code
stringlengths 87
55.2k
| code_codestyle
int64 0
349
| style_context
stringlengths 135
49.1k
| style_context_codestyle
int64 0
349
| label
int64 0
1
|
|---|---|---|---|---|
import math
import os
import sys
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = """"""
try:
with open(_a , """rb""" ) as binary_file:
UpperCAmelCase_ : Optional[Any] = binary_file.read()
for dat in data:
UpperCAmelCase_ : Union[str, Any] = F'''{dat:08b}'''
result += curr_byte
return result
except OSError:
print("""File not accessible""" )
sys.exit()
def lowerCamelCase_ ( _a : dict[str, str] , _a : str , _a : int , _a : str ):
'''simple docstring'''
lexicon.pop(_a )
UpperCAmelCase_ : Union[str, Any] = last_match_id
if math.loga(_a ).is_integer():
for curr_key in lexicon:
UpperCAmelCase_ : int = """0""" + lexicon[curr_key]
UpperCAmelCase_ : Tuple = bin(_a )[2:]
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
UpperCAmelCase_ : Any = {"""0""": """0""", """1""": """1"""}
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = """""", """"""
UpperCAmelCase_ : str = len(_a )
for i in range(len(_a ) ):
curr_string += data_bits[i]
if curr_string not in lexicon:
continue
UpperCAmelCase_ : str = lexicon[curr_string]
result += last_match_id
add_key_to_lexicon(_a , _a , _a , _a )
index += 1
UpperCAmelCase_ : int = """"""
while curr_string != "" and curr_string not in lexicon:
curr_string += "0"
if curr_string != "":
UpperCAmelCase_ : Optional[Any] = lexicon[curr_string]
result += last_match_id
return result
def lowerCamelCase_ ( _a : str , _a : str ):
'''simple docstring'''
UpperCAmelCase_ : str = os.path.getsize(_a )
UpperCAmelCase_ : Union[str, Any] = bin(_a )[2:]
UpperCAmelCase_ : Optional[Any] = len(_a )
return "0" * (length_length - 1) + file_length_binary + compressed
def lowerCamelCase_ ( _a : str , _a : str ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = 8
try:
with open(_a , """wb""" ) as opened_file:
UpperCAmelCase_ : Tuple = [
to_write[i : i + byte_length]
for i in range(0 , len(_a ) , _a )
]
if len(result_byte_array[-1] ) % byte_length == 0:
result_byte_array.append("""10000000""" )
else:
result_byte_array[-1] += "1" + "0" * (
byte_length - len(result_byte_array[-1] ) - 1
)
for elem in result_byte_array:
opened_file.write(int(_a , 2 ).to_bytes(1 , byteorder="""big""" ) )
except OSError:
print("""File not accessible""" )
sys.exit()
def lowerCamelCase_ ( _a : str , _a : str ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = read_file_binary(_a )
UpperCAmelCase_ : str = compress_data(_a )
UpperCAmelCase_ : List[Any] = add_file_length(_a , _a )
write_file_binary(_a , _a )
if __name__ == "__main__":
compress(sys.argv[1], sys.argv[2])
| 345 |
import pickle
import numpy as np
from matplotlib import pyplot as plt
class _snake_case :
'''simple docstring'''
def __init__( self: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple=0.2 ,lowerCamelCase_: Union[str, Any]=0.2 ) -> List[str]:
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : str = bp_numa
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : Optional[int] = conva_get[:2]
UpperCAmelCase_ : List[Any] = conva_get[2]
UpperCAmelCase_ : str = size_pa
UpperCAmelCase_ : Optional[int] = rate_w
UpperCAmelCase_ : Dict = rate_t
UpperCAmelCase_ : List[Any] = [
np.mat(-1 * np.random.rand(self.conva[0] ,self.conva[0] ) + 0.5 )
for i in range(self.conva[1] )
]
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : Dict = -2 * np.random.rand(self.conva[1] ) + 1
UpperCAmelCase_ : str = -2 * np.random.rand(self.num_bpa ) + 1
UpperCAmelCase_ : Union[str, Any] = -2 * np.random.rand(self.num_bpa ) + 1
def A__ ( self: str ,lowerCamelCase_: Optional[Any] ) -> Tuple:
# save model dict with pickle
UpperCAmelCase_ : Dict = {
"""num_bp1""": self.num_bpa,
"""num_bp2""": self.num_bpa,
"""num_bp3""": self.num_bpa,
"""conv1""": self.conva,
"""step_conv1""": self.step_conva,
"""size_pooling1""": self.size_poolinga,
"""rate_weight""": self.rate_weight,
"""rate_thre""": self.rate_thre,
"""w_conv1""": self.w_conva,
"""wkj""": self.wkj,
"""vji""": self.vji,
"""thre_conv1""": self.thre_conva,
"""thre_bp2""": self.thre_bpa,
"""thre_bp3""": self.thre_bpa,
}
with open(lowerCamelCase_ ,"""wb""" ) as f:
pickle.dump(lowerCamelCase_ ,lowerCamelCase_ )
print(F'''Model saved： {save_path}''' )
@classmethod
def A__ ( cls: List[str] ,lowerCamelCase_: str ) -> List[str]:
# read saved model
with open(lowerCamelCase_ ,"""rb""" ) as f:
UpperCAmelCase_ : Any = pickle.load(lowerCamelCase_ ) # noqa: S301
UpperCAmelCase_ : Union[str, Any] = model_dic.get("""conv1""" )
conv_get.append(model_dic.get("""step_conv1""" ) )
UpperCAmelCase_ : List[str] = model_dic.get("""size_pooling1""" )
UpperCAmelCase_ : Tuple = model_dic.get("""num_bp1""" )
UpperCAmelCase_ : Optional[Any] = model_dic.get("""num_bp2""" )
UpperCAmelCase_ : List[str] = model_dic.get("""num_bp3""" )
UpperCAmelCase_ : List[Any] = model_dic.get("""rate_weight""" )
UpperCAmelCase_ : Dict = model_dic.get("""rate_thre""" )
# create model instance
UpperCAmelCase_ : List[Any] = CNN(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# modify model parameter
UpperCAmelCase_ : Any = model_dic.get("""w_conv1""" )
UpperCAmelCase_ : int = model_dic.get("""wkj""" )
UpperCAmelCase_ : int = model_dic.get("""vji""" )
UpperCAmelCase_ : Optional[int] = model_dic.get("""thre_conv1""" )
UpperCAmelCase_ : List[str] = model_dic.get("""thre_bp2""" )
UpperCAmelCase_ : Dict = model_dic.get("""thre_bp3""" )
return conv_ins
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> Tuple:
return 1 / (1 + np.exp(-1 * x ))
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
return round(lowerCamelCase_ ,3 )
def A__ ( self: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Any:
# convolution process
UpperCAmelCase_ : Optional[Any] = convs[0]
UpperCAmelCase_ : int = convs[1]
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )[0]
# get the data slice of original image data, data_focus
UpperCAmelCase_ : Dict = []
for i_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
for j_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = data[
i_focus : i_focus + size_conv, j_focus : j_focus + size_conv
]
data_focus.append(lowerCamelCase_ )
# calculate the feature map of every single kernel, and saved as list of matrix
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = int((size_data - size_conv) / conv_step + 1 )
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[int] = []
for i_focus in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = (
np.sum(np.multiply(data_focus[i_focus] ,w_convs[i_map] ) )
- thre_convs[i_map]
)
featuremap.append(self.sig(lowerCamelCase_ ) )
UpperCAmelCase_ : Union[str, Any] = np.asmatrix(lowerCamelCase_ ).reshape(
lowerCamelCase_ ,lowerCamelCase_ )
data_featuremap.append(lowerCamelCase_ )
# expanding the data slice to One dimenssion
UpperCAmelCase_ : Optional[Any] = []
for each_focus in data_focus:
focusa_list.extend(self.Expand_Mat(lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[int] = np.asarray(lowerCamelCase_ )
return focus_list, data_featuremap
def A__ ( self: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any]="average_pool" ) -> List[Any]:
# pooling process
UpperCAmelCase_ : Optional[Any] = len(featuremaps[0] )
UpperCAmelCase_ : Any = int(size_map / size_pooling )
UpperCAmelCase_ : Optional[int] = []
for i_map in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Any = featuremaps[i_map]
UpperCAmelCase_ : Tuple = []
for i_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : str = feature_map[
i_focus : i_focus + size_pooling,
j_focus : j_focus + size_pooling,
]
if pooling_type == "average_pool":
# average pooling
map_pooled.append(np.average(lowerCamelCase_ ) )
elif pooling_type == "max_pooling":
# max pooling
map_pooled.append(np.max(lowerCamelCase_ ) )
UpperCAmelCase_ : int = np.asmatrix(lowerCamelCase_ ).reshape(lowerCamelCase_ ,lowerCamelCase_ )
featuremap_pooled.append(lowerCamelCase_ )
return featuremap_pooled
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> Optional[int]:
# expanding three dimension data to one dimension list
UpperCAmelCase_ : List[Any] = []
for i in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Tuple = np.shape(data[i] )
UpperCAmelCase_ : Optional[int] = data[i].reshape(1 ,shapes[0] * shapes[1] )
UpperCAmelCase_ : Optional[int] = data_listed.getA().tolist()[0]
data_expanded.extend(lowerCamelCase_ )
UpperCAmelCase_ : int = np.asarray(lowerCamelCase_ )
return data_expanded
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
# expanding matrix to one dimension list
UpperCAmelCase_ : List[Any] = np.asarray(lowerCamelCase_ )
UpperCAmelCase_ : str = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = data_mat.reshape(1 ,shapes[0] * shapes[1] )
return data_expanded
def A__ ( self: str ,lowerCamelCase_: Dict ,lowerCamelCase_: int ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = 0
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = np.ones((size_map, size_map) )
for i in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Any = pd_pool[
i_pool
]
UpperCAmelCase_ : List[str] = i_pool + 1
UpperCAmelCase_ : Optional[Any] = np.multiply(
lowerCamelCase_ ,np.multiply(out_map[i_map] ,(1 - out_map[i_map]) ) )
pd_all.append(lowerCamelCase_ )
return pd_all
def A__ ( self: str ,lowerCamelCase_: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any=bool ) -> Optional[int]:
# model traning
print("""----------------------Start Training-------------------------""" )
print((""" - - Shape: Train_Data """, np.shape(lowerCamelCase_ )) )
print((""" - - Shape: Teach_Data """, np.shape(lowerCamelCase_ )) )
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : Any = 10000
while rp < n_repeat and mse >= error_accuracy:
UpperCAmelCase_ : List[str] = 0
print(F'''-------------Learning Time {rp}--------------''' )
for p in range(len(lowerCamelCase_ ) ):
# print('------------Learning Image: %d--------------'%p)
UpperCAmelCase_ : str = np.asmatrix(datas_train[p] )
UpperCAmelCase_ : Optional[Any] = np.asarray(datas_teach[p] )
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : List[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = data_bp_input
UpperCAmelCase_ : Optional[Any] = np.dot(lowerCamelCase_ ,self.vji.T ) - self.thre_bpa
UpperCAmelCase_ : int = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = np.dot(lowerCamelCase_ ,self.wkj.T ) - self.thre_bpa
UpperCAmelCase_ : Optional[Any] = self.sig(lowerCamelCase_ )
# --------------Model Leaning ------------------------
# calculate error and gradient---------------
UpperCAmelCase_ : List[str] = np.multiply(
(data_teach - bp_outa) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : List[Any] = np.multiply(
np.dot(lowerCamelCase_ ,self.wkj ) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : Any = np.dot(lowerCamelCase_ ,self.vji )
UpperCAmelCase_ : Tuple = pd_i_all / (self.size_poolinga * self.size_poolinga)
UpperCAmelCase_ : List[str] = pd_conva_pooled.T.getA().tolist()
UpperCAmelCase_ : str = self._calculate_gradient_from_pool(
lowerCamelCase_ ,lowerCamelCase_ ,shape_featuremapa[0] ,shape_featuremapa[1] ,self.size_poolinga ,)
# weight and threshold learning process---------
# convolution layer
for k_conv in range(self.conva[1] ):
UpperCAmelCase_ : List[str] = self._expand_mat(pd_conva_all[k_conv] )
UpperCAmelCase_ : Optional[Any] = self.rate_weight * np.dot(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : int = self.w_conva[k_conv] + delta_w.reshape(
(self.conva[0], self.conva[0]) )
UpperCAmelCase_ : str = (
self.thre_conva[k_conv]
- np.sum(pd_conva_all[k_conv] ) * self.rate_thre
)
# all connected layer
UpperCAmelCase_ : int = self.wkj + pd_k_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : Tuple = self.vji + pd_j_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : int = self.thre_bpa - pd_k_all * self.rate_thre
UpperCAmelCase_ : str = self.thre_bpa - pd_j_all * self.rate_thre
# calculate the sum error of all single image
UpperCAmelCase_ : int = np.sum(abs(data_teach - bp_outa ) )
error_count += errors
# print(' ----Teach ',data_teach)
# print(' ----BP_output ',bp_out3)
UpperCAmelCase_ : int = rp + 1
UpperCAmelCase_ : Any = error_count / patterns
all_mse.append(lowerCamelCase_ )
def draw_error():
UpperCAmelCase_ : Any = [error_accuracy for i in range(int(n_repeat * 1.2 ) )]
plt.plot(lowerCamelCase_ ,"""+-""" )
plt.plot(lowerCamelCase_ ,"""r--""" )
plt.xlabel("""Learning Times""" )
plt.ylabel("""All_mse""" )
plt.grid(lowerCamelCase_ ,alpha=0.5 )
plt.show()
print("""------------------Training Complished---------------------""" )
print((""" - - Training epoch: """, rp, F''' - - Mse: {mse:.6f}''') )
if draw_e:
draw_error()
return mse
def A__ ( self: Optional[int] ,lowerCamelCase_: Any ) -> Tuple:
# model predict
UpperCAmelCase_ : Union[str, Any] = []
print("""-------------------Start Testing-------------------------""" )
print((""" - - Shape: Test_Data """, np.shape(lowerCamelCase_ )) )
for p in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = np.asmatrix(datas_test[p] )
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Optional[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : str = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : str = data_bp_input
UpperCAmelCase_ : Union[str, Any] = bp_outa * self.vji.T - self.thre_bpa
UpperCAmelCase_ : Optional[int] = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = bp_outa * self.wkj.T - self.thre_bpa
UpperCAmelCase_ : List[Any] = self.sig(lowerCamelCase_ )
produce_out.extend(bp_outa.getA().tolist() )
UpperCAmelCase_ : int = [list(map(self.do_round ,lowerCamelCase_ ) ) for each in produce_out]
return np.asarray(lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Tuple:
# return the data of image after convoluting process so we can check it out
UpperCAmelCase_ : Optional[int] = np.asmatrix(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Dict = self.pooling(lowerCamelCase_ ,self.size_poolinga )
return data_conveda, data_pooleda
if __name__ == "__main__":
pass
| 345 | 1 |
import argparse
import math
import os
from copy import deepcopy
import torch
from audio_diffusion.models import DiffusionAttnUnetaD
from diffusion import sampling
from torch import nn
from diffusers import DanceDiffusionPipeline, IPNDMScheduler, UNetaDModel
UpperCamelCase_ = {
'''gwf-440k''': {
'''url''': '''https://model-server.zqevans2.workers.dev/gwf-440k.ckpt''',
'''sample_rate''': 48000,
'''sample_size''': 65536,
},
'''jmann-small-190k''': {
'''url''': '''https://model-server.zqevans2.workers.dev/jmann-small-190k.ckpt''',
'''sample_rate''': 48000,
'''sample_size''': 65536,
},
'''jmann-large-580k''': {
'''url''': '''https://model-server.zqevans2.workers.dev/jmann-large-580k.ckpt''',
'''sample_rate''': 48000,
'''sample_size''': 131072,
},
'''maestro-uncond-150k''': {
'''url''': '''https://model-server.zqevans2.workers.dev/maestro-uncond-150k.ckpt''',
'''sample_rate''': 16000,
'''sample_size''': 65536,
},
'''unlocked-uncond-250k''': {
'''url''': '''https://model-server.zqevans2.workers.dev/unlocked-uncond-250k.ckpt''',
'''sample_rate''': 16000,
'''sample_size''': 65536,
},
'''honk-140k''': {
'''url''': '''https://model-server.zqevans2.workers.dev/honk-140k.ckpt''',
'''sample_rate''': 16000,
'''sample_size''': 65536,
},
}
def lowerCamelCase_ ( _a : Optional[Any] , _a : Optional[Any] ):
'''simple docstring'''
return torch.atana(_a , _a ) / math.pi * 2
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = torch.sin(t * math.pi / 2 ) ** 2
UpperCAmelCase_ : Union[str, Any] = (1 - sigma**2) ** 0.5
return alpha_sigma_to_t(_a , _a )
class _snake_case ( __snake_case ):
'''simple docstring'''
pass
class _snake_case ( nn.Module ):
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> int:
super().__init__()
UpperCAmelCase_ : Optional[Any] = DiffusionAttnUnetaD(lowerCamelCase_ ,n_attn_layers=4 )
UpperCAmelCase_ : Optional[Any] = deepcopy(self.diffusion )
UpperCAmelCase_ : List[str] = torch.quasirandom.SobolEngine(1 ,scramble=lowerCamelCase_ )
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
UpperCAmelCase_ : List[Any] = MODELS_MAP[model_name]["""url"""]
os.system(F'''wget {url} ./''' )
return F'''./{model_name}.ckpt'''
UpperCamelCase_ = {
'''1''': '''resnets.0''',
'''2''': '''attentions.0''',
'''3''': '''resnets.1''',
'''4''': '''attentions.1''',
'''5''': '''resnets.2''',
'''6''': '''attentions.2''',
}
UpperCamelCase_ = {
'''8''': '''resnets.0''',
'''9''': '''attentions.0''',
'''10''': '''resnets.1''',
'''11''': '''attentions.1''',
'''12''': '''resnets.2''',
'''13''': '''attentions.2''',
}
UpperCamelCase_ = {
'''1''': '''resnets.0''',
'''2''': '''attentions.0''',
'''3''': '''resnets.1''',
'''4''': '''attentions.1''',
'''5''': '''resnets.2''',
'''6''': '''attentions.2''',
'''8''': '''resnets.3''',
'''9''': '''attentions.3''',
'''10''': '''resnets.4''',
'''11''': '''attentions.4''',
'''12''': '''resnets.5''',
'''13''': '''attentions.5''',
}
UpperCamelCase_ = {
'''0''': '''resnets.0''',
'''1''': '''resnets.1''',
'''2''': '''resnets.2''',
'''4''': '''resnets.0''',
'''5''': '''resnets.1''',
'''6''': '''resnets.2''',
}
UpperCamelCase_ = {
'''skip''': '''conv_skip''',
'''main.0''': '''conv_1''',
'''main.1''': '''group_norm_1''',
'''main.3''': '''conv_2''',
'''main.4''': '''group_norm_2''',
}
UpperCamelCase_ = {
'''norm''': '''group_norm''',
'''qkv_proj''': ['''query''', '''key''', '''value'''],
'''out_proj''': ['''proj_attn'''],
}
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
if name.startswith("""skip""" ):
return name.replace("""skip""" , RES_CONV_MAP["""skip"""] )
# name has to be of format main.{digit}
if not name.startswith("""main.""" ):
raise ValueError(F'''ResConvBlock error with {name}''' )
return name.replace(name[:6] , RES_CONV_MAP[name[:6]] )
def lowerCamelCase_ ( _a : Optional[int] ):
'''simple docstring'''
for key, value in ATTN_MAP.items():
if name.startswith(_a ) and not isinstance(_a , _a ):
return name.replace(_a , _a )
elif name.startswith(_a ):
return [name.replace(_a , _a ) for v in value]
raise ValueError(F'''Attn error with {name}''' )
def lowerCamelCase_ ( _a : Dict , _a : Any=13 ):
'''simple docstring'''
UpperCAmelCase_ : int = input_string
if string.split(""".""" )[0] == "timestep_embed":
return string.replace("""timestep_embed""" , """time_proj""" )
UpperCAmelCase_ : Optional[int] = 0
if string.startswith("""net.3.""" ):
depth += 1
UpperCAmelCase_ : Union[str, Any] = string[6:]
elif string.startswith("""net.""" ):
UpperCAmelCase_ : Optional[int] = string[4:]
while string.startswith("""main.7.""" ):
depth += 1
UpperCAmelCase_ : Any = string[7:]
if string.startswith("""main.""" ):
UpperCAmelCase_ : int = string[5:]
# mid block
if string[:2].isdigit():
UpperCAmelCase_ : Any = string[:2]
UpperCAmelCase_ : List[Any] = string[2:]
else:
UpperCAmelCase_ : int = string[0]
UpperCAmelCase_ : int = string[1:]
if depth == max_depth:
UpperCAmelCase_ : List[Any] = MID_NUM_TO_LAYER[layer_num]
UpperCAmelCase_ : List[str] = """mid_block"""
elif depth > 0 and int(_a ) < 7:
UpperCAmelCase_ : Tuple = DOWN_NUM_TO_LAYER[layer_num]
UpperCAmelCase_ : Optional[Any] = F'''down_blocks.{depth}'''
elif depth > 0 and int(_a ) > 7:
UpperCAmelCase_ : Union[str, Any] = UP_NUM_TO_LAYER[layer_num]
UpperCAmelCase_ : Optional[int] = F'''up_blocks.{max_depth - depth - 1}'''
elif depth == 0:
UpperCAmelCase_ : List[Any] = DEPTH_0_TO_LAYER[layer_num]
UpperCAmelCase_ : int = F'''up_blocks.{max_depth - 1}''' if int(_a ) > 3 else """down_blocks.0"""
if not string_left.startswith(""".""" ):
raise ValueError(F'''Naming error with {input_string} and string_left: {string_left}.''' )
UpperCAmelCase_ : Any = string_left[1:]
if "resnets" in new_layer:
UpperCAmelCase_ : Dict = convert_resconv_naming(_a )
elif "attentions" in new_layer:
UpperCAmelCase_ : Optional[int] = convert_attn_naming(_a )
UpperCAmelCase_ : str = new_string_left
if not isinstance(_a , _a ):
UpperCAmelCase_ : Optional[int] = prefix + """.""" + new_layer + """.""" + string_left
else:
UpperCAmelCase_ : str = [prefix + """.""" + new_layer + """.""" + s for s in string_left]
return new_string
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = {}
for k, v in state_dict.items():
if k.endswith("""kernel""" ):
# up- and downsample layers, don't have trainable weights
continue
UpperCAmelCase_ : List[str] = rename(_a )
# check if we need to transform from Conv => Linear for attention
if isinstance(_a , _a ):
UpperCAmelCase_ : Union[str, Any] = transform_conv_attns(_a , _a , _a )
else:
UpperCAmelCase_ : str = v
return new_state_dict
def lowerCamelCase_ ( _a : int , _a : Dict , _a : Optional[int] ):
'''simple docstring'''
if len(_a ) == 1:
if len(v.shape ) == 3:
# weight
UpperCAmelCase_ : Tuple = v[:, :, 0]
else:
# bias
UpperCAmelCase_ : Optional[Any] = v
else:
# qkv matrices
UpperCAmelCase_ : int = v.shape[0]
UpperCAmelCase_ : Union[str, Any] = trippled_shape // 3
for i in range(3 ):
if len(v.shape ) == 3:
UpperCAmelCase_ : List[str] = v[i * single_shape : (i + 1) * single_shape, :, 0]
else:
UpperCAmelCase_ : List[Any] = v[i * single_shape : (i + 1) * single_shape]
return new_state_dict
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = torch.device("""cuda""" if torch.cuda.is_available() else """cpu""" )
UpperCAmelCase_ : Tuple = args.model_path.split("""/""" )[-1].split(""".""" )[0]
if not os.path.isfile(args.model_path ):
assert (
model_name == args.model_path
), F'''Make sure to provide one of the official model names {MODELS_MAP.keys()}'''
UpperCAmelCase_ : Dict = download(_a )
UpperCAmelCase_ : List[str] = MODELS_MAP[model_name]["""sample_rate"""]
UpperCAmelCase_ : List[str] = MODELS_MAP[model_name]["""sample_size"""]
UpperCAmelCase_ : List[Any] = Object()
UpperCAmelCase_ : Optional[int] = sample_size
UpperCAmelCase_ : str = sample_rate
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : Tuple = UNetaDModel(sample_size=_a , sample_rate=_a )
UpperCAmelCase_ : List[Any] = diffusers_model.state_dict()
UpperCAmelCase_ : Union[str, Any] = DiffusionUncond(_a )
orig_model.load_state_dict(torch.load(args.model_path , map_location=_a )["""state_dict"""] )
UpperCAmelCase_ : List[str] = orig_model.diffusion_ema.eval()
UpperCAmelCase_ : List[str] = orig_model.state_dict()
UpperCAmelCase_ : List[str] = rename_orig_weights(_a )
UpperCAmelCase_ : str = set(renamed_state_dict.keys() ) - set(diffusers_state_dict.keys() )
UpperCAmelCase_ : Dict = set(diffusers_state_dict.keys() ) - set(renamed_state_dict.keys() )
assert len(_a ) == 0, F'''Problem with {renamed_minus_diffusers}'''
assert all(k.endswith("""kernel""" ) for k in list(_a ) ), F'''Problem with {diffusers_minus_renamed}'''
for key, value in renamed_state_dict.items():
assert (
diffusers_state_dict[key].squeeze().shape == value.squeeze().shape
), F'''Shape for {key} doesn\'t match. Diffusers: {diffusers_state_dict[key].shape} vs. {value.shape}'''
if key == "time_proj.weight":
UpperCAmelCase_ : int = value.squeeze()
UpperCAmelCase_ : str = value
diffusers_model.load_state_dict(_a )
UpperCAmelCase_ : List[str] = 100
UpperCAmelCase_ : Tuple = 33
UpperCAmelCase_ : str = IPNDMScheduler(num_train_timesteps=_a )
UpperCAmelCase_ : Optional[Any] = torch.manual_seed(_a )
UpperCAmelCase_ : List[str] = torch.randn([1, 2, config.sample_size] , generator=_a ).to(_a )
UpperCAmelCase_ : List[Any] = torch.linspace(1 , 0 , steps + 1 , device=_a )[:-1]
UpperCAmelCase_ : str = get_crash_schedule(_a )
UpperCAmelCase_ : int = DanceDiffusionPipeline(unet=_a , scheduler=_a )
UpperCAmelCase_ : str = torch.manual_seed(33 )
UpperCAmelCase_ : str = pipe(num_inference_steps=_a , generator=_a ).audios
UpperCAmelCase_ : Optional[int] = sampling.iplms_sample(_a , _a , _a , {} )
UpperCAmelCase_ : int = generated.clamp(-1 , 1 )
UpperCAmelCase_ : Union[str, Any] = (generated - audio).abs().sum()
UpperCAmelCase_ : int = (generated - audio).abs().max()
if args.save:
pipe.save_pretrained(args.checkpoint_path )
print("""Diff sum""" , _a )
print("""Diff max""" , _a )
assert diff_max < 1E-3, F'''Diff max: {diff_max} is too much :-/'''
print(F'''Conversion for {model_name} successful!''' )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
parser.add_argument('''--model_path''', default=None, type=str, required=True, help='''Path to the model to convert.''')
parser.add_argument(
'''--save''', default=True, type=bool, required=False, help='''Whether to save the converted model or not.'''
