decoding.
retrieved_doc_embeds (`torch.FloatTensor` of shape `(batch_size, config.n_docs, hidden_size)`, *optional*, returned when *output_retrieved=True*):
Embedded documents retrieved by the retriever. Is used with `question_encoder_last_hidden_state` to compute
the `doc_scores`.
retrieved_doc_ids (`torch.LongTensor` of shape `(batch_size, config.n_docs)`, *optional*, returned when *output_retrieved=True*):
The indexes of the embedded documents retrieved by the retriever.
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input ids post-processed from the retrieved documents and the question encoder input_ids by the retriever.
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
question_encoder_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden states at the output of the last layer of the question encoder pooled output of the
model.
question_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the question encoder at the output of each layer plus the initial embedding outputs.
question_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the question encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_enc_last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
Sequence of hidden-states at the output of the last layer of the generator encoder of the model.
generator_enc_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator encoder at the output of each layer plus the initial embedding outputs.
generator_enc_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator encoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_dec_hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
Tuple of `torch.FloatTensor` (one for the output of the embeddings and one for the output of each layer) of
shape `(batch_size, sequence_length, hidden_size)`.
Hidden states of the generator decoder at the output of each layer plus the initial embedding outputs.
generator_dec_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Attentions weights of the generator decoder, after the attention softmax, used to compute the weighted
average in the self-attention heads.
generator_cross_attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
sequence_length)`.
Cross-attentions weights of the generator decoder, after the attention softmax, used to compute the
weighted average in the cross-attention heads.
"""
logits: torch.FloatTensor = None
doc_scores: torch.FloatTensor = None
past_key_values: Optional[List[torch.FloatTensor]] = None
retrieved_doc_embeds: Optional[torch.FloatTensor] = None
retrieved_doc_ids: Optional[torch.LongTensor] = None
context_input_ids: Optional[torch.LongTensor] = None
context_attention_mask: Optional[torch.LongTensor] = None
question_encoder_last_hidden_state: Optional[torch.FloatTensor] = None
question_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
question_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_last_hidden_state: Optional[torch.FloatTensor] = None
generator_enc_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_enc_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
generator_dec_attentions: Optional[Tuple[torch.FloatTensor]] = None
generator_cross_attentions: Optional[Tuple[torch.FloatTensor]] = None
class RagPreTrainedModel(PreTrainedModel):
r"""
RAG models were released with the paper [Retrieval-Augmented Generation for Knowledge-Intensive NLP
Tasks](https://arxiv.org/abs/2005.11401) by Patrick Lewis, Ethan Perez, Aleksandra Piktus et al.
RAG is a retriever augmented model and encapsulate three components: a question encoder, a dataset retriever and a
generator, the encoder and generator are trainable while the retriever is just an indexed dataset.
"""
config_class = RagConfig
base_model_prefix = "rag"
_keys_to_ignore_on_load_missing = [r"position_ids"]
@classmethod
def from_pretrained(cls, *args, **kwargs):
# At the moment fast initialization is not supported
# for composite models
kwargs["_fast_init"] = False
return super().from_pretrained(*args, **kwargs)
@classmethod
def from_pretrained_question_encoder_generator(
cls,
question_encoder_pretrained_model_name_or_path: str = None,
generator_pretrained_model_name_or_path: str = None,
retriever: RagRetriever = None,
**kwargs
) -> PreTrainedModel:
r"""
Instantiates an question encoder and a generator from one or two base classes of the library from pretrained
model checkpoints.
The model is set in evaluation mode by default using `model.eval()` (Dropout modules are deactivated). To train
the model, you need to first set it back in training mode with `model.train()`.
Params:
question_encoder_pretrained_model_name_or_path (`str`, *optional*, defaults to `None`):
Information necessary to initiate the question encoder. Can be either:
- A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing model weights saved using
[`~PreTrainedModel.save_pretrained`], e.g., `./my_model_directory/`.
- A path or url to a *tensorflow index checkpoint file* (e.g, `./tf_model/model.ckpt.index`). In
this case, `from_tf` should be set to `True` and a configuration object should be provided as
`config` argument. This loading path is slower than converting the TensorFlow checkpoint in a
PyTorch model using the provided conversion scripts and loading the PyTorch model afterwards.
generator_pretrained_model_name_or_path (`str`, *optional*, defaults to `None`):
Information necessary to initiate the generator. Can be either:
- A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co.
Valid model ids can be located at the root-level, like `bert-base-uncased`, or namespaced under a
user or organization name, like `dbmdz/bert-base-german-cased`.
- A path to a *directory* containing model weights saved using
[`~PreTrainedModel.save_pretrained`], e.g., `./my_model_directory/`.
- A path or url to a *tensorflow index checkpoint file* (e.g, `./tf_model/model.ckpt.index`). In
this case, `from_tf` should be set to `True` and a configuration object should be provided as
`config` argument. This loading path is slower than converting the TensorFlow checkpoint in a
PyTorch model using the provided conversion scripts and loading the PyTorch model afterwards.
model_args (remaining positional arguments, *optional*):
All remaining positional arguments will be passed to the underlying model's `__init__` method.
retriever ([`RagRetriever`], *optional*):
The retriever to use.
kwwargs (remaining dictionary of keyword arguments, *optional*):
Can be used to update the configuration object (after it being loaded) and initiate the model (e.g.,
`output_attentions=True`).
- To update the question_encoder configuration, use the prefix *question_encoder_* for each
configuration parameter.
- To update the generator configuration, use the prefix *generator_* for each configuration parameter.
- To update the parent model configuration, do not use a prefix for each configuration parameter.
Behaves differently depending on whether a `config` is provided or automatically loaded.
Example:
```python
>>> from transformers import RagModel
>>> # initialize a RAG from two pretrained models.
>>> model = RagModel.from_pretrained_question_encoder_generator(
... "facebook/dpr-question_encoder-single-nq-base", "t5-small"
... )
>>> # saving model after fine-tuning
>>> model.save_pretrained("./rag")
>>> # load fine-tuned model
>>> model = RagModel.from_pretrained("./rag")
```"""
kwargs_question_encoder = {
argument[len("question_encoder_") :]: value
for argument, value in kwargs.items()
if argument.startswith("question_encoder_")
}
kwargs_generator = {
argument[len("generator_") :]: value
for argument, value in kwargs.items()
if argument.startswith("generator_")
}
# remove question_encoder, generator kwargs from kwargs
for key in kwargs_question_encoder.keys():
del kwargs["question_encoder_" + key]
for key in kwargs_generator.keys():
del kwargs["generator_" + key]
