Datasets:

---

Francesco-A

/

github-issues_huggingface-datasets

like 0

[]

[ "It is probably worth noting here that the [documentation](https://huggingface.co/docs/datasets/process#concatenate) is misleading (indicating that it does work for IterableDatasets):\r\n\r\n> You can also mix several datasets together by taking alternating examples from each one to create a new dataset. This is known as interleaving, and you can use it with [interleave_datasets()](https://huggingface.co/docs/datasets/v2.2.1/en/package_reference/main_classes#datasets.interleave_datasets). **Both [interleave_datasets()](https://huggingface.co/docs/datasets/v2.2.1/en/package_reference/main_classes#datasets.interleave_datasets) and [concatenate_datasets()](https://huggingface.co/docs/datasets/v2.2.1/en/package_reference/main_classes#datasets.concatenate_datasets) will work with regular [Dataset](https://huggingface.co/docs/datasets/v2.2.1/en/package_reference/main_classes#datasets.Dataset) and [IterableDataset](https://huggingface.co/docs/datasets/v2.2.1/en/package_reference/main_classes#datasets.IterableDataset) objects**. Refer to the [Stream](https://huggingface.co/docs/datasets/stream#interleave) section for an example of how it’s used. ", "Thanks for the heads up, I'll fix that" ]

[]

[ "Seeing this in the CircleCI builds, this is what I was originally getting before I started messing around with the download URLS to try to fix this:\r\n\r\n`FileNotFoundError: [Errno 2] No such file or directory: '/tmp/tmpwvji917g/extracted/d6185140afb24ad8fee67392100a478269cba286b0d88915a137fdf88872de14/dummy_data/train__VARIABLE_MISUSE__SStuB.txt-00001-of-00300'`\r\n\r\nCould this be because of the files in my `dummy_data.zip`? I had to manually create it, and it looked like the test was looking for the following files, so I created the `.zip` with this structure:\r\n\r\n```\r\nArchive: dummy_data.zip\r\n creating: dummy_data/\r\n inflating: dummy_data/blobtest \r\n inflating: dummy_data/parsed.jsontrain \r\n inflating: dummy_data/parsed.jsonvalidation \r\n```", "Going to close this out and link to a new (cleaner) PR" ]

[]

[ "The parameter can also be named `split_by` with values \"line\", \"paragraph\" or \"document\" (no 's' at the end)", "> The parameter can also be named `split_by` with values \"line\", \"paragraph\" or \"document\" (no 's' at the end)\r\n\r\n@lhoestq @mariosasko I would avoid the term `split` in this context and keep it only for \"train\", \"validation\" and \"test\" splits.\r\n- https://huggingface.co/docs/datasets/process.html#split\r\n > datasets.Dataset.train_test_split() creates train and test splits, if your dataset doesn’t already have them.\r\n- https://huggingface.co/docs/datasets/process.html#process-multiple-splits\r\n > Many datasets have splits that you can process simultaneously with datasets.DatasetDict.map().\r\n\r\nPlease note that in the documentation, one of the terms more frequently used in this context is **\"row\"**:\r\n- https://huggingface.co/docs/datasets/access.html#features-and-columns\r\n > A dataset is a table of rows and typed columns.\r\n\r\n > Return the number of rows and columns with the following standard attributes:\r\n > dataset.num_columns\r\n > 4\r\n > dataset.num_rows\r\n > 3668\r\n\r\n- https://huggingface.co/docs/datasets/access.html#rows-slices-batches-and-columns\r\n > Get several rows of your dataset at a time with slice notation or a list of indices:\r\n- https://huggingface.co/docs/datasets/process.html#map\r\n > This function can even create new rows and columns.\r\n\r\nOther of the terms more frequently used in the docs (in the code as well) is **\"example\"**:\r\n- https://huggingface.co/docs/datasets/process.html#map\r\n > It allows you to apply a processing function to each example in a dataset, independently or in batches.\r\n- https://huggingface.co/docs/datasets/process.html#batch-processing\r\n > datasets.Dataset.map() also supports working with batches of examples.\r\n- https://huggingface.co/docs/datasets/process.html#split-long-examples\r\n > When your examples are too long, you may want to split them\r\n- https://huggingface.co/docs/datasets/process.html#data-augmentation\r\n > With batch processing, you can even augment your dataset with additional examples.\r\n\r\nLess frequently used: **\"item\"**:\r\n- https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.add_item\r\n > Add item to Dataset.\r\n\r\nOther term used in the docs (although less frequently) is **\"sample\"**. The advantage of this word is that it is also a verb, so we can use the parameter: \"sample_by\" (if you insist on using a verb instead of a noun).\r\n\r\nIn summary, these proposals:\r\n- config.row\r\n- config.example\r\n- config.item\r\n- config.sample\r\n- config.sample_by", "I like `sample_by`. Another idea I had was `separate_by`.\r\n\r\nIt could also be `sampling`, `sampling_method`, `separation_method`.\r\n\r\nNot a big fan of the proposed nouns alone since they are very generic, that's why I tried to have something more specific.\r\n\r\nI also agree that we actually should avoid `split` to avoid any confusion", "Thanks for the analysis of the used terms. I also like `sample_by` (`separate_by` is good too).", "Thank you !! :D " ]

[]

[ "Hi @saradhix have you had the chance to reduce the size of the dummy data ?\r\n\r\nFeel free to ping me when it's done so we can merge :) ", "@lhoestq I have reduced the size of the dummy data manually and pushed the changes.", "The CI errors are not related to your dataset.\r\nThey're fixed on master, you can ignore them", "merging since the CI is fixed on master" ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "# Steps to reproduce\r\n\r\nTo reproduce:\r\n\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('jnlpba')\r\n\r\ndataset['train'].features['ner_tags']\r\n```\r\nOutput:\r\n```python\r\nSequence(feature=ClassLabel(num_classes=3, names=['O', 'B', 'I'], names_file=None, id=None), length=-1, id=None)\r\n```\r\n\r\n", "Since I cannot create a branch here is the updated code:\r\n\r\n```python\r\n\r\n# coding=utf-8\r\n# Copyright 2020 HuggingFace Datasets Authors.\r\n#\r\n# Licensed under the Apache License, Version 2.0 (the \"License\");\r\n# you may not use this file except in compliance with the License.\r\n# You may obtain a copy of the License at\r\n#\r\n# http://www.apache.org/licenses/LICENSE-2.0\r\n#\r\n# Unless required by applicable law or agreed to in writing, software\r\n# distributed under the License is distributed on an \"AS IS\" BASIS,\r\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\r\n# See the License for the specific language governing permissions and\r\n# limitations under the License.\r\n\r\n# Lint as: python3\r\n\"\"\"Introduction to the Bio-Entity Recognition Task at JNLPBA\"\"\"\r\n\r\nimport os\r\n\r\nimport datasets\r\n\r\n\r\nlogger = datasets.logging.get_logger(__name__)\r\n\r\n\r\n_CITATION = \"\"\"\\\r\n@inproceedings{kim2004introduction,\r\n title={Introduction to the bio-entity recognition task at JNLPBA},\r\n author={Kim, Jin-Dong and Ohta, Tomoko and Tsuruoka, Yoshimasa and Tateisi, Yuka and Collier, Nigel},\r\n booktitle={Proceedings of the international joint workshop on natural language processing in biomedicine and its applications},\r\n pages={70--75},\r\n year={2004},\r\n organization={Citeseer}\r\n}\r\n\"\"\"\r\n\r\n_DESCRIPTION = \"\"\"\\\r\nThe data came from the GENIA version 3.02 corpus (Kim et al., 2003). This was formed from a controlled search\r\non MEDLINE using the MeSH terms \u0018human\u0019, \u0018blood cells\u0019 and \u0018transcription factors\u0019. From this search 2,000 abstracts\r\nwere selected and hand annotated according to a small taxonomy of 48 classes based on a chemical classification.\r\nAmong the classes, 36 terminal classes were used to annotate the GENIA corpus.\r\n\"\"\"\r\n\r\n_HOMEPAGE = \"http://www.geniaproject.org/shared-tasks/bionlp-jnlpba-shared-task-2004\"\r\n_TRAIN_URL = \"http://www.nactem.ac.uk/GENIA/current/Shared-tasks/JNLPBA/Train/Genia4ERtraining.tar.gz\"\r\n_VAL_URL = 'http://www.nactem.ac.uk/GENIA/current/Shared-tasks/JNLPBA/Evaluation/Genia4ERtest.tar.gz'\r\n\r\n\r\n_URLS = {\r\n \"train\": _TRAIN_URL,\r\n \"val\": _VAL_URL,\r\n}\r\n\r\n_TRAIN_DIRECTORY = \"Genia4ERtraining\"\r\n_VAL_DIRECTORY = \"Genia4ERtest\"\r\n\r\n_TRAIN_FILE = \"Genia4ERtask1.iob2\"\r\n_VAL_FILE = \"Genia4EReval1.iob2\"\r\n\r\n\r\nclass JNLPBAConfig(datasets.BuilderConfig):\r\n \"\"\"BuilderConfig for JNLPBA\"\"\"\r\n\r\n def __init__(self, **kwargs):\r\n \"\"\"BuilderConfig for JNLPBA.\r\n Args:\r\n **kwargs: keyword arguments forwarded to super.\r\n \"\"\"\r\n super(JNLPBAConfig, self).__init__(**kwargs)\r\n\r\n\r\nclass JNLPBA(datasets.GeneratorBasedBuilder):\r\n \"\"\"JNLPBA dataset.\"\"\"\r\n\r\n BUILDER_CONFIGS = [\r\n JNLPBAConfig(name=\"jnlpba\", version=datasets.Version(\"1.0.0\"), description=\"JNLPBA dataset\"),\r\n ]\r\n\r\n def _info(self):\r\n return datasets.DatasetInfo(\r\n description=_DESCRIPTION,\r\n features=datasets.Features(\r\n {\r\n \"id\": datasets.Value(\"string\"),\r\n \"tokens\": datasets.Sequence(datasets.Value(\"string\")),\r\n \"ner_tags\": datasets.Sequence(\r\n datasets.features.ClassLabel(\r\n names=[\r\n 'O',\r\n 'B-DNA',\r\n 'I-DNA', \r\n 'B-RNA',\r\n 'I-RNA',\r\n 'B-cell_line',\r\n 'I-cell_line',\r\n 'B-cell_type',\r\n 'I-cell_type',\r\n 'B-protein',\r\n 'I-protein',\r\n ]\r\n )\r\n ),\r\n }\r\n ),\r\n supervised_keys=None,\r\n homepage=_HOMEPAGE,\r\n citation=_CITATION,\r\n )\r\n\r\n def _split_generators(self, dl_manager):\r\n downloaded_files = dl_manager.download_and_extract(_URLS)\r\n \r\n return [\r\n datasets.SplitGenerator(name=datasets.Split.TRAIN, \r\n gen_kwargs={\"filepath\": os.path.join(downloaded_files['train'], _TRAIN_FILE)}),\r\n datasets.SplitGenerator(name=datasets.Split.VALIDATION, \r\n gen_kwargs={\"filepath\": os.path.join(downloaded_files['val'], _VAL_FILE)})\r\n ]\r\n \r\n\r\n def _generate_examples(self, filepath):\r\n logger.info(\"⏳ Generating examples from = %s\", filepath)\r\n with open(filepath, encoding=\"utf-8\") as f:\r\n guid = 0\r\n tokens = []\r\n ner_tags = []\r\n for line in f:\r\n if line.startswith('###'):\r\n continue\r\n if line == '' or line == '\\n':\r\n if tokens:\r\n yield guid, {\r\n \"id\": str(guid),\r\n \"tokens\": tokens,\r\n \"ner_tags\": ner_tags,\r\n }\r\n guid += 1\r\n tokens = []\r\n ner_tags = []\r\n else:\r\n # tokens are tab separated\r\n splits = line.split(\"\\t\")\r\n #print(splits)\r\n #print(len(splits))\r\n if len(splits) < 2:\r\n splits = splits[0].split()\r\n tokens.append(splits[0])\r\n ner_tags.append(splits[1].rstrip())\r\n # last example\r\n yield guid, {\r\n \"id\": str(guid),\r\n \"tokens\": tokens,\r\n \"ner_tags\": ner_tags,\r\n }\r\n```" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._" ]

[]

[ "It looks like an issue with wandb/tqdm here.\r\nWe're using the `multiprocess` library instead of the `multiprocessing` builtin python package to support various types of mapping functions. Maybe there's some sort of incompatibility.\r\n\r\nCould you make a minimal script to reproduce or a google colab ?", "hi facing the same issue here - \r\n\r\n`AssertionError: Caught AssertionError in DataLoader worker process 0.\r\nOriginal Traceback (most recent call last):\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 996, in emit\r\n stream.write(msg)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/lib/redirect.py\", line 100, in new_write\r\n cb(name, data)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/wandb_run.py\", line 723, in _console_callback\r\n self._backend.interface.publish_output(name, data)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/interface/interface.py\", line 153, in publish_output\r\n self._publish_output(o)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/interface/interface.py\", line 158, in _publish_output\r\n self._publish(rec)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/interface/interface.py\", line 456, in _publish\r\n if self._process and not self._process.is_alive():\r\n File \"/usr/lib/python3.6/multiprocessing/process.py\", line 134, in is_alive\r\n assert self._parent_pid == os.getpid(), 'can only test a child process'\r\nAssertionError: can only test a child process\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n File \"/usr/local/lib/python3.6/dist-packages/torch/utils/data/_utils/worker.py\", line 198, in _worker_loop\r\n data = fetcher.fetch(index)\r\n File \"/usr/local/lib/python3.6/dist-packages/torch/utils/data/_utils/fetch.py\", line 44, in fetch\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/usr/local/lib/python3.6/dist-packages/torch/utils/data/_utils/fetch.py\", line 44, in <listcomp>\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"<ipython-input-8-a4d9a08d114e>\", line 20, in __getitem__\r\n return_token_type_ids=True\r\n File \"/usr/local/lib/python3.6/dist-packages/transformers/tokenization_utils_base.py\", line 2405, in encode_plus\r\n **kwargs,\r\n File \"/usr/local/lib/python3.6/dist-packages/transformers/tokenization_utils_base.py\", line 2125, in _get_padding_truncation_strategies\r\n \"Truncation was not explicitly activated but `max_length` is provided a specific value, \"\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 1320, in warning\r\n self._log(WARNING, msg, args, **kwargs)\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 1444, in _log\r\n self.handle(record)\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 1454, in handle\r\n self.callHandlers(record)\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 1516, in callHandlers\r\n hdlr.handle(record)\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 865, in handle\r\n self.emit(record)\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 1000, in emit\r\n self.handleError(record)\r\n File \"/usr/lib/python3.6/logging/__init__.py\", line 917, in handleError\r\n sys.stderr.write('--- Logging error ---\\n')\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/lib/redirect.py\", line 100, in new_write\r\n cb(name, data)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/wandb_run.py\", line 723, in _console_callback\r\n self._backend.interface.publish_output(name, data)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/interface/interface.py\", line 153, in publish_output\r\n self._publish_output(o)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/interface/interface.py\", line 158, in _publish_output\r\n self._publish(rec)\r\n File \"/usr/local/lib/python3.6/dist-packages/wandb/sdk/interface/interface.py\", line 456, in _publish\r\n if self._process and not self._process.is_alive():\r\n File \"/usr/lib/python3.6/multiprocessing/process.py\", line 134, in is_alive\r\n assert self._parent_pid == os.getpid(), 'can only test a child process'\r\nAssertionError: can only test a child process`\r\n", "It looks like this warning : \r\n\"Truncation was not explicitly activated but max_length is provided a specific value, \"\r\nis not handled well by wandb.\r\n\r\nThe error occurs when calling the tokenizer.\r\nMaybe you can try to specify `truncation=True` when calling the tokenizer to remove the warning ?\r\nOtherwise I don't know why wandb would fail on a warning. Maybe one of its logging handlers have some issues with the logging of tokenizers. Maybe @n1t0 knows more about this ?", "I'm having a similar issue but when I try to do multiprocessing with the `DataLoader`\r\n\r\nCode to reproduce:\r\n\r\n```\r\nfrom datasets import load_dataset\r\n\r\nbook_corpus = load_dataset('bookcorpus', 'plain_text', cache_dir='/home/ad/Desktop/bookcorpus', split='train[:1%]')\r\nbook_corpus = book_corpus.map(encode, batched=True, num_proc=20, load_from_cache_file=True, batch_size=5000)\r\nbook_corpus.set_format(type='torch', columns=['text', \"input_ids\", \"attention_mask\", \"token_type_ids\"])\r\n\r\nfrom transformers import DataCollatorForWholeWordMask\r\nfrom transformers import Trainer, TrainingArguments\r\n\r\ndata_collator = DataCollatorForWholeWordMask(\r\n tokenizer=tokenizer, mlm=True, mlm_probability=0.15)\r\n\r\ntraining_args = TrainingArguments(\r\n output_dir=\"./mobile_linear_att_8L_128_128_03layerdrop_shared\",\r\n overwrite_output_dir=True,\r\n num_train_epochs=1,\r\n per_device_train_batch_size=64,\r\n save_steps=50,\r\n save_total_limit=2,\r\n logging_first_step=True,\r\n warmup_steps=100,\r\n logging_steps=50,\r\n gradient_accumulation_steps=1,\r\n fp16=True,\r\n **dataloader_num_workers=10**,\r\n)\r\n\r\ntrainer = Trainer(\r\n model=model,\r\n args=training_args,\r\n data_collator=data_collator,\r\n train_dataset=book_corpus,\r\n tokenizer=tokenizer)\r\n\r\ntrainer.train()\r\n```\r\n\r\n```\r\n---------------------------------------------------------------------------\r\nAssertionError Traceback (most recent call last)\r\n<timed eval> in <module>\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/transformers/trainer.py in train(self, model_path, trial)\r\n 869 self.control = self.callback_handler.on_epoch_begin(self.args, self.state, self.control)\r\n 870 \r\n--> 871 for step, inputs in enumerate(epoch_iterator):\r\n 872 \r\n 873 # Skip past any already trained steps if resuming training\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/dataloader.py in __next__(self)\r\n 433 if self._sampler_iter is None:\r\n 434 self._reset()\r\n--> 435 data = self._next_data()\r\n 436 self._num_yielded += 1\r\n 437 if self._dataset_kind == _DatasetKind.Iterable and \\\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/dataloader.py in _next_data(self)\r\n 1083 else:\r\n 1084 del self._task_info[idx]\r\n-> 1085 return self._process_data(data)\r\n 1086 \r\n 1087 def _try_put_index(self):\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/dataloader.py in _process_data(self, data)\r\n 1109 self._try_put_index()\r\n 1110 if isinstance(data, ExceptionWrapper):\r\n-> 1111 data.reraise()\r\n 1112 return data\r\n 1113 \r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/_utils.py in reraise(self)\r\n 426 # have message field\r\n 427 raise self.exc_type(message=msg)\r\n--> 428 raise self.exc_type(msg)\r\n 429 \r\n 430 \r\n\r\nAssertionError: Caught AssertionError in DataLoader worker process 0.\r\nOriginal Traceback (most recent call last):\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py\", line 198, in _worker_loop\r\n data = fetcher.fetch(index)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py\", line 44, in fetch\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py\", line 44, in <listcomp>\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 1087, in __getitem__\r\n format_kwargs=self._format_kwargs,\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 1074, in _getitem\r\n format_kwargs=format_kwargs,\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 890, in _convert_outputs\r\n v = map_nested(command, v, **map_nested_kwargs)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/utils/py_utils.py\", line 225, in map_nested\r\n return function(data_struct)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 851, in command\r\n return torch.tensor(x, **format_kwargs)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/warnings.py\", line 101, in _showwarnmsg\r\n _showwarnmsg_impl(msg)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/warnings.py\", line 30, in _showwarnmsg_impl\r\n file.write(text)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/lib/redirect.py\", line 100, in new_write\r\n cb(name, data)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/wandb_run.py\", line 723, in _console_callback\r\n self._backend.interface.publish_output(name, data)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py\", line 153, in publish_output\r\n self._publish_output(o)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py\", line 158, in _publish_output\r\n self._publish(rec)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py\", line 456, in _publish\r\n if self._process and not self._process.is_alive():\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/multiprocessing/process.py\", line 134, in is_alive\r\n assert self._parent_pid == os.getpid(), 'can only test a child process'\r\nAssertionError: can only test a child process\r\n```\r\n\r\nAs a workaround I have commented line 456 and 457 in `/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py`", "Isn't it more the pytorch warning on the use of non-writable memory for tensor that trigger this here @lhoestq? (since it seems to be a warning triggered in `torch.tensor()`", "Yep this time this is a warning from pytorch that causes wandb to not work properly.\r\nCould this by a wandb issue ?", "Hi @timothyjlaurent @gaceladri \r\nIf you're running `transformers` from `master` you can try setting the env var `WAND_DISABLE=true` (from https://github.com/huggingface/transformers/pull/9896) and try again ?\r\nThis issue might be related to https://github.com/huggingface/transformers/issues/9623 ", "I have commented the lines that cause my code break. I'm now seeing my reports on Wandb and my code does not break. I am training now, so I will check probably in 6 hours. I suppose that setting wandb disable will work as well.", "This seems to be a bug in `wandb` (see https://github.com/wandb/wandb/issues/1994)." ]

[]

[ "Also the dummy data files are quite big (500KB)\r\nIf you could reduce that that would be nice (just look at the files inside and remove unecessary chunks of texts)\r\nin general dummy data are just a few KB and we suggest to not get higher than 50KB\r\n\r\nHaving light dummy data makes the repo faster to clone" ]

[]

[ "I think the underlying issue is in https://github.com/huggingface/datasets/blob/c0ed6fdc29675b3565b01b77fde5ab5d9d8b60ec/src/datasets/commands/dummy_data.py#L124 - where `CSV`s are considered to be in the same class of file as text, jsonl, and tsv.\r\n\r\nI think this is an error because CSVs can have newlines within the rows of a file. I'm happy to make a PR to change how this handling works, or make the change within this PR. \r\n\r\nWe should figure out:\r\n1. Does this dummy data need to be generated more than once? (It looks like no)\r\n2. Should this be fixed generally? (needs a HF person to weigh in here)\r\n3. What is the right way for such a fix to exist permanently here; the [Contributing document](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md) doesn't provide guidance on any tests. Writing a test is several times more effort than fixing the underlying issue. (again needs a HF person)", "Would someone from HF mind taking a look at this PR? (@lhoestq)", "Hi ! Sorry for the delay in responding :)\r\n\r\nI don't think there's a big need to fix this in the general case for now, feel free to just generate the dummy data for this specific dataset :)\r\n\r\nThe `datasets-cli dummy_data datasets/mslr2022` command should tell you what dummy files to generate. In each dummy file you just need to include enough data to generate 4 or 5 examples", "_The documentation is not available anymore as the PR was closed or merged._", "Awesome! Generated the dummy data and the tests now pass. @jayded thanks for your help! If you and @lucylw are happy with this I think it's ready to be merged. @lhoestq this is ready for another look :)", "Hi @lhoestq, is there anything blocking this from being merged that I can address?", "Hi @JohnGiorgi ! Thanks for the changes, it looks all good now :)\r\n\r\nI think this dataset can be under the AllenAI page here: https://huggingface.co/allenai What do you think ?\r\nFeel free to create a new dataset repository on huggingface.co and upload your files (dataset script, readme, etc.)\r\n\r\nOnce the dataset is under the AllenAI org, we can close this PR\r\n", "> Hi @JohnGiorgi ! Thanks for the changes, it looks all good now :)\r\n> \r\n> I think this dataset can be under the AllenAI page here: https://huggingface.co/allenai What do you think ? Feel free to create a new dataset repository on huggingface.co and upload your files (dataset script, readme, etc.)\r\n> \r\n> Once the dataset is under the AllenAI org, we can close this PR\r\n\r\nSweet! It is uploaded here: https://huggingface.co/datasets/allenai/mslr2022", "Nice ! Thanks :)\r\n\r\nI think we can close this PR then.\r\n\r\nI noticed that the dataset preview is not available on this dataset, this is because we require datasets to work in streaming mode to show a preview. However TAR archives don't work well in streaming mode (you can't know in advance what files are inside a TAR archive without reading it completely). This can be fixed by using a ZIP archive instead.\r\n\r\nLet me know if you have questions or if I can help." ]