)
parser.add_argument('''--checkpoint_path''', default=None, type=str, required=True, help='''Path to the output model.''')
UpperCamelCase_ = parser.parse_args()
main(args)
| 345 |
import json
import os
import unittest
from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = CTRLTokenizer
A__ : Optional[Any] = False
A__ : str = False
def A__ ( self: Optional[int] ) -> List[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
UpperCAmelCase_ : Dict = ["""adapt""", """re@@""", """a@@""", """apt""", """c@@""", """t""", """<unk>"""]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : List[Any] = ["""#version: 0.2""", """a p""", """ap t</w>""", """r e""", """a d""", """ad apt</w>""", """"""]
UpperCAmelCase_ : Optional[Any] = {"""unk_token""": """<unk>"""}
UpperCAmelCase_ : Union[str, Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : Optional[Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase_ ) + """\n""" )
with open(self.merges_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase_ ) )
def A__ ( self: Optional[int] ,**lowerCamelCase_: Any ) -> str:
kwargs.update(self.special_tokens_map )
return CTRLTokenizer.from_pretrained(self.tmpdirname ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: int ) -> str:
UpperCAmelCase_ : List[str] = """adapt react readapt apt"""
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
return input_text, output_text
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = CTRLTokenizer(self.vocab_file ,self.merges_file ,**self.special_tokens_map )
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
UpperCAmelCase_ : Optional[int] = """adapt re@@ a@@ c@@ t re@@ adapt apt""".split()
UpperCAmelCase_ : Tuple = tokenizer.tokenize(lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = tokens + [tokenizer.unk_token]
UpperCAmelCase_ : List[str] = [0, 1, 2, 4, 5, 1, 0, 3, 6]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase_ ) ,lowerCamelCase_ )
| 345 | 1 |
def lowerCamelCase_ ( _a : list , _a : int , _a : int = 0 , _a : int = 0 ):
'''simple docstring'''
UpperCAmelCase_ : str = right or len(_a ) - 1
if left > right:
return -1
elif list_data[left] == key:
return left
elif list_data[right] == key:
return right
else:
return search(_a , _a , left + 1 , right - 1 )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 345 |
from __future__ import annotations
from typing import Dict
from ...configuration_utils import PretrainedConfig
UpperCamelCase_ = {
'''susnato/ernie-m-base_pytorch''': '''https://huggingface.co/susnato/ernie-m-base_pytorch/blob/main/config.json''',
'''susnato/ernie-m-large_pytorch''': '''https://huggingface.co/susnato/ernie-m-large_pytorch/blob/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = "ernie_m"
A__ : Dict[str, str] = {"dropout": "classifier_dropout", "num_classes": "num_labels"}
def __init__( self: str ,lowerCamelCase_: int = 250002 ,lowerCamelCase_: int = 768 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 3072 ,lowerCamelCase_: str = "gelu" ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: int = 514 ,lowerCamelCase_: float = 0.0_2 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: float = 1e-05 ,lowerCamelCase_: Any=None ,lowerCamelCase_: List[Any]=False ,lowerCamelCase_: Tuple=0.0 ,**lowerCamelCase_: Optional[int] ,) -> Optional[Any]:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = vocab_size
UpperCAmelCase_ : Any = hidden_size
UpperCAmelCase_ : Optional[Any] = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : List[Any] = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = initializer_range
UpperCAmelCase_ : Union[str, Any] = layer_norm_eps
UpperCAmelCase_ : List[Any] = classifier_dropout
UpperCAmelCase_ : str = is_decoder
UpperCAmelCase_ : List[str] = act_dropout
| 345 | 1 |
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
return numa ^ numa < 0
if __name__ == "__main__":
import doctest
doctest.testmod()
| 345 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
UpperCamelCase_ = logging.getLogger(__name__)
torch.set_grad_enabled(False)
UpperCamelCase_ = '''cuda''' if torch.cuda.is_available() else '''cpu'''
def lowerCamelCase_ ( _a : str , _a : Any=100 , _a : int=" " ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = text.split(_a )
return [character.join(text[i : i + n] ).strip() for i in range(0 , len(_a ) , _a )]
def lowerCamelCase_ ( _a : dict ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = [], []
for title, text in zip(documents["""title"""] , documents["""text"""] ):
if text is not None:
for passage in split_text(_a ):
titles.append(title if title is not None else """""" )
texts.append(_a )
return {"title": titles, "text": texts}
def lowerCamelCase_ ( _a : dict , _a : DPRContextEncoder , _a : DPRContextEncoderTokenizerFast ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = ctx_tokenizer(
documents["""title"""] , documents["""text"""] , truncation=_a , padding="""longest""" , return_tensors="""pt""" )["""input_ids"""]
UpperCAmelCase_ : Tuple = ctx_encoder(input_ids.to(device=_a ) , return_dict=_a ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def lowerCamelCase_ ( _a : "RagExampleArguments" , _a : "ProcessingArguments" , _a : "IndexHnswArguments" , ):
'''simple docstring'''
logger.info("""Step 1 - Create the dataset""" )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
UpperCAmelCase_ : Optional[int] = load_dataset(
"""csv""" , data_files=[rag_example_args.csv_path] , split="""train""" , delimiter="""\t""" , column_names=["""title""", """text"""] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
UpperCAmelCase_ : Tuple = dataset.map(_a , batched=_a , num_proc=processing_args.num_proc )
# And compute the embeddings
UpperCAmelCase_ : List[str] = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=_a )
UpperCAmelCase_ : Dict = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
UpperCAmelCase_ : Any = Features(
{"""text""": Value("""string""" ), """title""": Value("""string""" ), """embeddings""": Sequence(Value("""float32""" ) )} ) # optional, save as float32 instead of float64 to save space
UpperCAmelCase_ : List[str] = dataset.map(
partial(_a , ctx_encoder=_a , ctx_tokenizer=_a ) , batched=_a , batch_size=processing_args.batch_size , features=_a , )
# And finally save your dataset
UpperCAmelCase_ : Union[str, Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset""" )
dataset.save_to_disk(_a )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info("""Step 2 - Index the dataset""" )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
UpperCAmelCase_ : Union[str, Any] = faiss.IndexHNSWFlat(index_hnsw_args.d , index_hnsw_args.m , faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index("""embeddings""" , custom_index=_a )
# And save the index
UpperCAmelCase_ : Optional[Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset_hnsw_index.faiss""" )
dataset.get_index("""embeddings""" ).save(_a )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" / "my_knowledge_dataset.csv" ) , metadata={"help": "Path to a tab-separated csv file with columns 'title' and 'text'"} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."} , )
A__ : str = field(
default="facebook/rag-sequence-nq" , metadata={"help": "The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"} , )
A__ : str = field(
default="facebook/dpr-ctx_encoder-multiset-base" , metadata={
"help": (
"The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"
" 'facebook/dpr-ctx_encoder-multiset-base'"
)
} , )
A__ : Optional[str] = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" ) , metadata={"help": "Path to a directory where the dataset passages and the index will be saved"} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": "The number of processes to use to split the documents into passages. Default is single process."
} , )
A__ : int = field(
default=16 , metadata={
"help": "The batch size to use when computing the passages embeddings using the DPR context encoder."
} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : int = field(
default=768 , metadata={"help": "The dimension of the embeddings to pass to the HNSW Faiss index."} , )
A__ : int = field(
default=128 , metadata={
"help": (
"The number of bi-directional links created for every new element during the HNSW index construction."
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
UpperCamelCase_ = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
UpperCamelCase_ = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 345 | 1 |
import copy
from typing import Dict, List, Optional
from ...configuration_utils import PretrainedConfig
from ...utils import logging
from ..auto import CONFIG_MAPPING
UpperCamelCase_ = {
'''facebook/mask2former-swin-small-coco-instance''': (
'''https://huggingface.co/facebook/mask2former-swin-small-coco-instance/blob/main/config.json'''
)
# See all Mask2Former models at https://huggingface.co/models?filter=mask2former
}
UpperCamelCase_ = logging.get_logger(__name__)
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : List[str] = "mask2former"
A__ : Union[str, Any] = ["swin"]
A__ : int = {"hidden_size": "hidden_dim"}
def __init__( self: Any ,lowerCamelCase_: Optional[Dict] = None ,lowerCamelCase_: int = 256 ,lowerCamelCase_: int = 256 ,lowerCamelCase_: int = 256 ,lowerCamelCase_: int = 1024 ,lowerCamelCase_: str = "relu" ,lowerCamelCase_: int = 6 ,lowerCamelCase_: int = 10 ,lowerCamelCase_: int = 8 ,lowerCamelCase_: float = 0.0 ,lowerCamelCase_: int = 2048 ,lowerCamelCase_: bool = False ,lowerCamelCase_: bool = False ,lowerCamelCase_: int = 4 ,lowerCamelCase_: int = 255 ,lowerCamelCase_: int = 100 ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: float = 2.0 ,lowerCamelCase_: float = 5.0 ,lowerCamelCase_: float = 5.0 ,lowerCamelCase_: int = 12544 ,lowerCamelCase_: float = 3.0 ,lowerCamelCase_: float = 0.7_5 ,lowerCamelCase_: float = 0.0_2 ,lowerCamelCase_: float = 1.0 ,lowerCamelCase_: bool = True ,lowerCamelCase_: List[int] = [4, 8, 16, 32] ,lowerCamelCase_: bool = None ,**lowerCamelCase_: str ,) -> List[Any]:
if backbone_config is None:
logger.info("""`backbone_config` is `None`. Initializing the config with the default `Swin` backbone.""" )
UpperCAmelCase_ : int = CONFIG_MAPPING["""swin"""](
image_size=224 ,in_channels=3 ,patch_size=4 ,embed_dim=96 ,depths=[2, 2, 18, 2] ,num_heads=[3, 6, 12, 24] ,window_size=7 ,drop_path_rate=0.3 ,use_absolute_embeddings=lowerCamelCase_ ,out_features=["""stage1""", """stage2""", """stage3""", """stage4"""] ,)
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Dict = backbone_config.pop("""model_type""" )
UpperCAmelCase_ : Dict = CONFIG_MAPPING[backbone_model_type]
UpperCAmelCase_ : Dict = config_class.from_dict(lowerCamelCase_ )
# verify that the backbone is supported
if backbone_config.model_type not in self.backbones_supported:
logger.warning_once(
F'''Backbone {backbone_config.model_type} is not a supported model and may not be compatible with Mask2Former. '''
F'''Supported model types: {','.join(self.backbones_supported )}''' )
UpperCAmelCase_ : Optional[int] = backbone_config
UpperCAmelCase_ : List[Any] = feature_size
UpperCAmelCase_ : Union[str, Any] = mask_feature_size
UpperCAmelCase_ : Dict = hidden_dim
UpperCAmelCase_ : Union[str, Any] = encoder_feedforward_dim
UpperCAmelCase_ : int = activation_function
UpperCAmelCase_ : str = encoder_layers
UpperCAmelCase_ : List[str] = decoder_layers
UpperCAmelCase_ : Optional[Any] = num_attention_heads
UpperCAmelCase_ : Tuple = dropout
UpperCAmelCase_ : Optional[Any] = dim_feedforward
UpperCAmelCase_ : Optional[Any] = pre_norm
UpperCAmelCase_ : Optional[Any] = enforce_input_projection
UpperCAmelCase_ : Optional[Any] = common_stride
UpperCAmelCase_ : Any = ignore_value
UpperCAmelCase_ : Union[str, Any] = num_queries
UpperCAmelCase_ : str = no_object_weight
UpperCAmelCase_ : int = class_weight
UpperCAmelCase_ : Union[str, Any] = mask_weight
UpperCAmelCase_ : Optional[Any] = dice_weight
UpperCAmelCase_ : str = train_num_points
UpperCAmelCase_ : Any = oversample_ratio
UpperCAmelCase_ : str = importance_sample_ratio
UpperCAmelCase_ : Optional[Any] = init_std
UpperCAmelCase_ : Dict = init_xavier_std
UpperCAmelCase_ : Any = use_auxiliary_loss
UpperCAmelCase_ : Union[str, Any] = feature_strides
UpperCAmelCase_ : List[str] = output_auxiliary_logits
UpperCAmelCase_ : List[str] = decoder_layers
super().__init__(**lowerCamelCase_ )
@classmethod
def A__ ( cls: Dict ,lowerCamelCase_: PretrainedConfig ,**lowerCamelCase_: Optional[Any] ) -> Tuple:
return cls(
backbone_config=lowerCamelCase_ ,**lowerCamelCase_ ,)
def A__ ( self: Union[str, Any] ) -> Dict[str, any]:
UpperCAmelCase_ : Tuple = copy.deepcopy(self.__dict__ )
UpperCAmelCase_ : Dict = self.backbone_config.to_dict()
UpperCAmelCase_ : Any = self.__class__.model_type
return output
| 345 |
import gc
import unittest
import torch
from parameterized import parameterized
from diffusers import AutoencoderKL
from diffusers.utils import floats_tensor, load_hf_numpy, require_torch_gpu, slow, torch_all_close, torch_device
from diffusers.utils.import_utils import is_xformers_available
from diffusers.utils.testing_utils import enable_full_determinism
from .test_modeling_common import ModelTesterMixin, UNetTesterMixin
enable_full_determinism()
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = AutoencoderKL
A__ : Optional[int] = "sample"
A__ : Tuple = 1E-2
@property
def A__ ( self: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = 4
UpperCAmelCase_ : str = 3
UpperCAmelCase_ : Any = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ).to(lowerCamelCase_ )
return {"sample": image}
@property
def A__ ( self: List[str] ) -> Tuple:
return (3, 32, 32)
@property
def A__ ( self: Optional[Any] ) -> Any:
return (3, 32, 32)
def A__ ( self: Any ) -> Tuple:
UpperCAmelCase_ : List[Any] = {
"""block_out_channels""": [32, 64],
"""in_channels""": 3,
"""out_channels""": 3,
"""down_block_types""": ["""DownEncoderBlock2D""", """DownEncoderBlock2D"""],
"""up_block_types""": ["""UpDecoderBlock2D""", """UpDecoderBlock2D"""],
"""latent_channels""": 4,
}
UpperCAmelCase_ : int = self.dummy_input
return init_dict, inputs_dict
def A__ ( self: Optional[Any] ) -> int:
pass
def A__ ( self: str ) -> Any:
pass
@unittest.skipIf(torch_device == """mps""" ,"""Gradient checkpointing skipped on MPS""" )
def A__ ( self: Union[str, Any] ) -> Dict:
# enable deterministic behavior for gradient checkpointing
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.prepare_init_args_and_inputs_for_common()
UpperCAmelCase_ : List[Any] = self.model_class(**lowerCamelCase_ )
model.to(lowerCamelCase_ )
assert not model.is_gradient_checkpointing and model.training
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model.zero_grad()
UpperCAmelCase_ : Any = torch.randn_like(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = (out - labels).mean()
loss.backward()
# re-instantiate the model now enabling gradient checkpointing
UpperCAmelCase_ : str = self.model_class(**lowerCamelCase_ )
# clone model
model_a.load_state_dict(model.state_dict() )
model_a.to(lowerCamelCase_ )
model_a.enable_gradient_checkpointing()
assert model_a.is_gradient_checkpointing and model_a.training
UpperCAmelCase_ : Optional[int] = model_a(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model_a.zero_grad()
UpperCAmelCase_ : Dict = (out_a - labels).mean()
loss_a.backward()
# compare the output and parameters gradients
self.assertTrue((loss - loss_a).abs() < 1e-5 )
UpperCAmelCase_ : Dict = dict(model.named_parameters() )
UpperCAmelCase_ : Union[str, Any] = dict(model_a.named_parameters() )
for name, param in named_params.items():
self.assertTrue(torch_all_close(param.grad.data ,named_params_a[name].grad.data ,atol=5e-5 ) )
def A__ ( self: Optional[Any] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" ,output_loading_info=lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
self.assertEqual(len(loading_info["""missing_keys"""] ) ,0 )
model.to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model(**self.dummy_input )
assert image is not None, "Make sure output is not None"
def A__ ( self: Optional[int] ) -> int:
UpperCAmelCase_ : Dict = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" )
UpperCAmelCase_ : Tuple = model.to(lowerCamelCase_ )
model.eval()
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.manual_seed(0 )
else:
UpperCAmelCase_ : Optional[int] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = torch.randn(
1 ,model.config.in_channels ,model.config.sample_size ,model.config.sample_size ,generator=torch.manual_seed(0 ) ,)
UpperCAmelCase_ : int = image.to(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ,generator=lowerCamelCase_ ).sample
UpperCAmelCase_ : Optional[int] = output[0, -1, -3:, -3:].flatten().cpu()
# Since the VAE Gaussian prior's generator is seeded on the appropriate device,
# the expected output slices are not the same for CPU and GPU.
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.tensor(
[
-4.0078e-01,
-3.8323e-04,
-1.2681e-01,
-1.1462e-01,
2.0095e-01,
1.0893e-01,
-8.8247e-02,
-3.0361e-01,
-9.8644e-03,
] )
elif torch_device == "cpu":
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.1_3_5_2, 0.0_8_7_8, 0.0_4_1_9, -0.0_8_1_8, -0.1_0_6_9, 0.0_6_8_8, -0.1_4_5_8, -0.4_4_4_6, -0.0_0_2_6] )
else:
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.2_4_2_1, 0.4_6_4_2, 0.2_5_0_7, -0.0_4_3_8, 0.0_6_8_2, 0.3_1_6_0, -0.2_0_1_8, -0.0_7_2_7, 0.2_4_8_5] )
self.assertTrue(torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,rtol=1e-2 ) )
@slow
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
return F'''gaussian_noise_s={seed}_shape={'_'.join([str(lowerCamelCase_ ) for s in shape] )}.npy'''
def A__ ( self: Union[str, Any] ) -> Optional[int]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: List[Any]=(4, 3, 512, 512) ,lowerCamelCase_: Optional[Any]=False ) -> Optional[int]:
UpperCAmelCase_ : Tuple = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : Tuple = torch.from_numpy(load_hf_numpy(self.get_file_format(lowerCamelCase_ ,lowerCamelCase_ ) ) ).to(lowerCamelCase_ ).to(lowerCamelCase_ )
return image
def A__ ( self: List[Any] ,lowerCamelCase_: List[str]="CompVis/stable-diffusion-v1-4" ,lowerCamelCase_: Union[str, Any]=False ) -> Any:
UpperCAmelCase_ : Optional[Any] = """fp16""" if fpaa else None
UpperCAmelCase_ : str = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : int = AutoencoderKL.from_pretrained(
lowerCamelCase_ ,subfolder="""vae""" ,torch_dtype=lowerCamelCase_ ,revision=lowerCamelCase_ ,)
model.to(lowerCamelCase_ ).eval()
return model
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any]=0 ) -> Optional[int]:
if torch_device == "mps":
return torch.manual_seed(lowerCamelCase_ )
return torch.Generator(device=lowerCamelCase_ ).manual_seed(lowerCamelCase_ )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_3, 0.9_8_7_8, -0.0_4_9_5, -0.0_7_9_0, -0.2_7_0_9, 0.8_3_7_5, -0.2_0_6_0, -0.0_8_2_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_6, 0.1_1_6_8, 0.1_3_3_2, -0.4_8_4_0, -0.2_5_0_8, -0.0_7_9_1, -0.0_4_9_3, -0.4_0_8_9], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ) -> Tuple:
UpperCAmelCase_ : List[Any] = self.get_sd_vae_model()
UpperCAmelCase_ : int = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Optional[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Tuple = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[33, [-0.0_5_1_3, 0.0_2_8_9, 1.3_7_9_9, 0.2_1_6_6, -0.2_5_7_3, -0.0_8_7_1, 0.5_1_0_3, -0.0_9_9_9]],
[47, [-0.4_1_2_8, -0.1_3_2_0, -0.3_7_0_4, 0.1_9_6_5, -0.4_1_1_6, -0.2_3_3_2, -0.3_3_4_0, 0.2_2_4_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Tuple:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Tuple = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : Optional[int] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_9, 0.9_8_6_6, -0.0_4_8_7, -0.0_7_7_7, -0.2_7_1_6, 0.8_3_6_8, -0.2_0_5_5, -0.0_8_1_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_7, 0.1_1_4_7, 0.1_3_3_3, -0.4_8_4_1, -0.2_5_0_6, -0.0_8_0_5, -0.0_4_9_1, -0.4_0_8_5], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[int] = self.get_sd_vae_model()
UpperCAmelCase_ : Dict = self.get_sd_image(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : str = model(lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : List[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Any = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[13, [-0.2_0_5_1, -0.1_8_0_3, -0.2_3_1_1, -0.2_1_1_4, -0.3_2_9_2, -0.3_5_7_4, -0.2_9_5_3, -0.3_3_2_3]],
[37, [-0.2_6_3_2, -0.2_6_2_5, -0.2_1_9_9, -0.2_7_4_1, -0.4_5_3_9, -0.4_9_9_0, -0.3_7_2_0, -0.4_9_2_5]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: str ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[int] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : str = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : Any = sample[-1, -2:, :2, -2:].flatten().cpu()
UpperCAmelCase_ : Union[str, Any] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 )
@parameterized.expand(
[
# fmt: off
[27, [-0.0_3_6_9, 0.0_2_0_7, -0.0_7_7_6, -0.0_6_8_2, -0.1_7_4_7, -0.1_9_3_0, -0.1_4_6_5, -0.2_0_3_9]],
[16, [-0.1_6_2_8, -0.2_1_3_4, -0.2_7_4_7, -0.2_6_4_2, -0.3_7_7_4, -0.4_4_0_4, -0.3_6_8_7, -0.4_2_7_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: str ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : str = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : str = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=5e-3 )
@parameterized.expand([(13,), (16,), (27,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> int:
UpperCAmelCase_ : Optional[Any] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-1 )
@parameterized.expand([(13,), (16,), (37,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = self.get_sd_vae_model()
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.3_0_0_1, 0.0_9_1_8, -2.6_9_8_4, -3.9_7_2_0, -3.2_0_9_9, -5.0_3_5_3, 1.7_3_3_8, -0.2_0_6_5, 3.4_2_6_7]],
[47, [-1.5_0_3_0, -4.3_8_7_1, -6.0_3_5_5, -9.1_1_5_7, -1.6_6_6_1, -2.7_8_5_3, 2.1_6_0_7, -5.0_8_2_3, 2.5_6_3_3]],
# fmt: on
] )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[Any] = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : str = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : int = model.encode(lowerCamelCase_ ).latent_dist
UpperCAmelCase_ : Optional[Any] = dist.sample(generator=lowerCamelCase_ )
assert list(sample.shape ) == [image.shape[0], 4] + [i // 8 for i in image.shape[2:]]
UpperCAmelCase_ : Tuple = sample[0, -1, -3:, -3:].flatten().cpu()
UpperCAmelCase_ : Optional[Any] = torch.tensor(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = 3e-3 if torch_device != """mps""" else 1e-2
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=lowerCamelCase_ )
| 345 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
UpperCamelCase_ = {
'''configuration_distilbert''': [
'''DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''DistilBertConfig''',
'''DistilBertOnnxConfig''',
],
'''tokenization_distilbert''': ['''DistilBertTokenizer'''],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = ['''DistilBertTokenizerFast''']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''DistilBertForMaskedLM''',
'''DistilBertForMultipleChoice''',
'''DistilBertForQuestionAnswering''',
'''DistilBertForSequenceClassification''',
'''DistilBertForTokenClassification''',
'''DistilBertModel''',
'''DistilBertPreTrainedModel''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TFDistilBertForMaskedLM''',
'''TFDistilBertForMultipleChoice''',
'''TFDistilBertForQuestionAnswering''',
'''TFDistilBertForSequenceClassification''',
'''TFDistilBertForTokenClassification''',
'''TFDistilBertMainLayer''',
'''TFDistilBertModel''',
'''TFDistilBertPreTrainedModel''',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''FlaxDistilBertForMaskedLM''',
'''FlaxDistilBertForMultipleChoice''',
'''FlaxDistilBertForQuestionAnswering''',
'''FlaxDistilBertForSequenceClassification''',
'''FlaxDistilBertForTokenClassification''',
'''FlaxDistilBertModel''',
'''FlaxDistilBertPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_distilbert import (
DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP,
DistilBertConfig,
DistilBertOnnxConfig,
)
from .tokenization_distilbert import DistilBertTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_distilbert_fast import DistilBertTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_distilbert import (
DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
DistilBertForMaskedLM,
DistilBertForMultipleChoice,
DistilBertForQuestionAnswering,
DistilBertForSequenceClassification,
DistilBertForTokenClassification,
DistilBertModel,
DistilBertPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_distilbert import (
TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDistilBertForMaskedLM,
TFDistilBertForMultipleChoice,
TFDistilBertForQuestionAnswering,
TFDistilBertForSequenceClassification,
TFDistilBertForTokenClassification,
TFDistilBertMainLayer,
TFDistilBertModel,
TFDistilBertPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_distilbert import (
FlaxDistilBertForMaskedLM,
FlaxDistilBertForMultipleChoice,
FlaxDistilBertForQuestionAnswering,
FlaxDistilBertForSequenceClassification,
FlaxDistilBertForTokenClassification,
FlaxDistilBertModel,
FlaxDistilBertPreTrainedModel,
)
else:
import sys
UpperCamelCase_ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 345 |
import copy
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Audio, Features, Value
from .base import TaskTemplate
@dataclass(frozen=__snake_case )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = field(default="automatic-speech-recognition" , metadata={"include_in_asdict_even_if_is_default": True} )
A__ : ClassVar[Features] = Features({"audio": Audio()} )
A__ : ClassVar[Features] = Features({"transcription": Value("string" )} )
A__ : str = "audio"
A__ : str = "transcription"
def A__ ( self: int ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
if self.audio_column not in features:
raise ValueError(F'''Column {self.audio_column} is not present in features.''' )
if not isinstance(features[self.audio_column] ,lowerCamelCase_ ):
raise ValueError(F'''Column {self.audio_column} is not an Audio type.''' )
UpperCAmelCase_ : Any = copy.deepcopy(self )
UpperCAmelCase_ : Union[str, Any] = self.input_schema.copy()
UpperCAmelCase_ : Any = features[self.audio_column]
UpperCAmelCase_ : Union[str, Any] = input_schema
return task_template
@property
def A__ ( self: List[str] ) -> Dict[str, str]:
return {self.audio_column: "audio", self.transcription_column: "transcription"}
| 345 | 1 |
import datasets
from .evaluate import evaluate
UpperCamelCase_ = '''\
@inproceedings{Rajpurkar2016SQuAD10,
title={SQuAD: 100, 000+ Questions for Machine Comprehension of Text},
author={Pranav Rajpurkar and Jian Zhang and Konstantin Lopyrev and Percy Liang},
booktitle={EMNLP},
year={2016}
}
'''
UpperCamelCase_ = '''
This metric wrap the official scoring script for version 1 of the Stanford Question Answering Dataset (SQuAD).