# Load and initialize the question_encoder and generator
# The distinction between question_encoder and generator at the model level is made
# by the value of the flag `is_generator` that we need to set correctly.
question_encoder = kwargs_question_encoder.pop("model", None)
if question_encoder is None:
assert question_encoder_pretrained_model_name_or_path is not None, (
"If `model` is not defined as an argument, a `question_encoder_pretrained_model_name_or_path` has to"
" be defined"
)
from ..auto.modeling_auto import AutoModel
if "config" not in kwargs_question_encoder:
from ..auto.configuration_auto import AutoConfig
question_encoder_config, kwargs_question_encoder = AutoConfig.from_pretrained(
question_encoder_pretrained_model_name_or_path,
**kwargs_question_encoder,
return_unused_kwargs=True,
)
kwargs_question_encoder["config"] = question_encoder_config
question_encoder = AutoModel.from_pretrained(
question_encoder_pretrained_model_name_or_path, **kwargs_question_encoder
)
generator = kwargs_generator.pop("model", None)
if generator is None:
assert generator_pretrained_model_name_or_path is not None, (
"If `generator_model` is not defined as an argument, a `generator_pretrained_model_name_or_path` has"
" to be defined"
)
from ..auto.modeling_auto import AutoModelForSeq2SeqLM
if "config" not in kwargs_generator:
from ..auto.configuration_auto import AutoConfig
generator_config, kwargs_generator = AutoConfig.from_pretrained(
generator_pretrained_model_name_or_path, **kwargs_generator, return_unused_kwargs=True
)
kwargs_generator["config"] = generator_config
generator = AutoModelForSeq2SeqLM.from_pretrained(
generator_pretrained_model_name_or_path, **kwargs_generator
)
# instantiate config with corresponding kwargs
config = kwargs.get("config", None)
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
return cls(question_encoder=question_encoder, generator=generator, config=config, retriever=retriever)
RAG_START_DOCSTRING = r"""
RAG is a seq2seq model which encapsulates two core components: a question encoder and a generator. During a forward
pass, we encode the input with the question encoder and pass it to the retriever to extract relevant context
documents. The documents are then prepended to the input. Such contextualized inputs is passed to the generator.
The question encoder can be any *autoencoding* model, preferably [`DPRQuestionEncoder`], and the generator can be
any *seq2seq* model, preferably [`BartForConditionalGeneration`].
The model can be initialized with a [`RagRetriever`] for end-to-end generation or used in combination with the
outputs of a retriever in multiple steps---see examples for more details. The model is compatible any
*autoencoding* model as the `question_encoder` and any *seq2seq* model with language model head as the `generator`.
It has been tested with [`DPRQuestionEncoder`] as the `question_encoder` and [`BartForConditionalGeneration`] or
[`T5ForConditionalGeneration`] as the `generator`.
This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
etc.)
This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
and behavior.
Args:
config ([`RagConfig`]):
Model configuration class with all the parameters of the model. Initializing with a config file does not
load the weights associated with the model, only the configuration. Check out the
[`~PreTrainedModel.from_pretrained`] method to load the model weights.
question_encoder ([`PreTrainedModel`]):
An encoder model compatible with the faiss index encapsulated by the `retriever`.
generator ([`PreTrainedModel`]):
A seq2seq model used as the generator in the RAG architecture.
retriever ([`RagRetriever`]):
A retriever class encapsulating a faiss index queried to obtain context documents for current inputs.
"""
RAG_FORWARD_INPUTS_DOCSTRING = r"""
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary. [`RagConfig`], used to initialize the model, specifies
which generator to use, it also specifies a compatible generator tokenizer. Use that tokenizer class to
obtain the indices.
[What are input IDs?](../glossary#input-ids)
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
encoder_outputs (`tuple(tuple(torch.FloatTensor)`, *optional*)
Tuple consists of (`generator_enc_last_hidden_state`, *optional*: `generator_enc_hidden_states`,
*optional*: `generator_enc_attentions`). `generator_enc_last_hidden_state` of shape `(batch_size, n_docs *
sequence_length, hidden_size)` is a sequence of hidden-states at the output of the last layer of the
generator's encoder.
Used by the ([`RagModel`]) model during decoding.
decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Provide for generation tasks. `None` by default, construct as per instructions for the generator model
you're using with your RAG instance.
decoder_attention_mask (`torch.BoolTensor` of shape `(batch_size, target_sequence_length)`, *optional*):
Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
be used by default.
past_key_values (`tuple(tuple(torch.FloatTensor))`):
Tuple consists of two elements: `encoder_outputs` of the RAG model (see `encoder_outputs`) and
`past_key_values` of the underlying generator. Can be used to speed up decoding. `past_key_values` are used
in the ([`RagTokenForGeneration`]) model during decoding.
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`. If the model has is not initialized with a `retriever` `doc_scores`
has to be provided to the forward pass. `doc_scores` can be computed via
`question_encoder_last_hidden_state` and `retrieved_doc_embeds`, see examples for more information.
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input IDs post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model has is not initialized with a `retriever` ``context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`]. context_attention_mask
(`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*,
returned when *output_retrieved=True*): Attention mask post-processed from the retrieved documents and the
question encoder `input_ids` by the retriever.
If the model has is not initialized with a `retriever` `context_attention_mask` has to be provided to the
forward pass. `context_attention_mask` are returned by [`~RagRetriever.__call__`].
use_cache (`bool`, *optional*, defaults to `True`):
If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
`past_key_values`).
output_attentions (`bool`, *optional*):
Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
tensors for more detail.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
output_retrieved(`bool`, *optional*):
Whether or not to return the `retrieved_doc_embeds`, `retrieved_doc_ids`, `context_input_ids` and
`context_attention_mask`. See returned tensors for more detail.
n_docs (`int`, *optional*, defaults to `config.n_docs``)
Number of documents to retrieve and/or number of documents for which to generate an answer.
"""
@add_start_docstrings_to_model_forward(RAG_START_DOCSTRING)
class RagModel(RagPreTrainedModel):
def __init__(
self,
config: Optional[PretrainedConfig] = None,
question_encoder: Optional[PreTrainedModel] = None,
generator: Optional[PreTrainedModel] = None,
retriever: Optional[RagRetriever] = None, # or maybe just use a `set_retriever(...)` method
**kwargs,
):
assert config is not None or (
question_encoder is not None and generator is not None
), "Either a configuration or an question_encoder and a generator has to be provided."