[ { "color": "a2eeef", "default": true, "description": "New feature or request", "id": 1935892871, "name": "enhancement", "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement" } ]

[]

[]

[]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "Thanks for reporting. That's a bug indeed.\r\nApparently only the `data_files` parameter is taken into account right now in `DatasetBuilder._create_builder_config` but it should also be the case for `config_kwargs` (or at least the instantiated `builder_config`)", "Hi, does this bug be fixed? when I load JSON files, I get the same errors by the command \r\n`!python3 run.py --do_train --task qa --dataset squad-retrain-data/train-v2.0.json --output_dir ./re_trained_model/`\r\n\r\nchange the dateset to load json by refering to https://huggingface.co/docs/datasets/loading.html\r\n`dataset = datasets.load_dataset('json', data_files=args.dataset)`\r\n\r\nErrors:\r\n`Downloading and preparing dataset json/default (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /root/.cache/huggingface/datasets/json/default-c1e124ad488911b8/0.0.0/45636811569ec4a6630521c18235dfbbab83b7ab572e3393c5ba68ccabe98264...\r\n`", "```ds = load_dataset(\"csv\", data_files={'train': 'train.csv', 'test': 'test.csv'})```\r\n\r\nGives the output\r\n```Using custom data configuration default-5c8ae7c208631aca```\r\n\r\nand the code hangs there.", "> `ds = load_dataset(\"csv\", data_files={'train': 'train.csv', 'test': 'test.csv'})`\r\n> \r\n> Gives the output `Using custom data configuration default-5c8ae7c208631aca`\r\n> \r\n> and the code hangs there.\r\n\r\nHave you solved it? I met this problem too!", "Can you Ctrl+C to kill the process and share the stacktrace here ? It should show at which location in the code it was hanging", "I had the same issue and solved it by downgrading the datasets version from 2.7.0 -> 2.6.1\r\npip install -q datasets==2.6.1", "> I had the same issue and solved it by downgrading the datasets version from 2.7.0 -> 2.6.1 pip install -q datasets==2.6.1\r\n\r\nThanks, it works for me" ]

[]

[ "CI fails due to missing tags, but they will be added in https://github.com/huggingface/datasets/pull/2949" ]

[]

[ "Hi, \r\n\r\ncould you please post this question on our [Forum](https://discuss.huggingface.co/) as we keep issues for bugs and feature requests? ", "> Hi,\r\n> \r\n> could you please post this question on our [Forum](https://discuss.huggingface.co/) as we keep issues for bugs and feature requests?\r\n\r\nok." ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "Hi @villmow, thanks for reporting.\r\n\r\nCould you please try with the Datasets version 1.6? We released it yesterday and it fixes some issues about the processing speed. You can see the fix implemented by @lhoestq here: #2122.\r\n\r\nOnce you update Datasets, please confirm if the problem persists.", "Hi @albertvillanova, thanks for the reply. I just tried the new version and the problem still persists. \r\n\r\nDo I need to rebuild the saved dataset (which I load from disk) with the 1.6.0 version of datasets? My script loads this dataset and creates new datasets from it. I tried it without rebuilding.\r\n\r\nSee this short video of what happens. It does not create all processes at the same time:\r\n\r\nhttps://user-images.githubusercontent.com/2743060/115720139-0da3a500-a37d-11eb-833a-9bbacc70868d.mp4\r\n\r\n", "There can be a bit of delay between the creations of the processes but this delay should be the same for both your `map` calls. We should look into this.\r\nAlso if you hav some code that reproduces this issue on google colab that'd be really useful !\r\n\r\nRegarding the speed differences:\r\nThis looks like a similar issue as https://github.com/huggingface/datasets/issues/1992 who is experiencing the same speed differences between processes.\r\nThis is a known bug that we are investigating. As of now I've never managed to reproduce it on my machine so it's pretty hard for me to find where this issue comes from.\r\n", "Upgrade to 1.6.1 solved my problem somehow. I did not change any of my code, but now it starts all processes around the same time.", "Nice ! I'm glad this works now.\r\nClosing for now, but feel free to re-open if you experience this issue again." ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._" ]

[]

[ "You need to remove the `text` and `text_en` columns before passing the dataset to the `DataLoader` to avoid this error:\r\n```python\r\ntokenized_datasets = tokenized_datasets.remove_columns([\"text\", \"text_en\"])\r\n```\r\n", "Thanks @mariosasko. Now I am getting this error:\r\n\r\n```\r\nTraceback (most recent call last):\r\n File \"client_2.py\", line 138, in <module>\r\n main()\r\n File \"client_2.py\", line 134, in main\r\n fl.client.start_numpy_client(server_address=\"localhost:8080\", client=IMDBClient())\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/flwr/client/app.py\", line 208, in start_numpy_client\r\n start_client(\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/flwr/client/app.py\", line 142, in start_client\r\n client_message, sleep_duration, keep_going = handle(\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/flwr/client/grpc_client/message_handler.py\", line 68, in handle\r\n return _fit(client, server_msg.fit_ins), 0, True\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/flwr/client/grpc_client/message_handler.py\", line 157, in _fit\r\n fit_res = client.fit(fit_ins)\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/flwr/client/app.py\", line 252, in _fit\r\n results = self.numpy_client.fit(parameters, ins.config) # type: ignore\r\n File \"client_2.py\", line 124, in fit\r\n train(net, trainloader, epochs=1)\r\n File \"client_2.py\", line 78, in train\r\n for batch in trainloader:\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/torch/utils/data/dataloader.py\", line 652, in __next__\r\n data = self._next_data()\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/torch/utils/data/dataloader.py\", line 692, in _next_data\r\n data = self._dataset_fetcher.fetch(index) # may raise StopIteration\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py\", line 49, in fetch\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py\", line 49, in <listcomp>\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/datasets/arrow_dataset.py\", line 1525, in __getitem__\r\n return self._getitem(\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/datasets/arrow_dataset.py\", line 1517, in _getitem\r\n pa_subtable = query_table(self._data, key, indices=self._indices if self._indices is not None else None)\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/datasets/formatting/formatting.py\", line 373, in query_table\r\n pa_subtable = _query_table_with_indices_mapping(table, key, indices=indices)\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/datasets/formatting/formatting.py\", line 55, in _query_table_with_indices_mapping\r\n return _query_table(table, key)\r\n File \"/home/saurav/.local/lib/python3.8/site-packages/datasets/formatting/formatting.py\", line 79, in _query_table\r\n return table.fast_slice(key % table.num_rows, 1)\r\nZeroDivisionError: integer division or modulo by zero\r\n```\r\n\r\nThis is my code:\r\n\r\n```\r\nfrom collections import OrderedDict\r\nimport warnings\r\n\r\nimport flwr as fl\r\nimport torch\r\nimport numpy as np\r\n\r\nimport random\r\nfrom torch.utils.data import DataLoader\r\n\r\nfrom datasets import load_dataset, load_metric\r\n\r\nfrom transformers import AutoTokenizer, DataCollatorWithPadding\r\nfrom transformers import AutoModelForSequenceClassification\r\nfrom transformers import AdamW\r\n#from transformers import tokenized_datasets\r\n\r\n\r\nwarnings.filterwarnings(\"ignore\", category=UserWarning)\r\n# DEVICE = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\r\n\r\nDEVICE = \"cpu\"\r\n\r\nCHECKPOINT = \"distilbert-base-uncased\" # transformer model checkpoint\r\n\r\n\r\ndef load_data():\r\n \"\"\"Load IMDB data (training and eval)\"\"\"\r\n raw_datasets = load_dataset(\"yhavinga/imdb_dutch\")\r\n raw_datasets = raw_datasets.shuffle(seed=42)\r\n\r\n # remove unnecessary data split\r\n del raw_datasets[\"unsupervised\"]\r\n\r\n tokenizer = AutoTokenizer.from_pretrained(CHECKPOINT)\r\n\r\n def tokenize_function(examples):\r\n return tokenizer(examples[\"text\"], truncation=True)\r\n\r\n # random 100 samples\r\n population = random.sample(range(len(raw_datasets[\"train\"])), 100)\r\n\r\n tokenized_datasets = raw_datasets.map(tokenize_function, batched=True)\r\n tokenized_datasets[\"train\"] = tokenized_datasets[\"train\"].select(population)\r\n tokenized_datasets[\"test\"] = tokenized_datasets[\"test\"].select(population)\r\n\r\n # tokenized_datasets = tokenized_datasets.remove_columns(\"text\")\r\n # tokenized_datasets = tokenized_datasets.rename_column(\"label\", \"labels\")\r\n\r\n tokenized_datasets = tokenized_datasets.remove_columns(\"attention_mask\")\r\n tokenized_datasets = tokenized_datasets.remove_columns(\"input_ids\")\r\n tokenized_datasets = tokenized_datasets.remove_columns(\"label\")\r\n # tokenized_datasets = tokenized_datasets.remove_columns(\"text_en\")\r\n\r\n # tokenized_datasets = tokenized_datasets.remove_columns(raw_datasets[\"train\"].column_names)\r\n \r\n tokenized_datasets = tokenized_datasets.remove_columns([\"text\", \"text_en\"])\r\n \r\n data_collator = DataCollatorWithPadding(tokenizer=tokenizer)\r\n trainloader = DataLoader(\r\n tokenized_datasets[\"train\"],\r\n shuffle=True,\r\n batch_size=32,\r\n collate_fn=data_collator,\r\n )\r\n\r\n testloader = DataLoader(\r\n tokenized_datasets[\"test\"], batch_size=32, collate_fn=data_collator\r\n )\r\n\r\n return trainloader, testloader\r\n\r\n\r\ndef train(net, trainloader, epochs):\r\n optimizer = AdamW(net.parameters(), lr=5e-4)\r\n net.train()\r\n for _ in range(epochs):\r\n for batch in trainloader:\r\n batch = {k: v.to(DEVICE) for k, v in batch.items()}\r\n outputs = net(**batch)\r\n loss = outputs.loss\r\n loss.backward()\r\n optimizer.step()\r\n optimizer.zero_grad()\r\n\r\n\r\ndef test(net, testloader):\r\n metric = load_metric(\"accuracy\")\r\n loss = 0\r\n net.eval()\r\n for batch in testloader:\r\n batch = {k: v.to(DEVICE) for k, v in batch.items()}\r\n with torch.no_grad():\r\n outputs = net(**batch)\r\n logits = outputs.logits\r\n loss += outputs.loss.item()\r\n predictions = torch.argmax(logits, dim=-1)\r\n metric.add_batch(predictions=predictions, references=batch[\"labels\"])\r\n loss /= len(testloader.dataset)\r\n accuracy = metric.compute()[\"accuracy\"]\r\n return loss, accuracy\r\n\r\n\r\ndef main():\r\n net = AutoModelForSequenceClassification.from_pretrained(\r\n CHECKPOINT, num_labels=2\r\n ).to(DEVICE)\r\n\r\n trainloader, testloader = load_data()\r\n\r\n # Flower client\r\n class IMDBClient(fl.client.NumPyClient):\r\n def get_parameters(self, config):\r\n return [val.cpu().numpy() for _, val in net.state_dict().items()]\r\n\r\n def set_parameters(self, parameters):\r\n params_dict = zip(net.state_dict().keys(), parameters)\r\n state_dict = OrderedDict({k: torch.Tensor(v) for k, v in params_dict})\r\n net.load_state_dict(state_dict, strict=True)\r\n\r\n def fit(self, parameters, config):\r\n self.set_parameters(parameters)\r\n print(\"Training Started...\")\r\n train(net, trainloader, epochs=1)\r\n print(\"Training Finished.\")\r\n return self.get_parameters(config={}), len(trainloader), {}\r\n\r\n def evaluate(self, parameters, config):\r\n self.set_parameters(parameters)\r\n loss, accuracy = test(net, testloader)\r\n return float(loss), len(testloader), {\"accuracy\": float(accuracy)}\r\n\r\n # Start client\r\n fl.client.start_numpy_client(server_address=\"localhost:8080\", client=IMDBClient())\r\n\r\n\r\nif __name__ == \"__main__\":\r\n main()\r\n```", "Please also remove/comment these lines:\r\n```python\r\ntokenized_datasets = tokenized_datasets.remove_columns(\"attention_mask\")\r\ntokenized_datasets = tokenized_datasets.remove_columns(\"input_ids\")\r\ntokenized_datasets = tokenized_datasets.remove_columns(\"label\")\r\n```", "Thanks @mariosasko .\r\n\r\nNow, I am trying out this [tutorial](https://flower.dev/docs/quickstart-huggingface.html) which basically trains distil-BERT with IMDB dataset (very similar to this [tutorial](https://huggingface.co/docs/transformers/main/tasks/sequence_classification)). But I don't know why my accuracy isn't increasing even after training for a significant amount of time and also by using the entire dataset. Below I have attached `client.py` file:\r\n\r\n`client.py`:\r\n\r\n```\r\nfrom collections import OrderedDict\r\nimport warnings\r\n\r\nimport flwr as fl\r\nimport torch\r\nimport numpy as np\r\n\r\nimport random\r\nfrom torch.utils.data import DataLoader\r\n\r\nfrom datasets import load_dataset, load_metric\r\n\r\nfrom transformers import AutoTokenizer, DataCollatorWithPadding\r\nfrom transformers import AutoModelForSequenceClassification\r\nfrom transformers import AdamW\r\n\r\nwarnings.filterwarnings(\"ignore\", category=UserWarning)\r\n\r\nDEVICE = \"cuda:1\"\r\n\r\nCHECKPOINT = \"distilbert-base-uncased\" # transformer model checkpoint\r\n\r\n\r\ndef load_data():\r\n \"\"\"Load IMDB data (training and eval)\"\"\"\r\n raw_datasets = load_dataset(\"imdb\")\r\n raw_datasets = raw_datasets.shuffle(seed=42)\r\n\r\n # remove unnecessary data split\r\n del raw_datasets[\"unsupervised\"]\r\n\r\n tokenizer = AutoTokenizer.from_pretrained(CHECKPOINT)\r\n\r\n def tokenize_function(examples):\r\n return tokenizer(examples[\"text\"], truncation=True)\r\n\r\n tokenized_datasets = raw_datasets.map(tokenize_function, batched=True)\r\n\r\n tokenized_datasets = tokenized_datasets.remove_columns(\"text\")\r\n tokenized_datasets = tokenized_datasets.rename_column(\"label\", \"labels\")\r\n\r\n data_collator = DataCollatorWithPadding(tokenizer=tokenizer)\r\n trainloader = DataLoader(\r\n tokenized_datasets[\"train\"],\r\n shuffle=True,\r\n batch_size=32,\r\n collate_fn=data_collator,\r\n )\r\n\r\n testloader = DataLoader(\r\n tokenized_datasets[\"test\"], batch_size=32, collate_fn=data_collator\r\n )\r\n\r\n return trainloader, testloader\r\n\r\n\r\ndef train(net, trainloader, epochs):\r\n optimizer = AdamW(net.parameters(), lr=5e-5)\r\n net.train()\r\n for i in range(epochs):\r\n print(\"Epoch: \", i+1)\r\n j = 1\r\n print(\"####################### The length of the trainloader is: \", len(trainloader)) \r\n for batch in trainloader:\r\n print(\"####################### The batch number is: \", j)\r\n batch = {k: v.to(DEVICE) for k, v in batch.items()}\r\n outputs = net(**batch)\r\n loss = outputs.loss\r\n loss.backward()\r\n optimizer.step()\r\n optimizer.zero_grad()\r\n j += 1\r\n\r\n\r\ndef test(net, testloader):\r\n metric = load_metric(\"accuracy\")\r\n loss = 0\r\n net.eval()\r\n for batch in testloader:\r\n batch = {k: v.to(DEVICE) for k, v in batch.items()}\r\n with torch.no_grad():\r\n outputs = net(**batch)\r\n logits = outputs.logits\r\n loss += outputs.loss.item()\r\n predictions = torch.argmax(logits, dim=-1)\r\n metric.add_batch(predictions=predictions, references=batch[\"labels\"])\r\n loss /= len(testloader.dataset)\r\n accuracy = metric.compute()[\"accuracy\"]\r\n return loss, accuracy\r\n\r\n\r\ndef main():\r\n net = AutoModelForSequenceClassification.from_pretrained(\r\n CHECKPOINT, num_labels=2\r\n ).to(DEVICE)\r\n\r\n trainloader, testloader = load_data()\r\n\r\n # Flower client\r\n class IMDBClient(fl.client.NumPyClient):\r\n def get_parameters(self, config):\r\n return [val.cpu().numpy() for _, val in net.state_dict().items()]\r\n\r\n def set_parameters(self, parameters):\r\n params_dict = zip(net.state_dict().keys(), parameters)\r\n state_dict = OrderedDict({k: torch.Tensor(v) for k, v in params_dict})\r\n net.load_state_dict(state_dict, strict=True)\r\n\r\n def fit(self, parameters, config):\r\n self.set_parameters(parameters)\r\n print(\"Training Started...\")\r\n train(net, trainloader, epochs=1)\r\n print(\"Training Finished.\")\r\n return self.get_parameters(config={}), len(trainloader), {}\r\n\r\n def evaluate(self, parameters, config):\r\n self.set_parameters(parameters)\r\n loss, accuracy = test(net, testloader)\r\n print({\"loss\": float(loss), \"accuracy\": float(accuracy)})\r\n return float(loss), len(testloader), {\"loss\": float(loss), \"accuracy\": float(accuracy)}\r\n\r\n # Start client\r\n fl.client.start_numpy_client(server_address=\"localhost:5040\", client=IMDBClient())\r\n\r\n\r\nif __name__ == \"__main__\":\r\n main()\r\n```\r\n\r\nCan I get any help, please?" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "Are you sure it's safe to remove? https://github.com/huggingface/datasets/pull/3866", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009520 / 0.011353 (-0.001833) | 0.005319 / 0.011008 (-0.005690) | 0.099372 / 0.038508 (0.060863) | 0.036173 / 0.023109 (0.013064) | 0.295752 / 0.275898 (0.019853) | 0.362882 / 0.323480 (0.039402) | 0.008442 / 0.007986 (0.000456) | 0.004225 / 0.004328 (-0.000103) | 0.076645 / 0.004250 (0.072394) | 0.044198 / 0.037052 (0.007146) | 0.311948 / 0.258489 (0.053459) | 0.342963 / 0.293841 (0.049122) | 0.038613 / 0.128546 (-0.089933) | 0.012127 / 0.075646 (-0.063519) | 0.334427 / 0.419271 (-0.084844) | 0.048309 / 0.043533 (0.004776) | 0.297046 / 0.255139 (0.041907) | 0.314562 / 0.283200 (0.031363) | 0.105797 / 0.141683 (-0.035886) | 1.460967 / 1.452155 (0.008812) | 1.500907 / 1.492716 (0.008190) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.216185 / 0.018006 (0.198179) | 0.438924 / 0.000490 (0.438435) | 0.001210 / 0.000200 (0.001011) | 0.000081 / 0.000054 (0.000027) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026193 / 0.037411 (-0.011219) | 0.105888 / 0.014526 (0.091363) | 0.115812 / 0.176557 (-0.060744) | 0.158748 / 0.737135 (-0.578387) | 0.121514 / 0.296338 (-0.174824) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.399837 / 0.215209 (0.184628) | 3.996992 / 2.077655 (1.919338) | 1.784964 / 1.504120 (0.280844) | 1.591078 / 1.541195 (0.049883) | 1.666424 / 1.468490 (0.197934) | 0.711450 / 4.584777 (-3.873327) | 3.787814 / 3.745712 (0.042102) | 2.056776 / 5.269862 (-3.213085) | 1.332163 / 4.565676 (-3.233514) | 0.085755 / 0.424275 (-0.338520) | 0.012033 / 0.007607 (0.004426) | 0.511500 / 0.226044 (0.285455) | 5.098999 / 2.268929 (2.830071) | 2.288261 / 55.444624 (-53.156364) | 1.947483 / 6.876477 (-4.928994) | 1.987838 / 2.142072 (-0.154234) | 0.852241 / 4.805227 (-3.952986) | 0.164781 / 6.500664 (-6.335883) | 0.061825 / 0.075469 (-0.013644) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.202253 / 1.841788 (-0.639534) | 14.632608 / 8.074308 (6.558300) | 13.331320 / 10.191392 (3.139928) | 0.157944 / 0.680424 (-0.522480) | 0.029284 / 0.534201 (-0.504917) | 0.446636 / 0.579283 (-0.132647) | 0.437009 / 0.434364 (0.002645) | 0.521883 / 0.540337 (-0.018455) | 0.606687 / 1.386936 (-0.780249) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007528 / 0.011353 (-0.003825) | 0.005274 / 0.011008 (-0.005734) | 0.073524 / 0.038508 (0.035016) | 0.033893 / 0.023109 (0.010784) | 0.335432 / 0.275898 (0.059534) | 0.379981 / 0.323480 (0.056501) | 0.005954 / 0.007986 (-0.002031) | 0.004126 / 0.004328 (-0.000203) | 0.072891 / 0.004250 (0.068641) | 0.046517 / 0.037052 (0.009465) | 0.337241 / 0.258489 (0.078752) | 0.385562 / 0.293841 (0.091721) | 0.036410 / 0.128546 (-0.092136) | 0.012246 / 0.075646 (-0.063401) | 0.085974 / 0.419271 (-0.333298) | 0.049665 / 0.043533 (0.006133) | 0.330919 / 0.255139 (0.075780) | 0.352041 / 0.283200 (0.068841) | 0.103751 / 0.141683 (-0.037931) | 1.468851 / 1.452155 (0.016696) | 1.565380 / 1.492716 (0.072663) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.260431 / 0.018006 (0.242425) | 0.444554 / 0.000490 (0.444064) | 0.016055 / 0.000200 (0.015855) | 0.000283 / 0.000054 (0.000228) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029130 / 0.037411 (-0.008281) | 0.112002 / 0.014526 (0.097476) | 0.120769 / 0.176557 (-0.055788) | 0.169345 / 0.737135 (-0.567790) | 0.129609 / 0.296338 (-0.166730) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.432211 / 0.215209 (0.217002) | 4.293008 / 2.077655 (2.215353) | 2.071291 / 1.504120 (0.567171) | 1.859322 / 1.541195 (0.318127) | 1.971434 / 1.468490 (0.502943) | 0.704042 / 4.584777 (-3.880735) | 3.791696 / 3.745712 (0.045983) | 3.142632 / 5.269862 (-2.127230) | 1.735151 / 4.565676 (-2.830525) | 0.086203 / 0.424275 (-0.338072) | 0.012542 / 0.007607 (0.004935) | 0.534870 / 0.226044 (0.308826) | 5.326042 / 2.268929 (3.057113) | 2.547960 / 55.444624 (-52.896664) | 2.212730 / 6.876477 (-4.663747) | 2.296177 / 2.142072 (0.154105) | 0.840311 / 4.805227 (-3.964917) | 0.168353 / 6.500664 (-6.332311) | 0.065949 / 0.075469 (-0.009520) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.255589 / 1.841788 (-0.586199) | 14.947344 / 8.074308 (6.873036) | 13.253721 / 10.191392 (3.062329) | 0.162349 / 0.680424 (-0.518075) | 0.017579 / 0.534201 (-0.516622) | 0.420758 / 0.579283 (-0.158525) | 0.430030 / 0.434364 (-0.004334) | 0.524669 / 0.540337 (-0.015669) | 0.623920 / 1.386936 (-0.763016) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#35b789e8f6826b6b5a6b48fcc2416c890a1f326a \"CML watermark\")\n" ]