Stanford Question Answering Dataset (SQuAD) is a reading comprehension dataset, consisting of questions posed by
crowdworkers on a set of Wikipedia articles, where the answer to every question is a segment of text, or span,
from the corresponding reading passage, or the question might be unanswerable.
'''
UpperCamelCase_ = '''
Computes SQuAD scores (F1 and EM).
Args:
predictions: List of question-answers dictionaries with the following key-values:
- \'id\': id of the question-answer pair as given in the references (see below)
- \'prediction_text\': the text of the answer
references: List of question-answers dictionaries with the following key-values:
- \'id\': id of the question-answer pair (see above),
- \'answers\': a Dict in the SQuAD dataset format
{
\'text\': list of possible texts for the answer, as a list of strings
\'answer_start\': list of start positions for the answer, as a list of ints
}
Note that answer_start values are not taken into account to compute the metric.
Returns:
\'exact_match\': Exact match (the normalized answer exactly match the gold answer)
\'f1\': The F-score of predicted tokens versus the gold answer
Examples:
>>> predictions = [{\'prediction_text\': \'1976\', \'id\': \'56e10a3be3433e1400422b22\'}]
>>> references = [{\'answers\': {\'answer_start\': [97], \'text\': [\'1976\']}, \'id\': \'56e10a3be3433e1400422b22\'}]
>>> squad_metric = datasets.load_metric("squad")
>>> results = squad_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 100.0, \'f1\': 100.0}
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class _snake_case ( datasets.Metric ):
'''simple docstring'''
def A__ ( self: int ) -> Dict:
return datasets.MetricInfo(
description=_DESCRIPTION ,citation=_CITATION ,inputs_description=_KWARGS_DESCRIPTION ,features=datasets.Features(
{
"""predictions""": {"""id""": datasets.Value("""string""" ), """prediction_text""": datasets.Value("""string""" )},
"""references""": {
"""id""": datasets.Value("""string""" ),
"""answers""": datasets.features.Sequence(
{
"""text""": datasets.Value("""string""" ),
"""answer_start""": datasets.Value("""int32""" ),
} ),
},
} ) ,codebase_urls=["""https://rajpurkar.github.io/SQuAD-explorer/"""] ,reference_urls=["""https://rajpurkar.github.io/SQuAD-explorer/"""] ,)
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[Any] ) -> List[Any]:
UpperCAmelCase_ : Any = {prediction["""id"""]: prediction["""prediction_text"""] for prediction in predictions}
UpperCAmelCase_ : str = [
{
"""paragraphs""": [
{
"""qas""": [
{
"""answers""": [{"""text""": answer_text} for answer_text in ref["""answers"""]["""text"""]],
"""id""": ref["""id"""],
}
for ref in references
]
}
]
}
]
UpperCAmelCase_ : List[str] = evaluate(dataset=lowerCamelCase_ ,predictions=lowerCamelCase_ )
return score
| 345 |
from collections import OrderedDict
from typing import TYPE_CHECKING, Any, Mapping, Optional
from packaging import version
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...onnx.utils import compute_effective_axis_dimension
from ...utils import logging
if TYPE_CHECKING:
from ...processing_utils import ProcessorMixin
from ...utils import TensorType
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/layoutlmv3-base''': '''https://huggingface.co/microsoft/layoutlmv3-base/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "layoutlmv3"
def __init__( self: str ,lowerCamelCase_: Any=50265 ,lowerCamelCase_: int=768 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=3072 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Any=0.1 ,lowerCamelCase_: Tuple=512 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: List[str]=1e-5 ,lowerCamelCase_: int=1 ,lowerCamelCase_: int=0 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Dict=1024 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=32 ,lowerCamelCase_: Union[str, Any]=128 ,lowerCamelCase_: Tuple=64 ,lowerCamelCase_: Tuple=256 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Any=True ,lowerCamelCase_: Dict=224 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Optional[int]=16 ,lowerCamelCase_: Dict=None ,**lowerCamelCase_: str ,) -> List[Any]:
super().__init__(
vocab_size=lowerCamelCase_ ,hidden_size=lowerCamelCase_ ,num_hidden_layers=lowerCamelCase_ ,num_attention_heads=lowerCamelCase_ ,intermediate_size=lowerCamelCase_ ,hidden_act=lowerCamelCase_ ,hidden_dropout_prob=lowerCamelCase_ ,attention_probs_dropout_prob=lowerCamelCase_ ,max_position_embeddings=lowerCamelCase_ ,type_vocab_size=lowerCamelCase_ ,initializer_range=lowerCamelCase_ ,layer_norm_eps=lowerCamelCase_ ,pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = max_ad_position_embeddings
UpperCAmelCase_ : Optional[int] = coordinate_size
UpperCAmelCase_ : Optional[int] = shape_size
UpperCAmelCase_ : Optional[Any] = has_relative_attention_bias
UpperCAmelCase_ : Optional[int] = rel_pos_bins
UpperCAmelCase_ : Union[str, Any] = max_rel_pos
UpperCAmelCase_ : Dict = has_spatial_attention_bias
UpperCAmelCase_ : Optional[int] = rel_ad_pos_bins
UpperCAmelCase_ : Tuple = max_rel_ad_pos
UpperCAmelCase_ : Union[str, Any] = text_embed
UpperCAmelCase_ : Optional[Any] = visual_embed
UpperCAmelCase_ : List[str] = input_size
UpperCAmelCase_ : str = num_channels
UpperCAmelCase_ : Optional[int] = patch_size
UpperCAmelCase_ : Tuple = classifier_dropout
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = version.parse("1.12" )
@property
def A__ ( self: Dict ) -> Mapping[str, Mapping[int, str]]:
# The order of inputs is different for question answering and sequence classification
if self.task in ["question-answering", "sequence-classification"]:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels""", 2: """height""", 3: """width"""}),
] )
else:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels"""}),
] )
@property
def A__ ( self: Any ) -> float:
return 1e-5
@property
def A__ ( self: int ) -> int:
return 12
def A__ ( self: List[str] ,lowerCamelCase_: "ProcessorMixin" ,lowerCamelCase_: int = -1 ,lowerCamelCase_: int = -1 ,lowerCamelCase_: bool = False ,lowerCamelCase_: Optional["TensorType"] = None ,lowerCamelCase_: int = 3 ,lowerCamelCase_: int = 40 ,lowerCamelCase_: int = 40 ,) -> Mapping[str, Any]:
setattr(processor.image_processor ,"""apply_ocr""" ,lowerCamelCase_ )
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
UpperCAmelCase_ : List[str] = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_batch ,num_token_to_add=0 )
# If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX
UpperCAmelCase_ : int = processor.tokenizer.num_special_tokens_to_add(lowerCamelCase_ )
UpperCAmelCase_ : int = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_sequence ,num_token_to_add=lowerCamelCase_ )
# Generate dummy inputs according to compute batch and sequence
UpperCAmelCase_ : Optional[int] = [[""" """.join([processor.tokenizer.unk_token] ) * seq_length]] * batch_size
# Generate dummy bounding boxes
UpperCAmelCase_ : List[Any] = [[[48, 84, 73, 128]]] * batch_size
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
# batch_size = compute_effective_axis_dimension(batch_size, fixed_dimension=OnnxConfig.default_fixed_batch)
UpperCAmelCase_ : Any = self._generate_dummy_images(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = dict(
processor(
lowerCamelCase_ ,text=lowerCamelCase_ ,boxes=lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,) )
return inputs
| 345 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''studio-ousia/luke-base''': '''https://huggingface.co/studio-ousia/luke-base/resolve/main/config.json''',
'''studio-ousia/luke-large''': '''https://huggingface.co/studio-ousia/luke-large/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[int] = "luke"
def __init__( self: Optional[int] ,lowerCamelCase_: Union[str, Any]=50267 ,lowerCamelCase_: Any=500000 ,lowerCamelCase_: List[str]=768 ,lowerCamelCase_: Tuple=256 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=12 ,lowerCamelCase_: int=3072 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Tuple=0.1 ,lowerCamelCase_: Tuple=0.1 ,lowerCamelCase_: str=512 ,lowerCamelCase_: Any=2 ,lowerCamelCase_: str=0.0_2 ,lowerCamelCase_: Any=1e-12 ,lowerCamelCase_: Optional[Any]=True ,lowerCamelCase_: str=None ,lowerCamelCase_: Union[str, Any]=1 ,lowerCamelCase_: Tuple=0 ,lowerCamelCase_: Any=2 ,**lowerCamelCase_: Optional[Any] ,) -> int:
super().__init__(pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : List[str] = vocab_size
UpperCAmelCase_ : Tuple = entity_vocab_size
UpperCAmelCase_ : str = hidden_size
UpperCAmelCase_ : List[Any] = entity_emb_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : Optional[int] = num_attention_heads
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : int = intermediate_size
UpperCAmelCase_ : Optional[int] = hidden_dropout_prob
UpperCAmelCase_ : Dict = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : Optional[int] = initializer_range
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = use_entity_aware_attention
UpperCAmelCase_ : Dict = classifier_dropout
| 345 |
import argparse
from argparse import Namespace
import torch
from torch import nn
from transformers import XGLMConfig, XGLMForCausalLM
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = [
"""decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = emb.weight.shape
UpperCAmelCase_ : Tuple = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : List[Any] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = torch.load(_a , map_location="""cpu""" )
UpperCAmelCase_ : Dict = Namespace(**checkpoint["""cfg"""]["""model"""] )
UpperCAmelCase_ : Optional[int] = checkpoint["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = state_dict["""decoder.embed_tokens.weight"""].shape[0]
UpperCAmelCase_ : List[str] = {key.replace("""decoder""" , """model""" ): val for key, val in state_dict.items()}
UpperCAmelCase_ : int = XGLMConfig(
vocab_size=_a , max_position_embeddings=args.max_target_positions , num_layers=args.decoder_layers , attention_heads=args.decoder_attention_heads , ffn_dim=args.decoder_ffn_embed_dim , d_model=args.decoder_embed_dim , layerdrop=args.decoder_layerdrop , dropout=args.dropout , attention_dropout=args.attention_dropout , activation_dropout=args.activation_dropout , activation_function="""gelu""" , scale_embedding=not args.no_scale_embedding , tie_word_embeddings=args.share_decoder_input_output_embed , )
UpperCAmelCase_ : List[str] = XGLMForCausalLM(_a )
UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
print(_a )
UpperCAmelCase_ : Optional[Any] = make_linear_from_emb(model.model.embed_tokens )
return model
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument('''fairseq_path''', type=str, help='''path to a model.pt on local filesystem.''')
parser.add_argument('''pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = convert_fairseq_xglm_checkpoint_from_disk(args.fairseq_path)
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 | 1 |
from typing import Callable, Optional, Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/xprophetnet-large-wiki100-cased''': (
'''https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Tuple = "xlm-prophetnet"
A__ : Optional[int] = ["past_key_values"]
A__ : List[Any] = {
"num_attention_heads": "num_encoder_attention_heads",
}
def __init__( self: Union[str, Any] ,lowerCamelCase_: Optional[float] = 0.1 ,lowerCamelCase_: Optional[Union[str, Callable]] = "gelu" ,lowerCamelCase_: Optional[int] = 30522 ,lowerCamelCase_: Optional[int] = 1024 ,lowerCamelCase_: Optional[int] = 4096 ,lowerCamelCase_: Optional[int] = 12 ,lowerCamelCase_: Optional[int] = 16 ,lowerCamelCase_: Optional[int] = 4096 ,lowerCamelCase_: Optional[int] = 12 ,lowerCamelCase_: Optional[int] = 16 ,lowerCamelCase_: Optional[float] = 0.1 ,lowerCamelCase_: Optional[float] = 0.1 ,lowerCamelCase_: Optional[int] = 512 ,lowerCamelCase_: Optional[float] = 0.0_2 ,lowerCamelCase_: Optional[bool] = True ,lowerCamelCase_: Optional[bool] = True ,lowerCamelCase_: Optional[int] = 0 ,lowerCamelCase_: Optional[int] = 2 ,lowerCamelCase_: Optional[int] = 32 ,lowerCamelCase_: Optional[int] = 128 ,lowerCamelCase_: Optional[bool] = False ,lowerCamelCase_: Optional[float] = 0.0 ,lowerCamelCase_: Optional[bool] = True ,lowerCamelCase_: Optional[int] = 0 ,lowerCamelCase_: Optional[int] = 1 ,lowerCamelCase_: Optional[int] = 2 ,**lowerCamelCase_: Optional[int] ,) -> Dict:
UpperCAmelCase_ : List[str] = vocab_size
UpperCAmelCase_ : Optional[int] = hidden_size
UpperCAmelCase_ : Tuple = encoder_ffn_dim
UpperCAmelCase_ : Dict = num_encoder_layers
UpperCAmelCase_ : List[str] = num_encoder_attention_heads
UpperCAmelCase_ : Tuple = decoder_ffn_dim
UpperCAmelCase_ : Optional[Any] = num_decoder_layers
UpperCAmelCase_ : int = num_decoder_attention_heads
UpperCAmelCase_ : Optional[int] = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = init_std # Normal(0, this parameter)
UpperCAmelCase_ : Any = activation_function
# parameters for xlmprophetnet
UpperCAmelCase_ : str = ngram
UpperCAmelCase_ : str = num_buckets
UpperCAmelCase_ : Dict = relative_max_distance
UpperCAmelCase_ : List[Any] = disable_ngram_loss
UpperCAmelCase_ : int = eps
# 3 Types of Dropout
UpperCAmelCase_ : Any = attention_dropout
UpperCAmelCase_ : Dict = activation_dropout
UpperCAmelCase_ : int = dropout
UpperCAmelCase_ : List[str] = use_cache
super().__init__(
pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,is_encoder_decoder=lowerCamelCase_ ,add_cross_attention=lowerCamelCase_ ,decoder_start_token_id=lowerCamelCase_ ,**lowerCamelCase_ ,)
@property
def A__ ( self: Optional[Any] ) -> int:
return self.num_encoder_layers + self.num_decoder_layers
@num_hidden_layers.setter
def A__ ( self: Dict ,lowerCamelCase_: Optional[int] ) -> int:
raise NotImplementedError(
"""This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and"""
""" `num_decoder_layers`.""" )
| 345 |
import collections
import inspect
import unittest
from transformers import FocalNetConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
FocalNetBackbone,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetModel,
)
from transformers.models.focalnet.modeling_focalnet import FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=13 ,lowerCamelCase_: Optional[int]=32 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Optional[Any]=[32, 64, 128] ,lowerCamelCase_: Optional[int]=[1, 2, 1] ,lowerCamelCase_: Union[str, Any]=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[str]=2.0 ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: int=True ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: str=True ,lowerCamelCase_: Dict=10 ,lowerCamelCase_: str=8 ,lowerCamelCase_: Union[str, Any]=["stage1", "stage2"] ,lowerCamelCase_: Optional[Any]=[1, 2] ,) -> str:
UpperCAmelCase_ : List[Any] = parent
UpperCAmelCase_ : Tuple = batch_size
UpperCAmelCase_ : Any = image_size
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : List[str] = num_channels
UpperCAmelCase_ : Dict = embed_dim
UpperCAmelCase_ : Dict = hidden_sizes
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = num_heads
UpperCAmelCase_ : List[Any] = window_size
UpperCAmelCase_ : Union[str, Any] = mlp_ratio
UpperCAmelCase_ : int = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Union[str, Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[int] = drop_path_rate
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : List[Any] = use_absolute_embeddings
UpperCAmelCase_ : List[Any] = patch_norm
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Optional[Any] = scope
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = encoder_stride
UpperCAmelCase_ : Optional[int] = out_features
UpperCAmelCase_ : Optional[int] = out_indices
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : int = None
if self.use_labels:
UpperCAmelCase_ : str = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Any = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Tuple:
return FocalNetConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,hidden_sizes=self.hidden_sizes ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,out_features=self.out_features ,out_indices=self.out_indices ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : Dict = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : Optional[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[str]:
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size, 8, 8] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,len(config.out_features ) )
self.parent.assertListEqual(model.channels ,config.hidden_sizes[:-1] )
# verify backbone works with out_features=None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size * 2, 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,1 )
self.parent.assertListEqual(model.channels ,[config.hidden_sizes[-1]] )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Any = FocalNetForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.reconstruction.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : int = 1
UpperCAmelCase_ : List[str] = FocalNetForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.reconstruction.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ) -> int:
UpperCAmelCase_ : List[Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
# test greyscale images
UpperCAmelCase_ : List[Any] = 1
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = config_and_inputs
UpperCAmelCase_ : int = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (
(
FocalNetModel,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetBackbone,
)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (
{"feature-extraction": FocalNetModel, "image-classification": FocalNetForImageClassification}
if is_torch_available()
else {}
)
A__ : Optional[Any] = False
A__ : Any = False
A__ : List[str] = False
A__ : Any = False
A__ : Any = False
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : Dict = FocalNetModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: List[str] ) -> Union[str, Any]:
return
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCamelCase_ )
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@unittest.skip(reason="""FocalNet does not use inputs_embeds""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""FocalNet does not use feedforward chunking""" )
def A__ ( self: Optional[Any] ) -> Optional[Any]:
pass
def A__ ( self: Optional[Any] ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : List[Any] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Dict = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Any = [*signature.parameters.keys()]
UpperCAmelCase_ : List[str] = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.hidden_states
UpperCAmelCase_ : List[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# FocalNet has a different seq_length
UpperCAmelCase_ : int = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Optional[int] = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Union[str, Any] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = reshaped_hidden_states[0].shape
UpperCAmelCase_ : List[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Union[str, Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Tuple = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : Union[str, Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Union[str, Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Any = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Optional[int] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
@slow
def A__ ( self: Optional[int] ) -> Optional[Any]:
for model_name in FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Tuple = FocalNetModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[Any] = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Optional[int] ) -> str:
# TODO update organization
return AutoImageProcessor.from_pretrained("""microsoft/focalnet-tiny""" ) if is_vision_available() else None
@slow
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification.from_pretrained("""microsoft/focalnet-tiny""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.default_image_processor
UpperCAmelCase_ : Union[str, Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Dict = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : str = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_1_6_6, -0.4_3_6_8, 0.2_1_9_1] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
self.assertTrue(outputs.logits.argmax(dim=-1 ).item() ,281 )
@require_torch
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (FocalNetBackbone,) if is_torch_available() else ()
A__ : int = FocalNetConfig
A__ : List[str] = False
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : str = FocalNetModelTester(self )
| 345 | 1 |
import copy
import json
import os
import tempfile
from transformers import is_torch_available
from .test_configuration_utils import config_common_kwargs
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: int ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: int=True ,lowerCamelCase_: Any=None ,**lowerCamelCase_: List[str] ) -> Any:
UpperCAmelCase_ : Dict = parent
UpperCAmelCase_ : List[str] = config_class
UpperCAmelCase_ : str = has_text_modality
UpperCAmelCase_ : Union[str, Any] = kwargs
UpperCAmelCase_ : int = common_properties
def A__ ( self: Union[str, Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = self.config_class(**self.inputs_dict )
UpperCAmelCase_ : List[str] = (
["""hidden_size""", """num_attention_heads""", """num_hidden_layers"""]
if self.common_properties is None
else self.common_properties
)
# Add common fields for text models
if self.has_text_modality:
common_properties.extend(["""vocab_size"""] )
# Test that config has the common properties as getters
for prop in common_properties:
self.parent.assertTrue(hasattr(lowerCamelCase_ ,lowerCamelCase_ ) ,msg=F'''`{prop}` does not exist''' )
# Test that config has the common properties as setter
for idx, name in enumerate(lowerCamelCase_ ):
try:
setattr(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
self.parent.assertEqual(
getattr(lowerCamelCase_ ,lowerCamelCase_ ) ,lowerCamelCase_ ,msg=F'''`{name} value {idx} expected, but was {getattr(lowerCamelCase_ ,lowerCamelCase_ )}''' )
except NotImplementedError:
# Some models might not be able to implement setters for common_properties
# In that case, a NotImplementedError is raised
pass
# Test if config class can be called with Config(prop_name=..)
for idx, name in enumerate(lowerCamelCase_ ):
try:
UpperCAmelCase_ : Optional[Any] = self.config_class(**{name: idx} )
self.parent.assertEqual(
getattr(lowerCamelCase_ ,lowerCamelCase_ ) ,lowerCamelCase_ ,msg=F'''`{name} value {idx} expected, but was {getattr(lowerCamelCase_ ,lowerCamelCase_ )}''' )
except NotImplementedError:
# Some models might not be able to implement setters for common_properties
# In that case, a NotImplementedError is raised
pass
def A__ ( self: Optional[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Any = self.config_class(**self.inputs_dict )
UpperCAmelCase_ : Union[str, Any] = json.loads(config.to_json_string() )
for key, value in self.inputs_dict.items():
self.parent.assertEqual(obj[key] ,lowerCamelCase_ )
def A__ ( self: Tuple ) -> Tuple:
UpperCAmelCase_ : str = self.config_class(**self.inputs_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
UpperCAmelCase_ : List[str] = os.path.join(lowerCamelCase_ ,"""config.json""" )
config_first.to_json_file(lowerCamelCase_ )
UpperCAmelCase_ : Any = self.config_class.from_json_file(lowerCamelCase_ )
self.parent.assertEqual(config_second.to_dict() ,config_first.to_dict() )
def A__ ( self: List[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = self.config_class(**self.inputs_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
config_first.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = self.config_class.from_pretrained(lowerCamelCase_ )
self.parent.assertEqual(config_second.to_dict() ,config_first.to_dict() )
def A__ ( self: List[Any] ) -> Tuple:
UpperCAmelCase_ : Any = self.config_class(**self.inputs_dict )
UpperCAmelCase_ : Optional[Any] = """test"""
with tempfile.TemporaryDirectory() as tmpdirname:
UpperCAmelCase_ : str = os.path.join(lowerCamelCase_ ,lowerCamelCase_ )
config_first.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.config_class.from_pretrained(lowerCamelCase_ ,subfolder=lowerCamelCase_ )
self.parent.assertEqual(config_second.to_dict() ,config_first.to_dict() )
def A__ ( self: int ) -> List[Any]:
UpperCAmelCase_ : Any = self.config_class(**self.inputs_dict ,num_labels=5 )
self.parent.assertEqual(len(config.idalabel ) ,5 )
self.parent.assertEqual(len(config.labelaid ) ,5 )
UpperCAmelCase_ : int = 3
self.parent.assertEqual(len(config.idalabel ) ,3 )
self.parent.assertEqual(len(config.labelaid ) ,3 )
def A__ ( self: List[str] ) -> int:
if self.config_class.is_composition:
return
UpperCAmelCase_ : str = self.config_class()
self.parent.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Optional[Any] = copy.deepcopy(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.config_class(**lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = []
for key, value in config_common_kwargs.items():
if key == "torch_dtype":
if not is_torch_available():
continue
else:
import torch
if config.torch_dtype != torch.floataa:
wrong_values.append(("""torch_dtype""", config.torch_dtype, torch.floataa) )
elif getattr(lowerCamelCase_ ,lowerCamelCase_ ) != value:
wrong_values.append((key, getattr(lowerCamelCase_ ,lowerCamelCase_ ), value) )
if len(lowerCamelCase_ ) > 0:
UpperCAmelCase_ : Optional[Any] = """\n""".join([F'''- {v[0]}: got {v[1]} instead of {v[2]}''' for v in wrong_values] )
raise ValueError(F'''The following keys were not properly set in the config:\n{errors}''' )
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
self.create_and_test_config_common_properties()
self.create_and_test_config_to_json_string()
self.create_and_test_config_to_json_file()
self.create_and_test_config_from_and_save_pretrained()
self.create_and_test_config_from_and_save_pretrained_subfolder()
self.create_and_test_config_with_num_labels()
self.check_config_can_be_init_without_params()
self.check_config_arguments_init()
| 345 |
import collections
import inspect
import unittest
from transformers import SwinvaConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import SwinvaForImageClassification, SwinvaForMaskedImageModeling, SwinvaModel
from transformers.models.swinva.modeling_swinva import SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: int=13 ,lowerCamelCase_: int=32 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: Any=3 ,lowerCamelCase_: str=16 ,lowerCamelCase_: Optional[Any]=[1, 2, 1] ,lowerCamelCase_: Tuple=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[Any]=2.0 ,lowerCamelCase_: str=True ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[Any]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Union[str, Any]=False ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=10 ,lowerCamelCase_: Tuple=8 ,) -> List[Any]:
UpperCAmelCase_ : List[str] = parent
UpperCAmelCase_ : int = batch_size
UpperCAmelCase_ : int = image_size
UpperCAmelCase_ : Union[str, Any] = patch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : int = embed_dim
UpperCAmelCase_ : Union[str, Any] = depths
UpperCAmelCase_ : List[str] = num_heads
UpperCAmelCase_ : int = window_size
UpperCAmelCase_ : List[str] = mlp_ratio
UpperCAmelCase_ : Tuple = qkv_bias
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : str = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = drop_path_rate
UpperCAmelCase_ : List[str] = hidden_act
UpperCAmelCase_ : int = use_absolute_embeddings
UpperCAmelCase_ : Any = patch_norm
UpperCAmelCase_ : Optional[int] = layer_norm_eps
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Dict = scope
UpperCAmelCase_ : int = use_labels
UpperCAmelCase_ : Optional[Any] = type_sequence_label_size
UpperCAmelCase_ : List[str] = encoder_stride
def A__ ( self: Any ) -> int:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : List[Any] = None
if self.use_labels:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : str = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Union[str, Any]:
return SwinvaConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,)
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ) -> str:
UpperCAmelCase_ : str = SwinvaModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : List[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Any = SwinvaForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.logits.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : str = 1
UpperCAmelCase_ : Optional[Any] = SwinvaForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : int = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ) -> int:
UpperCAmelCase_ : Union[str, Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = SwinvaForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: str ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = config_and_inputs
UpperCAmelCase_ : Optional[int] = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Tuple = (
(SwinvaModel, SwinvaForImageClassification, SwinvaForMaskedImageModeling) if is_torch_available() else ()
)
A__ : Optional[Any] = (
{"feature-extraction": SwinvaModel, "image-classification": SwinvaForImageClassification}
if is_torch_available()
else {}
)
A__ : List[Any] = False
A__ : Tuple = False
A__ : int = False
A__ : Union[str, Any] = False
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = SwinvaModelTester(self )
UpperCAmelCase_ : str = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 )
def A__ ( self: Optional[int] ) -> List[Any]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: Any ) -> Dict:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
@unittest.skip(reason="""Got `CUDA error: misaligned address` with PyTorch 2.0.0.""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""Swinv2 does not use inputs_embeds""" )
def A__ ( self: Tuple ) -> List[str]:
pass
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : Tuple = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Any = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : int = [*signature.parameters.keys()]
UpperCAmelCase_ : Tuple = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Any = True
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
UpperCAmelCase_ : Union[str, Any] = False
UpperCAmelCase_ : str = True
UpperCAmelCase_ : List[Any] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[Any] = outputs.attentions
UpperCAmelCase_ : List[str] = len(self.model_tester.depths )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
UpperCAmelCase_ : str = True
UpperCAmelCase_ : Optional[Any] = config.window_size**2
UpperCAmelCase_ : Optional[int] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[Any] = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
UpperCAmelCase_ : Optional[Any] = len(lowerCamelCase_ )
# Check attention is always last and order is fine
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = True
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
if hasattr(self.model_tester ,"""num_hidden_states_types""" ):
UpperCAmelCase_ : List[Any] = self.model_tester.num_hidden_states_types
else:
# also another +1 for reshaped_hidden_states
UpperCAmelCase_ : List[str] = 2
self.assertEqual(out_len + added_hidden_states ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
def A__ ( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : int = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[str] = outputs.hidden_states
UpperCAmelCase_ : Optional[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# Swinv2 has a different seq_length
UpperCAmelCase_ : Optional[Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : int = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Optional[int] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = reshaped_hidden_states[0].shape
UpperCAmelCase_ : Optional[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Dict = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes:
UpperCAmelCase_ : Any = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Union[str, Any] = 3
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : List[str] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : List[Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Optional[Any] = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : List[str] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@slow
def A__ ( self: str ) -> Tuple:
for model_name in SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Dict = SwinvaModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : List[str] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and "logit_scale" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Dict ) -> Optional[Any]:
return (
AutoImageProcessor.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" )
if is_vision_available()
else None
)
@slow
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ : Tuple = SwinvaForImageClassification.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" ).to(
lowerCamelCase_ )
UpperCAmelCase_ : Any = self.default_image_processor
UpperCAmelCase_ : List[str] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Optional[int] = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : Dict = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.tensor([-0.3_9_4_7, -0.4_3_0_6, 0.0_0_2_6] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
| 345 | 1 |
from typing import List, Optional, Tuple, Union
import torch
from ...utils import logging, randn_tensor
from ..pipeline_utils import AudioPipelineOutput, DiffusionPipeline
UpperCamelCase_ = logging.get_logger(__name__) # pylint: disable=invalid-name
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: Optional[int] ) -> Any:
super().__init__()
self.register_modules(unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
@torch.no_grad()
def __call__( self: Dict ,lowerCamelCase_: int = 1 ,lowerCamelCase_: int = 100 ,lowerCamelCase_: Optional[Union[torch.Generator, List[torch.Generator]]] = None ,lowerCamelCase_: Optional[float] = None ,lowerCamelCase_: bool = True ,) -> Union[AudioPipelineOutput, Tuple]:
if audio_length_in_s is None:
UpperCAmelCase_ : int = self.unet.config.sample_size / self.unet.config.sample_rate
UpperCAmelCase_ : Tuple = audio_length_in_s * self.unet.config.sample_rate
UpperCAmelCase_ : Tuple = 2 ** len(self.unet.up_blocks )
if sample_size < 3 * down_scale_factor:
raise ValueError(
F'''{audio_length_in_s} is too small. Make sure it\'s bigger or equal to'''
F''' {3 * down_scale_factor / self.unet.config.sample_rate}.''' )
UpperCAmelCase_ : List[str] = int(lowerCamelCase_ )
if sample_size % down_scale_factor != 0:
UpperCAmelCase_ : Dict = (
(audio_length_in_s * self.unet.config.sample_rate) // down_scale_factor + 1
) * down_scale_factor
logger.info(
F'''{audio_length_in_s} is increased to {sample_size / self.unet.config.sample_rate} so that it can be handled'''
F''' by the model. It will be cut to {original_sample_size / self.unet.config.sample_rate} after the denoising'''
""" process.""" )
UpperCAmelCase_ : Dict = int(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = next(iter(self.unet.parameters() ) ).dtype
UpperCAmelCase_ : int = (batch_size, self.unet.config.in_channels, sample_size)
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) and len(lowerCamelCase_ ) != batch_size:
raise ValueError(
F'''You have passed a list of generators of length {len(lowerCamelCase_ )}, but requested an effective batch'''
F''' size of {batch_size}. Make sure the batch size matches the length of the generators.''' )
UpperCAmelCase_ : Dict = randn_tensor(lowerCamelCase_ ,generator=lowerCamelCase_ ,device=self.device ,dtype=lowerCamelCase_ )
# set step values
self.scheduler.set_timesteps(lowerCamelCase_ ,device=audio.device )
UpperCAmelCase_ : Union[str, Any] = self.scheduler.timesteps.to(lowerCamelCase_ )
for t in self.progress_bar(self.scheduler.timesteps ):
# 1. predict noise model_output
UpperCAmelCase_ : str = self.unet(lowerCamelCase_ ,lowerCamelCase_ ).sample
# 2. compute previous image: x_t -> t_t-1
UpperCAmelCase_ : Union[str, Any] = self.scheduler.step(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ).prev_sample
UpperCAmelCase_ : Any = audio.clamp(-1 ,1 ).float().cpu().numpy()
UpperCAmelCase_ : List[str] = audio[:, :, :original_sample_size]
if not return_dict:
return (audio,)
return AudioPipelineOutput(audios=lowerCamelCase_ )
| 345 |
import json
import os
from functools import lru_cache
from typing import Dict, List, Optional, Tuple, Union
import regex as re
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...tokenization_utils_base import BatchEncoding, EncodedInput
from ...utils import PaddingStrategy, logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {'''vocab_file''': '''vocab.json''', '''merges_file''': '''merges.txt'''}
# See all LED models at https://huggingface.co/models?filter=LED
UpperCamelCase_ = {
'''vocab_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/vocab.json''',
},
'''merges_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/merges.txt''',
},
'''tokenizer_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/tokenizer.json''',
},
}
UpperCamelCase_ = {
'''allenai/led-base-16384''': 16384,
}
@lru_cache()
# Copied from transformers.models.bart.tokenization_bart.bytes_to_unicode
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : int = (
list(range(ord("""!""" ) , ord("""~""" ) + 1 ) ) + list(range(ord("""¡""" ) , ord("""¬""" ) + 1 ) ) + list(range(ord("""®""" ) , ord("""ÿ""" ) + 1 ) )
)
UpperCAmelCase_ : Dict = bs[:]
UpperCAmelCase_ : Any = 0
for b in range(2**8 ):
if b not in bs:
bs.append(_a )
cs.append(2**8 + n )
n += 1
UpperCAmelCase_ : Any = [chr(_a ) for n in cs]
return dict(zip(_a , _a ) )
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = set()
UpperCAmelCase_ : List[Any] = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
UpperCAmelCase_ : Optional[int] = char
return pairs
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = VOCAB_FILES_NAMES
A__ : List[str] = PRETRAINED_VOCAB_FILES_MAP
A__ : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Optional[int] = ["input_ids", "attention_mask"]
def __init__( self: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any]="replace" ,lowerCamelCase_: Optional[Any]="<s>" ,lowerCamelCase_: List[Any]="</s>" ,lowerCamelCase_: List[str]="</s>" ,lowerCamelCase_: int="<s>" ,lowerCamelCase_: int="<unk>" ,lowerCamelCase_: str="<pad>" ,lowerCamelCase_: Optional[Any]="<mask>" ,lowerCamelCase_: List[str]=False ,**lowerCamelCase_: Tuple ,) -> Any:
UpperCAmelCase_ : Union[str, Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else bos_token
UpperCAmelCase_ : int = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else eos_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else sep_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else cls_token
UpperCAmelCase_ : Optional[Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else unk_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else pad_token
# Mask token behave like a normal word, i.e. include the space before it
UpperCAmelCase_ : str = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else mask_token
super().__init__(
errors=lowerCamelCase_ ,bos_token=lowerCamelCase_ ,eos_token=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,add_prefix_space=lowerCamelCase_ ,**lowerCamelCase_ ,)
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as vocab_handle:
UpperCAmelCase_ : Union[str, Any] = json.load(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {v: k for k, v in self.encoder.items()}
UpperCAmelCase_ : Any = errors # how to handle errors in decoding
UpperCAmelCase_ : int = bytes_to_unicode()
UpperCAmelCase_ : Dict = {v: k for k, v in self.byte_encoder.items()}
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as merges_handle:
UpperCAmelCase_ : Any = merges_handle.read().split("""\n""" )[1:-1]
UpperCAmelCase_ : int = [tuple(merge.split() ) for merge in bpe_merges]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : Tuple = {}
UpperCAmelCase_ : Optional[int] = add_prefix_space
# Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions
UpperCAmelCase_ : int = re.compile(R"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""" )
@property
# Copied from transformers.models.bart.tokenization_bart.BartTokenizer.vocab_size
def A__ ( self: List[str] ) -> List[str]:
return len(self.encoder )
def A__ ( self: Any ) -> Union[str, Any]:
return dict(self.encoder ,**self.added_tokens_encoder )
def A__ ( self: Tuple ,lowerCamelCase_: Dict ) -> Optional[Any]:
if token in self.cache:
return self.cache[token]
UpperCAmelCase_ : Union[str, Any] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = get_pairs(lowerCamelCase_ )
if not pairs:
return token
while True:
UpperCAmelCase_ : Union[str, Any] = min(lowerCamelCase_ ,key=lambda lowerCamelCase_ : self.bpe_ranks.get(lowerCamelCase_ ,float("""inf""" ) ) )
if bigram not in self.bpe_ranks:
break
UpperCAmelCase_ , UpperCAmelCase_ : Any = bigram
UpperCAmelCase_ : Optional[Any] = []
UpperCAmelCase_ : List[str] = 0
while i < len(lowerCamelCase_ ):
try:
UpperCAmelCase_ : str = word.index(lowerCamelCase_ ,lowerCamelCase_ )
except ValueError:
new_word.extend(word[i:] )
break
else:
new_word.extend(word[i:j] )
UpperCAmelCase_ : Union[str, Any] = j
if word[i] == first and i < len(lowerCamelCase_ ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
UpperCAmelCase_ : List[str] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = new_word
if len(lowerCamelCase_ ) == 1:
break
else:
UpperCAmelCase_ : List[str] = get_pairs(lowerCamelCase_ )
UpperCAmelCase_ : int = """ """.join(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = word
return word
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> List[str]:
UpperCAmelCase_ : str = []
for token in re.findall(self.pat ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = """""".join(
self.byte_encoder[b] for b in token.encode("""utf-8""" ) ) # Maps all our bytes to unicode strings, avoiding control tokens of the BPE (spaces in our case)
bpe_tokens.extend(bpe_token for bpe_token in self.bpe(lowerCamelCase_ ).split(""" """ ) )
return bpe_tokens
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ) -> Optional[int]:
return self.encoder.get(lowerCamelCase_ ,self.encoder.get(self.unk_token ) )
def A__ ( self: List[str] ,lowerCamelCase_: str ) -> Optional[Any]:
return self.decoder.get(lowerCamelCase_ )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ) -> List[Any]:
UpperCAmelCase_ : str = """""".join(lowerCamelCase_ )
UpperCAmelCase_ : int = bytearray([self.byte_decoder[c] for c in text] ).decode("""utf-8""" ,errors=self.errors )
return text
def A__ ( self: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCamelCase_ ):
logger.error(F'''Vocabulary path ({save_directory}) should be a directory''' )
return
UpperCAmelCase_ : List[Any] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : List[str] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""merges_file"""] )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as f:
f.write(json.dumps(self.encoder ,indent=2 ,sort_keys=lowerCamelCase_ ,ensure_ascii=lowerCamelCase_ ) + """\n""" )
UpperCAmelCase_ : str = 0
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as writer:
writer.write("""#version: 0.2\n""" )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() ,key=lambda lowerCamelCase_ : kv[1] ):
if index != token_index:
logger.warning(
F'''Saving vocabulary to {merge_file}: BPE merge indices are not consecutive.'''
""" Please check that the tokenizer is not corrupted!""" )
UpperCAmelCase_ : Tuple = token_index
writer.write(""" """.join(lowerCamelCase_ ) + """\n""" )
index += 1
return vocab_file, merge_file
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
UpperCAmelCase_ : int = [self.cls_token_id]
UpperCAmelCase_ : Optional[int] = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ,lowerCamelCase_: bool = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=lowerCamelCase_ ,token_ids_a=lowerCamelCase_ ,already_has_special_tokens=lowerCamelCase_ )
if token_ids_a is None:
return [1] + ([0] * len(lowerCamelCase_ )) + [1]
return [1] + ([0] * len(lowerCamelCase_ )) + [1, 1] + ([0] * len(lowerCamelCase_ )) + [1]
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Optional[Any] = [self.sep_token_id]
UpperCAmelCase_ : Optional[Any] = [self.cls_token_id]
if token_ids_a is None:
return len(cls + token_ids_a + sep ) * [0]
return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0]
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str=False ,**lowerCamelCase_: List[str] ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = kwargs.pop("""add_prefix_space""" ,self.add_prefix_space )
if (is_split_into_words or add_prefix_space) and (len(lowerCamelCase_ ) > 0 and not text[0].isspace()):
UpperCAmelCase_ : Dict = """ """ + text
return (text, kwargs)
def A__ ( self: List[str] ,lowerCamelCase_: Union[Dict[str, EncodedInput], BatchEncoding] ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: PaddingStrategy = PaddingStrategy.DO_NOT_PAD ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: Optional[bool] = None ,) -> dict:
UpperCAmelCase_ : Optional[int] = super()._pad(
encoded_inputs=lowerCamelCase_ ,max_length=lowerCamelCase_ ,padding_strategy=lowerCamelCase_ ,pad_to_multiple_of=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,)
# Load from model defaults
if return_attention_mask is None:
UpperCAmelCase_ : str = """attention_mask""" in self.model_input_names
if return_attention_mask and "global_attention_mask" in encoded_inputs:
UpperCAmelCase_ : str = encoded_inputs[self.model_input_names[0]]
# `global_attention_mask` need to have the same length as other (sequential) inputs.
UpperCAmelCase_ : List[Any] = len(encoded_inputs["""global_attention_mask"""] ) != len(lowerCamelCase_ )
if needs_to_be_padded:
UpperCAmelCase_ : Dict = len(lowerCamelCase_ ) - len(encoded_inputs["""global_attention_mask"""] )
if self.padding_side == "right":
# Use `-1` since `0` in `global_attention_mask` means `local attention` instead of `not to attend`
UpperCAmelCase_ : str = (
encoded_inputs["""global_attention_mask"""] + [-1] * difference
)
elif self.padding_side == "left":
UpperCAmelCase_ : List[str] = [-1] * difference + encoded_inputs[
"""global_attention_mask"""
]
else:
raise ValueError("""Invalid padding strategy:""" + str(self.padding_side ) )
return encoded_inputs
| 345 | 1 |
import logging
from transformers.configuration_utils import PretrainedConfig
UpperCamelCase_ = logging.getLogger(__name__)
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "masked_bert"
def __init__( self: Optional[Any] ,lowerCamelCase_: Union[str, Any]=30522 ,lowerCamelCase_: Union[str, Any]=768 ,lowerCamelCase_: List[Any]=12 ,lowerCamelCase_: List[Any]=12 ,lowerCamelCase_: Tuple=3072 ,lowerCamelCase_: Optional[Any]="gelu" ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Union[str, Any]=512 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: Optional[int]=0.0_2 ,lowerCamelCase_: Any=1e-12 ,lowerCamelCase_: Optional[Any]=0 ,lowerCamelCase_: Dict="topK" ,lowerCamelCase_: str="constant" ,lowerCamelCase_: Optional[int]=0.0 ,**lowerCamelCase_: List[str] ,) -> str:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = vocab_size
UpperCAmelCase_ : List[Any] = hidden_size
UpperCAmelCase_ : Dict = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : Any = intermediate_size
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Optional[int] = attention_probs_dropout_prob
UpperCAmelCase_ : int = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : List[str] = initializer_range
UpperCAmelCase_ : Optional[Any] = layer_norm_eps
UpperCAmelCase_ : int = pruning_method
UpperCAmelCase_ : Optional[Any] = mask_init
UpperCAmelCase_ : Tuple = mask_scale
| 345 |
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Optional[Any] = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[Any] ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : int = UNetaDConditionModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=4 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,)
return model
@property
def A__ ( self: str ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = AutoencoderKL(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Tuple:
def extract(*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: str ):
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = torch.ones([0] )
def A__ ( self: List[Any] ,lowerCamelCase_: str ) -> int:
self.pixel_values.to(lowerCamelCase_ )
return self
return Out()
return extract
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : int = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : int = self.dummy_cond_unet
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(
beta_start=0.0_0_0_8_5 ,beta_end=0.0_1_2 ,beta_schedule="""scaled_linear""" ,clip_sample=lowerCamelCase_ ,set_alpha_to_one=lowerCamelCase_ ,)
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : int = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : List[Any] = output.images
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Dict = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : int = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Dict = self.dummy_cond_unet
UpperCAmelCase_ : List[Any] = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : Union[str, Any] = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : str = output.images
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(
"""hf-internal-testing/tiny-stable-diffusion-lms-pipe""" ,safety_checker=lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(pipe.scheduler ,lowerCamelCase_ )
assert pipe.safety_checker is None
UpperCAmelCase_ : List[Any] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(lowerCamelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
UpperCAmelCase_ : Optional[int] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Tuple = self.dummy_cond_unet
UpperCAmelCase_ : Dict = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# put models in fp16
UpperCAmelCase_ : Optional[Any] = unet.half()
UpperCAmelCase_ : Optional[int] = vae.half()
UpperCAmelCase_ : int = bert.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = sd_pipe([prompt] ,num_inference_steps=2 ,output_type="""np""" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Tuple = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : str = (
"""portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"""
""" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"""
""" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"""
""" children from bahnhof zoo, detailed """
)
UpperCAmelCase_ : Optional[int] = 4003660346
UpperCAmelCase_ : int = 7
# without safety guidance (sld_guidance_scale = 0)
UpperCAmelCase_ : Dict = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Optional[int] = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Dict = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
# without safety guidance (strong configuration)
UpperCAmelCase_ : Union[str, Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Any = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Tuple = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : str = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Any = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = """padme amidala taking a bath artwork, safe for work, no nudity"""
UpperCAmelCase_ : List[Any] = 2734971755
UpperCAmelCase_ : Optional[Any] = 7
UpperCAmelCase_ : int = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : Tuple = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
UpperCAmelCase_ : Any = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : List[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" )
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = (
"""the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."""