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
else:
assert isinstance(config, self.config_class), f"config: {config} has to be of type {self.config_class}"
super().__init__(config)
if question_encoder is None:
from ..auto.modeling_auto import AutoModel
question_encoder = AutoModel.from_config(config.question_encoder)
if generator is None:
from ..auto.modeling_auto import AutoModelForSeq2SeqLM
generator = AutoModelForSeq2SeqLM.from_config(config.generator)
self.retriever = retriever
if self.retriever is not None:
assert isinstance(
retriever, RagRetriever
), f"`self.retriever` is of type {type(self.retriever)}, but should be of type `RagRetriever`"
self.retriever = retriever
self.question_encoder = question_encoder
self.generator = generator
self.ctx_encoder = None
self.context_encoder_training = False
@add_start_docstrings_to_model_forward(RAG_FORWARD_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=RetrievAugLMOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.BoolTensor] = None,
past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
doc_scores: Optional[torch.FloatTensor] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask=None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
output_retrieved: Optional[bool] = None,
n_docs: Optional[int] = None,
) -> Union[Tuple[torch.Tensor], RetrievAugLMOutput]:
r"""
Returns:
Example:
```python
>>> from transformers import RagTokenizer, RagRetriever, RagModel
>>> import torch
>>> tokenizer = RagTokenizer.from_pretrained("facebook/rag-token-base")
>>> retriever = RagRetriever.from_pretrained(
... "facebook/rag-token-base", index_name="exact", use_dummy_dataset=True
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = RagModel.from_pretrained("facebook/rag-token-base", retriever=retriever)
>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> outputs = model(input_ids=inputs["input_ids"])
```"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
use_cache = use_cache if use_cache is not None else self.config.use_cache
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
output_hidden_states = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
output_retrieved = output_retrieved if output_retrieved is not None else self.config.output_retrieved
# whether retriever has to be used
has_to_retrieve = (
self.retriever is not None
and (context_input_ids is None or context_attention_mask is None or doc_scores is None)
and encoder_outputs is None
)
# encoder_outputs are pre-computed during RAG-token generation
if encoder_outputs is None:
if has_to_retrieve:
question_enc_outputs = self.question_encoder(
input_ids, attention_mask=attention_mask, return_dict=True
)
question_encoder_last_hidden_state = question_enc_outputs[0] # hidden states of question encoder
retriever_outputs = self.retriever(
input_ids,
question_encoder_last_hidden_state.cpu().detach().to(torch.float32).numpy(),
prefix=self.generator.config.prefix,
n_docs=n_docs,
return_tensors="pt",
)
if self.context_encoder_training:
(
context_input_ids,
context_attention_mask,
retrieved_doc_embeds,
retrived_doc_input_ids,
retrived_doc_attention_mask,
retrieved_doc_ids,
) = (
retriever_outputs["context_input_ids"],
retriever_outputs["context_attention_mask"],
retriever_outputs["retrieved_doc_embeds"],
retriever_outputs["tokenized_doc_ids"],
retriever_outputs["tokenized_doc_attention_mask"],
retriever_outputs["doc_ids"],
)
context_input_ids = context_input_ids.to(input_ids)
context_attention_mask = context_attention_mask.to(input_ids)
retrived_doc_input_ids = retrived_doc_input_ids.to(input_ids)
retrived_doc_attention_mask = retrived_doc_attention_mask.to(input_ids)
retrieved_doc_embeds = self.ctx_encoder(
retrived_doc_input_ids, attention_mask=retrived_doc_attention_mask, return_dict=True
).pooler_output
retrieved_doc_embeds = retrieved_doc_embeds.view(
-1, n_docs, question_encoder_last_hidden_state.shape[1]
) # reshaping
# compute doc_scores involving ctx_encoder
doc_scores = torch.bmm(
question_encoder_last_hidden_state.unsqueeze(1), retrieved_doc_embeds.transpose(1, 2)
).squeeze(1)
else:
context_input_ids, context_attention_mask, retrieved_doc_embeds, retrieved_doc_ids = (
retriever_outputs["context_input_ids"],
retriever_outputs["context_attention_mask"],
retriever_outputs["retrieved_doc_embeds"],
retriever_outputs["doc_ids"],
)
# set to correct device
retrieved_doc_embeds = retrieved_doc_embeds.to(question_encoder_last_hidden_state)
context_input_ids = context_input_ids.to(input_ids)
context_attention_mask = context_attention_mask.to(input_ids)
# compute doc_scores
doc_scores = torch.bmm(
question_encoder_last_hidden_state.unsqueeze(1), retrieved_doc_embeds.transpose(1, 2)
).squeeze(1)
else:
assert context_input_ids is not None, (
"Make sure that `context_input_ids` are passed, if no `retriever` is set. Alternatively, you can"
" set a retriever using the `set_retriever(...)` function."
)
assert context_attention_mask is not None, (
"Make sure that `context_attention_mask` are passed, if no `retriever` is set. Alternatively, you"
" can set a retriever using the `set_retriever(...)` function."
)
assert doc_scores is not None, (
"Make sure that `doc_scores` are passed, if no `retriever` is set. Alternatively, you can set a"
" retriever using the `set_retriever(...)` function."
)
assert (
doc_scores is not None
), "Make sure that `doc_scores` are passed when passing `encoder_outputs` to the forward function."
assert (doc_scores.shape[1] % n_docs) == 0, (
f" The first dimension of `context_input_ids` should be a multiple of `n_docs`={n_docs}, but is"
f" {context_input_ids.shape[0]}."
)
# Decoder input without context documents
if decoder_input_ids is not None:
decoder_input_ids = decoder_input_ids.repeat_interleave(n_docs, dim=0)
if decoder_attention_mask is not None:
decoder_attention_mask = decoder_attention_mask.repeat_interleave(n_docs, dim=0)
gen_outputs = self.generator(
input_ids=context_input_ids,
attention_mask=context_attention_mask,
encoder_outputs=encoder_outputs,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
past_key_values=past_key_values,
use_cache=use_cache,
output_attentions=output_attentions,
return_dict=True,
)
if not has_to_retrieve:
question_encoder_last_hidden_state = None
question_enc_hidden_states = None
question_enc_attentions = None
retrieved_doc_embeds = None
retrieved_doc_ids = None
else:
question_enc_hidden_states = question_enc_outputs.hidden_states
question_enc_attentions = question_enc_outputs.attentions
if not has_to_retrieve or not output_retrieved:
# don't output retrieved docs
context_input_ids = (None,)
context_attention_mask = None
retrieved_doc_embeds = None
retrieved_doc_ids = None
return RetrievAugLMOutput(
logits=gen_outputs.logits,
doc_scores=doc_scores,
past_key_values=gen_outputs.past_key_values,
context_input_ids=context_input_ids,
context_attention_mask=context_attention_mask,
retrieved_doc_embeds=retrieved_doc_embeds,
retrieved_doc_ids=retrieved_doc_ids,
question_encoder_last_hidden_state=question_encoder_last_hidden_state,
question_enc_hidden_states=question_enc_hidden_states,
question_enc_attentions=question_enc_attentions,
generator_enc_last_hidden_state=gen_outputs.encoder_last_hidden_state,
generator_enc_hidden_states=gen_outputs.encoder_hidden_states,
generator_enc_attentions=gen_outputs.encoder_attentions,
generator_dec_hidden_states=gen_outputs.decoder_hidden_states,
generator_dec_attentions=gen_outputs.decoder_attentions,
generator_cross_attentions=gen_outputs.cross_attentions,
)
@add_start_docstrings_to_model_forward(
"""
A RAG-sequence model implementation. It performs RAG-sequence specific marginalization in the forward pass.
""",
RAG_START_DOCSTRING,
)
class RagSequenceForGeneration(RagPreTrainedModel):
def __init__(
self,
config: Optional[PretrainedConfig] = None,
question_encoder: Optional[PreTrainedModel] = None,
generator: Optional[PreTrainedModel] = None,
retriever: Optional[RagRetriever] = None,
**kwargs,
):
assert config is not None or (
question_encoder is not None and generator is not None
), "Either a configuration or an encoder and a generator has to be provided."
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
super().__init__(config)
# instantiate model
self.rag = RagModel(config=config, question_encoder=question_encoder, generator=generator, retriever=retriever)
def set_retriever(self, retriever: RagRetriever):
self.rag.retriever = retriever
def set_context_encoder_for_training(self, ctx_encoder: PreTrainedModel):
self.rag.context_encoder_training = True
self.rag.ctx_encoder = ctx_encoder
@add_start_docstrings_to_model_forward(RAG_FORWARD_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=RetrievAugLMMarginOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.Tensor] = None,
encoder_outputs: Optional[Tuple[Tuple[torch.Tensor]]] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.BoolTensor] = None,
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
output_retrieved: Optional[bool] = None,
exclude_bos_score: Optional[bool] = None,
reduce_loss: Optional[bool] = None,
labels: Optional[torch.LongTensor] = None,
n_docs: Optional[int] = None,
**kwargs # needs kwargs for generation
) -> RetrievAugLMMarginOutput:
r"""
exclude_bos_score (`bool`, *optional*):
Only relevant if `labels` is passed. If `True`, the score of the BOS token is disregarded when computing
the loss.
reduce_loss (`bool`, *optional*):
Only relevant if `labels` is passed. If `True`, the NLL loss is reduced using the `torch.Tensor.sum`
operation.
kwargs (`Dict[str, any]`, optional, defaults to *{}*):
Legacy dictionary, which is required so that model can use *generate()* function.