[]

[ "`amazon_polarity` is probably a subset of `amazon_us_reviews` but I am not entirely sure about that.\r\nI guess `amazon_polarity` will help in reproducing results of papers using this dataset since even if it is a subset from `amazon_us_reviews`, it is not trivial how to extract `amazon_polarity` from `amazon_us_reviews`, especially since `amazon_us_reviews` was released after `amazon_polarity`. ", "do you know what the problem would be ? should I pull the master before ? @lhoestq ", "The error just appeared on master. I will try to fix it today.\r\nYou can ignore them since it's not related to the dataset you added", "merging since the CI is fixed on master", "Great thanks for the help. " ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008581 / 0.011353 (-0.002772) | 0.004519 / 0.011008 (-0.006490) | 0.099721 / 0.038508 (0.061213) | 0.029217 / 0.023109 (0.006107) | 0.298229 / 0.275898 (0.022331) | 0.332605 / 0.323480 (0.009125) | 0.006880 / 0.007986 (-0.001106) | 0.003324 / 0.004328 (-0.001005) | 0.078143 / 0.004250 (0.073892) | 0.034262 / 0.037052 (-0.002790) | 0.304162 / 0.258489 (0.045673) | 0.342351 / 0.293841 (0.048510) | 0.033387 / 0.128546 (-0.095159) | 0.011397 / 0.075646 (-0.064249) | 0.321527 / 0.419271 (-0.097744) | 0.040886 / 0.043533 (-0.002647) | 0.299968 / 0.255139 (0.044829) | 0.322484 / 0.283200 (0.039285) | 0.083832 / 0.141683 (-0.057851) | 1.482241 / 1.452155 (0.030086) | 1.548438 / 1.492716 (0.055721) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.191002 / 0.018006 (0.172996) | 0.403423 / 0.000490 (0.402933) | 0.002493 / 0.000200 (0.002293) | 0.000074 / 0.000054 (0.000020) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.023720 / 0.037411 (-0.013691) | 0.100806 / 0.014526 (0.086281) | 0.105314 / 0.176557 (-0.071242) | 0.141490 / 0.737135 (-0.595645) | 0.108695 / 0.296338 (-0.187644) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.412250 / 0.215209 (0.197041) | 4.124830 / 2.077655 (2.047175) | 1.851948 / 1.504120 (0.347828) | 1.651597 / 1.541195 (0.110403) | 1.712486 / 1.468490 (0.243996) | 0.696634 / 4.584777 (-3.888143) | 3.304220 / 3.745712 (-0.441492) | 1.862776 / 5.269862 (-3.407086) | 1.159452 / 4.565676 (-3.406224) | 0.082930 / 0.424275 (-0.341345) | 0.012586 / 0.007607 (0.004979) | 0.524499 / 0.226044 (0.298455) | 5.249235 / 2.268929 (2.980307) | 2.293187 / 55.444624 (-53.151437) | 1.950101 / 6.876477 (-4.926376) | 2.008274 / 2.142072 (-0.133799) | 0.811641 / 4.805227 (-3.993586) | 0.148785 / 6.500664 (-6.351879) | 0.064461 / 0.075469 (-0.011008) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.232227 / 1.841788 (-0.609561) | 13.235896 / 8.074308 (5.161588) | 13.837420 / 10.191392 (3.646028) | 0.135586 / 0.680424 (-0.544838) | 0.028935 / 0.534201 (-0.505266) | 0.397064 / 0.579283 (-0.182220) | 0.393814 / 0.434364 (-0.040549) | 0.480450 / 0.540337 (-0.059887) | 0.561159 / 1.386936 (-0.825777) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006696 / 0.011353 (-0.004657) | 0.004528 / 0.011008 (-0.006480) | 0.077335 / 0.038508 (0.038827) | 0.027181 / 0.023109 (0.004072) | 0.345379 / 0.275898 (0.069481) | 0.372544 / 0.323480 (0.049064) | 0.006808 / 0.007986 (-0.001178) | 0.003284 / 0.004328 (-0.001045) | 0.077379 / 0.004250 (0.073129) | 0.039954 / 0.037052 (0.002901) | 0.348094 / 0.258489 (0.089605) | 0.382315 / 0.293841 (0.088474) | 0.031694 / 0.128546 (-0.096852) | 0.011714 / 0.075646 (-0.063933) | 0.086425 / 0.419271 (-0.332846) | 0.041778 / 0.043533 (-0.001754) | 0.342161 / 0.255139 (0.087022) | 0.363798 / 0.283200 (0.080599) | 0.091315 / 0.141683 (-0.050368) | 1.462066 / 1.452155 (0.009912) | 1.541417 / 1.492716 (0.048700) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.235840 / 0.018006 (0.217834) | 0.397096 / 0.000490 (0.396606) | 0.004597 / 0.000200 (0.004397) | 0.000079 / 0.000054 (0.000025) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024296 / 0.037411 (-0.013115) | 0.099167 / 0.014526 (0.084641) | 0.108257 / 0.176557 (-0.068299) | 0.143434 / 0.737135 (-0.593701) | 0.111933 / 0.296338 (-0.184406) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.440306 / 0.215209 (0.225096) | 4.374065 / 2.077655 (2.296410) | 2.072653 / 1.504120 (0.568533) | 1.864829 / 1.541195 (0.323635) | 1.927970 / 1.468490 (0.459479) | 0.710118 / 4.584777 (-3.874659) | 3.391216 / 3.745712 (-0.354496) | 1.888847 / 5.269862 (-3.381015) | 1.178740 / 4.565676 (-3.386936) | 0.083950 / 0.424275 (-0.340325) | 0.012567 / 0.007607 (0.004960) | 0.540557 / 0.226044 (0.314513) | 5.437621 / 2.268929 (3.168692) | 2.531165 / 55.444624 (-52.913460) | 2.181450 / 6.876477 (-4.695027) | 2.209108 / 2.142072 (0.067035) | 0.814236 / 4.805227 (-3.990991) | 0.153000 / 6.500664 (-6.347664) | 0.066769 / 0.075469 (-0.008700) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.301057 / 1.841788 (-0.540731) | 14.066786 / 8.074308 (5.992478) | 13.641455 / 10.191392 (3.450063) | 0.138838 / 0.680424 (-0.541586) | 0.016733 / 0.534201 (-0.517468) | 0.391823 / 0.579283 (-0.187460) | 0.390817 / 0.434364 (-0.043547) | 0.487682 / 0.540337 (-0.052656) | 0.581134 / 1.386936 (-0.805802) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#b065547654efa0ec633cf373ac1512884c68b2e1 \"CML watermark\")\n" ]

[]

[ "I still get this error on 2.9.0\r\n<img width=\"1925\" alt=\"image\" src=\"https://user-images.githubusercontent.com/7208470/215597359-2f253c76-c472-4612-8099-d3a74d16eb29.png\">\r\n", "Hi ! I just tested locally and or colab and it works fine for 2.9 on `sst2`.\r\n\r\nAlso the code that is shown in your stack trace is not present in the 2.9 source code - so I'm wondering how you installed `datasets` that could cause this ? (you can check by searching for `[0:{label_ids[-1] + 1}]` in the [2.9 codebase](https://github.dev/huggingface/datasets/tree/b5672a956d5de864e6f5550e493527d962d6ae55) - it doesn't find anything)\r\n\r\nAnyway you can try uninstalling `datasets` and install it again", "For what it's worth, I've also gotten this error on 2.9.0, and I've tried uninstalling an reinstalling\r\n\r\n\r\nI'm very new to this package (I was following this tutorial: https://huggingface.co/docs/transformers/training), so there's a good chance I was doing something wrong 😅 but thought I'd pass along the feedback", "@ntrpnr @mtwichel Did you install `datasets` with conda ?\r\n\r\nI suspect that `datasets` 2.9 on conda still have this issue for some reason. When I install `datasets` with `pip` I don't have this error.", "> @ntrpnr @mtwichel Did you install datasets with conda ?\r\n\r\nI did yeah, I wonder if that's the issue", "I just checked on conda at https://anaconda.org/HuggingFace/datasets/files\r\n\r\nand everything looks fine, I got\r\n```python\r\n\r\nf\"ClassLabel expected a value for all label ids [0:{int(label_ids[-1]) + 1}] but some ids are missing.\"\r\n```\r\nas expected in features.py line 1760 (notice the \"int()\") to not have the TypeError.\r\n\r\nFrom where on conda did you install `datasets` ? You should use the `HuggingFace` official channel\r\n\r\nedit: the conda-forge one [here](https://anaconda.org/conda-forge/datasets/files) seems ok as well", "Could you also try this in your notebook ? In case your python kernel doesn't match the `pip` environment in your shell\r\n```python\r\nimport datasets; datasets.__version__\r\n```\r\nand\r\n```\r\n!which python\r\n```\r\n```python\r\nimport sys; sys.executable\r\n```", "Mmmm, just a potential clue:\r\n\r\nWhere are you running your Python code? Is it the Spyder IDE?\r\n\r\nI have recently seen some users reporting conflicting Python environments while using Spyder...\r\n\r\nMaybe related:\r\n- #5487", "Other potential clue:\r\n- Had you already imported `datasets` before pip-updating it? You should first update datasets, before importing it. Otherwise, you need to restart the kernel after updating it.", "I installed `datasets` with Conda using `conda install datasets` and got this issue.\r\n\r\nThen I tried to reinstall using\r\n`\r\nconda install -c huggingface -c conda-forge datasets\r\n`\r\nThe issue is now fixed.", "I'm still getting this error on 2.13.0" ]

[ { "color": "a2eeef", "default": true, "description": "New feature or request", "id": 1935892871, "name": "enhancement", "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement" }, { "color": "c5def5", "default": false, "description": "Generic discussion on the library", "id": 2067400324, "name": "generic discussion", "node_id": "MDU6TGFiZWwyMDY3NDAwMzI0", "url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion" } ]

[ "No comments ?", "I think it would be very cool. I'm currently working on a cluster from Compute Canada, and I have internet access only when I'm not in the nodes where I run the scripts. So I was expecting to be able to use the wmt14 dataset until I realized I needed internet connection even if I downloaded the data already. I'm going to try option 2 you mention for now though! Thanks ;)", "Requiring online connection is a deal breaker in some cases unfortunately so it'd be great if offline mode is added similar to how `transformers` loads models offline fine.\r\n\r\n@mandubian's second bullet point suggests that there's a workaround allowing you to use your offline (custom?) dataset with `datasets`. Could you please elaborate on how that should look like?", "here is my way to load a dataset offline, but it **requires** an online machine\r\n1. (online machine)\r\n```\r\nimport datasets\r\ndata = datasets.load_dataset(...)\r\ndata.save_to_disk(/YOUR/DATASET/DIR)\r\n```\r\n2. copy the dir from online to the offline machine\r\n3. (offline machine)\r\n```\r\nimport datasets\r\ndata = datasets.load_from_disk(/SAVED/DATA/DIR)\r\n```\r\n\r\nHTH.", "> here is my way to load a dataset offline, but it **requires** an online machine\n> \n> 1. (online machine)\n> \n> ```\n> \n> import datasets\n> \n> data = datasets.load_dataset(...)\n> \n> data.save_to_disk(/YOUR/DATASET/DIR)\n> \n> ```\n> \n> 2. copy the dir from online to the offline machine\n> \n> 3. (offline machine)\n> \n> ```\n> \n> import datasets\n> \n> data = datasets.load_from_disk(/SAVED/DATA/DIR)\n> \n> ```\n> \n> \n> \n> HTH.\n\n", "I opened a PR that allows to reload modules that have already been loaded once even if there's no internet.\r\n\r\nLet me know if you know other ways that can make the offline mode experience better. I'd be happy to add them :) \r\n\r\nI already note the \"freeze\" modules option, to prevent local modules updates. It would be a cool feature.\r\n\r\n----------\r\n\r\n> @mandubian's second bullet point suggests that there's a workaround allowing you to use your offline (custom?) dataset with `datasets`. Could you please elaborate on how that should look like?\r\n\r\nIndeed `load_dataset` allows to load remote dataset script (squad, glue, etc.) but also you own local ones.\r\nFor example if you have a dataset script at `./my_dataset/my_dataset.py` then you can do\r\n```python\r\nload_dataset(\"./my_dataset\")\r\n```\r\nand the dataset script will generate your dataset once and for all.\r\n\r\n----------\r\n\r\nAbout I'm looking into having `csv`, `json`, `text`, `pandas` dataset builders already included in the `datasets` package, so that they are available offline by default, as opposed to the other datasets that require the script to be downloaded.\r\ncf #1724 ", "The local dataset builders (csv, text , json and pandas) are now part of the `datasets` package since #1726 :)\r\nYou can now use them offline\r\n```python\r\ndatasets = load_dataset('text', data_files=data_files)\r\n```\r\n\r\nWe'll do a new release soon", "Already fixed by:\r\n- #1726" ]

[]

[ "Hi @lhoestq,\r\nI have updated with the changes from the review.", "Thanks for approving :)" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._" ]

[]

[]

[]

[ "@lhoestq there are many things missing in the README.md file, but this correction is right despite not passing the validation tests...", "Yes indeed. We run the validation in all modified readme because we think that it is the time when contributors are the most likely to fix a dataset card - or it will never be done" ]

[ { "color": "a2eeef", "default": true, "description": "New feature or request", "id": 1935892871, "name": "enhancement", "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement" } ]

[ "This should also resolve: https://github.com/huggingface/datasets/issues/3185." ]

[ { "color": "7057ff", "default": true, "description": "Good for newcomers", "id": 1935892877, "name": "good first issue", "node_id": "MDU6TGFiZWwxOTM1ODkyODc3", "url": "https://api.github.com/repos/huggingface/datasets/labels/good%20first%20issue" } ]

[ "Here please find TFDS format of this dataset: https://www.tensorflow.org/datasets/catalog/wikiann\r\nwhere there is a span column, this is really necessary to be able to use the data, and I appreciate your help @lhoestq ", "Hi !\r\nApparently you can get the spans from the NER tags using `tags_to_spans` defined here:\r\n\r\nhttps://github.com/tensorflow/datasets/blob/c7096bd38e86ed240b8b2c11ecab9893715a7d55/tensorflow_datasets/text/wikiann/wikiann.py#L81-L126\r\n\r\nIt would be nice to include the `spans` field in this dataset as in TFDS. This could be a good first issue for new contributors !\r\n\r\nThe objective is to use `tags_to_spans` in the `_generate_examples` method [here](https://github.com/huggingface/nlp/blob/c98e4b8f23e3770c401c6d9326e243e1ffd599ec/datasets/wikiann/wikiann.py#L292-L316) to create he `spans` for each example.", "Hi @lhoestq \r\nthank you very much for the help, it would be very nice to have it included, here is the full code, one need to also convert tags to string first:\r\n\r\n```\r\nimport datasets \r\nfrom datasets import load_dataset\r\n\r\ndef tags_to_spans(tags):\r\n \"\"\"Convert tags to spans.\"\"\"\r\n spans = set()\r\n span_start = 0\r\n span_end = 0\r\n active_conll_tag = None\r\n for index, string_tag in enumerate(tags):\r\n # Actual BIO tag.\r\n bio_tag = string_tag[0]\r\n assert bio_tag in [\"B\", \"I\", \"O\"], \"Invalid Tag\"\r\n conll_tag = string_tag[2:]\r\n if bio_tag == \"O\":\r\n # The span has ended.\r\n if active_conll_tag:\r\n spans.add((active_conll_tag, (span_start, span_end)))\r\n active_conll_tag = None\r\n # We don't care about tags we are\r\n # told to ignore, so we do nothing.\r\n continue\r\n elif bio_tag == \"B\":\r\n # We are entering a new span; reset indices and active tag to new span.\r\n if active_conll_tag:\r\n spans.add((active_conll_tag, (span_start, span_end)))\r\n active_conll_tag = conll_tag\r\n span_start = index\r\n span_end = index\r\n elif bio_tag == \"I\" and conll_tag == active_conll_tag:\r\n # We're inside a span.\r\n span_end += 1\r\n else:\r\n # This is the case the bio label is an \"I\", but either:\r\n # 1) the span hasn't started - i.e. an ill formed span.\r\n # 2) We have IOB1 tagging scheme.\r\n # We'll process the previous span if it exists, but also include this\r\n # span. This is important, because otherwise, a model may get a perfect\r\n # F1 score whilst still including false positive ill-formed spans.\r\n if active_conll_tag:\r\n spans.add((active_conll_tag, (span_start, span_end)))\r\n active_conll_tag = conll_tag\r\n span_start = index\r\n span_end = index\r\n # Last token might have been a part of a valid span.\r\n if active_conll_tag:\r\n spans.add((active_conll_tag, (span_start, span_end)))\r\n # Return sorted list of spans\r\n return sorted(list(spans), key=lambda x: x[1][0])\r\n\r\ndataset = load_dataset('wikiann', 'en', split=\"train\")\r\nner_tags = {\r\n 0:\"O\",\r\n 1:\"B-PER\",\r\n 2:\"I-PER\",\r\n 3:\"B-ORG\",\r\n 4:\"I-ORG\",\r\n 5:\"B-LOC\",\r\n 6:\"I-LOC\"\r\n}\r\n\r\ndef get_spans(tokens, tags):\r\n \"\"\"Convert tags to textspans.\"\"\"\r\n spans = tags_to_spans(tags)\r\n text_spans = [\r\n x[0] + \": \" + \" \".join([tokens[i]\r\n for i in range(x[1][0], x[1][1] + 1)])\r\n for x in spans\r\n ]\r\n if not text_spans:\r\n text_spans = [\"None\"]\r\n return text_spans\r\n\r\n\r\nfor i, d in enumerate(dataset):\r\n tokens = d['tokens']\r\n tags = d['ner_tags']\r\n tags = [ner_tags[i] for i in tags]\r\n spans = get_spans(tokens, tags)\r\n print(\"spans \", spans)\r\n print(d)\r\n if i > 10:\r\n break; \r\n```\r\nI am not sure how to contribute to the repository and how things work, could you let me know how one can access the datasets to be able to contribute to the repository? Maybe I could do it then\r\nthanks \r\n", "Cool ! Let me give you some context:\r\n\r\n#### Contribution guide\r\n\r\nYou can find the contribution guide here:\r\n\r\nhttps://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md\r\n\r\nIt explains how to set up your dev environment in a few steps.\r\n\r\n#### Dataset loading\r\n\r\nEach Dataset is defined by a Table that have many rows (one row = one example) and columns (one column = one feature).\r\nTo change how a dataset is constructed, you have to modify its dataset script that you can find here:\r\n\r\nhttps://github.com/huggingface/datasets/blob/master/datasets/wikiann/wikiann.py\r\n\r\nIt includes everything needed to load the WikiANN dataset.\r\nYou can load locally a modified version of `wikiann.py` with `load_dataset(\"path/to/wikiann.py\")`.\r\n\r\n#### Define a new column\r\n\r\nEach column has a name and a type. You can see how the features of WikiANN are defined here:\r\n\r\nhttps://github.com/huggingface/datasets/blob/c98e4b8f23e3770c401c6d9326e243e1ffd599ec/datasets/wikiann/wikiann.py#L245-L263\r\n\r\nIdeally we would have one additional feature \"spans\":\r\n```python\r\n \"spans\": datasets.Sequence(datasets.Value(\"string\")),\r\n```\r\n\r\n#### Compute the content of each row\r\n\r\nTo build the WikiANN rows, the _generate_examples method from [here](https://github.com/huggingface/nlp/blob/c98e4b8f23e3770c401c6d9326e243e1ffd599ec/datasets/wikiann/wikiann.py#L292-L316) is used. This function `yield` one python dictionary for each example:\r\n```python\r\nyield guid_index, {\"tokens\": tokens, \"ner_tags\": ner_tags, \"langs\": langs}\r\n```\r\n\r\nThe objective would be to return instead something like\r\n```python\r\nspans = spans = get_spans(tokens, tags)\r\nyield guid_index, {\"tokens\": tokens, \"ner_tags\": ner_tags, \"langs\": langs, \"spans\": spans}\r\n```\r\n\r\nLet me know if you have questions !", "The PR was merged. Issue should be closed.\r\n\r\nCC: @lhoestq " ]

[]

[]

[ { "color": "ffffff", "default": true, "description": "This will not be worked on", "id": 1935892913, "name": "wontfix", "node_id": "MDU6TGFiZWwxOTM1ODkyOTEz", "url": "https://api.github.com/repos/huggingface/datasets/labels/wontfix" } ]

[ "As already replied by @lhoestq (private channel):\r\n> `.train_test_split` (as well as `.shard`, `.select`) doesn't create a new arrow table to save time and disk space. Instead, it uses an indices mapping on top of the table that locate which examples are part of train or test.", "This is an optimization that we don't plan to \"fix\", so I'm closing this issue." ]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_4828). All of your documentation changes will be reflected on that endpoint.", "Hey @mariosasko could we please merge this? I'm still getting the original error at #4796 ." ]

[]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_3907). All of your documentation changes will be reflected on that endpoint." ]

[]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_5247). All of your documentation changes will be reflected on that endpoint." ]

[ { "color": "0075ca", "default": true, "description": "Improvements or additions to documentation", "id": 1935892861, "name": "documentation", "node_id": "MDU6TGFiZWwxOTM1ODkyODYx", "url": "https://api.github.com/repos/huggingface/datasets/labels/documentation" } ]

[ "Good idea, I think I've seen this a couple of times before too on the forums. I can work on this :)", "Closed in #5693 " ]

[]

[]

[]

[ "Hmm the error doesn't seem related to data loading.\r\n\r\nRegarding `split_dataset_by_node`: it's generally used to split an iterable dataset (e.g. when streaming) in pytorch DDP. It's not needed if you use a regular dataset since the pytorch DataLoader already assigns a subset of the dataset indices to each node." ]

[]

[ "The CI fail about the meteor metric is unrelated to this PR " ]

[]

[ "@lhoestq thanks for the comments - I believe these are now addressed. I re-generated the datasets_info.json and dummy data" ]

[ { "color": "e99695", "default": false, "description": "Requesting to add a new dataset", "id": 2067376369, "name": "dataset request", "node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request" } ]

[]

[]

[]

[]