""" leyendecker"""
)
UpperCAmelCase_ : Optional[Any] = 1044355234
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : List[Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Any = output.images
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-7
UpperCAmelCase_ : Optional[int] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Any = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Any = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 345 | 1 |
import copy
import random
from transformers import CLIPTokenizer
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: str ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> str:
super().__init__(*lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = {}
def A__ ( self: str ,lowerCamelCase_: str ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : Union[str, Any] = super().add_tokens(lowerCamelCase_ ,*lowerCamelCase_ ,**lowerCamelCase_ )
if num_added_tokens == 0:
raise ValueError(
F'''The tokenizer already contains the token {placeholder_token}. Please pass a different'''
""" `placeholder_token` that is not already in the tokenizer.""" )
def A__ ( self: List[Any] ,lowerCamelCase_: Dict ,*lowerCamelCase_: List[str] ,lowerCamelCase_: List[Any]=1 ,**lowerCamelCase_: List[str] ) -> List[str]:
UpperCAmelCase_ : str = []
if num_vec_per_token == 1:
self.try_adding_tokens(lowerCamelCase_ ,*lowerCamelCase_ ,**lowerCamelCase_ )
output.append(lowerCamelCase_ )
else:
UpperCAmelCase_ : Optional[int] = []
for i in range(lowerCamelCase_ ):
UpperCAmelCase_ : int = placeholder_token + F'''_{i}'''
self.try_adding_tokens(lowerCamelCase_ ,*lowerCamelCase_ ,**lowerCamelCase_ )
output.append(lowerCamelCase_ )
# handle cases where there is a new placeholder token that contains the current placeholder token but is larger
for token in self.token_map:
if token in placeholder_token:
raise ValueError(
F'''The tokenizer already has placeholder token {token} that can get confused with'''
F''' {placeholder_token}keep placeholder tokens independent''' )
UpperCAmelCase_ : Optional[int] = output
def A__ ( self: int ,lowerCamelCase_: List[str] ,lowerCamelCase_: int=False ,lowerCamelCase_: Optional[Any]=1.0 ) -> Tuple:
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = []
for i in range(len(lowerCamelCase_ ) ):
output.append(self.replace_placeholder_tokens_in_text(text[i] ,vector_shuffle=lowerCamelCase_ ) )
return output
for placeholder_token in self.token_map:
if placeholder_token in text:
UpperCAmelCase_ : Tuple = self.token_map[placeholder_token]
UpperCAmelCase_ : List[str] = tokens[: 1 + int(len(lowerCamelCase_ ) * prop_tokens_to_load )]
if vector_shuffle:
UpperCAmelCase_ : Union[str, Any] = copy.copy(lowerCamelCase_ )
random.shuffle(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = text.replace(lowerCamelCase_ ,""" """.join(lowerCamelCase_ ) )
return text
def __call__( self: Any ,lowerCamelCase_: str ,*lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any]=False ,lowerCamelCase_: Union[str, Any]=1.0 ,**lowerCamelCase_: List[Any] ) -> Optional[Any]:
return super().__call__(
self.replace_placeholder_tokens_in_text(
lowerCamelCase_ ,vector_shuffle=lowerCamelCase_ ,prop_tokens_to_load=lowerCamelCase_ ) ,*lowerCamelCase_ ,**lowerCamelCase_ ,)
def A__ ( self: Tuple ,lowerCamelCase_: Any ,*lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Optional[Any]=False ,lowerCamelCase_: str=1.0 ,**lowerCamelCase_: Union[str, Any] ) -> str:
return super().encode(
self.replace_placeholder_tokens_in_text(
lowerCamelCase_ ,vector_shuffle=lowerCamelCase_ ,prop_tokens_to_load=lowerCamelCase_ ) ,*lowerCamelCase_ ,**lowerCamelCase_ ,)
| 345 |
import unittest
from transformers import MobileBertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
MobileBertModel,
)
class _snake_case :
'''simple docstring'''
def __init__( self: Optional[int] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Tuple=13 ,lowerCamelCase_: int=7 ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: str=True ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: int=99 ,lowerCamelCase_: List[str]=64 ,lowerCamelCase_: Tuple=32 ,lowerCamelCase_: List[str]=5 ,lowerCamelCase_: str=4 ,lowerCamelCase_: str=37 ,lowerCamelCase_: Union[str, Any]="gelu" ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: str=0.1 ,lowerCamelCase_: List[str]=512 ,lowerCamelCase_: Dict=16 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: List[str]=0.0_2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: Union[str, Any]=4 ,lowerCamelCase_: str=None ,) -> List[str]:
UpperCAmelCase_ : Any = parent
UpperCAmelCase_ : List[Any] = batch_size
UpperCAmelCase_ : Union[str, Any] = seq_length
UpperCAmelCase_ : Optional[int] = is_training
UpperCAmelCase_ : Dict = use_input_mask
UpperCAmelCase_ : Any = use_token_type_ids
UpperCAmelCase_ : Tuple = use_labels
UpperCAmelCase_ : List[Any] = vocab_size
UpperCAmelCase_ : str = hidden_size
UpperCAmelCase_ : List[str] = embedding_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : List[Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : str = hidden_dropout_prob
UpperCAmelCase_ : List[str] = attention_probs_dropout_prob
UpperCAmelCase_ : Any = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : Any = type_sequence_label_size
UpperCAmelCase_ : Optional[Any] = initializer_range
UpperCAmelCase_ : Optional[int] = num_labels
UpperCAmelCase_ : Optional[int] = num_choices
UpperCAmelCase_ : List[str] = scope
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : List[str] = None
if self.use_input_mask:
UpperCAmelCase_ : Tuple = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : Dict = None
if self.use_token_type_ids:
UpperCAmelCase_ : str = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : int = None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : Union[str, Any] = None
if self.use_labels:
UpperCAmelCase_ : List[str] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : int = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : Tuple = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: Any ) -> Dict:
return MobileBertConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,embedding_size=self.embedding_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> int:
UpperCAmelCase_ : Any = MobileBertModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Dict ) -> int:
UpperCAmelCase_ : Union[str, Any] = MobileBertForMaskedLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: str ,lowerCamelCase_: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : List[Any] = MobileBertForNextSentencePrediction(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 2) )
def A__ ( self: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Tuple = MobileBertForPreTraining(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,next_sentence_label=lowerCamelCase_ ,)
self.parent.assertEqual(result.prediction_logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape ,(self.batch_size, 2) )
def A__ ( self: Any ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = MobileBertForQuestionAnswering(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,start_positions=lowerCamelCase_ ,end_positions=lowerCamelCase_ ,)
self.parent.assertEqual(result.start_logits.shape ,(self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape ,(self.batch_size, self.seq_length) )
def A__ ( self: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> str:
UpperCAmelCase_ : Optional[Any] = self.num_labels
UpperCAmelCase_ : Union[str, Any] = MobileBertForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_labels) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Any:
UpperCAmelCase_ : str = self.num_labels
UpperCAmelCase_ : Optional[int] = MobileBertForTokenClassification(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Tuple ,lowerCamelCase_: str ,lowerCamelCase_: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Union[str, Any] = self.num_choices
UpperCAmelCase_ : Tuple = MobileBertForMultipleChoice(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = input_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Union[str, Any] = token_type_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : str = input_mask.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_choices) )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ : str = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Union[str, Any] = config_and_inputs
UpperCAmelCase_ : Dict = {"""input_ids""": input_ids, """token_type_ids""": token_type_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = (
(
MobileBertModel,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
)
if is_torch_available()
else ()
)
A__ : List[str] = (
{
"feature-extraction": MobileBertModel,
"fill-mask": MobileBertForMaskedLM,
"question-answering": MobileBertForQuestionAnswering,
"text-classification": MobileBertForSequenceClassification,
"token-classification": MobileBertForTokenClassification,
"zero-shot": MobileBertForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : List[str] = True
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: int=False ) -> Union[str, Any]:
UpperCAmelCase_ : List[Any] = super()._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ,return_labels=lowerCamelCase_ )
if return_labels:
if model_class in get_values(lowerCamelCase_ ):
UpperCAmelCase_ : Any = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = torch.zeros(
self.model_tester.batch_size ,dtype=torch.long ,device=lowerCamelCase_ )
return inputs_dict
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[str] = MobileBertModelTester(self )
UpperCAmelCase_ : Union[str, Any] = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Optional[Any] ) -> List[Any]:
self.config_tester.run_common_tests()
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_model(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_masked_lm(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Tuple:
UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_multiple_choice(*lowerCamelCase_ )
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_next_sequence_prediction(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_pretraining(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_question_answering(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_sequence_classification(*lowerCamelCase_ )
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_token_classification(*lowerCamelCase_ )
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return torch.tensor(
_a , dtype=torch.long , device=_a , )
UpperCamelCase_ = 1E-3
@require_torch
@require_sentencepiece
@require_tokenizers
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: List[Any] ) -> str:
UpperCAmelCase_ : Any = MobileBertModel.from_pretrained("""google/mobilebert-uncased""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = _long_tensor([[101, 7110, 1005, 1056, 2023, 11333, 17413, 1029, 102]] )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )[0]
UpperCAmelCase_ : Union[str, Any] = torch.Size((1, 9, 512) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = torch.tensor(
[
[
[-2.473_6526e07, 8.269_1656e04, 1.652_1838e05],
[-5.754_1704e-01, 3.905_6022e00, 4.401_1507e00],
[2.604_7359e00, 1.567_7652e00, -1.732_4188e-01],
]
] ,device=lowerCamelCase_ ,)
# MobileBERT results range from 10e0 to 10e8. Even a 0.0000001% difference with a value of 10e8 results in a
# ~1 difference, it's therefore not a good idea to measure using addition.
# Here, we instead divide the expected result with the result in order to obtain ~1. We then check that the
# result is held between bounds: 1 - TOLERANCE < expected_result / result < 1 + TOLERANCE
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) >= 1 - TOLERANCE )
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) <= 1 + TOLERANCE )
self.assertTrue(lower_bound and upper_bound )
| 345 | 1 |
import re
import jax.numpy as jnp
from flax.traverse_util import flatten_dict, unflatten_dict
from jax.random import PRNGKey
from ..utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
def lowerCamelCase_ ( _a : Optional[Any] ):
'''simple docstring'''
UpperCAmelCase_ : str = r"""\w+[.]\d+"""
UpperCAmelCase_ : List[Any] = re.findall(_a , _a )
for pat in pats:
UpperCAmelCase_ : Any = key.replace(_a , """_""".join(pat.split(""".""" ) ) )
return key
def lowerCamelCase_ ( _a : List[str] , _a : Optional[Any] , _a : int ):
'''simple docstring'''
UpperCAmelCase_ : str = pt_tuple_key[:-1] + ("""scale""",)
if (
any("""norm""" in str_ for str_ in pt_tuple_key )
and (pt_tuple_key[-1] == "bias")
and (pt_tuple_key[:-1] + ("bias",) not in random_flax_state_dict)
and (pt_tuple_key[:-1] + ("scale",) in random_flax_state_dict)
):
UpperCAmelCase_ : Tuple = pt_tuple_key[:-1] + ("""scale""",)
return renamed_pt_tuple_key, pt_tensor
elif pt_tuple_key[-1] in ["weight", "gamma"] and pt_tuple_key[:-1] + ("scale",) in random_flax_state_dict:
UpperCAmelCase_ : Union[str, Any] = pt_tuple_key[:-1] + ("""scale""",)
return renamed_pt_tuple_key, pt_tensor
# embedding
if pt_tuple_key[-1] == "weight" and pt_tuple_key[:-1] + ("embedding",) in random_flax_state_dict:
UpperCAmelCase_ : Optional[Any] = pt_tuple_key[:-1] + ("""embedding""",)
return renamed_pt_tuple_key, pt_tensor
# conv layer
UpperCAmelCase_ : str = pt_tuple_key[:-1] + ("""kernel""",)
if pt_tuple_key[-1] == "weight" and pt_tensor.ndim == 4:
UpperCAmelCase_ : List[Any] = pt_tensor.transpose(2 , 3 , 1 , 0 )
return renamed_pt_tuple_key, pt_tensor
# linear layer
UpperCAmelCase_ : Union[str, Any] = pt_tuple_key[:-1] + ("""kernel""",)
if pt_tuple_key[-1] == "weight":
UpperCAmelCase_ : Optional[Any] = pt_tensor.T
return renamed_pt_tuple_key, pt_tensor
# old PyTorch layer norm weight
UpperCAmelCase_ : str = pt_tuple_key[:-1] + ("""weight""",)
if pt_tuple_key[-1] == "gamma":
return renamed_pt_tuple_key, pt_tensor
# old PyTorch layer norm bias
UpperCAmelCase_ : Any = pt_tuple_key[:-1] + ("""bias""",)
if pt_tuple_key[-1] == "beta":
return renamed_pt_tuple_key, pt_tensor
return pt_tuple_key, pt_tensor
def lowerCamelCase_ ( _a : Optional[int] , _a : Union[str, Any] , _a : List[str]=42 ):
'''simple docstring'''
UpperCAmelCase_ : Any = {k: v.numpy() for k, v in pt_state_dict.items()}
# Step 2: Since the model is stateless, get random Flax params
UpperCAmelCase_ : str = flax_model.init_weights(PRNGKey(_a ) )
UpperCAmelCase_ : List[Any] = flatten_dict(_a )
UpperCAmelCase_ : Tuple = {}
# Need to change some parameters name to match Flax names
for pt_key, pt_tensor in pt_state_dict.items():
UpperCAmelCase_ : int = rename_key(_a )
UpperCAmelCase_ : Any = tuple(renamed_pt_key.split(""".""" ) )
# Correctly rename weight parameters
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = rename_key_and_reshape_tensor(_a , _a , _a )
if flax_key in random_flax_state_dict:
if flax_tensor.shape != random_flax_state_dict[flax_key].shape:
raise ValueError(
F'''PyTorch checkpoint seems to be incorrect. Weight {pt_key} was expected to be of shape '''
F'''{random_flax_state_dict[flax_key].shape}, but is {flax_tensor.shape}.''' )
# also add unexpected weight so that warning is thrown
UpperCAmelCase_ : Tuple = jnp.asarray(_a )
return unflatten_dict(_a )
| 345 |
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import AutoTokenizer, BarkProcessor
from transformers.testing_utils import require_torch, slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: str ) -> int:
UpperCAmelCase_ : List[Any] = """ylacombe/bark-small"""
UpperCAmelCase_ : Tuple = tempfile.mkdtemp()
UpperCAmelCase_ : Union[str, Any] = """en_speaker_1"""
UpperCAmelCase_ : Optional[Any] = """This is a test string"""
UpperCAmelCase_ : int = """speaker_embeddings_path.json"""
UpperCAmelCase_ : Any = """speaker_embeddings"""
def A__ ( self: Tuple ,**lowerCamelCase_: List[str] ) -> List[Any]:
return AutoTokenizer.from_pretrained(self.checkpoint ,**lowerCamelCase_ )
def A__ ( self: str ) -> Union[str, Any]:
shutil.rmtree(self.tmpdirname )
def A__ ( self: List[Any] ) -> int:
UpperCAmelCase_ : int = self.get_tokenizer()
UpperCAmelCase_ : Tuple = BarkProcessor(tokenizer=lowerCamelCase_ )
processor.save_pretrained(self.tmpdirname )
UpperCAmelCase_ : Optional[int] = BarkProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer.get_vocab() )
@slow
def A__ ( self: List[Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
processor.save_pretrained(
self.tmpdirname ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,speaker_embeddings_directory=self.speaker_embeddings_directory ,)
UpperCAmelCase_ : Optional[Any] = self.get_tokenizer(bos_token="""(BOS)""" ,eos_token="""(EOS)""" )
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
self.tmpdirname ,self.speaker_embeddings_dict_path ,bos_token="""(BOS)""" ,eos_token="""(EOS)""" ,)
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer_add_kwargs.get_vocab() )
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
UpperCAmelCase_ : Optional[int] = 35
UpperCAmelCase_ : Optional[int] = 2
UpperCAmelCase_ : Dict = 8
UpperCAmelCase_ : Optional[int] = {
"""semantic_prompt""": np.ones(lowerCamelCase_ ),
"""coarse_prompt""": np.ones((nb_codebooks_coarse, seq_len) ),
"""fine_prompt""": np.ones((nb_codebooks_total, seq_len) ),
}
# test providing already loaded voice_preset
UpperCAmelCase_ : str = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from npz file
UpperCAmelCase_ : List[Any] = os.path.join(self.tmpdirname ,"""file.npz""" )
np.savez(lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : int = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from the hub
UpperCAmelCase_ : Union[str, Any] = processor(text=self.input_string ,voice_preset=self.voice_preset )
def A__ ( self: Dict ) -> Tuple:
UpperCAmelCase_ : Any = self.get_tokenizer()
UpperCAmelCase_ : Dict = BarkProcessor(tokenizer=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string )
UpperCAmelCase_ : str = tokenizer(
self.input_string ,padding="""max_length""" ,max_length=256 ,add_special_tokens=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,return_token_type_ids=lowerCamelCase_ ,)
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] ,encoded_processor[key].squeeze().tolist() )
| 345 | 1 |
import argparse
import json
import os
import re
import torch
from transformers import BloomConfig, BloomModel
from transformers.file_utils import CONFIG_NAME, WEIGHTS_NAME
from transformers.utils import logging
logging.set_verbosity_info()
UpperCamelCase_ = [
'''word_embeddings_layernorm.weight''',
'''word_embeddings_layernorm.bias''',
'''input_layernorm.weight''',
'''input_layernorm.bias''',
'''post_attention_layernorm.weight''',
'''post_attention_layernorm.bias''',
'''self_attention.dense.bias''',
'''mlp.dense_4h_to_h.bias''',
'''ln_f.weight''',
'''ln_f.bias''',
]
UpperCamelCase_ = [
'''mlp.dense_4h_to_h.weight''',
'''self_attention.dense.weight''',
]
def lowerCamelCase_ ( _a : List[Any] , _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Dict = {
"""word_embeddings.weight""": """word_embeddings.weight""",
"""word_embeddings.norm.weight""": """word_embeddings_layernorm.weight""",
"""word_embeddings.norm.bias""": """word_embeddings_layernorm.bias""",
"""weight""": """ln_f.weight""",
"""bias""": """ln_f.bias""",
}
if key in layer_rename_map:
return layer_rename_map[key]
# Handle transformer blocks
UpperCAmelCase_ : Union[str, Any] = int(re.match(r""".*layer_(\d*).*""" , _a )[1] )
layer_number -= 3
return F'''h.{layer_number}.''' + key
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
if dtype == torch.bool:
return 1 / 8
UpperCAmelCase_ : str = re.search(r"""[^\d](\d+)$""" , str(_a ) )
if bit_search is None:
raise ValueError(F'''`dtype` is not a valid dtype: {dtype}.''' )
UpperCAmelCase_ : int = int(bit_search.groups()[0] )
return bit_size // 8
def lowerCamelCase_ ( _a : Optional[int] , _a : Dict , _a : Any , _a : Any , _a : str ):
'''simple docstring'''
if bloom_config_file == "":
UpperCAmelCase_ : Dict = BloomConfig()
else:
UpperCAmelCase_ : Optional[int] = BloomConfig.from_json_file(_a )
if shard_model:
UpperCAmelCase_ : Dict = os.listdir(_a )
UpperCAmelCase_ : List[str] = sorted(filter(lambda _a : s.startswith("""layer""" ) and "model_00" in s , _a ) )
UpperCAmelCase_ : Optional[Any] = {"""weight_map""": {}, """metadata""": {}}
UpperCAmelCase_ : Dict = 0
UpperCAmelCase_ : List[str] = None
UpperCAmelCase_ : int = BloomConfig()
for j, file in enumerate(_a ):
print("""Processing file: {}""".format(_a ) )
UpperCAmelCase_ : Any = None
for i in range(_a ):
# load all TP files
UpperCAmelCase_ : List[str] = file.replace("""model_00""" , F'''model_0{i}''' )
UpperCAmelCase_ : Any = torch.load(os.path.join(_a , _a ) , map_location="""cpu""" )
# Rename keys in the transformers names
UpperCAmelCase_ : List[Any] = list(temp.keys() )
for key in keys:
UpperCAmelCase_ : List[Any] = temp.pop(_a )
if tensors is None:
UpperCAmelCase_ : Union[str, Any] = temp
else:
for key in tensors.keys():
if any(key.endswith(_a ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
# We average (sum and then divide) some weights accross TP ranks (see https://github.com/bigscience-workshop/Megatron-DeepSpeed/blob/olruwase/sync_layer_norms/megatron/training.py#L425)
tensors[key] += temp[key]
else:
# Some weights are RowParallelLinear in Megatron-Deepspeed, others are ColumnParallel
UpperCAmelCase_ : List[str] = 1 if any(text in key for text in WEIGHTS_WITH_ROW_PARALLELISM_CONTAIN ) else 0
# We concatenate these weights accross TP ranks
UpperCAmelCase_ : Any = torch.cat([tensors[key], temp[key]] , dim=_a )
# Divide by the number of TP the weights we want to average
for key in tensors.keys():
if any(key.endswith(_a ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
UpperCAmelCase_ : int = tensors[key] / pretraining_tp
torch.save(
_a , os.path.join(
_a , """pytorch_model_{}-of-{}.bin""".format(str(j + 1 ).zfill(5 ) , str(len(_a ) ).zfill(5 ) ) , ) , )
for key in tensors.keys():
UpperCAmelCase_ : Optional[Any] = tensors[key]
total_size += value.numel() * get_dtype_size(value.dtype )
if key not in index_dict["weight_map"]:
UpperCAmelCase_ : Dict = """pytorch_model_{}-of-{}.bin""".format(
str(j + 1 ).zfill(5 ) , str(len(_a ) ).zfill(5 ) )
UpperCAmelCase_ : int = BloomConfig()
UpperCAmelCase_ : Tuple = pytorch_dump_folder_path + """/""" + CONFIG_NAME
UpperCAmelCase_ : Any = total_size
with open(_a , """w""" , encoding="""utf-8""" ) as f:
f.write(config.to_json_string() )
with open(os.path.join(_a , WEIGHTS_NAME + """.index.json""" ) , """w""" , encoding="""utf-8""" ) as f:
UpperCAmelCase_ : Optional[int] = json.dumps(_a , indent=2 , sort_keys=_a ) + """\n"""
f.write(_a )
else:
UpperCAmelCase_ : str = BloomModel(_a )
UpperCAmelCase_ : Any = os.listdir(_a )
UpperCAmelCase_ : Union[str, Any] = sorted(filter(lambda _a : s.startswith("""layer""" ) and "model_00" in s , _a ) )
UpperCAmelCase_ : Optional[Any] = None
for i, file in enumerate(_a ):
UpperCAmelCase_ : Dict = None
for i in range(_a ):
# load all TP files
UpperCAmelCase_ : Optional[Any] = file.replace("""model_00""" , F'''model_0{i}''' )
UpperCAmelCase_ : List[str] = torch.load(os.path.join(_a , _a ) , map_location="""cpu""" )
# Rename keys in the transformers names
UpperCAmelCase_ : Union[str, Any] = list(temp.keys() )
for key in keys:
UpperCAmelCase_ : Union[str, Any] = temp.pop(_a )
if tensors is None:
UpperCAmelCase_ : Optional[int] = temp
else:
for key in tensors.keys():
# We average (sum and then divide) some weights accross TP ranks (see https://github.com/bigscience-workshop/Megatron-DeepSpeed/blob/olruwase/sync_layer_norms/megatron/training.py#L425)
if any(key.endswith(_a ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
tensors[key] += temp[key]
else:
# Some weights are RowParallelLinear in Megatron-Deepspeed, others are ColumnParallel
UpperCAmelCase_ : Any = 1 if any(text in key for text in WEIGHTS_WITH_ROW_PARALLELISM_CONTAIN ) else 0
# We concatenate these weights accross TP ranks
UpperCAmelCase_ : str = torch.cat([tensors[key], temp[key]] , dim=_a )
# Divide by the number of TP the weights we want to average
for key in tensors.keys():
if any(key.endswith(_a ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
UpperCAmelCase_ : str = tensors[key] / pretraining_tp
UpperCAmelCase_ : List[Any] = model.load_state_dict(_a , strict=_a )
assert not other_keys.unexpected_keys, F'''The keys {other_keys.unexpected_keys} are unexpected'''
if missing_keys is None:
UpperCAmelCase_ : Optional[int] = set(other_keys.missing_keys )
else:
UpperCAmelCase_ : Union[str, Any] = missing_keys.intersection(set(other_keys.missing_keys ) )
assert not missing_keys, F'''The keys {missing_keys} are missing'''
# Save pytorch-model
os.makedirs(_a , exist_ok=_a )
UpperCAmelCase_ : Union[str, Any] = pytorch_dump_folder_path + """/""" + WEIGHTS_NAME
UpperCAmelCase_ : int = pytorch_dump_folder_path + """/""" + CONFIG_NAME
print(F'''Save PyTorch model to {pytorch_weights_dump_path} with dtype {config.torch_dtype}''' )
if config.torch_dtype is not None:
UpperCAmelCase_ : Optional[Any] = model.to(config.torch_dtype )
torch.save(model.state_dict() , _a )
print(F'''Save configuration file to {pytorch_config_dump_path}''' )
with open(_a , """w""" , encoding="""utf-8""" ) as f:
f.write(config.to_json_string() )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--bloom_checkpoint_path''',
default=None,
type=str,
required=True,
help='''Path to the Megatron-LM checkpoint path.''',
)
parser.add_argument(
'''--pytorch_dump_folder_path''', default=None, type=str, required=True, help='''Path to the output PyTorch model.'''
)
parser.add_argument(
'''--bloom_config_file''',
default='''''',
type=str,
help=(
'''An optional config json file corresponding to the pre-trained model. \n'''
'''This specifies the model architecture.'''