Returns:
Example:
```python
>>> from transformers import RagTokenizer, RagRetriever, RagSequenceForGeneration
>>> import torch
>>> tokenizer = RagTokenizer.from_pretrained("facebook/rag-sequence-nq")
>>> retriever = RagRetriever.from_pretrained(
... "facebook/rag-sequence-nq", index_name="exact", use_dummy_dataset=True
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = RagSequenceForGeneration.from_pretrained("facebook/rag-token-nq", retriever=retriever)
>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> targets = tokenizer(text_target="In Paris, there are 10 million people.", return_tensors="pt")
>>> input_ids = inputs["input_ids"]
>>> labels = targets["input_ids"]
>>> outputs = model(input_ids=input_ids, labels=labels)
>>> # or use retriever separately
>>> model = RagSequenceForGeneration.from_pretrained("facebook/rag-sequence-nq", use_dummy_dataset=True)
>>> # 1. Encode
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.detach().numpy(), return_tensors="pt")
>>> doc_scores = torch.bmm(
... question_hidden_states.unsqueeze(1), docs_dict["retrieved_doc_embeds"].float().transpose(1, 2)
... ).squeeze(1)
>>> # 3. Forward to generator
>>> outputs = model(
... context_input_ids=docs_dict["context_input_ids"],
... context_attention_mask=docs_dict["context_attention_mask"],
... doc_scores=doc_scores,
... decoder_input_ids=labels,
... )
```"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
exclude_bos_score = exclude_bos_score if exclude_bos_score is not None else self.config.exclude_bos_score
reduce_loss = reduce_loss if reduce_loss is not None else self.config.reduce_loss
if labels is not None:
if decoder_input_ids is None:
decoder_input_ids = labels
use_cache = False
outputs = self.rag(
input_ids=input_ids,
attention_mask=attention_mask,
encoder_outputs=encoder_outputs,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
context_input_ids=context_input_ids,
context_attention_mask=context_attention_mask,
doc_scores=doc_scores,
past_key_values=past_key_values,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
output_retrieved=output_retrieved,
n_docs=n_docs,
)
loss = None
if labels is not None:
loss = self.get_nll(
outputs.logits,
outputs.doc_scores,
decoder_input_ids,
reduce_loss=reduce_loss,
epsilon=self.config.label_smoothing,
exclude_bos_score=exclude_bos_score,
n_docs=n_docs,
)
return RetrievAugLMMarginOutput(
loss=loss,
logits=outputs.logits,
doc_scores=outputs.doc_scores,
past_key_values=outputs.past_key_values,
context_input_ids=outputs.context_input_ids,
context_attention_mask=outputs.context_attention_mask,
retrieved_doc_embeds=outputs.retrieved_doc_embeds,
retrieved_doc_ids=outputs.retrieved_doc_ids,
question_encoder_last_hidden_state=outputs.question_encoder_last_hidden_state,
question_enc_hidden_states=outputs.question_enc_hidden_states,
question_enc_attentions=outputs.question_enc_attentions,
generator_enc_last_hidden_state=outputs.generator_enc_last_hidden_state,
generator_enc_hidden_states=outputs.generator_enc_hidden_states,
generator_enc_attentions=outputs.generator_enc_attentions,
generator_dec_hidden_states=outputs.generator_dec_hidden_states,
generator_dec_attentions=outputs.generator_dec_attentions,
generator_cross_attentions=outputs.generator_cross_attentions,
)
@property
def retriever(self):
return self.rag.retriever
@property
def generator(self):
return self.rag.generator
@property
def question_encoder(self):
return self.rag.question_encoder
@torch.no_grad()
def generate(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.LongTensor] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
do_deduplication: Optional[bool] = None, # defaults to True
num_return_sequences: Optional[int] = None, # defaults to 1
num_beams: Optional[int] = None, # defaults to 1
n_docs: Optional[int] = None,
**model_kwargs
) -> torch.LongTensor:
"""
Implements RAG sequence "thorough" decoding. Read the [`~generation.GenerationMixin.generate`]` documentation
for more information on how to set other generate input parameters.
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
The sequence used as a prompt for the generation. If `input_ids` is not passed, then
`context_input_ids` has to be provided.
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input IDs post-processed from the retrieved documents and the question encoder input_ids by the
retriever.
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model is not initialized with a `retriever` or `input_ids` is not given, `context_input_ids` and
`context_attention_mask` have to be provided to the forward pass. They are returned by
[`~RagRetriever.__call__`].
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
If the model is not initialized with a `retriever` or `input_ids` is not given, `doc_scores` has to be
provided to the forward pass. `doc_scores` are returned by [`~RagRetriever.__call__`].
do_deduplication (`bool`, *optional*):
Whether or not to deduplicate the generations from different context documents for a given input. Has
to be set to `False` if used while training with distributed backend.
num_return_sequences(`int`, *optional*, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the `generator`'s `[`~generation.GenerationMixin.generate`]` function,
where we set `num_return_sequences` to `num_beams`.
num_beams (`int`, *optional*, defaults to 1):
Number of beams for beam search. 1 means no beam search.
n_docs (`int`, *optional*, defaults to `config.n_docs`)
Number of documents to retrieve and/or number of documents for which to generate an answer.
kwargs:
Additional kwargs will be passed to [`~generation.GenerationMixin.generate`].
Return:
`torch.LongTensor` of shape `(batch_size * num_return_sequences, sequence_length)`: The generated
sequences. The second dimension (sequence length) is either equal to `max_length` or shorter if all batches
finished early due to the `eos_token_id`.
"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
do_deduplication = do_deduplication if do_deduplication is not None else self.config.do_deduplication
num_doc_return_sequences = (
num_return_sequences if num_return_sequences is not None else self.config.num_return_sequences
)
num_beams = num_beams if num_beams is not None else self.config.num_beams
assert (
input_ids is not None or context_input_ids is not None
), " At least one of input_ids or context_input_ids must be given"
if self.retriever is not None and context_input_ids is None:
question_hidden_states = self.question_encoder(input_ids, attention_mask=attention_mask)[0]
context_input_ids = self.retriever(
input_ids,
question_hidden_states.cpu().detach().to(torch.float32).numpy(),
prefix=self.generator.config.prefix,
n_docs=n_docs,
return_tensors="pt",
)["context_input_ids"]
# set to correct device
context_input_ids = context_input_ids.to(input_ids)
hypos = []
model_kwargs["num_beams"] = num_beams
model_kwargs["num_return_sequences"] = num_beams
model_kwargs["attention_mask"] = None
batch_size = input_ids.shape[0] if input_ids is not None else context_input_ids.shape[0] // n_docs
for index in range(batch_size):
# first, generate beams from documents:
generator_input_ids = context_input_ids[index * n_docs : (index + 1) * n_docs] # (n_docs, max_len)
output_sequences = self.generator.generate(
generator_input_ids,
**model_kwargs,
) # n_docs * n_beam, tgt_len
if do_deduplication:
# do_deduplication, max_output_len
output_sequences = torch.stack(list({str(k.tolist()): k for k in output_sequences}.values()))
num_candidates = output_sequences.shape[
0
] # after deduplication, this number can be less than n_docs*n_beam
# then, run model forwards to get nll scores:
if input_ids is not None:
new_input_ids = input_ids[index : index + 1].repeat(num_candidates, 1)
outputs = self(new_input_ids, labels=output_sequences, exclude_bos_score=True)
else: # input_ids is None, need context_input_ids/mask and doc_scores
assert context_attention_mask is not None, (
"Make sure that `context_attention_mask` are passed, if no `input_ids` is set. Alternatively, you"
" can set a retriever using the `set_retriever(...)` function."