[ "this also happened to me with `nlp.load_dataset('glue', 'mnli')`", "Yes, this is because the test sets for glue are hidden so the labels are\nnot publicly available. You can read the glue paper for more details.\n\nOn Sat, 6 Jun 2020 at 18:16, Jack Morris <[email protected]> wrote:\n\n> this also happened to me with nlp.load_datasets('glue', 'mnli')\n>\n> —\n> You are receiving this because you are subscribed to this thread.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/nlp/issues/245#issuecomment-640083980>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/ABYDIHMVQD2EDX2HTZUXG5DRVJTWRANCNFSM4NVG3AKQ>\n> .\n>\n", "Thanks @thomwolf!", "@thomwolf shouldn't this be visible in the .info and/or in the .features?", "It should be in the datasets card (the README.md and on the hub) in my opinion. What do you think @yjernite?", "I checked both before I got to looking at issues, so that would be fine as well.\r\n\r\nSome additional thoughts on this: Is there a specific reason why the \"test\" split even has a \"label\" column if it isn't tagged. Shouldn't there just not be any. Seems more transparent", "I'm a little confused with the data size.\r\n`sst2` dataset is referenced to `Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank` and the link of the dataset in the paper is https://nlp.stanford.edu/sentiment/index.html which is often shown in GLUE/SST2 reference.\r\nFrom the original data, the standard train/dev/test splits split is 6920/872/1821 for binary classification. \r\nWhy in GLUE/SST2 the train/dev/test split is 67,349/872/1,821 ? \r\n\r\n", "> I'm a little confused with the data size.\r\n> `sst2` dataset is referenced to `Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank` and the link of the dataset in the paper is https://nlp.stanford.edu/sentiment/index.html which is often shown in GLUE/SST2 reference.\r\n> From the original data, the standard train/dev/test splits split is 6920/872/1821 for binary classification.\r\n> Why in GLUE/SST2 the train/dev/test split is 67,349/872/1,821 ?\r\n\r\nHave you figured out this problem? AFAIK, the original sst-2 dataset is totally different from the GLUE/sst-2. Do you think so?", "@yc1999 Sorry, I didn't solve this conflict. In the end, I just use a local data file provided by the previous work I followed(for consistent comparison), not use `datasets` package.\r\n\r\nRelated information: https://github.com/thunlp/OpenAttack/issues/146#issuecomment-766323571", "@yc1999 I find that the original SST-2 dataset (6,920/872/1,821) can be loaded from https://huggingface.co/datasets/gpt3mix/sst2 or built with SST data and the scripts in https://github.com/prrao87/fine-grained-sentiment/tree/master/data/sst.\r\nThe GLUE/SST-2 dataset (67,349/872/1,821) should be a completely different version.\r\n" ]

[]

[ "Moving `\"en\"` to the end of the path instead of passing it as a config name should fix the error:\r\n```python\r\nimport datasets\r\ndataset = datasets.load_dataset('/path/to/your/data/dir/en', streaming=True, split='train')\r\ndataset = dataset.shuffle(buffer_size=10_000, seed=42)\r\nnext(iter(dataset))\r\n```\r\n\r\nPS: https://github.com/huggingface/datasets/pull/5331, once merged, will allow us to define C4's configs in its README, making downloading it much more user-friendly." ]

[]

[ "Hi ! I don't remember exactly how it was done, but maybe you have to embed `f\"{title}<sep>{text}\"` ?\r\n\r\nUsing a HF tokenizer it corresponds to doing\r\n```python\r\ntokenized = tokenizer(titles, texts)\r\n```" ]

[]

[]

[]

[]

[ { "color": "2edb81", "default": false, "description": "A bug in a dataset script provided in the library", "id": 2067388877, "name": "dataset bug", "node_id": "MDU6TGFiZWwyMDY3Mzg4ODc3", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20bug" } ]

[ "It would help a lot to be able to preview the dataset - I'd like to see if the pronunciations are in the dataset, eg. for [\"word\"](https://en.wiktionary.org/wiki/word),\r\n\r\nPronunciation\r\n([Received Pronunciation](https://en.wikipedia.org/wiki/Received_Pronunciation)) [IPA](https://en.wiktionary.org/wiki/Wiktionary:International_Phonetic_Alphabet)([key](https://en.wiktionary.org/wiki/Appendix:English_pronunciation)): /wɜːd/\r\n([General American](https://en.wikipedia.org/wiki/General_American)) [enPR](https://en.wiktionary.org/wiki/Appendix:English_pronunciation): wûrd, [IPA](https://en.wiktionary.org/wiki/Wiktionary:International_Phonetic_Alphabet)([key](https://en.wiktionary.org/wiki/Appendix:English_pronunciation)): /wɝd/", "Hi @i-am-neo, thanks for reporting.\r\n\r\nNormally this dataset should be private and not accessible for public use. @cakiki, @lvwerra, any reason why is it public? I see many other Wikimedia datasets are also public.\r\n\r\nAlso note that last commit \"Add metadata\" (https://huggingface.co/datasets/bigscience-catalogue-lm-data/lm_en_wiktionary_filtered/commit/dc2f458dab50e00f35c94efb3cd4009996858609) introduced buggy data files (`data/file-01.jsonl.gz.lock`, `data/file-01.jsonl.gz.lock.lock`). The same bug appears in other datasets as well.\r\n\r\n@i-am-neo, please note that in the near future we are planning to make public all datasets used for the BigScience project (at least all of them whose license allows to do that). Once public, they will be accessible for all the NLP community.", "Ah this must be a bug introduced at creation time since the repos were created programmatically; I'll go ahead and make them private; sorry about that!", "All datasets are private now. \r\n\r\nRe:that bug I think we're currently avoiding it by avoiding verifications. (i.e. `ignore_verifications=True`)", "Thanks a lot, @cakiki.\r\n\r\n@i-am-neo, I'm closing this issue for now because the dataset is not publicly available yet. Just stay tuned, as we will soon release all the BigScience open-license datasets. ", "Thanks for letting me know, @albertvillanova @cakiki.\r\nAny chance of having a subset alpha version in the meantime? \r\nI only need two dicts out of wiktionary: 1) phoneme(as key): word, and 2) word(as key): its phonemes.\r\n\r\nWould like to use it for a mini-poc [Robust ASR](https://github.com/huggingface/transformers/issues/13162#issuecomment-1096881290) decoding, cc @patrickvonplaten. \r\n\r\n(Patrick, possible to email you so as not to litter github with comments? I have some observations after experiments training hubert on some YT AMI-like data (11.44% wer). Also wonder if a robust ASR is on your/HG's roadmap). Thanks!", "Hey @i-am-neo,\r\n\r\nCool to hear that you're working on Robust ASR! Feel free to drop me a mail :-)", "@i-am-neo This particular subset of the dataset was taken from the [CirrusSearch dumps](https://dumps.wikimedia.org/other/cirrussearch/current/)\r\nYou're specifically after the [enwiktionary-20220425-cirrussearch-content.json.gz](https://dumps.wikimedia.org/other/cirrussearch/current/enwiktionary-20220425-cirrussearch-content.json.gz) file", "thanks @cakiki ! <del>I could access the gz file yesterday (but neglected to tuck it away somewhere safe), and today the link is throwing a 404. Can you help? </del> Never mind, got it!", "thanks @patrickvonplaten. will do - getting my observations together." ]

[]

[ "I agree that a `.filter` method would be VERY useful and appreciated. I'm not a big fan of using `flatten_nested` as it completely breaks down the structure of the example and it may create bugs. Right now I think it may not work for nested structures. Maybe there's a simpler way that we've not figured out yet.", "Instead of flattening everything and rebuilding the example, maybe we can try to access the examples like this:\r\n```python\r\nfor i in range(num_examples):\r\n example = map_nested(lambda x: x[i], batch)\r\n # ... then test to keep it or not\r\n```", "> Instead of flattening everything and rebuilding the example, maybe we can try to access the examples like this:\r\n> \r\n> ```python\r\n> for i in range(num_examples):\r\n> example = map_nested(lambda x: x[i], batch)\r\n> # ... then test to keep it or not\r\n> ```\r\n\r\nAwesome I'll check it out :-) ", "> Instead of flattening everything and rebuilding the example, maybe we can try to access the examples like this:\r\n> \r\n> ```python\r\n> for i in range(num_examples):\r\n> example = map_nested(lambda x: x[i], batch)\r\n> # ... then test to keep it or not\r\n> ```\r\n\r\nAwesome this function is definitely much nicer!", "Actually I just realized that `map_nested` might not work either as it applies the function at the very last list of the structure. However we can imagine that a single example has also a list in its structure:\r\n```python\r\none_example = {\r\n \"title\": \"blabla\",\r\n \"paragraphs\": [\r\n \"p1\", \"p2\", ...\r\n ]\r\n}\r\n```", "We'll probably have to take into account the `dset._data.schema` to extract the examples from the batch.", "> Actually I just realized that `map_nested` might not work either as it applies the function at the very last list of the structure. However we can imagine that a single example has also a list in its structure:\r\n> \r\n> ```python\r\n> one_example = {\r\n> \"title\": \"blabla\",\r\n> \"paragraphs\": [\r\n> \"p1\", \"p2\", ...\r\n> ]\r\n> }\r\n> ```\r\n\r\nThey both work. I'm using it on trivia_qa which is pretty nested. If you use the option `dict_only=True` I think it's fine.", "> We'll probably have to take into account the `dset._data.schema` to extract the examples from the batch.\r\n\r\nWhy? ", "Actually it's fine. I guess this is going to be yet another thing to be unit-tested just to make sure ^^", "Yes, I will need to add tests and documentation! \r\n@thomwolf - would a function like this be ok? It abstracts `.map()` a bit which might be hard to understand. ", "I tried on some datasets with nested structure and it works fine ! Great work :D \r\n", "Awesome :-), I will add documentation and some simple unittests", "Ok merging!" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._" ]

[]

[ "@lhoestq Re-added the arg. The ConnectionError in CI seems unrelated to this PR (the same test fails on master as well).", "Thanks !\r\nI'm re-running the CI, maybe this was an issue with circleCI", "Looks all good now, merged :)" ]

[]

[ "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.005746 / 0.011353 (-0.005607) | 0.003461 / 0.011008 (-0.007548) | 0.078672 / 0.038508 (0.040164) | 0.056800 / 0.023109 (0.033691) | 0.312853 / 0.275898 (0.036955) | 0.346715 / 0.323480 (0.023235) | 0.004516 / 0.007986 (-0.003469) | 0.002872 / 0.004328 (-0.001457) | 0.061264 / 0.004250 (0.057013) | 0.046606 / 0.037052 (0.009553) | 0.320080 / 0.258489 (0.061591) | 0.350390 / 0.293841 (0.056550) | 0.026445 / 0.128546 (-0.102101) | 0.007710 / 0.075646 (-0.067936) | 0.259519 / 0.419271 (-0.159752) | 0.043935 / 0.043533 (0.000402) | 0.320015 / 0.255139 (0.064876) | 0.339799 / 0.283200 (0.056599) | 0.018638 / 0.141683 (-0.123044) | 1.463393 / 1.452155 (0.011239) | 1.496977 / 1.492716 (0.004261) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.185175 / 0.018006 (0.167168) | 0.420734 / 0.000490 (0.420245) | 0.002569 / 0.000200 (0.002369) | 0.000067 / 0.000054 (0.000012) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.022335 / 0.037411 (-0.015077) | 0.071686 / 0.014526 (0.057161) | 0.079906 / 0.176557 (-0.096650) | 0.140386 / 0.737135 (-0.596749) | 0.079712 / 0.296338 (-0.216627) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.392643 / 0.215209 (0.177434) | 3.917934 / 2.077655 (1.840279) | 1.906808 / 1.504120 (0.402688) | 1.729564 / 1.541195 (0.188369) | 1.751533 / 1.468490 (0.283043) | 0.496810 / 4.584777 (-4.087967) | 3.047405 / 3.745712 (-0.698307) | 4.361766 / 5.269862 (-0.908095) | 2.660845 / 4.565676 (-1.904832) | 0.056951 / 0.424275 (-0.367324) | 0.006277 / 0.007607 (-0.001330) | 0.466357 / 0.226044 (0.240312) | 4.660457 / 2.268929 (2.391529) | 2.328590 / 55.444624 (-53.116034) | 1.986140 / 6.876477 (-4.890337) | 2.096182 / 2.142072 (-0.045891) | 0.581685 / 4.805227 (-4.223542) | 0.123643 / 6.500664 (-6.377021) | 0.060286 / 0.075469 (-0.015183) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.237024 / 1.841788 (-0.604763) | 17.778533 / 8.074308 (9.704225) | 13.202205 / 10.191392 (3.010813) | 0.141301 / 0.680424 (-0.539123) | 0.016453 / 0.534201 (-0.517748) | 0.329173 / 0.579283 (-0.250110) | 0.349945 / 0.434364 (-0.084419) | 0.375319 / 0.540337 (-0.165018) | 0.530394 / 1.386936 (-0.856542) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.005863 / 0.011353 (-0.005489) | 0.003578 / 0.011008 (-0.007430) | 0.062719 / 0.038508 (0.024211) | 0.056192 / 0.023109 (0.033082) | 0.422812 / 0.275898 (0.146914) | 0.454316 / 0.323480 (0.130836) | 0.004446 / 0.007986 (-0.003540) | 0.002808 / 0.004328 (-0.001521) | 0.062819 / 0.004250 (0.058569) | 0.046243 / 0.037052 (0.009190) | 0.445858 / 0.258489 (0.187369) | 0.463750 / 0.293841 (0.169909) | 0.027504 / 0.128546 (-0.101042) | 0.007897 / 0.075646 (-0.067749) | 0.068248 / 0.419271 (-0.351024) | 0.041921 / 0.043533 (-0.001612) | 0.413314 / 0.255139 (0.158175) | 0.441619 / 0.283200 (0.158419) | 0.019246 / 0.141683 (-0.122437) | 1.457069 / 1.452155 (0.004914) | 1.524168 / 1.492716 (0.031452) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.237785 / 0.018006 (0.219779) | 0.418455 / 0.000490 (0.417965) | 0.002301 / 0.000200 (0.002101) | 0.000065 / 0.000054 (0.000011) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025630 / 0.037411 (-0.011781) | 0.076673 / 0.014526 (0.062147) | 0.084877 / 0.176557 (-0.091680) | 0.137528 / 0.737135 (-0.599607) | 0.085261 / 0.296338 (-0.211077) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.419040 / 0.215209 (0.203831) | 4.183022 / 2.077655 (2.105368) | 2.157852 / 1.504120 (0.653732) | 1.966177 / 1.541195 (0.424982) | 2.019612 / 1.468490 (0.551122) | 0.497415 / 4.584777 (-4.087362) | 3.102873 / 3.745712 (-0.642839) | 4.526336 / 5.269862 (-0.743525) | 2.991503 / 4.565676 (-1.574174) | 0.057235 / 0.424275 (-0.367040) | 0.006735 / 0.007607 (-0.000872) | 0.498255 / 0.226044 (0.272211) | 4.957364 / 2.268929 (2.688435) | 2.632643 / 55.444624 (-52.811981) | 2.249788 / 6.876477 (-4.626688) | 2.289134 / 2.142072 (0.147062) | 0.583581 / 4.805227 (-4.221646) | 0.126046 / 6.500664 (-6.374618) | 0.062966 / 0.075469 (-0.012504) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.295215 / 1.841788 (-0.546573) | 18.554020 / 8.074308 (10.479711) | 13.683273 / 10.191392 (3.491881) | 0.132266 / 0.680424 (-0.548158) | 0.016376 / 0.534201 (-0.517825) | 0.334495 / 0.579283 (-0.244788) | 0.347106 / 0.434364 (-0.087258) | 0.387531 / 0.540337 (-0.152806) | 0.525745 / 1.386936 (-0.861191) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#f22aa0bf179602e4bf3f44a9de5d180579bb377e \"CML watermark\")\n", "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_6036). All of your documentation changes will be reflected on that endpoint.", "I don't think `transformers` should have any dataset indexing code. So before deprecating I'd be in favor of finding a suitable replacement. Not sure about the stats of the RAG model that uses `datasets` indexing though", "The RAG downloads stats are decent (over 20k downloads last month).\r\n\r\nI think it's suboptimal to maintain an API that only a single model uses. One option is to put this code into a separate lib. However, `langchain` and `docarray` already provide a unified interface to vector stores, so I don't see this as an impactful project. Considering how specific this model is, I think we should go with the simplest solution and combine an index with a dataset in Transformers (this wouldn't require too much code).", "What about migrating to the [datasets-server](https://github.com/huggingface/datasets-server) search feature instead? Would make more sense from a product perspective ", "I don't think it's a good idea:\r\n- using datasets-server would require to upload the data and to not control the indexing, whereas the current feature is about using a local index that you control\r\n- faiss indexes are vector indexes that are not supported by datasets-server, and they are also very customised. For instance RAG uses DPR embeddings and cosine similarity\r\n- FTS is only done for the first 5GB of data for now in datasets-server\r\n\r\nI think a better option would be to integrate with open source search tools such as docarray.\r\nAnd if we want to make the datasets-server search available in python we can build an integration in docarray and/or in huggingface_hub.", "`llama_index` is another popular tool in this space.\r\n\r\n@lhoestq \r\n> I think a better option would be to integrate with open source search tools such as docarray.\r\nAnd if we want to make the datasets-server search available in python we can build an integration in docarray and/or in huggingface_hub.\r\n\r\nI don't think these integrations would be popular unless we integrate them with the Hub \"UI-wise\" (e.g., through a widget), so they can wait IMO. Also, FAISS supports `fsspec` already with the callback reader/writer, so this doesn't require a specific integration. ", "After discussing it a bit with @lhoestq, do we need to deprecate the search API? While I understand it's imperfect, it looks like this will result in significant work to update it everywhere, so I'd favor keeping it until there's an obviously better alternative; this way we can focus on different things in the meantime." ]

[]

[ "If you use huggingface trainer, you will find the trainer has wrapped a `IterableDatasetShard` to avoid duplication.\r\nSee:\r\nhttps://github.com/huggingface/transformers/blob/dfd818420dcbad68e05a502495cf666d338b2bfb/src/transformers/trainer.py#L835\r\n", "If you want to support it by datasets natively, maybe we also need to change the code in `transformers` ?", "Opened https://github.com/huggingface/transformers/issues/20770 to discuss this :)", "Maybe something like this then ?\r\n```python\r\nfrom datasets.distributed import split_dataset_by_node\r\nds = split_dataset_by_node(ds, rank=rank, world_size=world_size)\r\n```\r\n\r\nFor map-style datasets the implementation is trivial (it can simply use `.shard()`).\r\n\r\nFor iterable datasets we would need to implement a new ExamplesIterable that would only iterate on a subset of the (possibly shuffled and re-shuffled after each epoch) list of shards, based on the rank and world size.", "My plan is to skip examples by default to not end up with duplicates.\r\n\r\nAnd if a dataset has a number of shards that is a factor of the world size, then I'd make it more optimized by distributing the shards evenly across nodes instead.", "Opened a PR here: https://github.com/huggingface/datasets/pull/5369\r\n\r\nfeel free to play with it and share your feedbacks :)", "@lhoestq I add shuffle after split_dataset_by_node, duplicated data still exist. \r\nFor example, we have a directory named `mock_pretraining_data`, which has three files, `part-00000`, `part-00002`,`part-00002`. \r\nText in `part-00000` is like this: \r\n{\"id\": 0}\r\n{\"id\": 1}\r\n{\"id\": 2}\r\n{\"id\": 3}\r\n{\"id\": 4}\r\n{\"id\": 5}\r\n{\"id\": 6}\r\n{\"id\": 7}\r\n{\"id\": 8}\r\n{\"id\": 9}\r\n\r\nand `part-00001`\r\n{\"id\": 10}\r\n{\"id\": 11}\r\n{\"id\": 12}\r\n{\"id\": 13}\r\n{\"id\": 14}\r\n{\"id\": 15}\r\n{\"id\": 16}\r\n{\"id\": 17}\r\n{\"id\": 18}\r\n{\"id\": 19}\r\n\r\nand `part-00002`\r\n{\"id\": 20}\r\n{\"id\": 21}\r\n{\"id\": 22}\r\n{\"id\": 23}\r\n{\"id\": 24}\r\n{\"id\": 25}\r\n{\"id\": 26}\r\n{\"id\": 27}\r\n{\"id\": 28}\r\n{\"id\": 29}\r\n\r\nAnd code in `test_dist.py` like this,\r\n```python\r\nimport torch\r\nfrom torch.utils.data import Dataset, DataLoader\r\nfrom datasets import load_dataset\r\nimport os\r\nfrom transformers import AutoTokenizer, NezhaForPreTraining\r\nfrom transformers import AdamW, get_linear_schedule_with_warmup\r\nimport torch.nn.functional as F\r\nimport torch.nn as nn\r\nimport torch.distributed as dist\r\nfrom datasets.distributed import split_dataset_by_node\r\nfrom torch.nn.parallel import DistributedDataParallel as DDP\r\n\r\nos.environ[\"CUDA_VISIBLE_DEVICES\"] = '5,6,7'\r\n\r\ndist.init_process_group(\"nccl\")\r\nlocal_rank = int(os.environ['LOCAL_RANK'])\r\nworld_size = torch.distributed.get_world_size()\r\ndevice = torch.device('cuda', local_rank)\r\ndata_dir = './'\r\n\r\ndef load_trainset(train_path):\r\n dataset = load_dataset('json', data_dir=os.path.join(data_dir, train_path), split='train', streaming=True)\r\n return dataset\r\n\r\ndef collate_fn(examples):\r\n input_ids = []\r\n for example in examples:\r\n input_ids.append(example['id'])\r\n return torch.LongTensor(input_ids).to(device)\r\n\r\n\r\ndataset = load_trainset('mock_pretraining_data')\r\ndataset = split_dataset_by_node(dataset, rank=local_rank, world_size=world_size).shuffle(buffer_size=512)\r\n# train_sampler = torch.utils.data.distributed.DistributedSampler(dataset)\r\nbatch_size = 3\r\nprint('batch_size: {}'.format(batch_size))\r\ntrain_dataloader = DataLoader(dataset, batch_size=batch_size, collate_fn=collate_fn)\r\n\r\nfor x in train_dataloader:\r\n print({'rank': local_rank, 'id': x})\r\n```\r\nrun `python -m torch.distributed.launch --nproc_per_node=3 test_dist.py`\r\nThe output is\r\n```\r\n{'rank': 1, 'id': tensor([12, 15, 14], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([16, 10, 18], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([17, 13, 19], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([11], device='cuda:1')}\r\n{'rank': 0, 'id': tensor([0, 2, 9], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([4, 8, 1], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([5, 3, 6], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([7], device='cuda:0')}\r\n{'rank': 2, 'id': tensor([13, 15, 14], device='cuda:2')}\r\n{'rank': 2, 'id': tensor([19, 17, 18], device='cuda:2')}\r\n{'rank': 2, 'id': tensor([12, 16, 11], device='cuda:2')}\r\n{'rank': 2, 'id': tensor([10], device='cuda:2')}\r\n```\r\n`part-00001` is loaded twice, `part-00002` isn't loaded.\r\n\r\nIf I run `python -m torch.distributed.launch --nproc_per_node=2 test_dist.py`\r\nThe output is weirder，many numbers appear twice\r\n```\r\n{'rank': 1, 'id': tensor([26, 8, 13], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([22, 19, 20], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([12, 28, 11], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([24, 2, 14], device='cuda:1')}\r\n{'rank': 1, 'id': tensor([ 6, 27, 3], device='cuda:1')}\r\n{'rank': 0, 'id': tensor([ 8, 25, 1], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([20, 4, 12], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([14, 29, 5], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([ 7, 18, 23], device='cuda:0')}\r\n{'rank': 0, 'id': tensor([19, 17, 11], device='cuda:0')}\r\n``` ", "Hi ! Thanks for reporting, you need to pass `seed=` to `shuffle()` or the processes won't use the same seed to shuffle the shards order before assigning each shard to a node.\r\n\r\nThe issue is that the workers are not using the same seed to shuffle the shards before splitting the shards list by node.", "Opened https://github.com/huggingface/datasets/issues/5696", "I have the same issue\r\n```\r\nds['train'] = load_dataset(streaming=True)\r\nds['train'] = split_dataset_by_node(ds['train'], rank=int(os.environ[\"RANK\"]), world_size=int(os.environ[\"WORLD_SIZE\"]))\r\nvectorized_datasets = ds.map(\r\n prepare_dataset,\r\n remove_columns=raw_datasets_features,\r\n).with_format(\"torch\")\r\n\r\nvectorized_datasets[\"train\"] = vectorized_datasets[\"train\"].shuffle(\r\n buffer_size=500,\r\n seed=42,\r\n)\r\n\r\ndef prepare_dataset(batch):\r\n ....\r\n print(f\"sentence: {batch['sentence']}, target_text: {batch['target_text']}\")\r\n return batch\r\n```\r\nWhen using split_dataset_by_node(), the data being read is indeed different for each GPU ID.\r\n\r\n```\r\ntrainer = Trainer(\r\n model=model,\r\n data_collator=data_collator,\r\n args=training_args,\r\n compute_metrics=compute_metrics,\r\n train_dataset=vectorized_datasets[\"train\"] if training_args.do_train else None,\r\n eval_dataset=vectorized_datasets[\"eval\"] if training_args.do_eval else None,\r\n tokenizer=processor,\r\n callbacks=[ShuffleCallback()],\r\n )\r\n...\r\ntrain_result = trainer.train(resume_from_checkpoint=checkpoint)\r\n```\r\nHowever, when I execute trainer.train(), the data being read is different from what I expected.\r\nBecause I print the batch value in prepare_dataset() , I observe that the data is the same for each GPU ID.\r\n\r\nHow should I handle this issue?\r\n\r\n\r\n", "There are two ways an iterable dataset can be split by node:\r\n1. if the number of shards is a factor of number of GPUs: in that case the shards are evenly distributed per GPU\r\n2. otherwise, each GPU iterate on the data and at the end keeps 1 sample out of n(GPUs) - skipping the others.\r\n\r\nIn case 2. it's therefore possible to have the same examples passed to `prepare_dataset` for each GPU.\r\n\r\nThis doesn't sound optimized though, because it runs the preprocessing on samples that won't be used in the end.\r\n\r\nCould you open a new issue so that we can discuss about this and find a solution ?" ]