),
)
parser.add_argument(
'''--shard_model''',
action='''store_true''',
help='''An optional setting to shard the output model \nThis enables sharding the converted checkpoint''',
)
parser.add_argument(
'''--pretraining_tp''',
default=4,
type=int,
help='''Pretraining TP rank that has been used when training the model in Megatron-LM \n''',
)
UpperCamelCase_ = parser.parse_args()
convert_bloom_checkpoint_to_pytorch(
args.bloom_checkpoint_path,
args.bloom_config_file,
args.pytorch_dump_folder_path,
args.shard_model,
args.pretraining_tp,
)
| 345 |
import unittest
from queue import Empty
from threading import Thread
from transformers import AutoTokenizer, TextIteratorStreamer, TextStreamer, is_torch_available
from transformers.testing_utils import CaptureStdout, require_torch, torch_device
from ..test_modeling_common import ids_tensor
if is_torch_available():
import torch
from transformers import AutoModelForCausalLM
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Union[str, Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = -1
UpperCAmelCase_ : Dict = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Any = tokenizer.decode(greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : Optional[int] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Dict ) -> Optional[Any]:
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : List[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Dict = tokenizer.decode(greedy_ids[0] )
UpperCAmelCase_ : str = TextIteratorStreamer(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : str = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
UpperCAmelCase_ : int = """"""
for new_text in streamer:
streamer_text += new_text
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : Tuple = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : str = greedy_ids[:, input_ids.shape[1] :]
UpperCAmelCase_ : Dict = tokenizer.decode(new_greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ ,skip_prompt=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : List[str] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: str ) -> str:
# Tests that we can pass `decode_kwargs` to the streamer to control how the tokens are decoded. Must be tested
# with actual models -- the dummy models' tokenizers are not aligned with their models, and
# `skip_special_tokens=True` has no effect on them
UpperCAmelCase_ : Union[str, Any] = AutoTokenizer.from_pretrained("""distilgpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""distilgpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = -1
UpperCAmelCase_ : Union[str, Any] = torch.ones((1, 5) ,device=lowerCamelCase_ ).long() * model.config.bos_token_id
with CaptureStdout() as cs:
UpperCAmelCase_ : Union[str, Any] = TextStreamer(lowerCamelCase_ ,skip_special_tokens=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=1 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The prompt contains a special token, so the streamer should not print it. As such, the output text, when
# re-tokenized, must only contain one token
UpperCAmelCase_ : List[str] = cs.out[:-1] # Remove the final "\n"
UpperCAmelCase_ : Dict = tokenizer(lowerCamelCase_ ,return_tensors="""pt""" )
self.assertEqual(streamer_text_tokenized.input_ids.shape ,(1, 1) )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Any = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = -1
UpperCAmelCase_ : Optional[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = TextIteratorStreamer(lowerCamelCase_ ,timeout=0.0_0_1 )
UpperCAmelCase_ : Any = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : Dict = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
# The streamer will timeout after 0.001 seconds, so an exception will be raised
with self.assertRaises(lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = """"""
for new_text in streamer:
streamer_text += new_text
| 345 | 1 |
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
if n == 1 or not isinstance(_a , _a ):
return 0
elif n == 2:
return 1
else:
UpperCAmelCase_ : Union[str, Any] = [0, 1]
for i in range(2 , n + 1 ):
sequence.append(sequence[i - 1] + sequence[i - 2] )
return sequence[n]
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = 0
UpperCAmelCase_ : Any = 2
while digits < n:
index += 1
UpperCAmelCase_ : Any = len(str(fibonacci(_a ) ) )
return index
def lowerCamelCase_ ( _a : int = 1000 ):
'''simple docstring'''
return fibonacci_digits_index(_a )
if __name__ == "__main__":
print(solution(int(str(input()).strip())))
| 345 |
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel
from diffusers import DDIMScheduler, LDMPipeline, UNetaDModel, VQModel
from diffusers.utils.testing_utils import enable_full_determinism, require_torch, slow, torch_device
enable_full_determinism()
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = UNetaDModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=3 ,out_channels=3 ,down_block_types=("""DownBlock2D""", """AttnDownBlock2D""") ,up_block_types=("""AttnUpBlock2D""", """UpBlock2D""") ,)
return model
@property
def A__ ( self: Tuple ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : List[str] = VQModel(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=3 ,)
return model
@property
def A__ ( self: Tuple ) -> Any:
torch.manual_seed(0 )
UpperCAmelCase_ : int = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
def A__ ( self: str ) -> Optional[Any]:
UpperCAmelCase_ : str = self.dummy_uncond_unet
UpperCAmelCase_ : List[Any] = DDIMScheduler()
UpperCAmelCase_ : List[Any] = self.dummy_vq_model
UpperCAmelCase_ : Optional[int] = LDMPipeline(unet=lowerCamelCase_ ,vqvae=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.manual_seed(0 )
UpperCAmelCase_ : int = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ).images
UpperCAmelCase_ : List[str] = torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ,return_dict=lowerCamelCase_ )[0]
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : str = np.array([0.8_5_1_2, 0.8_1_8, 0.6_4_1_1, 0.6_8_0_8, 0.4_4_6_5, 0.5_6_1_8, 0.4_6, 0.6_2_3_1, 0.5_1_7_2] )
UpperCAmelCase_ : Tuple = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < tolerance
@slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = LDMPipeline.from_pretrained("""CompVis/ldm-celebahq-256""" )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = ldm(generator=lowerCamelCase_ ,num_inference_steps=5 ,output_type="""numpy""" ).images
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
assert image.shape == (1, 256, 256, 3)
UpperCAmelCase_ : int = np.array([0.4_3_9_9, 0.4_4_9_7_5, 0.4_6_8_2_5, 0.4_7_4, 0.4_3_5_9, 0.4_5_8_1, 0.4_5_0_9_5, 0.4_3_4_1, 0.4_4_4_7] )
UpperCAmelCase_ : Union[str, Any] = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
| 345 | 1 |
import torch
def lowerCamelCase_ ( ):
'''simple docstring'''
if torch.cuda.is_available():
UpperCAmelCase_ : Tuple = torch.cuda.device_count()
else:
UpperCAmelCase_ : Optional[int] = 0
print(F'''Successfully ran on {num_gpus} GPUs''' )
if __name__ == "__main__":
main()
| 345 |
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = [0] * len(_a )
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Dict = 0
for values in graph.values():
for i in values:
indegree[i] += 1
for i in range(len(_a ) ):
if indegree[i] == 0:
queue.append(_a )
while queue:
UpperCAmelCase_ : List[str] = queue.pop(0 )
cnt += 1
topo.append(_a )
for x in graph[vertex]:
indegree[x] -= 1
if indegree[x] == 0:
queue.append(_a )
if cnt != len(_a ):
print("""Cycle exists""" )
else:
print(_a )
# Adjacency List of Graph
UpperCamelCase_ = {0: [1, 2], 1: [3], 2: [3], 3: [4, 5], 4: [], 5: []}
topological_sort(graph)
| 345 | 1 |
import timeit
import numpy as np
import datasets
from datasets.arrow_writer import ArrowWriter
from datasets.features.features import _ArrayXD
def lowerCamelCase_ ( _a : Tuple ):
'''simple docstring'''
def wrapper(*_a : Tuple , **_a : Union[str, Any] ):
UpperCAmelCase_ : str = timeit.default_timer()
UpperCAmelCase_ : List[Any] = func(*_a , **_a )
UpperCAmelCase_ : str = timeit.default_timer() - starttime
return delta
UpperCAmelCase_ : Any = func.__name__
return wrapper
def lowerCamelCase_ ( _a : dict , _a : Union[str, Any]=100 , _a : List[Any]=None ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : Optional[Any] = seq_shapes or {}
for i in range(_a ):
UpperCAmelCase_ : List[Any] = {}
for col_id, (k, v) in enumerate(features.items() ):
if isinstance(_a , _ArrayXD ):
UpperCAmelCase_ : Optional[Any] = np.random.rand(*v.shape ).astype(v.dtype )
elif isinstance(_a , datasets.Value ):
if v.dtype == "string":
UpperCAmelCase_ : Dict = """The small grey turtle was surprisingly fast when challenged."""
else:
UpperCAmelCase_ : Optional[Any] = np.random.randint(10 , size=1 ).astype(v.dtype ).item()
elif isinstance(_a , datasets.Sequence ):
while isinstance(_a , datasets.Sequence ):
UpperCAmelCase_ : Union[str, Any] = v.feature
UpperCAmelCase_ : int = seq_shapes[k]
UpperCAmelCase_ : Dict = np.random.rand(*_a ).astype(v.dtype )
UpperCAmelCase_ : List[str] = data
dummy_data.append((i, example) )
return dummy_data
def lowerCamelCase_ ( _a : int , _a : Tuple , _a : Optional[Any]=100 , _a : int=None ):
'''simple docstring'''
UpperCAmelCase_ : Dict = generate_examples(_a , num_examples=_a , seq_shapes=_a )
with ArrowWriter(features=_a , path=_a ) as writer:
for key, record in dummy_data:
UpperCAmelCase_ : str = features.encode_example(_a )
writer.write(_a )
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = writer.finalize()
if not num_final_examples == num_examples:
raise ValueError(
F'''Error writing the dataset, wrote {num_final_examples} examples but should have written {num_examples}.''' )
UpperCAmelCase_ : List[str] = datasets.Dataset.from_file(filename=_a , info=datasets.DatasetInfo(features=_a ) )
return dataset
| 345 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/swinv2-tiny-patch4-window8-256''': (
'''https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "swinv2"
A__ : int = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self: List[str] ,lowerCamelCase_: List[str]=224 ,lowerCamelCase_: List[str]=4 ,lowerCamelCase_: List[Any]=3 ,lowerCamelCase_: Optional[Any]=96 ,lowerCamelCase_: Any=[2, 2, 6, 2] ,lowerCamelCase_: Dict=[3, 6, 12, 24] ,lowerCamelCase_: str=7 ,lowerCamelCase_: Optional[Any]=4.0 ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: str=False ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: Union[str, Any]=1e-5 ,lowerCamelCase_: str=32 ,**lowerCamelCase_: List[str] ,) -> Tuple:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = image_size
UpperCAmelCase_ : Tuple = patch_size
UpperCAmelCase_ : Dict = num_channels
UpperCAmelCase_ : List[Any] = embed_dim
UpperCAmelCase_ : Dict = depths
UpperCAmelCase_ : Dict = len(lowerCamelCase_ )
UpperCAmelCase_ : str = num_heads
UpperCAmelCase_ : Tuple = window_size
UpperCAmelCase_ : int = mlp_ratio
UpperCAmelCase_ : str = qkv_bias
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Tuple = attention_probs_dropout_prob
UpperCAmelCase_ : int = drop_path_rate
UpperCAmelCase_ : Optional[Any] = hidden_act
UpperCAmelCase_ : List[str] = use_absolute_embeddings
UpperCAmelCase_ : Dict = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Union[str, Any] = encoder_stride
# we set the hidden_size attribute in order to make Swinv2 work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
UpperCAmelCase_ : List[str] = int(embed_dim * 2 ** (len(lowerCamelCase_ ) - 1) )
UpperCAmelCase_ : Any = (0, 0, 0, 0)
| 345 | 1 |
from importlib import import_module
from .logging import get_logger
UpperCamelCase_ = get_logger(__name__)
class _snake_case :
'''simple docstring'''
def __init__( self: str ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any=None ) -> int:
UpperCAmelCase_ : int = attrs or []
if module is not None:
for key in module.__dict__:
if key in attrs or not key.startswith("""__""" ):
setattr(self ,lowerCamelCase_ ,getattr(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[Any] = module._original_module if isinstance(lowerCamelCase_ ,_PatchedModuleObj ) else module
class _snake_case :
'''simple docstring'''
A__ : List[Any] = []
def __init__( self: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple=None ) -> Dict:
UpperCAmelCase_ : Optional[Any] = obj
UpperCAmelCase_ : Optional[Any] = target
UpperCAmelCase_ : List[str] = new
UpperCAmelCase_ : Union[str, Any] = target.split(""".""" )[0]
UpperCAmelCase_ : List[Any] = {}
UpperCAmelCase_ : Any = attrs or []
def __enter__( self: Tuple ) -> Dict:
*UpperCAmelCase_ , UpperCAmelCase_ : int = self.target.split(""".""" )
# Patch modules:
# it's used to patch attributes of submodules like "os.path.join";
# in this case we need to patch "os" and "os.path"
for i in range(len(lowerCamelCase_ ) ):
try:
UpperCAmelCase_ : int = import_module(""".""".join(submodules[: i + 1] ) )
except ModuleNotFoundError:
continue
# We iterate over all the globals in self.obj in case we find "os" or "os.path"
for attr in self.obj.__dir__():
UpperCAmelCase_ : Any = getattr(self.obj ,lowerCamelCase_ )
# We don't check for the name of the global, but rather if its value *is* "os" or "os.path".
# This allows to patch renamed modules like "from os import path as ospath".
if obj_attr is submodule or (
(isinstance(lowerCamelCase_ ,_PatchedModuleObj ) and obj_attr._original_module is submodule)
):
UpperCAmelCase_ : str = obj_attr
# patch at top level
setattr(self.obj ,lowerCamelCase_ ,_PatchedModuleObj(lowerCamelCase_ ,attrs=self.attrs ) )
UpperCAmelCase_ : str = getattr(self.obj ,lowerCamelCase_ )
# construct lower levels patches
for key in submodules[i + 1 :]:
setattr(lowerCamelCase_ ,lowerCamelCase_ ,_PatchedModuleObj(getattr(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ) ,attrs=self.attrs ) )
UpperCAmelCase_ : str = getattr(lowerCamelCase_ ,lowerCamelCase_ )
# finally set the target attribute
setattr(lowerCamelCase_ ,lowerCamelCase_ ,self.new )
# Patch attribute itself:
# it's used for builtins like "open",
# and also to patch "os.path.join" we may also need to patch "join"
# itself if it was imported as "from os.path import join".
if submodules: # if it's an attribute of a submodule like "os.path.join"
try:
UpperCAmelCase_ : Dict = getattr(import_module(""".""".join(lowerCamelCase_ ) ) ,lowerCamelCase_ )
except (AttributeError, ModuleNotFoundError):
return
# We iterate over all the globals in self.obj in case we find "os.path.join"
for attr in self.obj.__dir__():
# We don't check for the name of the global, but rather if its value *is* "os.path.join".
# This allows to patch renamed attributes like "from os.path import join as pjoin".
if getattr(self.obj ,lowerCamelCase_ ) is attr_value:
UpperCAmelCase_ : Dict = getattr(self.obj ,lowerCamelCase_ )
setattr(self.obj ,lowerCamelCase_ ,self.new )
elif target_attr in globals()["__builtins__"]: # if it'a s builtin like "open"
UpperCAmelCase_ : Any = globals()["""__builtins__"""][target_attr]
setattr(self.obj ,lowerCamelCase_ ,self.new )
else:
raise RuntimeError(F'''Tried to patch attribute {target_attr} instead of a submodule.''' )
def __exit__( self: int ,*lowerCamelCase_: str ) -> Optional[Any]:
for attr in list(self.original ):
setattr(self.obj ,lowerCamelCase_ ,self.original.pop(lowerCamelCase_ ) )
def A__ ( self: str ) -> int:
self.__enter__()
self._active_patches.append(self )
def A__ ( self: Any ) -> Any:
try:
self._active_patches.remove(self )
except ValueError:
# If the patch hasn't been started this will fail
return None
return self.__exit__()
| 345 |
import os
import sys
import tempfile
import unittest
import unittest.mock as mock
from pathlib import Path
from huggingface_hub import HfFolder, delete_repo
from huggingface_hub.file_download import http_get
from requests.exceptions import HTTPError
from transformers import (
AlbertTokenizer,
AutoTokenizer,
BertTokenizer,
BertTokenizerFast,
GPTaTokenizerFast,
is_tokenizers_available,
)
from transformers.testing_utils import TOKEN, USER, is_staging_test, require_tokenizers
from transformers.tokenization_utils import Trie
sys.path.append(str(Path(__file__).parent.parent / '''utils'''))
from test_module.custom_tokenization import CustomTokenizer # noqa E402
if is_tokenizers_available():
from test_module.custom_tokenization_fast import CustomTokenizerFast
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: int ) -> str:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : List[str] = mock.Mock()
UpperCAmelCase_ : List[Any] = 500
UpperCAmelCase_ : Union[str, Any] = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : Any = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Union[str, Any] = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# This check we did call the fake head request
mock_head.assert_called()
@require_tokenizers
def A__ ( self: str ) -> int:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : str = mock.Mock()
UpperCAmelCase_ : Optional[int] = 500
UpperCAmelCase_ : int = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : List[Any] = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Optional[int] = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# This check we did call the fake head request
mock_head.assert_called()
def A__ ( self: str ) -> Dict:
# This test is for deprecated behavior and can be removed in v5
try:
UpperCAmelCase_ : Any = tempfile.mktemp()
with open(lowerCamelCase_ ,"""wb""" ) as f:
http_get("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = AlbertTokenizer.from_pretrained(lowerCamelCase_ )
finally:
os.remove(lowerCamelCase_ )
# Supporting this legacy load introduced a weird bug where the tokenizer would load local files if they are in
# the current folder and have the right name.
if os.path.isfile("""tokenizer.json""" ):
# We skip the test if the user has a `tokenizer.json` in this folder to avoid deleting it.
return
try:
with open("""tokenizer.json""" ,"""wb""" ) as f:
http_get("""https://huggingface.co/hf-internal-testing/tiny-random-bert/blob/main/tokenizer.json""" ,lowerCamelCase_ )
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
# The tiny random BERT has a vocab size of 1024, tiny gpt2 as a vocab size of 1000
self.assertEqual(tokenizer.vocab_size ,1000 )
# Tokenizer should depend on the remote checkpoint, not the local tokenizer.json file.
finally:
os.remove("""tokenizer.json""" )
def A__ ( self: List[str] ) -> Tuple:
# This test is for deprecated behavior and can be removed in v5
UpperCAmelCase_ : str = AlbertTokenizer.from_pretrained("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" )
@is_staging_test
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
A__ : str = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "bla", "blou"]
@classmethod
def A__ ( cls: Dict ) -> Optional[int]:
UpperCAmelCase_ : List[str] = TOKEN
HfFolder.save_token(lowerCamelCase_ )
@classmethod
def A__ ( cls: Optional[Any] ) -> List[str]:
try:
delete_repo(token=cls._token ,repo_id="""test-tokenizer""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""valid_org/test-tokenizer-org""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""test-dynamic-tokenizer""" )
except HTTPError:
pass
def A__ ( self: Any ) -> Optional[int]:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Tuple = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : List[Any] = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""test-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""test-tokenizer""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(lowerCamelCase_ ,repo_id="""test-tokenizer""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
def A__ ( self: Optional[int] ) -> Any:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[Any] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Dict = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""valid_org/test-tokenizer-org""" ,use_auth_token=self._token )
UpperCAmelCase_ : Dict = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""valid_org/test-tokenizer-org""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(
lowerCamelCase_ ,repo_id="""valid_org/test-tokenizer-org""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
@require_tokenizers
def A__ ( self: Optional[int] ) -> Optional[Any]:
CustomTokenizer.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Any = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Optional[Any] = CustomTokenizer(lowerCamelCase_ )
# No fast custom tokenizer
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : Optional[Any] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the CustomTokenizer class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
# Fast and slow custom tokenizer
CustomTokenizerFast.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[str] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : str = BertTokenizerFast.from_pretrained(lowerCamelCase_ )
bert_tokenizer.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = CustomTokenizerFast.from_pretrained(lowerCamelCase_ )
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizerFast""" )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(
F'''{USER}/test-dynamic-tokenizer''' ,use_fast=lowerCamelCase_ ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Any = Trie()
trie.add("""Hello 友達""" )
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {""" """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
trie.add("""Hello""" )
trie.data
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {"""""": 1, """ """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
def A__ ( self: Tuple ) -> Optional[int]:
UpperCAmelCase_ : str = Trie()
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS] This is a extra_id_100"""] )
trie.add("""[CLS]""" )
trie.add("""extra_id_1""" )
trie.add("""extra_id_100""" )
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS]""", """ This is a """, """extra_id_100"""] )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ : Dict = Trie()
trie.add("""A""" )
self.assertEqual(trie.split("""ABC""" ) ,["""A""", """BC"""] )
self.assertEqual(trie.split("""BCA""" ) ,["""BC""", """A"""] )
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""TOKEN]""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> Union[str, Any]:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""A""" )
trie.add("""P""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> List[str]:
UpperCAmelCase_ : int = Trie()
trie.add("""AB""" )
trie.add("""B""" )
trie.add("""C""" )
self.assertEqual(trie.split("""ABC""" ) ,["""AB""", """C"""] )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = Trie()
trie.add("""ABC""" )
trie.add("""B""" )
trie.add("""CD""" )
self.assertEqual(trie.split("""ABCD""" ) ,["""ABC""", """D"""] )
def A__ ( self: List[Any] ) -> Any:
# Even if the offsets are wrong, we necessarily output correct string
# parts.
UpperCAmelCase_ : Tuple = Trie()
UpperCAmelCase_ : Optional[Any] = trie.cut_text("""ABC""" ,[0, 0, 2, 1, 2, 3] )
self.assertEqual(lowerCamelCase_ ,["""AB""", """C"""] )
| 345 | 1 |
import math
import unittest
from transformers import BioGptConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
BioGptForCausalLM,
BioGptForSequenceClassification,
BioGptForTokenClassification,
BioGptModel,
BioGptTokenizer,
)
from transformers.models.biogpt.modeling_biogpt import BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST
class _snake_case :
'''simple docstring'''
def __init__( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: List[str]=13 ,lowerCamelCase_: Any=7 ,lowerCamelCase_: Any=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Tuple=False ,lowerCamelCase_: int=True ,lowerCamelCase_: Optional[Any]=99 ,lowerCamelCase_: Tuple=32 ,lowerCamelCase_: Optional[int]=5 ,lowerCamelCase_: Dict=4 ,lowerCamelCase_: str=37 ,lowerCamelCase_: Any="gelu" ,lowerCamelCase_: Optional[Any]=0.1 ,lowerCamelCase_: Dict=0.1 ,lowerCamelCase_: List[str]=512 ,lowerCamelCase_: Tuple=16 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: int=3 ,lowerCamelCase_: Union[str, Any]=4 ,lowerCamelCase_: int=None ,) -> Any:
UpperCAmelCase_ : Union[str, Any] = parent
UpperCAmelCase_ : Any = batch_size
UpperCAmelCase_ : Union[str, Any] = seq_length
UpperCAmelCase_ : Tuple = is_training
UpperCAmelCase_ : Optional[Any] = use_input_mask
UpperCAmelCase_ : Optional[int] = use_token_type_ids
UpperCAmelCase_ : Tuple = use_labels
UpperCAmelCase_ : str = vocab_size
UpperCAmelCase_ : Any = hidden_size
UpperCAmelCase_ : int = num_hidden_layers
UpperCAmelCase_ : Optional[Any] = num_attention_heads
UpperCAmelCase_ : Union[str, Any] = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Any = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : int = type_vocab_size
UpperCAmelCase_ : Any = type_sequence_label_size
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Tuple = num_labels
UpperCAmelCase_ : int = num_choices
UpperCAmelCase_ : Optional[Any] = scope
def A__ ( self: Optional[int] ) -> List[str]:
UpperCAmelCase_ : Union[str, Any] = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : Optional[Any] = None
if self.use_input_mask:
UpperCAmelCase_ : Dict = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : Union[str, Any] = None
if self.use_token_type_ids:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : Any = None
UpperCAmelCase_ : Optional[int] = None
UpperCAmelCase_ : str = None
if self.use_labels:
UpperCAmelCase_ : int = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Union[str, Any] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : Any = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : Optional[int] = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: int ) -> List[Any]:
return BioGptConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: str ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : List[Any] = BioGptModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int ,) -> Optional[Any]:
UpperCAmelCase_ : List[str] = BioGptForCausalLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Any = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: Any ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,*lowerCamelCase_: str ) -> Tuple:
UpperCAmelCase_ : List[Any] = BioGptModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
# create attention mask
UpperCAmelCase_ : Dict = torch.ones(input_ids.shape ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : str = self.seq_length // 2
UpperCAmelCase_ : Any = 0
# first forward pass
UpperCAmelCase_ , UpperCAmelCase_ : Any = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ).to_tuple()
# create hypothetical next token and extent to next_input_ids
UpperCAmelCase_ : Union[str, Any] = ids_tensor((self.batch_size, 1) ,config.vocab_size )
# change a random masked slice from input_ids
UpperCAmelCase_ : Optional[Any] = ids_tensor((1,) ,lowerCamelCase_ ).item() + 1
UpperCAmelCase_ : List[Any] = ids_tensor((self.batch_size, 1) ,config.vocab_size ).squeeze(-1 )
UpperCAmelCase_ : int = random_other_next_tokens
# append to next input_ids and attn_mask
UpperCAmelCase_ : Tuple = torch.cat([input_ids, next_tokens] ,dim=-1 )
UpperCAmelCase_ : Optional[int] = torch.cat(
[attn_mask, torch.ones((attn_mask.shape[0], 1) ,dtype=torch.long ,device=lowerCamelCase_ )] ,dim=1 ,)
# get two different outputs
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ )["""last_hidden_state"""]
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ ,past_key_values=lowerCamelCase_ ,attention_mask=lowerCamelCase_ )["""last_hidden_state"""]
# select random slice
UpperCAmelCase_ : Optional[Any] = ids_tensor((1,) ,output_from_past.shape[-1] ).item()
UpperCAmelCase_ : Dict = output_from_no_past[:, -1, random_slice_idx].detach()
UpperCAmelCase_ : str = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 ) )
def A__ ( self: Any ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,*lowerCamelCase_: Optional[int] ) -> Dict:
UpperCAmelCase_ : Dict = BioGptModel(config=lowerCamelCase_ ).to(lowerCamelCase_ ).eval()
UpperCAmelCase_ : List[Any] = torch.ones(input_ids.shape ,dtype=torch.long ,device=lowerCamelCase_ )
# first forward pass
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,use_cache=lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = outputs.to_tuple()
# create hypothetical multiple next token and extent to next_input_ids
UpperCAmelCase_ : Any = ids_tensor((self.batch_size, 3) ,config.vocab_size )
UpperCAmelCase_ : List[str] = ids_tensor((self.batch_size, 3) ,2 )
# append to next input_ids and
UpperCAmelCase_ : Optional[Any] = torch.cat([input_ids, next_tokens] ,dim=-1 )
UpperCAmelCase_ : Dict = torch.cat([attention_mask, next_attn_mask] ,dim=-1 )
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ )["""last_hidden_state"""]
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,past_key_values=lowerCamelCase_ )[
"""last_hidden_state"""
]
# select random slice
UpperCAmelCase_ : int = ids_tensor((1,) ,output_from_past.shape[-1] ).item()
UpperCAmelCase_ : Tuple = output_from_no_past[:, -3:, random_slice_idx].detach()
UpperCAmelCase_ : Union[str, Any] = output_from_past[:, :, random_slice_idx].detach()
self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1] )
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 ) )
def A__ ( self: int ,lowerCamelCase_: List[str] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ,*lowerCamelCase_: Optional[int] ,lowerCamelCase_: Union[str, Any]=False ) -> Tuple:
UpperCAmelCase_ : Tuple = BioGptForCausalLM(lowerCamelCase_ )
model.to(lowerCamelCase_ )
if gradient_checkpointing:
model.gradient_checkpointing_enable()
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.loss.shape ,() )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
result.loss.backward()
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ,*lowerCamelCase_: Optional[Any] ) -> Any:
UpperCAmelCase_ : Union[str, Any] = BioGptModel(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = model.config.initializer_range / math.sqrt(2 * model.config.num_hidden_layers )
for key in model.state_dict().keys():
if "c_proj" in key and "weight" in key:
self.parent.assertLessEqual(abs(torch.std(model.state_dict()[key] ) - model_std ) ,0.0_0_1 )
self.parent.assertLessEqual(abs(torch.mean(model.state_dict()[key] ) - 0.0 ) ,0.0_1 )
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: int ,lowerCamelCase_: Dict ,lowerCamelCase_: Union[str, Any] ,*lowerCamelCase_: int ) -> List[Any]:
UpperCAmelCase_ : Dict = self.num_labels
UpperCAmelCase_ : Any = BioGptForTokenClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Tuple = config_and_inputs
UpperCAmelCase_ : str = {"""input_ids""": input_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : int = (
(BioGptModel, BioGptForCausalLM, BioGptForSequenceClassification, BioGptForTokenClassification)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (BioGptForCausalLM,) if is_torch_available() else ()
A__ : Any = (
{
"feature-extraction": BioGptModel,
"text-classification": BioGptForSequenceClassification,
"text-generation": BioGptForCausalLM,
"token-classification": BioGptForTokenClassification,
"zero-shot": BioGptForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : str = False
def A__ ( self: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : Optional[Any] = BioGptModelTester(self )
UpperCAmelCase_ : Dict = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Tuple ) -> Optional[int]:
self.config_tester.run_common_tests()
def A__ ( self: Dict ) -> int:
UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Any ) -> int:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
UpperCAmelCase_ : int = type
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: int ) -> List[Any]:
UpperCAmelCase_ : Any = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_model_attention_mask_past(*lowerCamelCase_ )
def A__ ( self: Any ) -> Optional[Any]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_forward_and_backwards(*lowerCamelCase_ ,gradient_checkpointing=lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_model_past_large_inputs(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_weight_initialization(*lowerCamelCase_ )
def A__ ( self: Any ) -> str:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_for_token_classification(*lowerCamelCase_ )
@slow
def A__ ( self: int ) -> Optional[Any]:
UpperCAmelCase_ : Any = BioGptForCausalLM.from_pretrained("""microsoft/biogpt""" )
model.to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = BioGptTokenizer.from_pretrained("""microsoft/biogpt""" )
UpperCAmelCase_ : Union[str, Any] = """left"""
# Define PAD Token = EOS Token = 50256
UpperCAmelCase_ : Union[str, Any] = tokenizer.eos_token
UpperCAmelCase_ : Optional[int] = model.config.eos_token_id
# use different length sentences to test batching
UpperCAmelCase_ : str = [
"""Hello, my dog is a little""",
"""Today, I""",
]
UpperCAmelCase_ : str = tokenizer(lowerCamelCase_ ,return_tensors="""pt""" ,padding=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = inputs["""input_ids"""].to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model.generate(
input_ids=lowerCamelCase_ ,attention_mask=inputs["""attention_mask"""].to(lowerCamelCase_ ) ,)
UpperCAmelCase_ : List[str] = tokenizer(sentences[0] ,return_tensors="""pt""" ).input_ids.to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = model.generate(input_ids=lowerCamelCase_ )
UpperCAmelCase_ : Dict = inputs_non_padded.shape[-1] - inputs["""attention_mask"""][-1].long().sum().cpu().item()
UpperCAmelCase_ : Optional[int] = tokenizer(sentences[1] ,return_tensors="""pt""" ).input_ids.to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = model.generate(input_ids=lowerCamelCase_ ,max_length=model.config.max_length - num_paddings )
UpperCAmelCase_ : Any = tokenizer.batch_decode(lowerCamelCase_ ,skip_special_tokens=lowerCamelCase_ )
UpperCAmelCase_ : int = tokenizer.decode(output_non_padded[0] ,skip_special_tokens=lowerCamelCase_ )
UpperCAmelCase_ : int = tokenizer.decode(output_padded[0] ,skip_special_tokens=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = [
"""Hello, my dog is a little bit bigger than a little bit.""",
"""Today, I have a good idea of how to use the information""",
]
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,[non_padded_sentence, padded_sentence] )
@slow
def A__ ( self: Tuple ) -> List[str]:
for model_name in BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : List[str] = BioGptModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Tuple:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Dict = 3
UpperCAmelCase_ : Tuple = input_dict["""input_ids"""]
UpperCAmelCase_ : Optional[Any] = input_ids.ne(1 ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = ids_tensor([self.model_tester.batch_size] ,self.model_tester.type_sequence_label_size )
UpperCAmelCase_ : str = BioGptForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.assertEqual(result.logits.shape ,(self.model_tester.batch_size, self.model_tester.num_labels) )
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : List[str] = 3
UpperCAmelCase_ : Tuple = """multi_label_classification"""
UpperCAmelCase_ : Tuple = input_dict["""input_ids"""]
UpperCAmelCase_ : Dict = input_ids.ne(1 ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = ids_tensor(
[self.model_tester.batch_size, config.num_labels] ,self.model_tester.type_sequence_label_size ).to(torch.float )
UpperCAmelCase_ : Union[str, Any] = BioGptForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.assertEqual(result.logits.shape ,(self.model_tester.batch_size, self.model_tester.num_labels) )
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : int = BioGptForCausalLM.from_pretrained("""microsoft/biogpt""" )
UpperCAmelCase_ : int = torch.tensor([[2, 4805, 9, 656, 21]] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )[0]
UpperCAmelCase_ : Union[str, Any] = 42384
UpperCAmelCase_ : Tuple = torch.Size((1, 5, vocab_size) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.tensor(
[[[-9.5_2_3_6, -9.8_9_1_8, 1_0.4_5_5_7], [-1_1.0_4_6_9, -9.6_4_2_3, 8.1_0_2_2], [-8.8_6_6_4, -7.8_8_2_6, 5.5_3_2_5]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
@slow
def A__ ( self: List[Any] ) -> Optional[int]:
UpperCAmelCase_ : Tuple = BioGptTokenizer.from_pretrained("""microsoft/biogpt""" )
UpperCAmelCase_ : List[Any] = BioGptForCausalLM.from_pretrained("""microsoft/biogpt""" )
model.to(lowerCamelCase_ )
torch.manual_seed(0 )
UpperCAmelCase_ : Any = tokenizer("""COVID-19 is""" ,return_tensors="""pt""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = model.generate(
**lowerCamelCase_ ,min_length=100 ,max_length=1024 ,num_beams=5 ,early_stopping=lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = tokenizer.decode(output_ids[0] ,skip_special_tokens=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = (
"""COVID-19 is a global pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the"""
""" causative agent of coronavirus disease 2019 (COVID-19), which has spread to more than 200 countries and"""
""" territories, including the United States (US), Canada, Australia, New Zealand, the United Kingdom (UK),"""
""" and the United States of America (USA), as of March 11, 2020, with more than 800,000 confirmed cases and"""
""" more than 800,000 deaths."""
)
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
| 345 |
from ..utils import DummyObject, requires_backends
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: int ,**lowerCamelCase_: List[str] ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: List[str] ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: int ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Any = ["flax"]
def __init__( self: int ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[str] ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Dict = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[Any] ) -> Any:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: List[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[str] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : int = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[int] ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Optional[Any] ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[Any] = ["flax"]
def __init__( self: Union[str, Any] ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> List[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Dict:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Any ,**lowerCamelCase_: int ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: Union[str, Any] ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Dict ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : str = ["flax"]
def __init__( self: Optional[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: int ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: int ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Optional[int] ) -> List[str]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Dict ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: List[str] ) -> int:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[Any] = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Optional[int] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Dict ) -> int:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Optional[Any]:
requires_backends(cls ,["""flax"""] )
| 345 | 1 |
import json
import pathlib
import unittest
import numpy as np
from transformers.testing_utils import require_torch, require_vision, slow
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import ConditionalDetrImageProcessor
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def __init__( self: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=7 ,lowerCamelCase_: Any=3 ,lowerCamelCase_: Tuple=30 ,lowerCamelCase_: Optional[int]=400 ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Optional[Any]=None ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Dict=[0.5, 0.5, 0.5] ,lowerCamelCase_: int=[0.5, 0.5, 0.5] ,lowerCamelCase_: str=True ,lowerCamelCase_: List[str]=1 / 255 ,lowerCamelCase_: Optional[Any]=True ,) -> Dict:
# by setting size["longest_edge"] > max_resolution we're effectively not testing this :p
UpperCAmelCase_ : List[str] = size if size is not None else {"""shortest_edge""": 18, """longest_edge""": 1333}
UpperCAmelCase_ : Optional[Any] = parent
UpperCAmelCase_ : Optional[Any] = batch_size
UpperCAmelCase_ : List[Any] = num_channels
UpperCAmelCase_ : List[str] = min_resolution
UpperCAmelCase_ : Dict = max_resolution
UpperCAmelCase_ : Any = do_resize
UpperCAmelCase_ : Union[str, Any] = size
UpperCAmelCase_ : List[Any] = do_normalize
UpperCAmelCase_ : int = image_mean
UpperCAmelCase_ : str = image_std
UpperCAmelCase_ : List[str] = do_rescale
UpperCAmelCase_ : Dict = rescale_factor
UpperCAmelCase_ : int = do_pad
def A__ ( self: Union[str, Any] ) -> Dict:
return {
"do_resize": self.do_resize,
"size": self.size,
"do_normalize": self.do_normalize,
"image_mean": self.image_mean,
"image_std": self.image_std,
"do_rescale": self.do_rescale,
"rescale_factor": self.rescale_factor,
"do_pad": self.do_pad,
}
def A__ ( self: Optional[int] ,lowerCamelCase_: Dict ,lowerCamelCase_: List[Any]=False ) -> int:
if not batched:
UpperCAmelCase_ : Union[str, Any] = image_inputs[0]
if isinstance(lowerCamelCase_ ,Image.Image ):
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = image.size
else:
UpperCAmelCase_ , UpperCAmelCase_ : Any = image.shape[1], image.shape[2]
if w < h:
UpperCAmelCase_ : List[Any] = int(self.size["""shortest_edge"""] * h / w )
UpperCAmelCase_ : List[str] = self.size["""shortest_edge"""]
elif w > h:
UpperCAmelCase_ : Any = self.size["""shortest_edge"""]
UpperCAmelCase_ : Dict = int(self.size["""shortest_edge"""] * w / h )
else:
UpperCAmelCase_ : Dict = self.size["""shortest_edge"""]
UpperCAmelCase_ : Dict = self.size["""shortest_edge"""]
else:
UpperCAmelCase_ : List[Any] = []
for image in image_inputs:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.get_expected_values([image] )
expected_values.append((expected_height, expected_width) )
UpperCAmelCase_ : Union[str, Any] = max(lowerCamelCase_ ,key=lambda lowerCamelCase_ : item[0] )[0]
UpperCAmelCase_ : Optional[int] = max(lowerCamelCase_ ,key=lambda lowerCamelCase_ : item[1] )[1]
return expected_height, expected_width
@require_torch
@require_vision
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[str] = ConditionalDetrImageProcessor if is_vision_available() else None
def A__ ( self: Dict ) -> str:
UpperCAmelCase_ : int = ConditionalDetrImageProcessingTester(self )
@property
def A__ ( self: int ) -> Optional[int]:
return self.image_processor_tester.prepare_image_processor_dict()
def A__ ( self: Optional[Any] ) -> str:
UpperCAmelCase_ : Dict = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(lowerCamelCase_ ,"""image_mean""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""image_std""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""do_normalize""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""do_resize""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""size""" ) )
def A__ ( self: int ) -> Optional[Any]:
UpperCAmelCase_ : Optional[int] = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size ,{"""shortest_edge""": 18, """longest_edge""": 1333} )
self.assertEqual(image_processor.do_pad ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = self.image_processing_class.from_dict(
self.image_processor_dict ,size=42 ,max_size=84 ,pad_and_return_pixel_mask=lowerCamelCase_ )
self.assertEqual(image_processor.size ,{"""shortest_edge""": 42, """longest_edge""": 84} )
self.assertEqual(image_processor.do_pad ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
pass
def A__ ( self: str ) -> Any:
# Initialize image_processing
UpperCAmelCase_ : Dict = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
UpperCAmelCase_ : int = prepare_image_inputs(self.image_processor_tester ,equal_resolution=lowerCamelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCamelCase_ ,Image.Image )
# Test not batched input
UpperCAmelCase_ : Any = image_processing(image_inputs[0] ,return_tensors="""pt""" ).pixel_values
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.image_processor_tester.get_expected_values(lowerCamelCase_ )
self.assertEqual(
encoded_images.shape ,(1, self.image_processor_tester.num_channels, expected_height, expected_width) ,)
# Test batched
UpperCAmelCase_ , UpperCAmelCase_ : str = self.image_processor_tester.get_expected_values(lowerCamelCase_ ,batched=lowerCamelCase_ )
UpperCAmelCase_ : Any = image_processing(lowerCamelCase_ ,return_tensors="""pt""" ).pixel_values
self.assertEqual(
encoded_images.shape ,(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
expected_height,
expected_width,
) ,)
def A__ ( self: Dict ) -> List[str]:
# Initialize image_processing
UpperCAmelCase_ : Optional[Any] = self.image_processing_class(**self.image_processor_dict )
# create random numpy tensors
UpperCAmelCase_ : List[Any] = prepare_image_inputs(self.image_processor_tester ,equal_resolution=lowerCamelCase_ ,numpify=lowerCamelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCamelCase_ ,np.ndarray )
# Test not batched input
UpperCAmelCase_ : Optional[int] = image_processing(image_inputs[0] ,return_tensors="""pt""" ).pixel_values
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.image_processor_tester.get_expected_values(lowerCamelCase_ )
self.assertEqual(
encoded_images.shape ,(1, self.image_processor_tester.num_channels, expected_height, expected_width) ,)
# Test batched
UpperCAmelCase_ : Tuple = image_processing(lowerCamelCase_ ,return_tensors="""pt""" ).pixel_values
UpperCAmelCase_ , UpperCAmelCase_ : Dict = self.image_processor_tester.get_expected_values(lowerCamelCase_ ,batched=lowerCamelCase_ )
self.assertEqual(
encoded_images.shape ,(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
expected_height,
expected_width,
) ,)
def A__ ( self: Any ) -> Dict:
# Initialize image_processing
UpperCAmelCase_ : Union[str, Any] = self.image_processing_class(**self.image_processor_dict )
# create random PyTorch tensors
UpperCAmelCase_ : str = prepare_image_inputs(self.image_processor_tester ,equal_resolution=lowerCamelCase_ ,torchify=lowerCamelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCamelCase_ ,torch.Tensor )
# Test not batched input
UpperCAmelCase_ : Optional[int] = image_processing(image_inputs[0] ,return_tensors="""pt""" ).pixel_values
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.image_processor_tester.get_expected_values(lowerCamelCase_ )
self.assertEqual(
encoded_images.shape ,(1, self.image_processor_tester.num_channels, expected_height, expected_width) ,)
# Test batched
UpperCAmelCase_ : str = image_processing(lowerCamelCase_ ,return_tensors="""pt""" ).pixel_values
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.image_processor_tester.get_expected_values(lowerCamelCase_ ,batched=lowerCamelCase_ )
self.assertEqual(
encoded_images.shape ,(
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
expected_height,
expected_width,
) ,)
@slow
def A__ ( self: Tuple ) -> int:
# prepare image and target
UpperCAmelCase_ : str = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
with open("""./tests/fixtures/tests_samples/COCO/coco_annotations.txt""" ,"""r""" ) as f:
UpperCAmelCase_ : List[str] = json.loads(f.read() )
UpperCAmelCase_ : Optional[Any] = {"""image_id""": 39769, """annotations""": target}
# encode them
UpperCAmelCase_ : Any = ConditionalDetrImageProcessor.from_pretrained("""microsoft/conditional-detr-resnet-50""" )
UpperCAmelCase_ : List[str] = image_processing(images=lowerCamelCase_ ,annotations=lowerCamelCase_ ,return_tensors="""pt""" )
# verify pixel values
UpperCAmelCase_ : Any = torch.Size([1, 3, 800, 1066] )
self.assertEqual(encoding["""pixel_values"""].shape ,lowerCamelCase_ )
UpperCAmelCase_ : int = torch.tensor([0.2_7_9_6, 0.3_1_3_8, 0.3_4_8_1] )
self.assertTrue(torch.allclose(encoding["""pixel_values"""][0, 0, 0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
# verify area
UpperCAmelCase_ : int = torch.tensor([5_8_8_7.9_6_0_0, 1_1_2_5_0.2_0_6_1, 4_8_9_3_5_3.8_4_3_8, 8_3_7_1_2_2.7_5_0_0, 1_4_7_9_6_7.5_1_5_6, 1_6_5_7_3_2.3_4_3_8] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""area"""] ,lowerCamelCase_ ) )
# verify boxes
UpperCAmelCase_ : Optional[int] = torch.Size([6, 4] )
self.assertEqual(encoding["""labels"""][0]["""boxes"""].shape ,lowerCamelCase_ )
UpperCAmelCase_ : Dict = torch.tensor([0.5_5_0_3, 0.2_7_6_5, 0.0_6_0_4, 0.2_2_1_5] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""boxes"""][0] ,lowerCamelCase_ ,atol=1e-3 ) )
# verify image_id
UpperCAmelCase_ : str = torch.tensor([39769] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""image_id"""] ,lowerCamelCase_ ) )
# verify is_crowd
UpperCAmelCase_ : Any = torch.tensor([0, 0, 0, 0, 0, 0] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""iscrowd"""] ,lowerCamelCase_ ) )
# verify class_labels
UpperCAmelCase_ : Any = torch.tensor([75, 75, 63, 65, 17, 17] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""class_labels"""] ,lowerCamelCase_ ) )
# verify orig_size
UpperCAmelCase_ : Union[str, Any] = torch.tensor([480, 640] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""orig_size"""] ,lowerCamelCase_ ) )
# verify size
UpperCAmelCase_ : Dict = torch.tensor([800, 1066] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""size"""] ,lowerCamelCase_ ) )
@slow
def A__ ( self: Optional[int] ) -> Dict:
# prepare image, target and masks_path
UpperCAmelCase_ : Optional[Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
with open("""./tests/fixtures/tests_samples/COCO/coco_panoptic_annotations.txt""" ,"""r""" ) as f:
UpperCAmelCase_ : Union[str, Any] = json.loads(f.read() )
UpperCAmelCase_ : Tuple = {"""file_name""": """000000039769.png""", """image_id""": 39769, """segments_info""": target}
UpperCAmelCase_ : Union[str, Any] = pathlib.Path("""./tests/fixtures/tests_samples/COCO/coco_panoptic""" )
# encode them
UpperCAmelCase_ : int = ConditionalDetrImageProcessor(format="""coco_panoptic""" )
UpperCAmelCase_ : Optional[int] = image_processing(images=lowerCamelCase_ ,annotations=lowerCamelCase_ ,masks_path=lowerCamelCase_ ,return_tensors="""pt""" )
# verify pixel values
UpperCAmelCase_ : Optional[int] = torch.Size([1, 3, 800, 1066] )
self.assertEqual(encoding["""pixel_values"""].shape ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.tensor([0.2_7_9_6, 0.3_1_3_8, 0.3_4_8_1] )
self.assertTrue(torch.allclose(encoding["""pixel_values"""][0, 0, 0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
# verify area
UpperCAmelCase_ : Optional[int] = torch.tensor([1_4_7_9_7_9.6_8_7_5, 1_6_5_5_2_7.0_4_6_9, 4_8_4_6_3_8.5_9_3_8, 1_1_2_9_2.9_3_7_5, 5_8_7_9.6_5_6_2, 7_6_3_4.1_1_4_7] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""area"""] ,lowerCamelCase_ ) )
# verify boxes
UpperCAmelCase_ : str = torch.Size([6, 4] )
self.assertEqual(encoding["""labels"""][0]["""boxes"""].shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_6_2_5, 0.5_4_3_7, 0.4_6_8_8, 0.8_6_2_5] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""boxes"""][0] ,lowerCamelCase_ ,atol=1e-3 ) )
# verify image_id
UpperCAmelCase_ : List[str] = torch.tensor([39769] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""image_id"""] ,lowerCamelCase_ ) )
# verify is_crowd
UpperCAmelCase_ : Tuple = torch.tensor([0, 0, 0, 0, 0, 0] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""iscrowd"""] ,lowerCamelCase_ ) )
# verify class_labels
UpperCAmelCase_ : Any = torch.tensor([17, 17, 63, 75, 75, 93] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""class_labels"""] ,lowerCamelCase_ ) )
# verify masks
UpperCAmelCase_ : Tuple = 822873
self.assertEqual(encoding["""labels"""][0]["""masks"""].sum().item() ,lowerCamelCase_ )
# verify orig_size
UpperCAmelCase_ : List[Any] = torch.tensor([480, 640] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""orig_size"""] ,lowerCamelCase_ ) )
# verify size
UpperCAmelCase_ : Optional[Any] = torch.tensor([800, 1066] )
self.assertTrue(torch.allclose(encoding["""labels"""][0]["""size"""] ,lowerCamelCase_ ) )
| 345 |
import random
from typing import Any
def lowerCamelCase_ ( _a : list ):
'''simple docstring'''
for _ in range(len(_a ) ):
UpperCAmelCase_ : Tuple = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ : List[Any] = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ , UpperCAmelCase_ : int = data[b], data[a]
return data
if __name__ == "__main__":
UpperCamelCase_ = [0, 1, 2, 3, 4, 5, 6, 7]
UpperCamelCase_ = ['''python''', '''says''', '''hello''', '''!''']