)
assert doc_scores is not None, (
"Make sure that `doc_scores` are passed, if no `input_ids` is set. Alternatively, you can set a"
" retriever using the `set_retriever(...)` function."
)
individual_input_ids = generator_input_ids.repeat(
num_candidates, 1
) # (num_candidates*n_docs, max_len)
individual_attention_mask = context_attention_mask[index * n_docs : (index + 1) * n_docs]
individual_attention_mask = individual_attention_mask.repeat(num_candidates, 1)
individual_doc_scores = doc_scores[index : (index + 1), :] # doc_scores.shape = [batch, n_docs]
individual_doc_scores = individual_doc_scores.repeat(num_candidates, 1) # [num_candidates, n_docs]
outputs = self(
context_input_ids=individual_input_ids,
context_attention_mask=individual_attention_mask,
doc_scores=individual_doc_scores,
labels=output_sequences,
exclude_bos_score=True,
)
top_cand_inds = (-outputs["loss"]).topk(num_doc_return_sequences)[1]
# add hypothesis
hypos.append(output_sequences[top_cand_inds])
return self._cat_and_pad(hypos, pad_token_id=self.config.generator.pad_token_id)
def get_nll(
self, seq_logits, doc_scores, target, reduce_loss=False, epsilon=0.0, exclude_bos_score=False, n_docs=None
):
# shift tokens left
target = torch.cat(
[target[:, 1:], target.new(target.shape[0], 1).fill_(self.config.generator.pad_token_id)], 1
)
n_docs = n_docs if n_docs is not None else self.config.n_docs
# bos_token_id is None for T5
bos_token_id = self.config.bos_token_id or self.config.generator.bos_token_id
use_bos = bos_token_id is not None and target[:, 0].eq(bos_token_id).all()
def _mask_pads(ll, smooth_obj):
pad_mask = target.eq(self.config.generator.pad_token_id)
if pad_mask.any():
ll.masked_fill_(pad_mask, 0.0)
smooth_obj.masked_fill_(pad_mask, 0.0)
return ll.squeeze(-1), smooth_obj.squeeze(-1)
# seq_logits dim = (batch*n_docs, tgt_len , #vocabs)
seq_logprobs = nn.functional.log_softmax(seq_logits, dim=-1).view(
seq_logits.shape[0] // n_docs, n_docs, -1, seq_logits.size(-1)
) # batch_size x n_docs x tgt_len x #vocab_size
doc_logprobs = nn.functional.log_softmax(doc_scores, dim=1).unsqueeze(-1).unsqueeze(-1)
# RAG-sequence marginalization
first_token_scores = seq_logprobs[:, :, :1, :]
second_token_scores = seq_logprobs[:, :, 1:2, :]
remainder = seq_logprobs[:, :, 2:, :]
rag_logprobs = torch.cat([first_token_scores, second_token_scores + doc_logprobs, remainder], dim=2)
# calculate loss
target = target.unsqueeze(1).unsqueeze(-1).repeat(1, n_docs, 1, 1)
assert target.dim() == rag_logprobs.dim()
ll = rag_logprobs.gather(dim=-1, index=target)
smooth_obj = rag_logprobs.sum(dim=-1, keepdim=True) # total sum of all (normalised) logits
ll, smooth_obj = _mask_pads(ll, smooth_obj)
# sum over tokens, exclude bos while scoring
ll = ll[:, :, 1:].sum(2) if exclude_bos_score and use_bos else ll.sum(2)
smooth_obj = smooth_obj.sum(2)
ll = ll.logsumexp(1) # logsumexp over docs
smooth_obj = smooth_obj.logsumexp(1)
nll_loss = -ll
smooth_loss = -smooth_obj
if reduce_loss:
nll_loss = nll_loss.sum()
smooth_loss = smooth_loss.sum()
eps_i = epsilon / rag_logprobs.size(-1)
loss = (1.0 - epsilon) * nll_loss + eps_i * smooth_loss
return loss
@staticmethod
def _cat_and_pad(tensors, pad_token_id):
output = (
tensors[0].new(sum([t.shape[0] for t in tensors]), max([t.shape[1] for t in tensors])).fill_(pad_token_id)
)
ind = 0
for t in tensors:
output[ind : ind + t.shape[0], : t.shape[1]] = t
ind += t.shape[0]
return output
@add_start_docstrings_to_model_forward(
"""
A RAG-token model implementation. It performs RAG-token specific marginalization in the forward pass.
""",
RAG_START_DOCSTRING,
)
class RagTokenForGeneration(RagPreTrainedModel):
def __init__(
self,
config: Optional[PretrainedConfig] = None,
question_encoder: Optional[PreTrainedModel] = None,
generator: Optional[PreTrainedModel] = None,
retriever: Optional[RagRetriever] = None,
**kwargs,
):
assert config is not None or (
question_encoder is not None and generator is not None
), "Either a configuration or an encoder and a generator has to be provided."