[]

[ "thanks!" ]

[]

[]

[]

[]

[]

[ "Please note that this dataset is actually behind a form that one needs to fill. However, the link is direct. I'm not sure what should the approach be in this case.", "also pinging @thomwolf \r\nThe dataset webpage returns a form when trying to download the dataset (form here : http://www.cfilt.iitb.ac.in/iitb_parallel/dataset.html).\r\nHowever the url we get with the form can be used for the dataset script.\r\nShould we ask the authors or use the urls this way ?", "> also pinging @thomwolf\r\n> The dataset webpage returns a form when trying to download the dataset (form here : http://www.cfilt.iitb.ac.in/iitb_parallel/dataset.html).\r\n> However the url we get with the form can be used for the dataset script.\r\n> Should we ask the authors or use the urls this way ?\r\n\r\nI had discussion on this with @thomwolf . We have already sent email to author of this dataset.", "Hi @spatil6 !\r\nAny news from the authors ?", "IIT B folks will add this dataset to repo." ]

[]

[]

[]

[]

[]

[ "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009812 / 0.011353 (-0.001540) | 0.005290 / 0.011008 (-0.005719) | 0.099728 / 0.038508 (0.061220) | 0.036712 / 0.023109 (0.013602) | 0.305924 / 0.275898 (0.030026) | 0.349844 / 0.323480 (0.026365) | 0.008353 / 0.007986 (0.000368) | 0.004464 / 0.004328 (0.000135) | 0.075329 / 0.004250 (0.071079) | 0.046146 / 0.037052 (0.009094) | 0.304197 / 0.258489 (0.045708) | 0.354245 / 0.293841 (0.060404) | 0.039270 / 0.128546 (-0.089276) | 0.012496 / 0.075646 (-0.063151) | 0.334390 / 0.419271 (-0.084882) | 0.049428 / 0.043533 (0.005896) | 0.297318 / 0.255139 (0.042179) | 0.315646 / 0.283200 (0.032447) | 0.106746 / 0.141683 (-0.034937) | 1.443562 / 1.452155 (-0.008593) | 1.546022 / 1.492716 (0.053305) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.303419 / 0.018006 (0.285413) | 0.536971 / 0.000490 (0.536481) | 0.001335 / 0.000200 (0.001135) | 0.000088 / 0.000054 (0.000033) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030484 / 0.037411 (-0.006927) | 0.110043 / 0.014526 (0.095518) | 0.125265 / 0.176557 (-0.051291) | 0.171410 / 0.737135 (-0.565725) | 0.128978 / 0.296338 (-0.167361) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.398354 / 0.215209 (0.183145) | 3.984180 / 2.077655 (1.906526) | 1.781134 / 1.504120 (0.277014) | 1.589656 / 1.541195 (0.048462) | 1.704192 / 1.468490 (0.235702) | 0.682271 / 4.584777 (-3.902506) | 3.731504 / 3.745712 (-0.014208) | 2.243520 / 5.269862 (-3.026342) | 1.511334 / 4.565676 (-3.054343) | 0.084243 / 0.424275 (-0.340032) | 0.012261 / 0.007607 (0.004654) | 0.507499 / 0.226044 (0.281454) | 5.066037 / 2.268929 (2.797109) | 2.246107 / 55.444624 (-53.198517) | 1.921032 / 6.876477 (-4.955444) | 2.144111 / 2.142072 (0.002039) | 0.845233 / 4.805227 (-3.959995) | 0.165392 / 6.500664 (-6.335272) | 0.064201 / 0.075469 (-0.011268) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.217649 / 1.841788 (-0.624138) | 15.890487 / 8.074308 (7.816179) | 14.772039 / 10.191392 (4.580647) | 0.192901 / 0.680424 (-0.487523) | 0.029119 / 0.534201 (-0.505082) | 0.442904 / 0.579283 (-0.136380) | 0.451035 / 0.434364 (0.016671) | 0.520788 / 0.540337 (-0.019550) | 0.623588 / 1.386936 (-0.763348) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007452 / 0.011353 (-0.003901) | 0.005426 / 0.011008 (-0.005582) | 0.096488 / 0.038508 (0.057980) | 0.033575 / 0.023109 (0.010465) | 0.375688 / 0.275898 (0.099790) | 0.412393 / 0.323480 (0.088913) | 0.006050 / 0.007986 (-0.001936) | 0.004424 / 0.004328 (0.000095) | 0.073102 / 0.004250 (0.068852) | 0.052672 / 0.037052 (0.015620) | 0.379352 / 0.258489 (0.120862) | 0.436065 / 0.293841 (0.142224) | 0.036594 / 0.128546 (-0.091952) | 0.012380 / 0.075646 (-0.063266) | 0.332899 / 0.419271 (-0.086373) | 0.048859 / 0.043533 (0.005326) | 0.373215 / 0.255139 (0.118076) | 0.386990 / 0.283200 (0.103791) | 0.105166 / 0.141683 (-0.036517) | 1.490762 / 1.452155 (0.038607) | 1.611310 / 1.492716 (0.118593) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.333142 / 0.018006 (0.315136) | 0.537137 / 0.000490 (0.536647) | 0.000452 / 0.000200 (0.000252) | 0.000063 / 0.000054 (0.000009) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030368 / 0.037411 (-0.007043) | 0.109608 / 0.014526 (0.095083) | 0.124220 / 0.176557 (-0.052336) | 0.162834 / 0.737135 (-0.574301) | 0.128037 / 0.296338 (-0.168302) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.440991 / 0.215209 (0.225782) | 4.400825 / 2.077655 (2.323170) | 2.158768 / 1.504120 (0.654648) | 1.968158 / 1.541195 (0.426963) | 2.085115 / 1.468490 (0.616625) | 0.710757 / 4.584777 (-3.874020) | 3.835441 / 3.745712 (0.089729) | 2.204118 / 5.269862 (-3.065744) | 1.378909 / 4.565676 (-3.186767) | 0.089149 / 0.424275 (-0.335126) | 0.013066 / 0.007607 (0.005459) | 0.539165 / 0.226044 (0.313121) | 5.414176 / 2.268929 (3.145248) | 2.677020 / 55.444624 (-52.767604) | 2.328334 / 6.876477 (-4.548143) | 2.518933 / 2.142072 (0.376860) | 0.840902 / 4.805227 (-3.964325) | 0.170365 / 6.500664 (-6.330299) | 0.063909 / 0.075469 (-0.011561) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.237205 / 1.841788 (-0.604583) | 15.678776 / 8.074308 (7.604468) | 14.118576 / 10.191392 (3.927184) | 0.167236 / 0.680424 (-0.513188) | 0.018177 / 0.534201 (-0.516024) | 0.426680 / 0.579283 (-0.152603) | 0.425126 / 0.434364 (-0.009238) | 0.501755 / 0.540337 (-0.038582) | 0.592754 / 1.386936 (-0.794182) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png \"CML watermark\")\n", "_The documentation is not available anymore as the PR was closed or merged._", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008708 / 0.011353 (-0.002645) | 0.004462 / 0.011008 (-0.006546) | 0.100159 / 0.038508 (0.061651) | 0.029543 / 0.023109 (0.006434) | 0.304056 / 0.275898 (0.028158) | 0.367098 / 0.323480 (0.043618) | 0.007049 / 0.007986 (-0.000937) | 0.003294 / 0.004328 (-0.001034) | 0.076954 / 0.004250 (0.072703) | 0.036850 / 0.037052 (-0.000202) | 0.307556 / 0.258489 (0.049067) | 0.348327 / 0.293841 (0.054486) | 0.033520 / 0.128546 (-0.095026) | 0.011312 / 0.075646 (-0.064334) | 0.317588 / 0.419271 (-0.101684) | 0.040196 / 0.043533 (-0.003337) | 0.298330 / 0.255139 (0.043191) | 0.333821 / 0.283200 (0.050622) | 0.086584 / 0.141683 (-0.055099) | 1.480205 / 1.452155 (0.028050) | 1.520975 / 1.492716 (0.028259) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.186641 / 0.018006 (0.168635) | 0.414420 / 0.000490 (0.413930) | 0.003021 / 0.000200 (0.002821) | 0.000073 / 0.000054 (0.000018) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.022953 / 0.037411 (-0.014458) | 0.097338 / 0.014526 (0.082812) | 0.104985 / 0.176557 (-0.071572) | 0.139208 / 0.737135 (-0.597927) | 0.108031 / 0.296338 (-0.188307) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.417969 / 0.215209 (0.202759) | 4.173189 / 2.077655 (2.095534) | 1.862813 / 1.504120 (0.358693) | 1.653226 / 1.541195 (0.112031) | 1.725917 / 1.468490 (0.257426) | 0.701038 / 4.584777 (-3.883739) | 3.350500 / 3.745712 (-0.395213) | 1.913156 / 5.269862 (-3.356705) | 1.267597 / 4.565676 (-3.298079) | 0.082197 / 0.424275 (-0.342078) | 0.012499 / 0.007607 (0.004892) | 0.520173 / 0.226044 (0.294128) | 5.219981 / 2.268929 (2.951053) | 2.306029 / 55.444624 (-53.138595) | 1.948169 / 6.876477 (-4.928307) | 2.013160 / 2.142072 (-0.128912) | 0.813325 / 4.805227 (-3.991902) | 0.149729 / 6.500664 (-6.350935) | 0.065492 / 0.075469 (-0.009977) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.194163 / 1.841788 (-0.647625) | 13.739562 / 8.074308 (5.665254) | 13.881988 / 10.191392 (3.690596) | 0.138180 / 0.680424 (-0.542244) | 0.029031 / 0.534201 (-0.505170) | 0.387858 / 0.579283 (-0.191425) | 0.395171 / 0.434364 (-0.039193) | 0.446349 / 0.540337 (-0.093988) | 0.527073 / 1.386936 (-0.859863) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006504 / 0.011353 (-0.004849) | 0.004564 / 0.011008 (-0.006444) | 0.099108 / 0.038508 (0.060599) | 0.027420 / 0.023109 (0.004311) | 0.340712 / 0.275898 (0.064814) | 0.391613 / 0.323480 (0.068133) | 0.004977 / 0.007986 (-0.003009) | 0.003375 / 0.004328 (-0.000953) | 0.076403 / 0.004250 (0.072152) | 0.036650 / 0.037052 (-0.000402) | 0.341948 / 0.258489 (0.083459) | 0.392065 / 0.293841 (0.098224) | 0.031802 / 0.128546 (-0.096745) | 0.011659 / 0.075646 (-0.063987) | 0.320099 / 0.419271 (-0.099173) | 0.041615 / 0.043533 (-0.001918) | 0.342125 / 0.255139 (0.086986) | 0.372833 / 0.283200 (0.089633) | 0.089032 / 0.141683 (-0.052650) | 1.486691 / 1.452155 (0.034536) | 1.567326 / 1.492716 (0.074610) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.193123 / 0.018006 (0.175117) | 0.404062 / 0.000490 (0.403573) | 0.003460 / 0.000200 (0.003260) | 0.000079 / 0.000054 (0.000024) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024565 / 0.037411 (-0.012846) | 0.098958 / 0.014526 (0.084432) | 0.108701 / 0.176557 (-0.067855) | 0.142567 / 0.737135 (-0.594569) | 0.111048 / 0.296338 (-0.185290) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.474549 / 0.215209 (0.259340) | 4.753776 / 2.077655 (2.676121) | 2.435528 / 1.504120 (0.931409) | 2.234491 / 1.541195 (0.693297) | 2.269474 / 1.468490 (0.800984) | 0.695636 / 4.584777 (-3.889141) | 3.367816 / 3.745712 (-0.377896) | 1.854828 / 5.269862 (-3.415034) | 1.159729 / 4.565676 (-3.405948) | 0.082267 / 0.424275 (-0.342008) | 0.012483 / 0.007607 (0.004876) | 0.578490 / 0.226044 (0.352446) | 5.814490 / 2.268929 (3.545561) | 2.893310 / 55.444624 (-52.551314) | 2.540555 / 6.876477 (-4.335922) | 2.573705 / 2.142072 (0.431633) | 0.800545 / 4.805227 (-4.004682) | 0.151306 / 6.500664 (-6.349358) | 0.067925 / 0.075469 (-0.007544) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.294645 / 1.841788 (-0.547142) | 13.641842 / 8.074308 (5.567534) | 14.015200 / 10.191392 (3.823808) | 0.128829 / 0.680424 (-0.551595) | 0.016870 / 0.534201 (-0.517331) | 0.389137 / 0.579283 (-0.190146) | 0.388384 / 0.434364 (-0.045980) | 0.447711 / 0.540337 (-0.092627) | 0.540637 / 1.386936 (-0.846299) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#45ad185b9040a68285080b6099ed3af58442ccb2 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.012282 / 0.011353 (0.000929) | 0.006328 / 0.011008 (-0.004680) | 0.129666 / 0.038508 (0.091158) | 0.039403 / 0.023109 (0.016294) | 0.375464 / 0.275898 (0.099566) | 0.463167 / 0.323480 (0.139687) | 0.010329 / 0.007986 (0.002344) | 0.005111 / 0.004328 (0.000782) | 0.108727 / 0.004250 (0.104476) | 0.047156 / 0.037052 (0.010103) | 0.381869 / 0.258489 (0.123380) | 0.441936 / 0.293841 (0.148095) | 0.054750 / 0.128546 (-0.073796) | 0.019809 / 0.075646 (-0.055837) | 0.436389 / 0.419271 (0.017118) | 0.066585 / 0.043533 (0.023052) | 0.402108 / 0.255139 (0.146969) | 0.424571 / 0.283200 (0.141371) | 0.118326 / 0.141683 (-0.023357) | 1.870175 / 1.452155 (0.418020) | 1.878720 / 1.492716 (0.386004) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.012863 / 0.018006 (-0.005144) | 0.528670 / 0.000490 (0.528181) | 0.006057 / 0.000200 (0.005857) | 0.000124 / 0.000054 (0.000069) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030091 / 0.037411 (-0.007320) | 0.136143 / 0.014526 (0.121618) | 0.148931 / 0.176557 (-0.027626) | 0.179578 / 0.737135 (-0.557558) | 0.144528 / 0.296338 (-0.151810) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.594080 / 0.215209 (0.378871) | 6.029101 / 2.077655 (3.951446) | 2.443084 / 1.504120 (0.938964) | 2.123949 / 1.541195 (0.582754) | 2.183021 / 1.468490 (0.714531) | 1.235453 / 4.584777 (-3.349324) | 5.585121 / 3.745712 (1.839408) | 3.208510 / 5.269862 (-2.061351) | 2.090334 / 4.565676 (-2.475342) | 0.150353 / 0.424275 (-0.273922) | 0.016787 / 0.007607 (0.009180) | 0.797561 / 0.226044 (0.571516) | 7.756291 / 2.268929 (5.487363) | 3.283638 / 55.444624 (-52.160986) | 2.527441 / 6.876477 (-4.349036) | 2.590765 / 2.142072 (0.448692) | 1.446818 / 4.805227 (-3.358409) | 0.250563 / 6.500664 (-6.250101) | 0.077919 / 0.075469 (0.002450) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.612022 / 1.841788 (-0.229765) | 18.363316 / 8.074308 (10.289008) | 22.578570 / 10.191392 (12.387178) | 0.232801 / 0.680424 (-0.447623) | 0.048232 / 0.534201 (-0.485969) | 0.549518 / 0.579283 (-0.029766) | 0.624663 / 0.434364 (0.190299) | 0.674745 / 0.540337 (0.134408) | 0.803489 / 1.386936 (-0.583447) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009872 / 0.011353 (-0.001481) | 0.006593 / 0.011008 (-0.004415) | 0.139248 / 0.038508 (0.100740) | 0.035708 / 0.023109 (0.012598) | 0.551335 / 0.275898 (0.275437) | 0.544995 / 0.323480 (0.221515) | 0.007085 / 0.007986 (-0.000900) | 0.004742 / 0.004328 (0.000413) | 0.095823 / 0.004250 (0.091572) | 0.051674 / 0.037052 (0.014621) | 0.463405 / 0.258489 (0.204916) | 0.640392 / 0.293841 (0.346551) | 0.055242 / 0.128546 (-0.073304) | 0.022602 / 0.075646 (-0.053044) | 0.419171 / 0.419271 (-0.000100) | 0.062986 / 0.043533 (0.019453) | 0.503683 / 0.255139 (0.248544) | 0.568719 / 0.283200 (0.285519) | 0.113906 / 0.141683 (-0.027777) | 1.825248 / 1.452155 (0.373094) | 1.985667 / 1.492716 (0.492951) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.237478 / 0.018006 (0.219472) | 0.528861 / 0.000490 (0.528371) | 0.008507 / 0.000200 (0.008307) | 0.000158 / 0.000054 (0.000103) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.033536 / 0.037411 (-0.003875) | 0.144202 / 0.014526 (0.129677) | 0.139472 / 0.176557 (-0.037084) | 0.184540 / 0.737135 (-0.552596) | 0.147818 / 0.296338 (-0.148520) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.671654 / 0.215209 (0.456445) | 6.616368 / 2.077655 (4.538713) | 2.805634 / 1.504120 (1.301514) | 2.482890 / 1.541195 (0.941695) | 2.547686 / 1.468490 (1.079195) | 1.289169 / 4.584777 (-3.295608) | 5.551436 / 3.745712 (1.805724) | 5.228500 / 5.269862 (-0.041362) | 2.456706 / 4.565676 (-2.108970) | 0.148556 / 0.424275 (-0.275720) | 0.015290 / 0.007607 (0.007683) | 0.837090 / 0.226044 (0.611045) | 8.373561 / 2.268929 (6.104632) | 3.663910 / 55.444624 (-51.780714) | 2.927117 / 6.876477 (-3.949360) | 2.976785 / 2.142072 (0.834712) | 1.501618 / 4.805227 (-3.303609) | 0.263321 / 6.500664 (-6.237343) | 0.082644 / 0.075469 (0.007175) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.707419 / 1.841788 (-0.134368) | 18.371117 / 8.074308 (10.296809) | 22.015154 / 10.191392 (11.823762) | 0.232066 / 0.680424 (-0.448357) | 0.027149 / 0.534201 (-0.507052) | 0.544450 / 0.579283 (-0.034833) | 0.605134 / 0.434364 (0.170770) | 0.656063 / 0.540337 (0.115725) | 0.788121 / 1.386936 (-0.598815) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#f1e0ec31e07e4bc0469f4acfed601d9c71e9a459 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008952 / 0.011353 (-0.002401) | 0.005592 / 0.011008 (-0.005416) | 0.101138 / 0.038508 (0.062630) | 0.035573 / 0.023109 (0.012464) | 0.295959 / 0.275898 (0.020060) | 0.365347 / 0.323480 (0.041867) | 0.008136 / 0.007986 (0.000150) | 0.004479 / 0.004328 (0.000150) | 0.078806 / 0.004250 (0.074556) | 0.045180 / 0.037052 (0.008127) | 0.321687 / 0.258489 (0.063198) | 0.345874 / 0.293841 (0.052033) | 0.038720 / 0.128546 (-0.089826) | 0.012534 / 0.075646 (-0.063112) | 0.335571 / 0.419271 (-0.083700) | 0.049048 / 0.043533 (0.005515) | 0.294756 / 0.255139 (0.039617) | 0.327496 / 0.283200 (0.044296) | 0.109181 / 0.141683 (-0.032502) | 1.417068 / 1.452155 (-0.035087) | 1.455473 / 1.492716 (-0.037244) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.267774 / 0.018006 (0.249768) | 0.538546 / 0.000490 (0.538056) | 0.001755 / 0.000200 (0.001555) | 0.000090 / 0.000054 (0.000035) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026839 / 0.037411 (-0.010572) | 0.105862 / 0.014526 (0.091336) | 0.118278 / 0.176557 (-0.058279) | 0.157926 / 0.737135 (-0.579209) | 0.124700 / 0.296338 (-0.171638) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.399060 / 0.215209 (0.183851) | 3.991409 / 2.077655 (1.913754) | 1.763569 / 1.504120 (0.259449) | 1.579602 / 1.541195 (0.038407) | 1.652928 / 1.468490 (0.184438) | 0.692962 / 4.584777 (-3.891815) | 3.784635 / 3.745712 (0.038922) | 3.249341 / 5.269862 (-2.020521) | 1.815711 / 4.565676 (-2.749966) | 0.084384 / 0.424275 (-0.339891) | 0.012546 / 0.007607 (0.004939) | 0.521397 / 0.226044 (0.295352) | 5.075824 / 2.268929 (2.806895) | 2.258353 / 55.444624 (-53.186272) | 1.925220 / 6.876477 (-4.951256) | 2.002821 / 2.142072 (-0.139252) | 0.830507 / 4.805227 (-3.974720) | 0.165845 / 6.500664 (-6.334819) | 0.063905 / 0.075469 (-0.011565) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.198726 / 1.841788 (-0.643061) | 14.804448 / 8.074308 (6.730139) | 12.855167 / 10.191392 (2.663775) | 0.167932 / 0.680424 (-0.512492) | 0.028643 / 0.534201 (-0.505558) | 0.441224 / 0.579283 (-0.138059) | 0.434924 / 0.434364 (0.000560) | 0.516188 / 0.540337 (-0.024150) | 0.605017 / 1.386936 (-0.781919) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007031 / 0.011353 (-0.004322) | 0.005157 / 0.011008 (-0.005851) | 0.086943 / 0.038508 (0.048434) | 0.031377 / 0.023109 (0.008268) | 0.334810 / 0.275898 (0.058912) | 0.368590 / 0.323480 (0.045110) | 0.005973 / 0.007986 (-0.002013) | 0.004173 / 0.004328 (-0.000155) | 0.067033 / 0.004250 (0.062783) | 0.054070 / 0.037052 (0.017018) | 0.332232 / 0.258489 (0.073743) | 0.384982 / 0.293841 (0.091141) | 0.034023 / 0.128546 (-0.094524) | 0.011301 / 0.075646 (-0.064345) | 0.295644 / 0.419271 (-0.123628) | 0.045589 / 0.043533 (0.002056) | 0.330739 / 0.255139 (0.075600) | 0.352841 / 0.283200 (0.069642) | 0.104829 / 0.141683 (-0.036854) | 1.329360 / 1.452155 (-0.122794) | 1.437956 / 1.492716 (-0.054760) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.299187 / 0.018006 (0.281181) | 0.563407 / 0.000490 (0.562917) | 0.004179 / 0.000200 (0.003979) | 0.000114 / 0.000054 (0.000060) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027405 / 0.037411 (-0.010006) | 0.097498 / 0.014526 (0.082972) | 0.114265 / 0.176557 (-0.062292) | 0.146823 / 0.737135 (-0.590313) | 0.117948 / 0.296338 (-0.178391) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.378756 / 0.215209 (0.163547) | 3.774804 / 2.077655 (1.697150) | 1.804149 / 1.504120 (0.300029) | 1.626312 / 1.541195 (0.085117) | 1.731111 / 1.468490 (0.262620) | 0.633493 / 4.584777 (-3.951284) | 3.488220 / 3.745712 (-0.257492) | 3.064710 / 5.269862 (-2.205151) | 1.690647 / 4.565676 (-2.875029) | 0.076093 / 0.424275 (-0.348182) | 0.010820 / 0.007607 (0.003213) | 0.465091 / 0.226044 (0.239046) | 4.676842 / 2.268929 (2.407913) | 2.297381 / 55.444624 (-53.147244) | 1.960355 / 6.876477 (-4.916122) | 1.983742 / 2.142072 (-0.158330) | 0.739525 / 4.805227 (-4.065702) | 0.152663 / 6.500664 (-6.348001) | 0.057316 / 0.075469 (-0.018153) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.104721 / 1.841788 (-0.737067) | 14.577171 / 8.074308 (6.502863) | 13.680402 / 10.191392 (3.489010) | 0.182234 / 0.680424 (-0.498190) | 0.018853 / 0.534201 (-0.515348) | 0.426194 / 0.579283 (-0.153089) | 0.429202 / 0.434364 (-0.005162) | 0.543125 / 0.540337 (0.002788) | 0.645887 / 1.386936 (-0.741049) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#f830952573bdc59f8732b8f1a13f70d9187e0a65 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.010055 / 0.011353 (-0.001298) | 0.005576 / 0.011008 (-0.005432) | 0.100059 / 0.038508 (0.061551) | 0.038535 / 0.023109 (0.015425) | 0.297538 / 0.275898 (0.021640) | 0.368117 / 0.323480 (0.044637) | 0.008540 / 0.007986 (0.000555) | 0.004469 / 0.004328 (0.000141) | 0.075801 / 0.004250 (0.071551) | 0.046604 / 0.037052 (0.009552) | 0.307242 / 0.258489 (0.048753) | 0.343949 / 0.293841 (0.050108) | 0.039353 / 0.128546 (-0.089194) | 0.012446 / 0.075646 (-0.063200) | 0.334628 / 0.419271 (-0.084643) | 0.051628 / 0.043533 (0.008095) | 0.298726 / 0.255139 (0.043587) | 0.316010 / 0.283200 (0.032810) | 0.120564 / 0.141683 (-0.021119) | 1.459396 / 1.452155 (0.007241) | 1.493682 / 1.492716 (0.000965) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.011702 / 0.018006 (-0.006304) | 0.570261 / 0.000490 (0.569771) | 0.003760 / 0.000200 (0.003560) | 0.000091 / 0.000054 (0.000037) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.028806 / 0.037411 (-0.008605) | 0.112150 / 0.014526 (0.097625) | 0.123140 / 0.176557 (-0.053417) | 0.173055 / 0.737135 (-0.564080) | 0.130060 / 0.296338 (-0.166279) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.398216 / 0.215209 (0.183007) | 3.978677 / 2.077655 (1.901022) | 1.754229 / 1.504120 (0.250109) | 1.561892 / 1.541195 (0.020697) | 1.679138 / 1.468490 (0.210648) | 0.690254 / 4.584777 (-3.894523) | 3.817698 / 3.745712 (0.071986) | 2.177854 / 5.269862 (-3.092008) | 1.361860 / 4.565676 (-3.203816) | 0.084108 / 0.424275 (-0.340167) | 0.012640 / 0.007607 (0.005033) | 0.504385 / 0.226044 (0.278341) | 5.034103 / 2.268929 (2.765174) | 2.254032 / 55.444624 (-53.190593) | 1.910439 / 6.876477 (-4.966038) | 2.003515 / 2.142072 (-0.138558) | 0.839747 / 4.805227 (-3.965480) | 0.165654 / 6.500664 (-6.335010) | 0.063483 / 0.075469 (-0.011986) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.187521 / 1.841788 (-0.654267) | 15.381121 / 8.074308 (7.306812) | 14.579418 / 10.191392 (4.388026) | 0.199221 / 0.680424 (-0.481202) | 0.029335 / 0.534201 (-0.504866) | 0.443159 / 0.579283 (-0.136124) | 0.447772 / 0.434364 (0.013408) | 0.545071 / 0.540337 (0.004733) | 0.650494 / 1.386936 (-0.736442) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007675 / 0.011353 (-0.003677) | 0.005364 / 0.011008 (-0.005644) | 0.097921 / 0.038508 (0.059413) | 0.033645 / 0.023109 (0.010536) | 0.404818 / 0.275898 (0.128920) | 0.429983 / 0.323480 (0.106503) | 0.006106 / 0.007986 (-0.001879) | 0.005281 / 0.004328 (0.000953) | 0.073762 / 0.004250 (0.069512) | 0.053065 / 0.037052 (0.016012) | 0.400657 / 0.258489 (0.142168) | 0.447743 / 0.293841 (0.153902) | 0.036782 / 0.128546 (-0.091765) | 0.012593 / 0.075646 (-0.063054) | 0.332825 / 0.419271 (-0.086446) | 0.049424 / 0.043533 (0.005891) | 0.400397 / 0.255139 (0.145258) | 0.414794 / 0.283200 (0.131594) | 0.106555 / 0.141683 (-0.035128) | 1.466917 / 1.452155 (0.014762) | 1.571351 / 1.492716 (0.078635) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.254337 / 0.018006 (0.236331) | 0.568360 / 0.000490 (0.567870) | 0.000445 / 0.000200 (0.000245) | 0.000059 / 0.000054 (0.000004) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.031044 / 0.037411 (-0.006367) | 0.112282 / 0.014526 (0.097756) | 0.127205 / 0.176557 (-0.049352) | 0.166551 / 0.737135 (-0.570584) | 0.130520 / 0.296338 (-0.165818) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.442906 / 0.215209 (0.227697) | 4.430218 / 2.077655 (2.352563) | 2.287251 / 1.504120 (0.783132) | 2.112345 / 1.541195 (0.571150) | 2.240952 / 1.468490 (0.772462) | 0.713800 / 4.584777 (-3.870977) | 3.884161 / 3.745712 (0.138449) | 2.166901 / 5.269862 (-3.102960) | 1.374490 / 4.565676 (-3.191187) | 0.087548 / 0.424275 (-0.336727) | 0.012369 / 0.007607 (0.004761) | 0.540783 / 0.226044 (0.314739) | 5.396187 / 2.268929 (3.127258) | 2.779636 / 55.444624 (-52.664988) | 2.434220 / 6.876477 (-4.442257) | 2.508180 / 2.142072 (0.366107) | 0.852470 / 4.805227 (-3.952757) | 0.171266 / 6.500664 (-6.329398) | 0.065463 / 0.075469 (-0.010006) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.241720 / 1.841788 (-0.600067) | 15.332568 / 8.074308 (7.258260) | 13.688723 / 10.191392 (3.497331) | 0.145150 / 0.680424 (-0.535273) | 0.017694 / 0.534201 (-0.516507) | 0.426078 / 0.579283 (-0.153205) | 0.441189 / 0.434364 (0.006825) | 0.540284 / 0.540337 (-0.000054) | 0.657548 / 1.386936 (-0.729388) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#c47ecf71362f6b6290b6471b30e77184a5e1df31 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008604 / 0.011353 (-0.002749) | 0.004566 / 0.011008 (-0.006442) | 0.099607 / 0.038508 (0.061099) | 0.029628 / 0.023109 (0.006519) | 0.300481 / 0.275898 (0.024583) | 0.342596 / 0.323480 (0.019116) | 0.007003 / 0.007986 (-0.000982) | 0.003408 / 0.004328 (-0.000920) | 0.079076 / 0.004250 (0.074826) | 0.034104 / 0.037052 (-0.002948) | 0.303856 / 0.258489 (0.045367) | 0.348729 / 0.293841 (0.054888) | 0.033752 / 0.128546 (-0.094794) | 0.011497 / 0.075646 (-0.064149) | 0.321568 / 0.419271 (-0.097704) | 0.041472 / 0.043533 (-0.002061) | 0.303396 / 0.255139 (0.048257) | 0.331121 / 0.283200 (0.047921) | 0.086203 / 0.141683 (-0.055480) | 1.476995 / 1.452155 (0.024840) | 1.539428 / 1.492716 (0.046712) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.215810 / 0.018006 (0.197803) | 0.414292 / 0.000490 (0.413802) | 0.000388 / 0.000200 (0.000188) | 0.000058 / 0.000054 (0.000004) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.023441 / 0.037411 (-0.013970) | 0.098463 / 0.014526 (0.083938) | 0.105435 / 0.176557 (-0.071121) | 0.139736 / 0.737135 (-0.597399) | 0.109467 / 0.296338 (-0.186872) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.418244 / 0.215209 (0.203035) | 4.160693 / 2.077655 (2.083039) | 1.878895 / 1.504120 (0.374775) | 1.679338 / 1.541195 (0.138143) | 1.730384 / 1.468490 (0.261894) | 0.688603 / 4.584777 (-3.896174) | 3.393542 / 3.745712 (-0.352170) | 1.901337 / 5.269862 (-3.368525) | 1.447269 / 4.565676 (-3.118408) | 0.083003 / 0.424275 (-0.341272) | 0.012574 / 0.007607 (0.004967) | 0.526363 / 0.226044 (0.300318) | 5.275159 / 2.268929 (3.006230) | 2.323642 / 55.444624 (-53.120982) | 1.982929 / 6.876477 (-4.893548) | 2.014081 / 2.142072 (-0.127991) | 0.809466 / 4.805227 (-3.995761) | 0.149038 / 6.500664 (-6.351626) | 0.064394 / 0.075469 (-0.011075) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.207439 / 1.841788 (-0.634349) | 13.691048 / 8.074308 (5.616740) | 13.880965 / 10.191392 (3.689573) | 0.148553 / 0.680424 (-0.531871) | 0.028397 / 0.534201 (-0.505804) | 0.391818 / 0.579283 (-0.187465) | 0.407181 / 0.434364 (-0.027183) | 0.481163 / 0.540337 (-0.059175) | 0.570689 / 1.386936 (-0.816247) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006361 / 0.011353 (-0.004992) | 0.004520 / 0.011008 (-0.006488) | 0.097679 / 0.038508 (0.059171) | 0.027223 / 0.023109 (0.004113) | 0.407966 / 0.275898 (0.132068) | 0.439868 / 0.323480 (0.116388) | 0.004625 / 0.007986 (-0.003360) | 0.004039 / 0.004328 (-0.000289) | 0.074548 / 0.004250 (0.070298) | 0.034957 / 0.037052 (-0.002095) | 0.412762 / 0.258489 (0.154273) | 0.449716 / 0.293841 (0.155875) | 0.031272 / 0.128546 (-0.097274) | 0.011598 / 0.075646 (-0.064049) | 0.320922 / 0.419271 (-0.098349) | 0.041250 / 0.043533 (-0.002283) | 0.411439 / 0.255139 (0.156300) | 0.429722 / 0.283200 (0.146523) | 0.087161 / 0.141683 (-0.054522) | 1.512573 / 1.452155 (0.060418) | 1.569385 / 1.492716 (0.076668) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.222612 / 0.018006 (0.204606) | 0.409086 / 0.000490 (0.408596) | 0.004246 / 0.000200 (0.004046) | 0.000083 / 0.000054 (0.000028) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024324 / 0.037411 (-0.013087) | 0.099055 / 0.014526 (0.084530) | 0.106809 / 0.176557 (-0.069748) | 0.141275 / 0.737135 (-0.595860) | 0.109426 / 0.296338 (-0.186913) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.469736 / 0.215209 (0.254527) | 4.686900 / 2.077655 (2.609246) | 2.413392 / 1.504120 (0.909272) | 2.217366 / 1.541195 (0.676171) | 2.266957 / 1.468490 (0.798467) | 0.698647 / 4.584777 (-3.886129) | 3.389317 / 3.745712 (-0.356395) | 1.862315 / 5.269862 (-3.407546) | 1.160931 / 4.565676 (-3.404746) | 0.082829 / 0.424275 (-0.341446) | 0.012627 / 0.007607 (0.005020) | 0.568027 / 0.226044 (0.341983) | 5.683220 / 2.268929 (3.414291) | 2.865701 / 55.444624 (-52.578924) | 2.522401 / 6.876477 (-4.354076) | 2.542395 / 2.142072 (0.400323) | 0.801224 / 4.805227 (-4.004003) | 0.149946 / 6.500664 (-6.350718) | 0.065447 / 0.075469 (-0.010023) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.283756 / 1.841788 (-0.558032) | 13.903662 / 8.074308 (5.829354) | 13.238389 / 10.191392 (3.046997) | 0.142304 / 0.680424 (-0.538120) | 0.016922 / 0.534201 (-0.517279) | 0.377797 / 0.579283 (-0.201487) | 0.382460 / 0.434364 (-0.051904) | 0.464645 / 0.540337 (-0.075692) | 0.556270 / 1.386936 (-0.830666) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#675cf2910c5e6f083ed6664a7bffba9a58f78309 \"CML watermark\")\n", "> I think this would be more of a Conceptual Guide doc since this is more explanatory and compares the differences between a Dataset and an IterableDataset\r\n\r\nsounds good to me !\r\n\r\n> There are definitely places in the docs where we can add a nice and link to this doc though to build up the user's understanding of this topic. For example, in the Know your dataset [tutorial](https://huggingface.co/docs/datasets/access), we only introduce the regular Dataset object and not the IterableDataset. We can add a section there for IterableDataset and then link to this doc that explains the difference between the two 🙂\r\n\r\ngood idea, thanks :)", "I'll open a PR to add a section on `IterableDataset`'s in the tutorial, and once you're done editing this doc I can give it a final polish! 😄 ", "I moved the doc page to conceptual guides and took your suggestions into account :)\r\n\r\nI think this is ready for final review now", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009890 / 0.011353 (-0.001463) | 0.005156 / 0.011008 (-0.005852) | 0.099493 / 0.038508 (0.060984) | 0.036671 / 0.023109 (0.013562) | 0.304686 / 0.275898 (0.028788) | 0.339070 / 0.323480 (0.015590) | 0.008466 / 0.007986 (0.000481) | 0.005863 / 0.004328 (0.001534) | 0.075082 / 0.004250 (0.070832) | 0.045926 / 0.037052 (0.008874) | 0.303157 / 0.258489 (0.044668) | 0.363710 / 0.293841 (0.069870) | 0.038497 / 0.128546 (-0.090049) | 0.012063 / 0.075646 (-0.063583) | 0.334463 / 0.419271 (-0.084808) | 0.048161 / 0.043533 (0.004628) | 0.300431 / 0.255139 (0.045292) | 0.330344 / 0.283200 (0.047145) | 0.105509 / 0.141683 (-0.036174) | 1.475242 / 1.452155 (0.023087) | 1.550624 / 1.492716 (0.057908) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.245749 / 0.018006 (0.227743) | 0.575091 / 0.000490 (0.574601) | 0.001556 / 0.000200 (0.001357) | 0.000089 / 0.000054 (0.000035) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030447 / 0.037411 (-0.006964) | 0.110982 / 0.014526 (0.096456) | 0.126760 / 0.176557 (-0.049797) | 0.173375 / 0.737135 (-0.563760) | 0.128799 / 0.296338 (-0.167539) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.392861 / 0.215209 (0.177651) | 3.911231 / 2.077655 (1.833576) | 1.757413 / 1.504120 (0.253293) | 1.563287 / 1.541195 (0.022093) | 1.658678 / 1.468490 (0.190188) | 0.677244 / 4.584777 (-3.907533) | 3.754917 / 3.745712 (0.009205) | 3.779417 / 5.269862 (-1.490444) | 1.993159 / 4.565676 (-2.572517) | 0.084425 / 0.424275 (-0.339850) | 0.012500 / 0.007607 (0.004893) | 0.501788 / 0.226044 (0.275743) | 5.003173 / 2.268929 (2.734244) | 2.273547 / 55.444624 (-53.171077) | 1.909766 / 6.876477 (-4.966711) | 1.968287 / 2.142072 (-0.173785) | 0.834895 / 4.805227 (-3.970332) | 0.165312 / 6.500664 (-6.335352) | 0.062202 / 0.075469 (-0.013267) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.203080 / 1.841788 (-0.638708) | 15.158284 / 8.074308 (7.083976) | 14.174484 / 10.191392 (3.983092) | 0.171540 / 0.680424 (-0.508883) | 0.028604 / 0.534201 (-0.505597) | 0.438379 / 0.579283 (-0.140904) | 0.429447 / 0.434364 (-0.004917) | 0.540979 / 0.540337 (0.000642) | 0.630322 / 1.386936 (-0.756614) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007600 / 0.011353 (-0.003753) | 0.005400 / 0.011008 (-0.005608) | 0.097983 / 0.038508 (0.059475) | 0.033407 / 0.023109 (0.010297) | 0.384429 / 0.275898 (0.108531) | 0.415880 / 0.323480 (0.092400) | 0.006085 / 0.007986 (-0.001900) | 0.004330 / 0.004328 (0.000002) | 0.074654 / 0.004250 (0.070403) | 0.053076 / 0.037052 (0.016024) | 0.383958 / 0.258489 (0.125469) | 0.427289 / 0.293841 (0.133448) | 0.036710 / 0.128546 (-0.091836) | 0.012400 / 0.075646 (-0.063246) | 0.332712 / 0.419271 (-0.086560) | 0.058390 / 0.043533 (0.014857) | 0.377747 / 0.255139 (0.122608) | 0.398997 / 0.283200 (0.115798) | 0.117370 / 0.141683 (-0.024313) | 1.464211 / 1.452155 (0.012057) | 1.596465 / 1.492716 (0.103749) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.212989 / 0.018006 (0.194983) | 0.554968 / 0.000490 (0.554479) | 0.004305 / 0.000200 (0.004105) | 0.000116 / 0.000054 (0.000061) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029167 / 0.037411 (-0.008244) | 0.109156 / 0.014526 (0.094631) | 0.122575 / 0.176557 (-0.053982) | 0.163058 / 0.737135 (-0.574077) | 0.127431 / 0.296338 (-0.168908) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.445395 / 0.215209 (0.230185) | 4.447534 / 2.077655 (2.369879) | 2.259186 / 1.504120 (0.755066) | 2.082956 / 1.541195 (0.541761) | 2.259126 / 1.468490 (0.790636) | 0.692271 / 4.584777 (-3.892506) | 3.795759 / 3.745712 (0.050047) | 3.603000 / 5.269862 (-1.666862) | 1.948556 / 4.565676 (-2.617120) | 0.084589 / 0.424275 (-0.339687) | 0.012751 / 0.007607 (0.005144) | 0.544783 / 0.226044 (0.318738) | 5.452278 / 2.268929 (3.183349) | 2.809467 / 55.444624 (-52.635157) | 2.479297 / 6.876477 (-4.397180) | 2.587756 / 2.142072 (0.445683) | 0.832258 / 4.805227 (-3.972970) | 0.167424 / 6.500664 (-6.333240) | 0.066064 / 0.075469 (-0.009405) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.262719 / 1.841788 (-0.579069) | 15.917869 / 8.074308 (7.843561) | 13.879301 / 10.191392 (3.687909) | 0.187712 / 0.680424 (-0.492712) | 0.018175 / 0.534201 (-0.516026) | 0.425840 / 0.579283 (-0.153443) | 0.426164 / 0.434364 (-0.008200) | 0.527465 / 0.540337 (-0.012872) | 0.629478 / 1.386936 (-0.757458) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#5f7e178d6373e7d66a60662a22fd60af117f0885 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009064 / 0.011353 (-0.002289) | 0.004824 / 0.011008 (-0.006184) | 0.100869 / 0.038508 (0.062361) | 0.030803 / 0.023109 (0.007694) | 0.350880 / 0.275898 (0.074982) | 0.423816 / 0.323480 (0.100336) | 0.007581 / 0.007986 (-0.000405) | 0.003642 / 0.004328 (-0.000686) | 0.077682 / 0.004250 (0.073432) | 0.039856 / 0.037052 (0.002803) | 0.366097 / 0.258489 (0.107608) | 0.409226 / 0.293841 (0.115385) | 0.033698 / 0.128546 (-0.094848) | 0.011730 / 0.075646 (-0.063916) | 0.321683 / 0.419271 (-0.097588) | 0.041794 / 0.043533 (-0.001739) | 0.351175 / 0.255139 (0.096036) | 0.374328 / 0.283200 (0.091128) | 0.091833 / 0.141683 (-0.049850) | 1.507082 / 1.452155 (0.054927) | 1.543289 / 1.492716 (0.050572) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.010670 / 0.018006 (-0.007337) | 0.429674 / 0.000490 (0.429184) | 0.003246 / 0.000200 (0.003046) | 0.000081 / 0.000054 (0.000026) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025015 / 0.037411 (-0.012397) | 0.102155 / 0.014526 (0.087629) | 0.107010 / 0.176557 (-0.069546) | 0.144265 / 0.737135 (-0.592870) | 0.110635 / 0.296338 (-0.185703) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.414211 / 0.215209 (0.199002) | 4.125582 / 2.077655 (2.047928) | 1.997856 / 1.504120 (0.493736) | 1.847676 / 1.541195 (0.306481) | 1.994100 / 1.468490 (0.525610) | 0.694975 / 4.584777 (-3.889802) | 3.373629 / 3.745712 (-0.372083) | 2.863255 / 5.269862 (-2.406606) | 1.565723 / 4.565676 (-2.999953) | 0.082539 / 0.424275 (-0.341736) | 0.012650 / 0.007607 (0.005043) | 0.522989 / 0.226044 (0.296945) | 5.205720 / 2.268929 (2.936792) | 2.352292 / 55.444624 (-53.092332) | 2.080467 / 6.876477 (-4.796010) | 2.231014 / 2.142072 (0.088942) | 0.811252 / 4.805227 (-3.993975) | 0.149171 / 6.500664 (-6.351493) | 0.065207 / 0.075469 (-0.010262) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.203137 / 1.841788 (-0.638651) | 14.244903 / 8.074308 (6.170595) | 14.454368 / 10.191392 (4.262976) | 0.139090 / 0.680424 (-0.541334) | 0.028738 / 0.534201 (-0.505463) | 0.396394 / 0.579283 (-0.182889) | 0.407207 / 0.434364 (-0.027156) | 0.478036 / 0.540337 (-0.062302) | 0.568488 / 1.386936 (-0.818448) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006878 / 0.011353 (-0.004475) | 0.004636 / 0.011008 (-0.006372) | 0.099118 / 0.038508 (0.060610) | 0.028076 / 0.023109 (0.004967) | 0.416097 / 0.275898 (0.140199) | 0.451722 / 0.323480 (0.128242) | 0.005364 / 0.007986 (-0.002622) | 0.003506 / 0.004328 (-0.000822) | 0.075791 / 0.004250 (0.071541) | 0.041373 / 0.037052 (0.004321) | 0.416358 / 0.258489 (0.157869) | 