if config is None:
config = RagConfig.from_question_encoder_generator_configs(
question_encoder.config, generator.config, **kwargs
)
super().__init__(config)
# instantiate model
self.rag = RagModel(config=config, question_encoder=question_encoder, generator=generator, retriever=retriever)
def set_retriever(self, retriever: RagRetriever):
self.rag.retriever = retriever
def set_context_encoder_for_training(self, ctx_encoder: PreTrainedModel):
self.rag.context_encoder_training = True
self.rag.ctx_encoder = ctx_encoder
def prepare_inputs_for_generation(
self,
decoder_input_ids,
past=None,
attention_mask=None,
use_cache=None,
encoder_outputs=None,
doc_scores=None,
n_docs=None,
**kwargs
):
if past is not None:
# if past is defined use only last decoder_input_ids
decoder_input_ids = decoder_input_ids[:, -1:]
return {
"input_ids": None,
"encoder_outputs": encoder_outputs,
"doc_scores": doc_scores,
"context_attention_mask": attention_mask,
"decoder_input_ids": decoder_input_ids,
"past_key_values": past,
"use_cache": use_cache,
"do_marginalize": True,
"n_docs": n_docs,
}
@property
def retriever(self):
return self.rag.retriever
@property
def generator(self):
return self.rag.generator
@property
def question_encoder(self):
return self.rag.question_encoder
@staticmethod
def _reorder_cache(past, beam_idx):
"""Reorders cache for generation. BART-inspired but we need to take care of the extra dimension for docs"""
def _reorder_stacked(hidden_states, new_order):
n_docs = hidden_states.shape[0] // new_order.shape[0]
hidden_states = hidden_states.view(-1, n_docs, *hidden_states.shape[1:])
hidden_states = hidden_states.index_select(0, new_order)
result = hidden_states.view(-1, *hidden_states.shape[2:])
return result
reordered_past = ()
for layer_past in past:
# get the correct batch idx from decoder layer's batch dim for cross and self-attn
reordered_past += (tuple(_reorder_stacked(past_state, beam_idx) for past_state in layer_past),)
return reordered_past
def marginalize(self, seq_logits, doc_scores, n_docs=None):
n_docs = n_docs if n_docs is not None else self.config.n_docs
# RAG-token marginalization
seq_logprobs = nn.functional.log_softmax(seq_logits, dim=-1).view(
seq_logits.shape[0] // n_docs, n_docs, -1, seq_logits.size(-1)
)
doc_logprobs = torch.log_softmax(doc_scores, dim=1)
log_prob_sum = seq_logprobs + doc_logprobs.unsqueeze(-1).unsqueeze(-1)
return torch.logsumexp(log_prob_sum, dim=1)
@add_start_docstrings_to_model_forward(RAG_FORWARD_INPUTS_DOCSTRING)
@replace_return_docstrings(output_type=RetrievAugLMMarginOutput, config_class=_CONFIG_FOR_DOC)
def forward(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.FloatTensor] = None,
encoder_outputs: Optional[Tuple[Tuple[torch.Tensor]]] = None,
decoder_input_ids: Optional[torch.LongTensor] = None,
decoder_attention_mask: Optional[torch.BoolTensor] = None,
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
use_cache: Optional[bool] = None,
output_attentions: Optional[bool] = None,
output_hidden_states: Optional[bool] = None,
output_retrieved: Optional[bool] = None,
do_marginalize: Optional[bool] = None,
reduce_loss: Optional[bool] = None,
labels: Optional[torch.LongTensor] = None,
n_docs: Optional[int] = None,
**kwargs # needs kwargs for generation
) -> RetrievAugLMMarginOutput:
r"""
do_marginalize (`bool`, *optional*):
If `True`, the logits are marginalized over all documents by making use of
`torch.nn.functional.log_softmax`.
reduce_loss (`bool`, *optional*):
Only relevant if `labels` is passed. If `True`, the NLL loss is reduced using the `torch.Tensor.sum`
operation.
kwargs (`Dict[str, any]`, optional, defaults to *{}*):
Legacy dictionary, which is required so that model can use *generate()* function.
Returns:
Example:
```python
>>> from transformers import RagTokenizer, RagRetriever, RagTokenForGeneration
>>> import torch
>>> tokenizer = RagTokenizer.from_pretrained("facebook/rag-token-nq")
>>> retriever = RagRetriever.from_pretrained(
... "facebook/rag-token-nq", index_name="exact", use_dummy_dataset=True
... )
>>> # initialize with RagRetriever to do everything in one forward call
>>> model = RagTokenForGeneration.from_pretrained("facebook/rag-token-nq", retriever=retriever)
>>> inputs = tokenizer("How many people live in Paris?", return_tensors="pt")
>>> targets = tokenizer(text_target="In Paris, there are 10 million people.", return_tensors="pt")
>>> input_ids = inputs["input_ids"]
>>> labels = targets["input_ids"]
>>> outputs = model(input_ids=input_ids, labels=labels)
>>> # or use retriever separately
>>> model = RagTokenForGeneration.from_pretrained("facebook/rag-token-nq", use_dummy_dataset=True)
>>> # 1. Encode
>>> question_hidden_states = model.question_encoder(input_ids)[0]
>>> # 2. Retrieve
>>> docs_dict = retriever(input_ids.numpy(), question_hidden_states.detach().numpy(), return_tensors="pt")
>>> doc_scores = torch.bmm(
... question_hidden_states.unsqueeze(1), docs_dict["retrieved_doc_embeds"].float().transpose(1, 2)
... ).squeeze(1)
>>> # 3. Forward to generator
>>> outputs = model(
... context_input_ids=docs_dict["context_input_ids"],
... context_attention_mask=docs_dict["context_attention_mask"],
... doc_scores=doc_scores,
... decoder_input_ids=labels,
... )
>>> # or directly generate
>>> generated = model.generate(
... context_input_ids=docs_dict["context_input_ids"],
... context_attention_mask=docs_dict["context_attention_mask"],
... doc_scores=doc_scores,
... )
>>> generated_string = tokenizer.batch_decode(generated, skip_special_tokens=True)
```"""
n_docs = n_docs if n_docs is not None else self.config.n_docs
do_marginalize = do_marginalize if do_marginalize is not None else self.config.do_marginalize
reduce_loss = reduce_loss if reduce_loss is not None else self.config.reduce_loss
if labels is not None:
if decoder_input_ids is None:
decoder_input_ids = labels
use_cache = False
outputs = self.rag(
input_ids=input_ids,
attention_mask=attention_mask,
encoder_outputs=encoder_outputs,
decoder_input_ids=decoder_input_ids,
decoder_attention_mask=decoder_attention_mask,
context_input_ids=context_input_ids,
context_attention_mask=context_attention_mask,
doc_scores=doc_scores,
past_key_values=past_key_values,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
output_retrieved=output_retrieved,
n_docs=n_docs,
)
loss = None
logits = outputs.logits
if labels is not None:
assert decoder_input_ids is not None
loss = self.get_nll(
outputs.logits,
outputs.doc_scores,
labels,
reduce_loss=reduce_loss,
epsilon=self.config.label_smoothing,
n_docs=n_docs,
)
if do_marginalize:
logits = self.marginalize(logits, outputs.doc_scores, n_docs)
return RetrievAugLMMarginOutput(
loss=loss,
logits=logits,
doc_scores=outputs.doc_scores,
past_key_values=outputs.past_key_values,
context_input_ids=outputs.context_input_ids,
context_attention_mask=outputs.context_attention_mask,
retrieved_doc_embeds=outputs.retrieved_doc_embeds,
retrieved_doc_ids=outputs.retrieved_doc_ids,
question_encoder_last_hidden_state=outputs.question_encoder_last_hidden_state,
question_enc_hidden_states=outputs.question_enc_hidden_states,
question_enc_attentions=outputs.question_enc_attentions,
generator_enc_last_hidden_state=outputs.generator_enc_last_hidden_state,
generator_enc_hidden_states=outputs.generator_enc_hidden_states,
generator_enc_attentions=outputs.generator_enc_attentions,
generator_dec_hidden_states=outputs.generator_dec_hidden_states,
generator_dec_attentions=outputs.generator_dec_attentions,
generator_cross_attentions=outputs.generator_cross_attentions,
)
@torch.no_grad()
def generate(
self,
input_ids: Optional[torch.LongTensor] = None,
attention_mask: Optional[torch.LongTensor] = None,
context_input_ids: Optional[torch.LongTensor] = None,
context_attention_mask: Optional[torch.LongTensor] = None,
doc_scores: Optional[torch.FloatTensor] = None,
max_length: Optional[int] = None,
min_length: Optional[int] = None,
early_stopping: Optional[bool] = None,
use_cache: Optional[bool] = None,
num_beams: Optional[int] = None,
num_beam_groups: Optional[int] = None,
diversity_penalty: Optional[float] = None,
bos_token_id: Optional[int] = None,
pad_token_id: Optional[int] = None,
eos_token_id: Optional[int] = None,
length_penalty: Optional[float] = None,
no_repeat_ngram_size: Optional[int] = None,
encoder_no_repeat_ngram_size: Optional[int] = None,
repetition_penalty: Optional[float] = None,
bad_words_ids: Optional[List[List[int]]] = None,
num_return_sequences: Optional[int] = None,
decoder_start_token_id: Optional[int] = None,
n_docs: Optional[int] = None,
prefix_allowed_tokens_fn: Callable[[int, torch.Tensor], List[int]] = None,
logits_processor: Optional[LogitsProcessorList] = LogitsProcessorList(),
renormalize_logits: Optional[bool] = None,
stopping_criteria: Optional[StoppingCriteriaList] = StoppingCriteriaList(),
forced_bos_token_id: Optional[int] = None,
forced_eos_token_id: Optional[int] = None,
remove_invalid_values: Optional[bool] = None,
exponential_decay_length_penalty: Optional[Tuple[Union[int, float]]] = None,
**model_kwargs
) -> torch.LongTensor:
"""
Implements RAG token decoding.
Args:
input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
The sequence used as a prompt for the generation. If `input_ids` is not passed, then
`context_input_ids` has to be provided.
attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
- 1 for tokens that are **not masked**,
- 0 for tokens that are **masked**.
[What are attention masks?](../glossary#attention-mask)
context_input_ids (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Input IDs post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model has is not initialized with a `retriever`, `context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`].
context_attention_mask (`torch.LongTensor` of shape `(batch_size * config.n_docs, config.max_combined_length)`, *optional*, returned when *output_retrieved=True*):
Attention mask post-processed from the retrieved documents and the question encoder `input_ids` by the
retriever.
If the model has is not initialized with a `retriever`, `context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`].
doc_scores (`torch.FloatTensor` of shape `(batch_size, config.n_docs)`):
Score between each retrieved document embeddings (see `retrieved_doc_embeds`) and
`question_encoder_last_hidden_state`.
If the model has is not initialized with a `retriever`, `context_input_ids` has to be provided to the
forward pass. `context_input_ids` are returned by [`~RagRetriever.__call__`].
max_length (`int`, *optional*, defaults to 20):
The maximum length of the sequence to be generated.
min_length (`int`, *optional*, defaults to 10):
The minimum length of the sequence to be generated.
early_stopping (`bool`, *optional*, defaults to `False`):
Whether or not to stop the beam search when at least `num_beams` sentences are finished per batch or
not.
use_cache: (`bool`, *optional*, defaults to `True`):
Whether or not the model should use the past last key/values attentions (if applicable to the model) to
speed up decoding.
pad_token_id (`int`, *optional*):
The id of the *padding* token.
bos_token_id (`int`, *optional*):
The id of the *beginning-of-sequence* token.
eos_token_id (`int`, *optional*):
The id of the *end-of-sequence* token.
length_penalty (`float`, *optional*, defaults to 1.0):
Exponential penalty to the length that is used with beam-based generation. It is applied as an exponent
to the sequence length, which in turn is used to divide the score of the sequence. Since the score is
the log likelihood of the sequence (i.e. negative), `length_penalty` > 0.0 promotes longer sequences,
while `length_penalty` < 0.0 encourages shorter sequences.
no_repeat_ngram_size (`int`, *optional*, defaults to 0):
If set to int > 0, all ngrams of that size can only occur once.
encoder_no_repeat_ngram_size (`int`, *optional*, defaults to 0):
If set to int > 0, all ngrams of that size that occur in the `encoder_input_ids` cannot occur in the
`decoder_input_ids`.
bad_words_ids(`List[int]`, *optional*):
List of token ids that are not allowed to be generated. In order to get the tokens of the words that
should not appear in the generated text, use `tokenizer.encode(bad_word, add_prefix_space=True)`.
num_beams (`int`, *optional*, defaults to 1):
Number of beams for beam search. 1 means no beam search.
num_beam_groups (`int`, *optional*, defaults to 1):
Number of groups to divide `num_beams` into in order to ensure diversity among different groups of
beams. [this paper](https://arxiv.org/pdf/1610.02424.pdf) for more details.
diversity_penalty (`float`, *optional*, defaults to 0.0):
This value is subtracted from a beam's score if it generates a token same as any beam from other group
at a particular time. Note that `diversity_penalty` is only effective if `group beam search` is
enabled.
num_return_sequences(`int`, *optional*, defaults to 1):
The number of independently computed returned sequences for each element in the batch. Note that this
is not the value we pass to the `generator`'s `[`~generation.GenerationMixin.generate`] function, where
we set `num_return_sequences` to `num_beams`. decoder_start_token_id (`int`, *optional*): If an
encoder-decoder model starts decoding with a different token than *bos*, the id of that token.
n_docs (`int`, *optional*, defaults to `config.n_docs`)
Number of documents to retrieve and/or number of documents for which to generate an answer.
prefix_allowed_tokens_fn: (`Callable[[int, torch.Tensor], List[int]]`, *optional*):
If provided, this function constraints the beam search to allowed tokens only at each step. If not
provided no constraint is applied. This function takes 2 arguments `inputs_ids` and the batch ID
`batch_id`. It has to return a list with the allowed tokens for the next generation step conditioned on
the previously generated tokens `inputs_ids` and the batch ID `batch_id`. This argument is useful for
constrained generation conditioned on the prefix, as described in [Autoregressive Entity
Retrieval](https://arxiv.org/abs/2010.00904).
logits_processor (`LogitsProcessorList`, *optional*):
Custom logits processors that complement the default logits processors built from arguments and a
model's config. If a logit processor is passed that is already created with the arguments or a model's
config an error is thrown.
stopping_criteria (`StoppingCriteriaList`, *optional*):
Custom stopping criteria that complement the default stopping criteria built from arguments and a
model's config. If a stopping criteria is passed that is already created with the arguments or a
model's config an error is thrown.
forced_bos_token_id (`int`, *optional*):
The id of the token to force as the first generated token after the `decoder_start_token_id`. Useful
for multilingual models like [mBART](../model_doc/mbart) where the first generated token needs to be
the target language token.
forced_eos_token_id (`int`, *optional*):
The id of the token to force as the last generated token when `max_length` is reached.
remove_invalid_values (`bool`, *optional*):
Whether to remove possible *nan* and *inf* outputs of the model to prevent the generation method to
crash. Note that using `remove_invalid_values` can slow down generation.
Return:
`torch.LongTensor` of shape `(batch_size * num_return_sequences, sequence_length)`: The generated
sequences. The second dimension (sequence_length) is either equal to `max_length` or shorter if all batches
finished early due to the `eos_token_id`.