0.458440 / 0.293841 (0.164599) | 0.031870 / 0.128546 (-0.096676) | 0.011751 / 0.075646 (-0.063896) | 0.321748 / 0.419271 (-0.097524) | 0.041780 / 0.043533 (-0.001752) | 0.425037 / 0.255139 (0.169898) | 0.444169 / 0.283200 (0.160969) | 0.093145 / 0.141683 (-0.048538) | 1.472151 / 1.452155 (0.019996) | 1.542942 / 1.492716 (0.050226) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.224287 / 0.018006 (0.206281) | 0.415303 / 0.000490 (0.414813) | 0.003180 / 0.000200 (0.002980) | 0.000082 / 0.000054 (0.000027) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026377 / 0.037411 (-0.011035) | 0.106222 / 0.014526 (0.091696) | 0.113873 / 0.176557 (-0.062684) | 0.143255 / 0.737135 (-0.593880) | 0.112642 / 0.296338 (-0.183697) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.444149 / 0.215209 (0.228940) | 4.421434 / 2.077655 (2.343779) | 2.082198 / 1.504120 (0.578078) | 1.879909 / 1.541195 (0.338715) | 1.968526 / 1.468490 (0.500036) | 0.697230 / 4.584777 (-3.887546) | 3.430800 / 3.745712 (-0.314912) | 1.893353 / 5.269862 (-3.376509) | 1.173271 / 4.565676 (-3.392406) | 0.082636 / 0.424275 (-0.341639) | 0.012357 / 0.007607 (0.004750) | 0.544008 / 0.226044 (0.317964) | 5.465472 / 2.268929 (3.196543) | 2.530017 / 55.444624 (-52.914608) | 2.178462 / 6.876477 (-4.698014) | 2.279570 / 2.142072 (0.137498) | 0.804890 / 4.805227 (-4.000337) | 0.152091 / 6.500664 (-6.348573) | 0.069442 / 0.075469 (-0.006027) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.256722 / 1.841788 (-0.585065) | 14.554131 / 8.074308 (6.479823) | 13.499913 / 10.191392 (3.308521) | 0.144350 / 0.680424 (-0.536074) | 0.016977 / 0.534201 (-0.517224) | 0.378836 / 0.579283 (-0.200447) | 0.392004 / 0.434364 (-0.042360) | 0.468423 / 0.540337 (-0.071914) | 0.584711 / 1.386936 (-0.802225) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#1e4894fcdf2a82b3355bb6a2dc5557c8e23f8144 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008542 / 0.011353 (-0.002811) | 0.004552 / 0.011008 (-0.006456) | 0.100543 / 0.038508 (0.062035) | 0.029717 / 0.023109 (0.006608) | 0.301948 / 0.275898 (0.026050) | 0.360211 / 0.323480 (0.036731) | 0.006881 / 0.007986 (-0.001105) | 0.003433 / 0.004328 (-0.000896) | 0.077760 / 0.004250 (0.073510) | 0.037069 / 0.037052 (0.000017) | 0.314084 / 0.258489 (0.055595) | 0.347759 / 0.293841 (0.053918) | 0.033255 / 0.128546 (-0.095291) | 0.011487 / 0.075646 (-0.064160) | 0.323873 / 0.419271 (-0.095399) | 0.041203 / 0.043533 (-0.002330) | 0.298397 / 0.255139 (0.043258) | 0.327174 / 0.283200 (0.043974) | 0.088892 / 0.141683 (-0.052791) | 1.560114 / 1.452155 (0.107959) | 1.532475 / 1.492716 (0.039759) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.226080 / 0.018006 (0.208074) | 0.467492 / 0.000490 (0.467003) | 0.002198 / 0.000200 (0.001998) | 0.000074 / 0.000054 (0.000019) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.023627 / 0.037411 (-0.013784) | 0.096696 / 0.014526 (0.082170) | 0.106196 / 0.176557 (-0.070360) | 0.140496 / 0.737135 (-0.596639) | 0.108859 / 0.296338 (-0.187480) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.422335 / 0.215209 (0.207126) | 4.214879 / 2.077655 (2.137224) | 1.865866 / 1.504120 (0.361747) | 1.660914 / 1.541195 (0.119719) | 1.691869 / 1.468490 (0.223379) | 0.688164 / 4.584777 (-3.896613) | 3.432708 / 3.745712 (-0.313004) | 1.856852 / 5.269862 (-3.413010) | 1.243685 / 4.565676 (-3.321991) | 0.081552 / 0.424275 (-0.342723) | 0.012491 / 0.007607 (0.004884) | 0.524331 / 0.226044 (0.298287) | 5.255090 / 2.268929 (2.986162) | 2.269705 / 55.444624 (-53.174919) | 1.936722 / 6.876477 (-4.939755) | 2.018958 / 2.142072 (-0.123114) | 0.800658 / 4.805227 (-4.004569) | 0.148665 / 6.500664 (-6.351999) | 0.064210 / 0.075469 (-0.011259) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.235422 / 1.841788 (-0.606365) | 14.156755 / 8.074308 (6.082447) | 14.005916 / 10.191392 (3.814524) | 0.150983 / 0.680424 (-0.529441) | 0.028500 / 0.534201 (-0.505701) | 0.393013 / 0.579283 (-0.186270) | 0.408191 / 0.434364 (-0.026173) | 0.481017 / 0.540337 (-0.059320) | 0.581711 / 1.386936 (-0.805225) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006950 / 0.011353 (-0.004403) | 0.004575 / 0.011008 (-0.006434) | 0.076702 / 0.038508 (0.038194) | 0.028050 / 0.023109 (0.004941) | 0.342916 / 0.275898 (0.067018) | 0.378861 / 0.323480 (0.055381) | 0.005315 / 0.007986 (-0.002671) | 0.004822 / 0.004328 (0.000494) | 0.075560 / 0.004250 (0.071310) | 0.040441 / 0.037052 (0.003388) | 0.344284 / 0.258489 (0.085795) | 0.386519 / 0.293841 (0.092678) | 0.032122 / 0.128546 (-0.096424) | 0.011843 / 0.075646 (-0.063803) | 0.085798 / 0.419271 (-0.333473) | 0.043027 / 0.043533 (-0.000506) | 0.342910 / 0.255139 (0.087771) | 0.366618 / 0.283200 (0.083418) | 0.094766 / 0.141683 (-0.046917) | 1.492981 / 1.452155 (0.040827) | 1.566994 / 1.492716 (0.074278) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.166083 / 0.018006 (0.148076) | 0.409315 / 0.000490 (0.408826) | 0.003189 / 0.000200 (0.002989) | 0.000127 / 0.000054 (0.000072) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024753 / 0.037411 (-0.012658) | 0.099112 / 0.014526 (0.084586) | 0.106668 / 0.176557 (-0.069889) | 0.142562 / 0.737135 (-0.594573) | 0.110648 / 0.296338 (-0.185690) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.452668 / 0.215209 (0.237459) | 4.501188 / 2.077655 (2.423534) | 2.086197 / 1.504120 (0.582077) | 1.873955 / 1.541195 (0.332761) | 1.935610 / 1.468490 (0.467120) | 0.708290 / 4.584777 (-3.876487) | 3.426986 / 3.745712 (-0.318726) | 2.805852 / 5.269862 (-2.464009) | 1.516918 / 4.565676 (-3.048759) | 0.084067 / 0.424275 (-0.340208) | 0.012776 / 0.007607 (0.005169) | 0.548853 / 0.226044 (0.322809) | 5.488198 / 2.268929 (3.219270) | 2.704464 / 55.444624 (-52.740161) | 2.377817 / 6.876477 (-4.498660) | 2.366152 / 2.142072 (0.224079) | 0.818192 / 4.805227 (-3.987035) | 0.152649 / 6.500664 (-6.348015) | 0.066914 / 0.075469 (-0.008555) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.273803 / 1.841788 (-0.567985) | 14.071633 / 8.074308 (5.997325) | 13.655586 / 10.191392 (3.464194) | 0.149471 / 0.680424 (-0.530953) | 0.016745 / 0.534201 (-0.517456) | 0.386850 / 0.579283 (-0.192434) | 0.393595 / 0.434364 (-0.040769) | 0.480396 / 0.540337 (-0.059942) | 0.573708 / 1.386936 (-0.813228) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#8b2c7de67b326a635c0dc39ea5dd1ae982c958d6 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008173 / 0.011353 (-0.003180) | 0.004461 / 0.011008 (-0.006547) | 0.100284 / 0.038508 (0.061776) | 0.028900 / 0.023109 (0.005791) | 0.293639 / 0.275898 (0.017741) | 0.359450 / 0.323480 (0.035971) | 0.007567 / 0.007986 (-0.000418) | 0.003434 / 0.004328 (-0.000894) | 0.077913 / 0.004250 (0.073663) | 0.036313 / 0.037052 (-0.000740) | 0.308484 / 0.258489 (0.049995) | 0.347575 / 0.293841 (0.053734) | 0.033367 / 0.128546 (-0.095179) | 0.011508 / 0.075646 (-0.064138) | 0.323490 / 0.419271 (-0.095782) | 0.042285 / 0.043533 (-0.001248) | 0.295696 / 0.255139 (0.040557) | 0.332475 / 0.283200 (0.049276) | 0.089980 / 0.141683 (-0.051703) | 1.461851 / 1.452155 (0.009697) | 1.493030 / 1.492716 (0.000314) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.191068 / 0.018006 (0.173062) | 0.396768 / 0.000490 (0.396278) | 0.002355 / 0.000200 (0.002155) | 0.000080 / 0.000054 (0.000025) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.023117 / 0.037411 (-0.014294) | 0.096155 / 0.014526 (0.081630) | 0.102424 / 0.176557 (-0.074132) | 0.142148 / 0.737135 (-0.594987) | 0.105954 / 0.296338 (-0.190384) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.421227 / 0.215209 (0.206018) | 4.200403 / 2.077655 (2.122748) | 1.899410 / 1.504120 (0.395290) | 1.684091 / 1.541195 (0.142896) | 1.698084 / 1.468490 (0.229594) | 0.696195 / 4.584777 (-3.888582) | 3.364116 / 3.745712 (-0.381596) | 1.899133 / 5.269862 (-3.370728) | 1.281405 / 4.565676 (-3.284272) | 0.082958 / 0.424275 (-0.341317) | 0.012433 / 0.007607 (0.004826) | 0.521856 / 0.226044 (0.295812) | 5.217626 / 2.268929 (2.948698) | 2.309228 / 55.444624 (-53.135396) | 1.956828 / 6.876477 (-4.919648) | 2.018964 / 2.142072 (-0.123108) | 0.816855 / 4.805227 (-3.988373) | 0.152867 / 6.500664 (-6.347798) | 0.064764 / 0.075469 (-0.010705) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.219020 / 1.841788 (-0.622768) | 13.509058 / 8.074308 (5.434750) | 13.637826 / 10.191392 (3.446434) | 0.156620 / 0.680424 (-0.523804) | 0.028518 / 0.534201 (-0.505683) | 0.399138 / 0.579283 (-0.180146) | 0.399931 / 0.434364 (-0.034433) | 0.482902 / 0.540337 (-0.057435) | 0.574089 / 1.386936 (-0.812847) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006232 / 0.011353 (-0.005121) | 0.004467 / 0.011008 (-0.006542) | 0.075494 / 0.038508 (0.036986) | 0.026891 / 0.023109 (0.003782) | 0.356603 / 0.275898 (0.080705) | 0.371977 / 0.323480 (0.048497) | 0.004709 / 0.007986 (-0.003276) | 0.003230 / 0.004328 (-0.001099) | 0.074338 / 0.004250 (0.070088) | 0.035588 / 0.037052 (-0.001464) | 0.349554 / 0.258489 (0.091065) | 0.389672 / 0.293841 (0.095831) | 0.031524 / 0.128546 (-0.097022) | 0.011493 / 0.075646 (-0.064153) | 0.084584 / 0.419271 (-0.334688) | 0.041945 / 0.043533 (-0.001588) | 0.341057 / 0.255139 (0.085918) | 0.367876 / 0.283200 (0.084677) | 0.090113 / 0.141683 (-0.051569) | 1.507104 / 1.452155 (0.054949) | 1.567810 / 1.492716 (0.075094) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.210939 / 0.018006 (0.192933) | 0.392600 / 0.000490 (0.392110) | 0.002188 / 0.000200 (0.001988) | 0.000073 / 0.000054 (0.000018) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024294 / 0.037411 (-0.013118) | 0.100325 / 0.014526 (0.085799) | 0.104027 / 0.176557 (-0.072530) | 0.141189 / 0.737135 (-0.595947) | 0.107438 / 0.296338 (-0.188901) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.443314 / 0.215209 (0.228105) | 4.429612 / 2.077655 (2.351957) | 2.129275 / 1.504120 (0.625156) | 1.940016 / 1.541195 (0.398821) | 2.008975 / 1.468490 (0.540485) | 0.695434 / 4.584777 (-3.889343) | 3.355137 / 3.745712 (-0.390575) | 2.606262 / 5.269862 (-2.663600) | 1.451283 / 4.565676 (-3.114394) | 0.082875 / 0.424275 (-0.341400) | 0.012398 / 0.007607 (0.004791) | 0.544262 / 0.226044 (0.318218) | 5.450829 / 2.268929 (3.181900) | 2.582074 / 55.444624 (-52.862550) | 2.220037 / 6.876477 (-4.656439) | 2.232473 / 2.142072 (0.090401) | 0.802094 / 4.805227 (-4.003134) | 0.150188 / 6.500664 (-6.350476) | 0.066543 / 0.075469 (-0.008926) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.269098 / 1.841788 (-0.572690) | 13.764780 / 8.074308 (5.690472) | 13.461490 / 10.191392 (3.270098) | 0.143841 / 0.680424 (-0.536583) | 0.016687 / 0.534201 (-0.517514) | 0.388548 / 0.579283 (-0.190736) | 0.385229 / 0.434364 (-0.049135) | 0.478966 / 0.540337 (-0.061371) | 0.570355 / 1.386936 (-0.816581) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#0ba81f5b299f0918cb0c0c071412feadd0ea3ef5 \"CML watermark\")\n", "I took your comments into account :)\r\n\r\n> Regarding the docs, I think it would be better to add this info as notes/tips/sections to the existing docs (Process/Stream; e.g. a tip under Dataset.shuffle that explains how to make this operation more performant by using to_iterable + shuffle, etc.) rather than introducing a new doc page.\r\n\r\nI added a paragraph in the Dataset.shuffle docstring, and a note in the Process doc page", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.010906 / 0.011353 (-0.000447) | 0.005995 / 0.011008 (-0.005014) | 0.120183 / 0.038508 (0.081675) | 0.042166 / 0.023109 (0.019057) | 0.350945 / 0.275898 (0.075046) | 0.433055 / 0.323480 (0.109575) | 0.009093 / 0.007986 (0.001107) | 0.004695 / 0.004328 (0.000366) | 0.090362 / 0.004250 (0.086112) | 0.051402 / 0.037052 (0.014350) | 0.368677 / 0.258489 (0.110188) | 0.410926 / 0.293841 (0.117086) | 0.044471 / 0.128546 (-0.084075) | 0.014051 / 0.075646 (-0.061595) | 0.397765 / 0.419271 (-0.021507) | 0.057227 / 0.043533 (0.013694) | 0.357587 / 0.255139 (0.102448) | 0.377470 / 0.283200 (0.094270) | 0.119482 / 0.141683 (-0.022201) | 1.719799 / 1.452155 (0.267645) | 1.758228 / 1.492716 (0.265511) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.224385 / 0.018006 (0.206379) | 0.505070 / 0.000490 (0.504580) | 0.004863 / 0.000200 (0.004663) | 0.000379 / 0.000054 (0.000324) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030366 / 0.037411 (-0.007046) | 0.130481 / 0.014526 (0.115955) | 0.136429 / 0.176557 (-0.040128) | 0.182263 / 0.737135 (-0.554872) | 0.142871 / 0.296338 (-0.153468) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.467623 / 0.215209 (0.252414) | 4.665522 / 2.077655 (2.587868) | 2.130885 / 1.504120 (0.626766) | 1.903810 / 1.541195 (0.362615) | 2.019077 / 1.468490 (0.550587) | 0.820868 / 4.584777 (-3.763909) | 4.543118 / 3.745712 (0.797406) | 2.491541 / 5.269862 (-2.778321) | 1.585377 / 4.565676 (-2.980299) | 0.101850 / 0.424275 (-0.322426) | 0.014737 / 0.007607 (0.007129) | 0.597241 / 0.226044 (0.371197) | 5.938445 / 2.268929 (3.669516) | 2.695799 / 55.444624 (-52.748825) | 2.286890 / 6.876477 (-4.589587) | 2.363064 / 2.142072 (0.220991) | 0.986670 / 4.805227 (-3.818557) | 0.194407 / 6.500664 (-6.306257) | 0.074767 / 0.075469 (-0.000702) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.420630 / 1.841788 (-0.421158) | 17.537702 / 8.074308 (9.463394) | 16.521804 / 10.191392 (6.330412) | 0.173622 / 0.680424 (-0.506802) | 0.033944 / 0.534201 (-0.500257) | 0.520461 / 0.579283 (-0.058822) | 0.541283 / 0.434364 (0.106919) | 0.651906 / 0.540337 (0.111569) | 0.771724 / 1.386936 (-0.615212) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008448 / 0.011353 (-0.002905) | 0.005893 / 0.011008 (-0.005115) | 0.087995 / 0.038508 (0.049487) | 0.038602 / 0.023109 (0.015493) | 0.400048 / 0.275898 (0.124150) | 0.436998 / 0.323480 (0.113518) | 0.006414 / 0.007986 (-0.001572) | 0.004478 / 0.004328 (0.000149) | 0.086444 / 0.004250 (0.082194) | 0.056535 / 0.037052 (0.019483) | 0.402066 / 0.258489 (0.143577) | 0.458730 / 0.293841 (0.164889) | 0.041622 / 0.128546 (-0.086924) | 0.014014 / 0.075646 (-0.061632) | 0.101382 / 0.419271 (-0.317889) | 0.056986 / 0.043533 (0.013453) | 0.404527 / 0.255139 (0.149388) | 0.428105 / 0.283200 (0.144906) | 0.118321 / 0.141683 (-0.023361) | 1.716940 / 1.452155 (0.264785) | 1.834683 / 1.492716 (0.341967) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.252917 / 0.018006 (0.234910) | 0.485950 / 0.000490 (0.485461) | 0.000489 / 0.000200 (0.000289) | 0.000066 / 0.000054 (0.000011) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.035023 / 0.037411 (-0.002388) | 0.139055 / 0.014526 (0.124529) | 0.144165 / 0.176557 (-0.032392) | 0.189559 / 0.737135 (-0.547577) | 0.153213 / 0.296338 (-0.143126) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.505069 / 0.215209 (0.289860) | 5.024620 / 2.077655 (2.946965) | 2.429469 / 1.504120 (0.925349) | 2.186210 / 1.541195 (0.645015) | 2.275971 / 1.468490 (0.807481) | 0.829432 / 4.584777 (-3.755345) | 4.518600 / 3.745712 (0.772888) | 2.466418 / 5.269862 (-2.803443) | 1.558910 / 4.565676 (-3.006767) | 0.102017 / 0.424275 (-0.322258) | 0.015191 / 0.007607 (0.007584) | 0.619092 / 0.226044 (0.393048) | 6.241105 / 2.268929 (3.972176) | 3.044213 / 55.444624 (-52.400411) | 2.630194 / 6.876477 (-4.246282) | 2.723685 / 2.142072 (0.581613) | 0.994018 / 4.805227 (-3.811210) | 0.198722 / 6.500664 (-6.301942) | 0.075812 / 0.075469 (0.000343) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.545497 / 1.841788 (-0.296291) | 18.305250 / 8.074308 (10.230942) | 16.035275 / 10.191392 (5.843883) | 0.209339 / 0.680424 (-0.471085) | 0.020903 / 0.534201 (-0.513298) | 0.499909 / 0.579283 (-0.079374) | 0.488775 / 0.434364 (0.054411) | 0.581990 / 0.540337 (0.041653) | 0.697786 / 1.386936 (-0.689150) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#78dca62e8aaddb9e0cf0212841f2c8d861fe74c8 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.011706 / 0.011353 (0.000353) | 0.008406 / 0.011008 (-0.002602) | 0.130887 / 0.038508 (0.092379) | 0.037468 / 0.023109 (0.014359) | 0.385043 / 0.275898 (0.109145) | 0.458837 / 0.323480 (0.135357) | 0.013400 / 0.007986 (0.005414) | 0.004885 / 0.004328 (0.000557) | 0.107156 / 0.004250 (0.102905) | 0.046958 / 0.037052 (0.009906) | 0.419314 / 0.258489 (0.160825) | 0.456061 / 0.293841 (0.162220) | 0.058859 / 0.128546 (-0.069687) | 0.016682 / 0.075646 (-0.058965) | 0.428401 / 0.419271 (0.009129) | 0.062908 / 0.043533 (0.019376) | 0.370902 / 0.255139 (0.115763) | 0.433897 / 0.283200 (0.150697) | 0.125672 / 0.141683 (-0.016011) | 1.818279 / 1.452155 (0.366124) | 1.935767 / 1.492716 (0.443050) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.011928 / 0.018006 (-0.006078) | 0.591995 / 0.000490 (0.591506) | 0.008416 / 0.000200 (0.008216) | 0.000122 / 0.000054 (0.000067) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029640 / 0.037411 (-0.007772) | 0.121044 / 0.014526 (0.106518) | 0.141840 / 0.176557 (-0.034716) | 0.195856 / 0.737135 (-0.541280) | 0.146460 / 0.296338 (-0.149879) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.591838 / 0.215209 (0.376629) | 5.817309 / 2.077655 (3.739654) | 2.411864 / 1.504120 (0.907744) | 2.098517 / 1.541195 (0.557323) | 2.214609 / 1.468490 (0.746119) | 1.217542 / 4.584777 (-3.367235) | 5.658394 / 3.745712 (1.912682) | 5.155807 / 5.269862 (-0.114055) | 2.797313 / 4.565676 (-1.768363) | 0.141309 / 0.424275 (-0.282967) | 0.014462 / 0.007607 (0.006855) | 0.772274 / 0.226044 (0.546230) | 7.547357 / 2.268929 (5.278429) | 3.150178 / 55.444624 (-52.294446) | 2.500130 / 6.876477 (-4.376347) | 2.572036 / 2.142072 (0.429964) | 1.434498 / 4.805227 (-3.370729) | 0.257355 / 6.500664 (-6.243309) | 0.087491 / 0.075469 (0.012022) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.483899 / 1.841788 (-0.357889) | 17.990741 / 8.074308 (9.916433) | 20.398965 / 10.191392 (10.207573) | 0.239529 / 0.680424 (-0.440895) | 0.046118 / 0.534201 (-0.488083) | 0.528349 / 0.579283 (-0.050934) | 0.614333 / 0.434364 (0.179969) | 0.653621 / 0.540337 (0.113284) | 0.794654 / 1.386936 (-0.592282) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008732 / 0.011353 (-0.002621) | 0.006432 / 0.011008 (-0.004576) | 0.090811 / 0.038508 (0.052303) | 0.030154 / 0.023109 (0.007045) | 0.407885 / 0.275898 (0.131987) | 0.452457 / 0.323480 (0.128977) | 0.006966 / 0.007986 (-0.001020) | 0.006449 / 0.004328 (0.002120) | 0.094439 / 0.004250 (0.090188) | 0.050628 / 0.037052 (0.013576) | 0.401815 / 0.258489 (0.143326) | 0.451814 / 0.293841 (0.157973) | 0.047456 / 0.128546 (-0.081090) | 0.019019 / 0.075646 (-0.056628) | 0.112941 / 0.419271 (-0.306331) | 0.057677 / 0.043533 (0.014145) | 0.406160 / 0.255139 (0.151021) | 0.434469 / 0.283200 (0.151269) | 0.110515 / 0.141683 (-0.031167) | 1.601393 / 1.452155 (0.149238) | 1.745581 / 1.492716 (0.252865) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.280264 / 0.018006 (0.262258) | 0.630074 / 0.000490 (0.629585) | 0.006900 / 0.000200 (0.006700) | 0.000112 / 0.000054 (0.000058) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027338 / 0.037411 (-0.010073) | 0.114772 / 0.014526 (0.100246) | 0.130436 / 0.176557 (-0.046121) | 0.168990 / 0.737135 (-0.568145) | 0.135842 / 0.296338 (-0.160496) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.666739 / 0.215209 (0.451530) | 6.212953 / 2.077655 (4.135298) | 2.781716 / 1.504120 (1.277596) | 2.369975 / 1.541195 (0.828781) | 2.338807 / 1.468490 (0.870317) | 1.174138 / 4.584777 (-3.410639) | 5.420297 / 3.745712 (1.674585) | 4.972669 / 5.269862 (-0.297192) | 2.214294 / 4.565676 (-2.351382) | 0.135429 / 0.424275 (-0.288846) | 0.013877 / 0.007607 (0.006270) | 0.750805 / 0.226044 (0.524761) | 7.145429 / 2.268929 (4.876500) | 3.215081 / 55.444624 (-52.229544) | 2.598307 / 6.876477 (-4.278170) | 2.690479 / 2.142072 (0.548406) | 1.344673 / 4.805227 (-3.460554) | 0.241536 / 6.500664 (-6.259128) | 0.075544 / 0.075469 (0.000074) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.473595 / 1.841788 (-0.368192) | 17.372237 / 8.074308 (9.297929) | 18.586588 / 10.191392 (8.395196) | 0.209300 / 0.680424 (-0.471124) | 0.030878 / 0.534201 (-0.503323) | 0.509131 / 0.579283 (-0.070152) | 0.617884 / 0.434364 (0.183520) | 0.633721 / 0.540337 (0.093383) | 0.727624 / 1.386936 (-0.659312) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#87f2062d47fdbec3fadf5b39bab0801f59c0f4a3 \"CML watermark\")\n", "Took your last comments into account !