"""
# set default parameters
n_docs = n_docs if n_docs is not None else self.config.n_docs
num_beams = num_beams if num_beams is not None else self.config.num_beams
num_beam_groups = num_beam_groups if num_beam_groups is not None else self.config.num_beam_groups
max_length = max_length if max_length is not None else self.config.max_length
num_return_sequences = (
num_return_sequences if num_return_sequences is not None else self.config.num_return_sequences
)
bos_token_id = bos_token_id if bos_token_id is not None else self.config.generator.bos_token_id
eos_token_id = eos_token_id if eos_token_id is not None else self.config.generator.eos_token_id
pad_token_id = pad_token_id if pad_token_id is not None else self.config.generator.pad_token_id
use_cache = use_cache if use_cache is not None else self.config.use_cache
decoder_start_token_id = (
decoder_start_token_id
if decoder_start_token_id is not None
else self.config.generator.decoder_start_token_id
)
remove_invalid_values = (
remove_invalid_values if remove_invalid_values is not None else self.config.remove_invalid_values
)
exponential_decay_length_penalty = (
exponential_decay_length_penalty
if exponential_decay_length_penalty is not None
else self.config.exponential_decay_length_penalty
)
# retrieve docs
if self.retriever is not None and context_input_ids is None:
question_hidden_states = self.question_encoder(input_ids, attention_mask=attention_mask)[0]
out = self.retriever(
input_ids,
question_hidden_states.cpu().detach().to(torch.float32).numpy(),
prefix=self.generator.config.prefix,
n_docs=n_docs,
return_tensors="pt",
)
context_input_ids, context_attention_mask, retrieved_doc_embeds = (
out["context_input_ids"],
out["context_attention_mask"],
out["retrieved_doc_embeds"],
)
# set to correct device
retrieved_doc_embeds = retrieved_doc_embeds.to(question_hidden_states)
context_input_ids = context_input_ids.to(input_ids)
context_attention_mask = context_attention_mask.to(input_ids)
# compute doc_scores
doc_scores = torch.bmm(question_hidden_states.unsqueeze(1), retrieved_doc_embeds.transpose(1, 2)).squeeze(
1
)
assert (context_input_ids.shape[0] % n_docs) == 0, (
f" The first dimension of `context_input_ids` should be a multiple of `n_docs`={n_docs}, but is"
f" {context_input_ids.shape[0]}."
)
# batch_size
batch_size = context_input_ids.shape[0] // n_docs
encoder = self.rag.generator.get_encoder()
encoder_outputs = encoder(input_ids=context_input_ids, attention_mask=context_attention_mask, return_dict=True)
input_ids = torch.full(
(batch_size * num_beams, 1),
decoder_start_token_id,
dtype=torch.long,
device=next(self.parameters()).device,
)
input_ids_seq_length = input_ids.shape[-1]
last_hidden_state = encoder_outputs["last_hidden_state"]
def extend_enc_output(tensor, num_beams=None):
# split into `batch_size`, `num_beams`, `num_docs`
tensor = tensor[None, None, :].reshape((batch_size, 1, n_docs) + tensor.shape[1:])
# repeat same last hidden states over `num_beams` dimension
tensor = tensor.expand((batch_size, num_beams, n_docs) + tensor.shape[3:])
# merge `batch_size`, `num_beams`, `num_docs` dims again
return tensor.reshape((batch_size * num_beams * n_docs,) + tensor.shape[3:])
# correctly extend last_hidden_state and attention mask
context_attention_mask = extend_enc_output(context_attention_mask, num_beams=num_beams)
encoder_outputs["last_hidden_state"] = extend_enc_output(last_hidden_state, num_beams=num_beams)
doc_scores = doc_scores.repeat_interleave(num_beams, dim=0)
# define start_len & additional parameters
model_kwargs["doc_scores"] = doc_scores
model_kwargs["encoder_outputs"] = encoder_outputs
model_kwargs["attention_mask"] = context_attention_mask
model_kwargs["n_docs"] = n_docs
pre_processor = self._get_logits_processor(
repetition_penalty=repetition_penalty,
no_repeat_ngram_size=no_repeat_ngram_size,
encoder_no_repeat_ngram_size=encoder_no_repeat_ngram_size,
input_ids_seq_length=input_ids_seq_length,
encoder_input_ids=context_input_ids,
bad_words_ids=bad_words_ids,
min_length=min_length,
max_length=max_length,
eos_token_id=eos_token_id,
forced_bos_token_id=forced_bos_token_id,
forced_eos_token_id=forced_eos_token_id,
prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
num_beams=num_beams,
num_beam_groups=num_beam_groups,
diversity_penalty=diversity_penalty,
remove_invalid_values=remove_invalid_values,
exponential_decay_length_penalty=exponential_decay_length_penalty,
logits_processor=logits_processor,
renormalize_logits=renormalize_logits,
)
if num_beams == 1:
if num_return_sequences > 1:
raise ValueError(
f"num_return_sequences has to be 1, but is {num_return_sequences} when doing greedy search."
)
return self.greedy_search(
input_ids,
logits_processor=pre_processor,
max_length=max_length,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
**model_kwargs,
)
elif num_beams > 1:
length_penalty = length_penalty if length_penalty is not None else self.config.length_penalty
early_stopping = early_stopping if early_stopping is not None else self.config.early_stopping
if num_return_sequences > num_beams:
raise ValueError("`num_return_sequences` has to be smaller or equal to `num_beams`.")
beam_scorer = BeamSearchScorer(
batch_size=batch_size,
num_beams=num_beams,
device=self.device,
length_penalty=length_penalty,
do_early_stopping=early_stopping,
num_beam_hyps_to_keep=num_return_sequences,
)
return self.beam_search(
input_ids,
beam_scorer,
logits_processor=pre_processor,
max_length=max_length,
pad_token_id=pad_token_id,
eos_token_id=eos_token_id,
**model_kwargs,
)
else:
raise ValueError(f"`num_beams` has to be an integer strictly superior to 0 (≥ 1), but is {num_beams}")
def get_input_embeddings(self):
return self.rag.generator.get_input_embeddings()
def get_output_embeddings(self):
return self.rag.generator.get_output_embeddings()
def set_output_embeddings(self, new_embeddings):
return self.rag.generator.set_output_embeddings(new_embeddings)
def shift_tokens_right(self, input_ids, start_token_id=None):
"""Shift input ids one token to the right, and pad with start_token_id"""
if start_token_id is None:
start_token_id = self.config.decoder_start_token_id
shifted_input_ids = input_ids.new_zeros(input_ids.shape)
shifted_input_ids[:, 1:] = input_ids[:, :-1].clone()
shifted_input_ids[:, 0] = start_token_id
return shifted_input_ids
def get_nll(self, seq_logits, doc_scores, target, reduce_loss=False, epsilon=0.0, n_docs=None):
n_docs = n_docs if n_docs is not None else self.config.n_docs
# shift tokens left
target = torch.cat(
[target[:, 1:], target.new(target.shape[0], 1).fill_(self.config.generator.pad_token_id)], 1
)
def _mask_pads(ll, smooth_obj):
pad_mask = target.eq(self.config.generator.pad_token_id)
if pad_mask.any():
ll.masked_fill_(pad_mask, 0.0)
smooth_obj.masked_fill_(pad_mask, 0.0)
return ll.squeeze(-1), smooth_obj.squeeze(-1)
rag_logprobs = self.marginalize(seq_logits, doc_scores, n_docs)
target = target.unsqueeze(-1)
assert target.dim() == rag_logprobs.dim()
ll = rag_logprobs.gather(dim=-1, index=target)
smooth_obj = rag_logprobs.sum(dim=-1, keepdim=True) # total sum of all (normalised) logits
ll, smooth_obj = _mask_pads(ll, smooth_obj)
ll = ll.sum(1) # sum over tokens
smooth_obj = smooth_obj.sum(1)
nll_loss = -ll
smooth_loss = -smooth_obj
if reduce_loss:
nll_loss = nll_loss.sum()
smooth_loss = smooth_loss.sum()
eps_i = epsilon / rag_logprobs.size(-1)
loss = (1.0 - epsilon) * nll_loss + eps_i * smooth_loss
return loss
| -1 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.