\r\n\r\n> so maybe a better title for it would be \"Optimize processing\" (or \"Working with datasets at scale\" as I mentioned earlier on Slack)\r\n\r\nI think the content would be slightly different, e.g. focus more on multiprocessing/sharding or what data formats to use. This can be a complementary page IMO\r\n\r\n> PS: I think it would be a good idea to add links to the Guide pages for better discoverability and to somewhat \"justify their presence in the docs\" (from the tutorial/how-to pages to the guides; some guides are not referenced at all)\r\n\r\nAdded a link in the how-to stream page. We may want to include it in the tutorial at one point at well - right now none of the tutorials mention streaming", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009167 / 0.011353 (-0.002186) | 0.005345 / 0.011008 (-0.005663) | 0.098302 / 0.038508 (0.059794) | 0.035649 / 0.023109 (0.012540) | 0.295597 / 0.275898 (0.019699) | 0.358843 / 0.323480 (0.035364) | 0.008011 / 0.007986 (0.000025) | 0.004229 / 0.004328 (-0.000100) | 0.075123 / 0.004250 (0.070872) | 0.046098 / 0.037052 (0.009046) | 0.310581 / 0.258489 (0.052092) | 0.343230 / 0.293841 (0.049389) | 0.038318 / 0.128546 (-0.090229) | 0.011954 / 0.075646 (-0.063693) | 0.331056 / 0.419271 (-0.088216) | 0.052875 / 0.043533 (0.009342) | 0.302758 / 0.255139 (0.047619) | 0.340596 / 0.283200 (0.057396) | 0.113676 / 0.141683 (-0.028007) | 1.448272 / 1.452155 (-0.003883) | 1.498008 / 1.492716 (0.005291) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.240524 / 0.018006 (0.222518) | 0.555823 / 0.000490 (0.555333) | 0.003143 / 0.000200 (0.002943) | 0.000098 / 0.000054 (0.000044) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027764 / 0.037411 (-0.009647) | 0.105006 / 0.014526 (0.090480) | 0.120550 / 0.176557 (-0.056007) | 0.167052 / 0.737135 (-0.570084) | 0.124521 / 0.296338 (-0.171818) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.401758 / 0.215209 (0.186549) | 3.989629 / 2.077655 (1.911974) | 1.767307 / 1.504120 (0.263187) | 1.579451 / 1.541195 (0.038257) | 1.637642 / 1.468490 (0.169152) | 0.702524 / 4.584777 (-3.882253) | 3.714326 / 3.745712 (-0.031386) | 2.131829 / 5.269862 (-3.138033) | 1.487410 / 4.565676 (-3.078267) | 0.084901 / 0.424275 (-0.339374) | 0.012292 / 0.007607 (0.004685) | 0.505211 / 0.226044 (0.279166) | 5.074479 / 2.268929 (2.805551) | 2.243068 / 55.444624 (-53.201556) | 1.880199 / 6.876477 (-4.996278) | 2.003757 / 2.142072 (-0.138315) | 0.870719 / 4.805227 (-3.934508) | 0.167626 / 6.500664 (-6.333039) | 0.062024 / 0.075469 (-0.013445) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.192969 / 1.841788 (-0.648819) | 14.830812 / 8.074308 (6.756504) | 14.331178 / 10.191392 (4.139786) | 0.199222 / 0.680424 (-0.481202) | 0.029292 / 0.534201 (-0.504909) | 0.440427 / 0.579283 (-0.138857) | 0.437893 / 0.434364 (0.003529) | 0.547155 / 0.540337 (0.006818) | 0.645255 / 1.386936 (-0.741681) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007465 / 0.011353 (-0.003888) | 0.005386 / 0.011008 (-0.005622) | 0.073609 / 0.038508 (0.035100) | 0.033550 / 0.023109 (0.010440) | 0.341730 / 0.275898 (0.065832) | 0.371518 / 0.323480 (0.048038) | 0.005986 / 0.007986 (-0.001999) | 0.004264 / 0.004328 (-0.000065) | 0.073749 / 0.004250 (0.069498) | 0.051452 / 0.037052 (0.014399) | 0.347385 / 0.258489 (0.088896) | 0.392284 / 0.293841 (0.098444) | 0.036981 / 0.128546 (-0.091566) | 0.012431 / 0.075646 (-0.063216) | 0.086421 / 0.419271 (-0.332850) | 0.053014 / 0.043533 (0.009481) | 0.336660 / 0.255139 (0.081521) | 0.359155 / 0.283200 (0.075956) | 0.107666 / 0.141683 (-0.034017) | 1.424324 / 1.452155 (-0.027830) | 1.543027 / 1.492716 (0.050310) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.260862 / 0.018006 (0.242855) | 0.552057 / 0.000490 (0.551567) | 0.000449 / 0.000200 (0.000249) | 0.000059 / 0.000054 (0.000005) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029184 / 0.037411 (-0.008227) | 0.108799 / 0.014526 (0.094274) | 0.125136 / 0.176557 (-0.051421) | 0.157436 / 0.737135 (-0.579699) | 0.126333 / 0.296338 (-0.170005) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.424054 / 0.215209 (0.208845) | 4.227847 / 2.077655 (2.150192) | 2.051102 / 1.504120 (0.546983) | 1.848651 / 1.541195 (0.307457) | 1.922728 / 1.468490 (0.454238) | 0.705903 / 4.584777 (-3.878874) | 3.800977 / 3.745712 (0.055265) | 2.099345 / 5.269862 (-3.170517) | 1.342919 / 4.565676 (-3.222757) | 0.086128 / 0.424275 (-0.338147) | 0.012539 / 0.007607 (0.004932) | 0.528767 / 0.226044 (0.302723) | 5.299989 / 2.268929 (3.031061) | 2.534280 / 55.444624 (-52.910345) | 2.229532 / 6.876477 (-4.646945) | 2.326704 / 2.142072 (0.184632) | 0.838533 / 4.805227 (-3.966694) | 0.168446 / 6.500664 (-6.332218) | 0.065158 / 0.075469 (-0.010311) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.250091 / 1.841788 (-0.591697) | 14.988651 / 8.074308 (6.914343) | 13.655103 / 10.191392 (3.463711) | 0.165079 / 0.680424 (-0.515345) | 0.017829 / 0.534201 (-0.516372) | 0.425903 / 0.579283 (-0.153381) | 0.419771 / 0.434364 (-0.014593) | 0.534309 / 0.540337 (-0.006028) | 0.635563 / 1.386936 (-0.751373) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#f7d17ccc9b9dde2d94803b1305226c5a58d916c5 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.010569 / 0.011353 (-0.000784) | 0.005790 / 0.011008 (-0.005218) | 0.118626 / 0.038508 (0.080118) | 0.040455 / 0.023109 (0.017346) | 0.342309 / 0.275898 (0.066411) | 0.411828 / 0.323480 (0.088349) | 0.008824 / 0.007986 (0.000839) | 0.005426 / 0.004328 (0.001098) | 0.088740 / 0.004250 (0.084489) | 0.050042 / 0.037052 (0.012990) | 0.352350 / 0.258489 (0.093861) | 0.396030 / 0.293841 (0.102189) | 0.043385 / 0.128546 (-0.085162) | 0.013805 / 0.075646 (-0.061841) | 0.396489 / 0.419271 (-0.022783) | 0.055667 / 0.043533 (0.012135) | 0.336165 / 0.255139 (0.081026) | 0.372912 / 0.283200 (0.089713) | 0.115343 / 0.141683 (-0.026340) | 1.656412 / 1.452155 (0.204257) | 1.708993 / 1.492716 (0.216277) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.011650 / 0.018006 (-0.006357) | 0.444415 / 0.000490 (0.443926) | 0.003985 / 0.000200 (0.003785) | 0.000136 / 0.000054 (0.000082) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.031718 / 0.037411 (-0.005693) | 0.119640 / 0.014526 (0.105114) | 0.138519 / 0.176557 (-0.038037) | 0.188847 / 0.737135 (-0.548288) | 0.137891 / 0.296338 (-0.158448) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.447540 / 0.215209 (0.232331) | 4.577189 / 2.077655 (2.499534) | 2.106992 / 1.504120 (0.602872) | 1.889631 / 1.541195 (0.348436) | 1.972256 / 1.468490 (0.503766) | 0.778209 / 4.584777 (-3.806568) | 4.430279 / 3.745712 (0.684567) | 2.401226 / 5.269862 (-2.868636) | 1.481251 / 4.565676 (-3.084425) | 0.094244 / 0.424275 (-0.330031) | 0.013961 / 0.007607 (0.006354) | 0.570962 / 0.226044 (0.344917) | 5.809224 / 2.268929 (3.540295) | 2.663290 / 55.444624 (-52.781334) | 2.201228 / 6.876477 (-4.675249) | 2.319240 / 2.142072 (0.177168) | 0.938340 / 4.805227 (-3.866887) | 0.185546 / 6.500664 (-6.315118) | 0.069087 / 0.075469 (-0.006382) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.448597 / 1.841788 (-0.393191) | 17.188573 / 8.074308 (9.114265) | 16.197532 / 10.191392 (6.006140) | 0.194064 / 0.680424 (-0.486360) | 0.033694 / 0.534201 (-0.500507) | 0.507585 / 0.579283 (-0.071699) | 0.505470 / 0.434364 (0.071106) | 0.623270 / 0.540337 (0.082932) | 0.729964 / 1.386936 (-0.656972) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008529 / 0.011353 (-0.002824) | 0.005705 / 0.011008 (-0.005304) | 0.085594 / 0.038508 (0.047086) | 0.038377 / 0.023109 (0.015268) | 0.384221 / 0.275898 (0.108323) | 0.414678 / 0.323480 (0.091199) | 0.006195 / 0.007986 (-0.001791) | 0.004549 / 0.004328 (0.000221) | 0.082710 / 0.004250 (0.078460) | 0.054899 / 0.037052 (0.017847) | 0.404017 / 0.258489 (0.145528) | 0.450309 / 0.293841 (0.156468) | 0.040620 / 0.128546 (-0.087926) | 0.013774 / 0.075646 (-0.061872) | 0.099231 / 0.419271 (-0.320041) | 0.057183 / 0.043533 (0.013650) | 0.390806 / 0.255139 (0.135667) | 0.419334 / 0.283200 (0.136134) | 0.116449 / 0.141683 (-0.025234) | 1.709124 / 1.452155 (0.256969) | 1.812769 / 1.492716 (0.320052) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.225206 / 0.018006 (0.207199) | 0.440530 / 0.000490 (0.440040) | 0.002982 / 0.000200 (0.002782) | 0.000102 / 0.000054 (0.000048) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.032256 / 0.037411 (-0.005155) | 0.127086 / 0.014526 (0.112560) | 0.138133 / 0.176557 (-0.038424) | 0.176168 / 0.737135 (-0.560968) | 0.146072 / 0.296338 (-0.150267) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.474374 / 0.215209 (0.259165) | 4.785106 / 2.077655 (2.707452) | 2.319344 / 1.504120 (0.815225) | 2.075239 / 1.541195 (0.534045) | 2.179231 / 1.468490 (0.710741) | 0.832124 / 4.584777 (-3.752653) | 4.376302 / 3.745712 (0.630590) | 3.966837 / 5.269862 (-1.303024) | 1.820230 / 4.565676 (-2.745446) | 0.100692 / 0.424275 (-0.323583) | 0.014748 / 0.007607 (0.007141) | 0.568702 / 0.226044 (0.342657) | 5.771548 / 2.268929 (3.502619) | 2.747431 / 55.444624 (-52.697193) | 2.448482 / 6.876477 (-4.427994) | 2.497206 / 2.142072 (0.355133) | 0.960842 / 4.805227 (-3.844385) | 0.192855 / 6.500664 (-6.307809) | 0.072494 / 0.075469 (-0.002975) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.474542 / 1.841788 (-0.367245) | 17.344804 / 8.074308 (9.270496) | 15.336082 / 10.191392 (5.144690) | 0.200134 / 0.680424 (-0.480290) | 0.020728 / 0.534201 (-0.513473) | 0.488854 / 0.579283 (-0.090429) | 0.490781 / 0.434364 (0.056418) | 0.626288 / 0.540337 (0.085950) | 0.721130 / 1.386936 (-0.665806) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#cd7877892aa48a2470b01f52013390c54aca8a49 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008542 / 0.011353 (-0.002811) | 0.004624 / 0.011008 (-0.006384) | 0.100749 / 0.038508 (0.062241) | 0.029587 / 0.023109 (0.006478) | 0.298680 / 0.275898 (0.022782) | 0.359659 / 0.323480 (0.036180) | 0.007001 / 0.007986 (-0.000984) | 0.003398 / 0.004328 (-0.000930) | 0.078654 / 0.004250 (0.074404) | 0.036440 / 0.037052 (-0.000612) | 0.313245 / 0.258489 (0.054756) | 0.342776 / 0.293841 (0.048936) | 0.033195 / 0.128546 (-0.095352) | 0.011500 / 0.075646 (-0.064146) | 0.323957 / 0.419271 (-0.095314) | 0.039878 / 0.043533 (-0.003655) | 0.298189 / 0.255139 (0.043050) | 0.325488 / 0.283200 (0.042289) | 0.087276 / 0.141683 (-0.054407) | 1.480846 / 1.452155 (0.028691) | 1.507016 / 1.492716 (0.014300) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.189570 / 0.018006 (0.171564) | 0.406407 / 0.000490 (0.405917) | 0.003062 / 0.000200 (0.002862) | 0.000073 / 0.000054 (0.000019) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.022865 / 0.037411 (-0.014546) | 0.096103 / 0.014526 (0.081578) | 0.106462 / 0.176557 (-0.070094) | 0.140888 / 0.737135 (-0.596247) | 0.108172 / 0.296338 (-0.188167) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.415951 / 0.215209 (0.200742) | 4.172187 / 2.077655 (2.094532) | 1.842210 / 1.504120 (0.338090) | 1.636997 / 1.541195 (0.095802) | 1.706078 / 1.468490 (0.237588) | 0.695825 / 4.584777 (-3.888952) | 3.337354 / 3.745712 (-0.408358) | 1.877880 / 5.269862 (-3.391982) | 1.153882 / 4.565676 (-3.411794) | 0.082923 / 0.424275 (-0.341352) | 0.012814 / 0.007607 (0.005207) | 0.521793 / 0.226044 (0.295748) | 5.275980 / 2.268929 (3.007051) | 2.279230 / 55.444624 (-53.165394) | 1.941777 / 6.876477 (-4.934700) | 1.981297 / 2.142072 (-0.160775) | 0.809669 / 4.805227 (-3.995558) | 0.148753 / 6.500664 (-6.351911) | 0.064909 / 0.075469 (-0.010560) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.226757 / 1.841788 (-0.615031) | 13.717354 / 8.074308 (5.643046) | 12.925885 / 10.191392 (2.734493) | 0.137926 / 0.680424 (-0.542498) | 0.028788 / 0.534201 (-0.505413) | 0.396654 / 0.579283 (-0.182630) | 0.401931 / 0.434364 (-0.032432) | 0.460515 / 0.540337 (-0.079823) | 0.537903 / 1.386936 (-0.849033) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006757 / 0.011353 (-0.004596) | 0.004474 / 0.011008 (-0.006534) | 0.076571 / 0.038508 (0.038063) | 0.027580 / 0.023109 (0.004471) | 0.348231 / 0.275898 (0.072333) | 0.398403 / 0.323480 (0.074923) | 0.005089 / 0.007986 (-0.002897) | 0.004676 / 0.004328 (0.000347) | 0.076444 / 0.004250 (0.072194) | 0.038508 / 0.037052 (0.001456) | 0.348515 / 0.258489 (0.090026) | 0.401456 / 0.293841 (0.107615) | 0.031630 / 0.128546 (-0.096916) | 0.011698 / 0.075646 (-0.063949) | 0.085805 / 0.419271 (-0.333467) | 0.041962 / 0.043533 (-0.001570) | 0.343415 / 0.255139 (0.088276) | 0.383001 / 0.283200 (0.099801) | 0.090231 / 0.141683 (-0.051452) | 1.488114 / 1.452155 (0.035960) | 1.569039 / 1.492716 (0.076323) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.261751 / 0.018006 (0.243745) | 0.411354 / 0.000490 (0.410865) | 0.015103 / 0.000200 (0.014903) | 0.000262 / 0.000054 (0.000208) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025423 / 0.037411 (-0.011988) | 0.101334 / 0.014526 (0.086808) | 0.108835 / 0.176557 (-0.067722) | 0.143995 / 0.737135 (-0.593140) | 0.111751 / 0.296338 (-0.184588) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.446507 / 0.215209 (0.231298) | 4.461543 / 2.077655 (2.383888) | 2.104648 / 1.504120 (0.600528) | 1.895900 / 1.541195 (0.354706) | 1.985481 / 1.468490 (0.516991) | 0.699029 / 4.584777 (-3.885748) | 3.371064 / 3.745712 (-0.374648) | 1.883445 / 5.269862 (-3.386416) | 1.166150 / 4.565676 (-3.399527) | 0.082639 / 0.424275 (-0.341636) | 0.012605 / 0.007607 (0.004998) | 0.544860 / 0.226044 (0.318815) | 5.513223 / 2.268929 (3.244294) | 2.570661 / 55.444624 (-52.873963) | 2.206066 / 6.876477 (-4.670411) | 2.256346 / 2.142072 (0.114273) | 0.801142 / 4.805227 (-4.004085) | 0.150412 / 6.500664 (-6.350252) | 0.067742 / 0.075469 (-0.007727) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.303477 / 1.841788 (-0.538310) | 14.287767 / 8.074308 (6.213458) | 13.525563 / 10.191392 (3.334171) | 0.148202 / 0.680424 (-0.532222) | 0.016868 / 0.534201 (-0.517333) | 0.380729 / 0.579283 (-0.198555) | 0.388177 / 0.434364 (-0.046187) | 0.477410 / 0.540337 (-0.062927) | 0.569343 / 1.386936 (-0.817593) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#79c18b77113da3f2e31af0570ec119877ca2a390 \"CML watermark\")\n", "> PS: I think it would be a good idea to add links to the Guide pages for better discoverability and to somewhat \"justify their presence in the docs\" (from the tutorial/how-to pages to the guides; some guides are not referenced at all)\r\n\r\nJust merged #5485, which references this new doc! Will look for other pages in the docs where it'd make sense to add them :)" ]

[ { "color": "ffffff", "default": true, "description": "This will not be worked on", "id": 1935892913, "name": "wontfix", "node_id": "MDU6TGFiZWwxOTM1ODkyOTEz", "url": "https://api.github.com/repos/huggingface/datasets/labels/wontfix" } ]

[ "We won't support in-place modifications since `datasets` is based on the Apache Arrow format which doesn't support in-place modifications.\r\n\r\nIn your case the old dataset is garbage collected pretty quickly so you won't have memory issues.\r\n\r\nNote that datasets loaded from disk (memory mapped) are not loaded in memory, and therefore the new dataset actually use the same buffers as the old one.", "Thank you for your detailed reply.\r\n\r\n> In your case the old dataset is garbage collected pretty quickly so you won't have memory issues.\r\n\r\nI understand this, but it still copies the old dataset to create the new one, is this correct? So maybe it is not memory-consuming, but time-consuming?", "Indeed, and because of that it is more efficient to add multiple rows at once instead of one by one, using `concatenate_datasets` for example." ]

[]

[]

[]

[]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._" ]

[]