print('''Fisher-Yates Shuffle:''')
print('''List''', integers, strings)
print('''FY Shuffle''', fisher_yates_shuffle(integers), fisher_yates_shuffle(strings))
| 345 | 1 |
from typing import List, Optional, Union
import torch
from ...models import UNetaDConditionModel, VQModel
from ...pipelines import DiffusionPipeline
from ...pipelines.pipeline_utils import ImagePipelineOutput
from ...schedulers import DDPMScheduler
from ...utils import (
is_accelerate_available,
is_accelerate_version,
logging,
randn_tensor,
replace_example_docstring,
)
UpperCamelCase_ = logging.get_logger(__name__) # pylint: disable=invalid-name
UpperCamelCase_ = '''
Examples:
```py
>>> import torch
>>> import numpy as np
>>> from diffusers import KandinskyV22PriorPipeline, KandinskyV22ControlnetPipeline
>>> from transformers import pipeline
>>> from diffusers.utils import load_image
>>> def make_hint(image, depth_estimator):
... image = depth_estimator(image)["depth"]
... image = np.array(image)
... image = image[:, :, None]
... image = np.concatenate([image, image, image], axis=2)
... detected_map = torch.from_numpy(image).float() / 255.0
... hint = detected_map.permute(2, 0, 1)
... return hint
>>> depth_estimator = pipeline("depth-estimation")
>>> pipe_prior = KandinskyV22PriorPipeline.from_pretrained(
... "kandinsky-community/kandinsky-2-2-prior", torch_dtype=torch.float16
... )
>>> pipe_prior = pipe_prior.to("cuda")
>>> pipe = KandinskyV22ControlnetPipeline.from_pretrained(
... "kandinsky-community/kandinsky-2-2-controlnet-depth", torch_dtype=torch.float16
... )
>>> pipe = pipe.to("cuda")
>>> img = load_image(
... "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"
... "/kandinsky/cat.png"
... ).resize((768, 768))
>>> hint = make_hint(img, depth_estimator).unsqueeze(0).half().to("cuda")
>>> prompt = "A robot, 4k photo"
>>> negative_prior_prompt = "lowres, text, error, cropped, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, username, watermark, signature"
>>> generator = torch.Generator(device="cuda").manual_seed(43)
>>> image_emb, zero_image_emb = pipe_prior(
... prompt=prompt, negative_prompt=negative_prior_prompt, generator=generator
... ).to_tuple()
>>> images = pipe(
... image_embeds=image_emb,
... negative_image_embeds=zero_image_emb,
... hint=hint,
... num_inference_steps=50,
... generator=generator,
... height=768,
... width=768,
... ).images
>>> images[0].save("robot_cat.png")
```
'''
def lowerCamelCase_ ( _a : Optional[Any] , _a : Any , _a : Dict=8 ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = height // scale_factor**2
if height % scale_factor**2 != 0:
new_height += 1
UpperCAmelCase_ : Tuple = width // scale_factor**2
if width % scale_factor**2 != 0:
new_width += 1
return new_height * scale_factor, new_width * scale_factor
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: Any ,lowerCamelCase_: UNetaDConditionModel ,lowerCamelCase_: DDPMScheduler ,lowerCamelCase_: VQModel ,) -> Dict:
super().__init__()
self.register_modules(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,movq=lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = 2 ** (len(self.movq.config.block_out_channels ) - 1)
def A__ ( self: Tuple ,lowerCamelCase_: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
if latents is None:
UpperCAmelCase_ : Union[str, Any] = randn_tensor(lowerCamelCase_ ,generator=lowerCamelCase_ ,device=lowerCamelCase_ ,dtype=lowerCamelCase_ )
else:
if latents.shape != shape:
raise ValueError(F'''Unexpected latents shape, got {latents.shape}, expected {shape}''' )
UpperCAmelCase_ : Optional[int] = latents.to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = latents * scheduler.init_noise_sigma
return latents
def A__ ( self: Optional[Any] ,lowerCamelCase_: int=0 ) -> List[str]:
if is_accelerate_available():
from accelerate import cpu_offload
else:
raise ImportError("""Please install accelerate via `pip install accelerate`""" )
UpperCAmelCase_ : Union[str, Any] = torch.device(F'''cuda:{gpu_id}''' )
UpperCAmelCase_ : Union[str, Any] = [
self.unet,
self.movq,
]
for cpu_offloaded_model in models:
if cpu_offloaded_model is not None:
cpu_offload(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[Any]=0 ) -> List[Any]:
if is_accelerate_available() and is_accelerate_version(""">=""" ,"""0.17.0.dev0""" ):
from accelerate import cpu_offload_with_hook
else:
raise ImportError("""`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher.""" )
UpperCAmelCase_ : Optional[int] = torch.device(F'''cuda:{gpu_id}''' )
if self.device.type != "cpu":
self.to("""cpu""" ,silence_dtype_warnings=lowerCamelCase_ )
torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist)
UpperCAmelCase_ : Dict = None
for cpu_offloaded_model in [self.unet, self.movq]:
UpperCAmelCase_ , UpperCAmelCase_ : Any = cpu_offload_with_hook(lowerCamelCase_ ,lowerCamelCase_ ,prev_module_hook=lowerCamelCase_ )
# We'll offload the last model manually.
UpperCAmelCase_ : str = hook
@property
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._execution_device
def A__ ( self: str ) -> Union[str, Any]:
if not hasattr(self.unet ,"""_hf_hook""" ):
return self.device
for module in self.unet.modules():
if (
hasattr(lowerCamelCase_ ,"""_hf_hook""" )
and hasattr(module._hf_hook ,"""execution_device""" )
and module._hf_hook.execution_device is not None
):
return torch.device(module._hf_hook.execution_device )
return self.device
@torch.no_grad()
@replace_example_docstring(lowerCamelCase_ )
def __call__( self: List[Any] ,lowerCamelCase_: Union[torch.FloatTensor, List[torch.FloatTensor]] ,lowerCamelCase_: Union[torch.FloatTensor, List[torch.FloatTensor]] ,lowerCamelCase_: torch.FloatTensor ,lowerCamelCase_: int = 512 ,lowerCamelCase_: int = 512 ,lowerCamelCase_: int = 100 ,lowerCamelCase_: float = 4.0 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: Optional[Union[torch.Generator, List[torch.Generator]]] = None ,lowerCamelCase_: Optional[torch.FloatTensor] = None ,lowerCamelCase_: Optional[str] = "pil" ,lowerCamelCase_: bool = True ,) -> Optional[Any]:
UpperCAmelCase_ : Optional[Any] = self._execution_device
UpperCAmelCase_ : str = guidance_scale > 1.0
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Optional[int] = torch.cat(lowerCamelCase_ ,dim=0 )
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : List[str] = torch.cat(lowerCamelCase_ ,dim=0 )
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = torch.cat(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : int = image_embeds.shape[0] * num_images_per_prompt
if do_classifier_free_guidance:
UpperCAmelCase_ : Union[str, Any] = image_embeds.repeat_interleave(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : Dict = negative_image_embeds.repeat_interleave(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : str = hint.repeat_interleave(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : Any = torch.cat([negative_image_embeds, image_embeds] ,dim=0 ).to(dtype=self.unet.dtype ,device=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = torch.cat([hint, hint] ,dim=0 ).to(dtype=self.unet.dtype ,device=lowerCamelCase_ )
self.scheduler.set_timesteps(lowerCamelCase_ ,device=lowerCamelCase_ )
UpperCAmelCase_ : int = self.scheduler.timesteps
UpperCAmelCase_ : str = self.movq.config.latent_channels
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = downscale_height_and_width(lowerCamelCase_ ,lowerCamelCase_ ,self.movq_scale_factor )
# create initial latent
UpperCAmelCase_ : List[str] = self.prepare_latents(
(batch_size, num_channels_latents, height, width) ,image_embeds.dtype ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,self.scheduler ,)
for i, t in enumerate(self.progress_bar(lowerCamelCase_ ) ):
# expand the latents if we are doing classifier free guidance
UpperCAmelCase_ : Any = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents
UpperCAmelCase_ : Tuple = {"""image_embeds""": image_embeds, """hint""": hint}
UpperCAmelCase_ : Tuple = self.unet(
sample=lowerCamelCase_ ,timestep=lowerCamelCase_ ,encoder_hidden_states=lowerCamelCase_ ,added_cond_kwargs=lowerCamelCase_ ,return_dict=lowerCamelCase_ ,)[0]
if do_classifier_free_guidance:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = noise_pred.split(latents.shape[1] ,dim=1 )
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = noise_pred.chunk(2 )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = variance_pred.chunk(2 )
UpperCAmelCase_ : Tuple = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
UpperCAmelCase_ : Dict = torch.cat([noise_pred, variance_pred_text] ,dim=1 )
if not (
hasattr(self.scheduler.config ,"""variance_type""" )
and self.scheduler.config.variance_type in ["learned", "learned_range"]
):
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = noise_pred.split(latents.shape[1] ,dim=1 )
# compute the previous noisy sample x_t -> x_t-1
UpperCAmelCase_ : Dict = self.scheduler.step(
lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,generator=lowerCamelCase_ ,)[0]
# post-processing
UpperCAmelCase_ : Dict = self.movq.decode(lowerCamelCase_ ,force_not_quantize=lowerCamelCase_ )["""sample"""]
if output_type not in ["pt", "np", "pil"]:
raise ValueError(F'''Only the output types `pt`, `pil` and `np` are supported not output_type={output_type}''' )
if output_type in ["np", "pil"]:
UpperCAmelCase_ : List[str] = image * 0.5 + 0.5
UpperCAmelCase_ : Optional[int] = image.clamp(0 ,1 )
UpperCAmelCase_ : Optional[Any] = image.cpu().permute(0 ,2 ,3 ,1 ).float().numpy()
if output_type == "pil":
UpperCAmelCase_ : Union[str, Any] = self.numpy_to_pil(lowerCamelCase_ )
if not return_dict:
return (image,)
return ImagePipelineOutput(images=lowerCamelCase_ )
| 345 |
import flax.linen as nn
import jax.numpy as jnp
from .attention_flax import FlaxTransformeraDModel
from .resnet_flax import FlaxDownsampleaD, FlaxResnetBlockaD, FlaxUpsampleaD
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Optional[int] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : List[Any] = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : List[Any] = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : int = resnets
UpperCAmelCase_ : Tuple = attentions
if self.add_downsample:
UpperCAmelCase_ : List[Any] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int=True ) -> int:
UpperCAmelCase_ : List[Any] = ()
for resnet, attn in zip(self.resnets ,self.attentions ):
UpperCAmelCase_ : str = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[Any] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> int:
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : Dict = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnets
if self.add_downsample:
UpperCAmelCase_ : List[str] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any]=True ) -> Any:
UpperCAmelCase_ : Union[str, Any] = ()
for resnet in self.resnets:
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[str] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: str ) -> Any:
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : int = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Optional[Any] = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : int = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = resnets
UpperCAmelCase_ : Dict = attentions
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: Any ,lowerCamelCase_: str ,lowerCamelCase_: List[str]=True ) -> List[str]:
for resnet, attn in zip(self.resnets ,self.attentions ):
# pop res hidden states
UpperCAmelCase_ : List[str] = res_hidden_states_tuple[-1]
UpperCAmelCase_ : Union[str, Any] = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : Optional[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Dict = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> Dict:
UpperCAmelCase_ : Any = []
for i in range(self.num_layers ):
UpperCAmelCase_ : str = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : Optional[int] = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : str = resnets
if self.add_upsample:
UpperCAmelCase_ : Union[str, Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Dict ,lowerCamelCase_: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any=True ) -> List[str]:
for resnet in self.resnets:
# pop res hidden states
UpperCAmelCase_ : Dict = res_hidden_states_tuple[-1]
UpperCAmelCase_ : str = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : List[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : List[str] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
# there is always at least one resnet
UpperCAmelCase_ : List[Any] = [
FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
]
UpperCAmelCase_ : Any = []
for _ in range(self.num_layers ):
UpperCAmelCase_ : Optional[Any] = FlaxTransformeraDModel(
in_channels=self.in_channels ,n_heads=self.num_attention_heads ,d_head=self.in_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Dict = resnets
UpperCAmelCase_ : Any = attentions
def __call__( self: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any]=True ) -> List[Any]:
UpperCAmelCase_ : List[Any] = self.resnets[0](lowerCamelCase_ ,lowerCamelCase_ )
for attn, resnet in zip(self.attentions ,self.resnets[1:] ):
UpperCAmelCase_ : Optional[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
return hidden_states
| 345 | 1 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
UpperCamelCase_ = logging.getLogger(__name__)
torch.set_grad_enabled(False)
UpperCamelCase_ = '''cuda''' if torch.cuda.is_available() else '''cpu'''
def lowerCamelCase_ ( _a : str , _a : Any=100 , _a : int=" " ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = text.split(_a )
return [character.join(text[i : i + n] ).strip() for i in range(0 , len(_a ) , _a )]
def lowerCamelCase_ ( _a : dict ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = [], []
for title, text in zip(documents["""title"""] , documents["""text"""] ):
if text is not None:
for passage in split_text(_a ):
titles.append(title if title is not None else """""" )
texts.append(_a )
return {"title": titles, "text": texts}
def lowerCamelCase_ ( _a : dict , _a : DPRContextEncoder , _a : DPRContextEncoderTokenizerFast ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = ctx_tokenizer(
documents["""title"""] , documents["""text"""] , truncation=_a , padding="""longest""" , return_tensors="""pt""" )["""input_ids"""]
UpperCAmelCase_ : Tuple = ctx_encoder(input_ids.to(device=_a ) , return_dict=_a ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def lowerCamelCase_ ( _a : "RagExampleArguments" , _a : "ProcessingArguments" , _a : "IndexHnswArguments" , ):
'''simple docstring'''
logger.info("""Step 1 - Create the dataset""" )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
UpperCAmelCase_ : Optional[int] = load_dataset(
"""csv""" , data_files=[rag_example_args.csv_path] , split="""train""" , delimiter="""\t""" , column_names=["""title""", """text"""] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
UpperCAmelCase_ : Tuple = dataset.map(_a , batched=_a , num_proc=processing_args.num_proc )
# And compute the embeddings
UpperCAmelCase_ : List[str] = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=_a )
UpperCAmelCase_ : Dict = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
UpperCAmelCase_ : Any = Features(
{"""text""": Value("""string""" ), """title""": Value("""string""" ), """embeddings""": Sequence(Value("""float32""" ) )} ) # optional, save as float32 instead of float64 to save space
UpperCAmelCase_ : List[str] = dataset.map(
partial(_a , ctx_encoder=_a , ctx_tokenizer=_a ) , batched=_a , batch_size=processing_args.batch_size , features=_a , )
# And finally save your dataset
UpperCAmelCase_ : Union[str, Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset""" )
dataset.save_to_disk(_a )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info("""Step 2 - Index the dataset""" )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
UpperCAmelCase_ : Union[str, Any] = faiss.IndexHNSWFlat(index_hnsw_args.d , index_hnsw_args.m , faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index("""embeddings""" , custom_index=_a )
# And save the index
UpperCAmelCase_ : Optional[Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset_hnsw_index.faiss""" )
dataset.get_index("""embeddings""" ).save(_a )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" / "my_knowledge_dataset.csv" ) , metadata={"help": "Path to a tab-separated csv file with columns 'title' and 'text'"} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."} , )
A__ : str = field(
default="facebook/rag-sequence-nq" , metadata={"help": "The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"} , )
A__ : str = field(
default="facebook/dpr-ctx_encoder-multiset-base" , metadata={
"help": (
"The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"
" 'facebook/dpr-ctx_encoder-multiset-base'"
)
} , )
A__ : Optional[str] = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" ) , metadata={"help": "Path to a directory where the dataset passages and the index will be saved"} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": "The number of processes to use to split the documents into passages. Default is single process."
} , )
A__ : int = field(
default=16 , metadata={
"help": "The batch size to use when computing the passages embeddings using the DPR context encoder."
} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : int = field(
default=768 , metadata={"help": "The dimension of the embeddings to pass to the HNSW Faiss index."} , )
A__ : int = field(
default=128 , metadata={
"help": (
"The number of bi-directional links created for every new element during the HNSW index construction."
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
UpperCamelCase_ = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
UpperCamelCase_ = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 345 |
import pickle
import numpy as np
from matplotlib import pyplot as plt
class _snake_case :
'''simple docstring'''
def __init__( self: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple=0.2 ,lowerCamelCase_: Union[str, Any]=0.2 ) -> List[str]:
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : str = bp_numa
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : Optional[int] = conva_get[:2]
UpperCAmelCase_ : List[Any] = conva_get[2]
UpperCAmelCase_ : str = size_pa
UpperCAmelCase_ : Optional[int] = rate_w
UpperCAmelCase_ : Dict = rate_t
UpperCAmelCase_ : List[Any] = [
np.mat(-1 * np.random.rand(self.conva[0] ,self.conva[0] ) + 0.5 )
for i in range(self.conva[1] )
]
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : Dict = -2 * np.random.rand(self.conva[1] ) + 1
UpperCAmelCase_ : str = -2 * np.random.rand(self.num_bpa ) + 1
UpperCAmelCase_ : Union[str, Any] = -2 * np.random.rand(self.num_bpa ) + 1
def A__ ( self: str ,lowerCamelCase_: Optional[Any] ) -> Tuple:
# save model dict with pickle
UpperCAmelCase_ : Dict = {
"""num_bp1""": self.num_bpa,
"""num_bp2""": self.num_bpa,
"""num_bp3""": self.num_bpa,
"""conv1""": self.conva,
"""step_conv1""": self.step_conva,
"""size_pooling1""": self.size_poolinga,
"""rate_weight""": self.rate_weight,
"""rate_thre""": self.rate_thre,
"""w_conv1""": self.w_conva,
"""wkj""": self.wkj,
"""vji""": self.vji,
"""thre_conv1""": self.thre_conva,
"""thre_bp2""": self.thre_bpa,
"""thre_bp3""": self.thre_bpa,
}
with open(lowerCamelCase_ ,"""wb""" ) as f:
pickle.dump(lowerCamelCase_ ,lowerCamelCase_ )
print(F'''Model saved： {save_path}''' )
@classmethod
def A__ ( cls: List[str] ,lowerCamelCase_: str ) -> List[str]:
# read saved model
with open(lowerCamelCase_ ,"""rb""" ) as f:
UpperCAmelCase_ : Any = pickle.load(lowerCamelCase_ ) # noqa: S301
UpperCAmelCase_ : Union[str, Any] = model_dic.get("""conv1""" )
conv_get.append(model_dic.get("""step_conv1""" ) )
UpperCAmelCase_ : List[str] = model_dic.get("""size_pooling1""" )
UpperCAmelCase_ : Tuple = model_dic.get("""num_bp1""" )
UpperCAmelCase_ : Optional[Any] = model_dic.get("""num_bp2""" )
UpperCAmelCase_ : List[str] = model_dic.get("""num_bp3""" )
UpperCAmelCase_ : List[Any] = model_dic.get("""rate_weight""" )
UpperCAmelCase_ : Dict = model_dic.get("""rate_thre""" )
# create model instance
UpperCAmelCase_ : List[Any] = CNN(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# modify model parameter
UpperCAmelCase_ : Any = model_dic.get("""w_conv1""" )
UpperCAmelCase_ : int = model_dic.get("""wkj""" )
UpperCAmelCase_ : int = model_dic.get("""vji""" )
UpperCAmelCase_ : Optional[int] = model_dic.get("""thre_conv1""" )
UpperCAmelCase_ : List[str] = model_dic.get("""thre_bp2""" )
UpperCAmelCase_ : Dict = model_dic.get("""thre_bp3""" )
return conv_ins
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> Tuple:
return 1 / (1 + np.exp(-1 * x ))
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
return round(lowerCamelCase_ ,3 )
def A__ ( self: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Any:
# convolution process
UpperCAmelCase_ : Optional[Any] = convs[0]
UpperCAmelCase_ : int = convs[1]
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )[0]
# get the data slice of original image data, data_focus
UpperCAmelCase_ : Dict = []
for i_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
for j_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = data[
i_focus : i_focus + size_conv, j_focus : j_focus + size_conv
]
data_focus.append(lowerCamelCase_ )
# calculate the feature map of every single kernel, and saved as list of matrix
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = int((size_data - size_conv) / conv_step + 1 )
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[int] = []
for i_focus in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = (
np.sum(np.multiply(data_focus[i_focus] ,w_convs[i_map] ) )
- thre_convs[i_map]
)
featuremap.append(self.sig(lowerCamelCase_ ) )
UpperCAmelCase_ : Union[str, Any] = np.asmatrix(lowerCamelCase_ ).reshape(
lowerCamelCase_ ,lowerCamelCase_ )
data_featuremap.append(lowerCamelCase_ )
# expanding the data slice to One dimenssion
UpperCAmelCase_ : Optional[Any] = []
for each_focus in data_focus:
focusa_list.extend(self.Expand_Mat(lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[int] = np.asarray(lowerCamelCase_ )
return focus_list, data_featuremap
def A__ ( self: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any]="average_pool" ) -> List[Any]:
# pooling process
UpperCAmelCase_ : Optional[Any] = len(featuremaps[0] )
UpperCAmelCase_ : Any = int(size_map / size_pooling )
UpperCAmelCase_ : Optional[int] = []
for i_map in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Any = featuremaps[i_map]
UpperCAmelCase_ : Tuple = []
for i_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : str = feature_map[
i_focus : i_focus + size_pooling,
j_focus : j_focus + size_pooling,
]
if pooling_type == "average_pool":
# average pooling
map_pooled.append(np.average(lowerCamelCase_ ) )
elif pooling_type == "max_pooling":
# max pooling
map_pooled.append(np.max(lowerCamelCase_ ) )
UpperCAmelCase_ : int = np.asmatrix(lowerCamelCase_ ).reshape(lowerCamelCase_ ,lowerCamelCase_ )
featuremap_pooled.append(lowerCamelCase_ )
return featuremap_pooled
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> Optional[int]:
# expanding three dimension data to one dimension list
UpperCAmelCase_ : List[Any] = []
for i in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Tuple = np.shape(data[i] )
UpperCAmelCase_ : Optional[int] = data[i].reshape(1 ,shapes[0] * shapes[1] )
UpperCAmelCase_ : Optional[int] = data_listed.getA().tolist()[0]
data_expanded.extend(lowerCamelCase_ )
UpperCAmelCase_ : int = np.asarray(lowerCamelCase_ )
return data_expanded
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
# expanding matrix to one dimension list
UpperCAmelCase_ : List[Any] = np.asarray(lowerCamelCase_ )
UpperCAmelCase_ : str = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = data_mat.reshape(1 ,shapes[0] * shapes[1] )
return data_expanded
def A__ ( self: str ,lowerCamelCase_: Dict ,lowerCamelCase_: int ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = 0
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = np.ones((size_map, size_map) )
for i in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Any = pd_pool[
i_pool
]
UpperCAmelCase_ : List[str] = i_pool + 1
UpperCAmelCase_ : Optional[Any] = np.multiply(
lowerCamelCase_ ,np.multiply(out_map[i_map] ,(1 - out_map[i_map]) ) )
pd_all.append(lowerCamelCase_ )
return pd_all
def A__ ( self: str ,lowerCamelCase_: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any=bool ) -> Optional[int]:
# model traning
print("""----------------------Start Training-------------------------""" )
print((""" - - Shape: Train_Data """, np.shape(lowerCamelCase_ )) )
print((""" - - Shape: Teach_Data """, np.shape(lowerCamelCase_ )) )
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : Any = 10000
while rp < n_repeat and mse >= error_accuracy:
UpperCAmelCase_ : List[str] = 0
print(F'''-------------Learning Time {rp}--------------''' )
for p in range(len(lowerCamelCase_ ) ):
# print('------------Learning Image: %d--------------'%p)
UpperCAmelCase_ : str = np.asmatrix(datas_train[p] )
UpperCAmelCase_ : Optional[Any] = np.asarray(datas_teach[p] )
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : List[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = data_bp_input
UpperCAmelCase_ : Optional[Any] = np.dot(lowerCamelCase_ ,self.vji.T ) - self.thre_bpa
UpperCAmelCase_ : int = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = np.dot(lowerCamelCase_ ,self.wkj.T ) - self.thre_bpa
UpperCAmelCase_ : Optional[Any] = self.sig(lowerCamelCase_ )
# --------------Model Leaning ------------------------
# calculate error and gradient---------------
UpperCAmelCase_ : List[str] = np.multiply(
(data_teach - bp_outa) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : List[Any] = np.multiply(
np.dot(lowerCamelCase_ ,self.wkj ) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : Any = np.dot(lowerCamelCase_ ,self.vji )
UpperCAmelCase_ : Tuple = pd_i_all / (self.size_poolinga * self.size_poolinga)
UpperCAmelCase_ : List[str] = pd_conva_pooled.T.getA().tolist()
UpperCAmelCase_ : str = self._calculate_gradient_from_pool(
lowerCamelCase_ ,lowerCamelCase_ ,shape_featuremapa[0] ,shape_featuremapa[1] ,self.size_poolinga ,)
# weight and threshold learning process---------
# convolution layer
for k_conv in range(self.conva[1] ):
UpperCAmelCase_ : List[str] = self._expand_mat(pd_conva_all[k_conv] )
UpperCAmelCase_ : Optional[Any] = self.rate_weight * np.dot(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : int = self.w_conva[k_conv] + delta_w.reshape(
(self.conva[0], self.conva[0]) )
UpperCAmelCase_ : str = (
self.thre_conva[k_conv]
- np.sum(pd_conva_all[k_conv] ) * self.rate_thre
)
# all connected layer
UpperCAmelCase_ : int = self.wkj + pd_k_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : Tuple = self.vji + pd_j_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : int = self.thre_bpa - pd_k_all * self.rate_thre
UpperCAmelCase_ : str = self.thre_bpa - pd_j_all * self.rate_thre
# calculate the sum error of all single image
UpperCAmelCase_ : int = np.sum(abs(data_teach - bp_outa ) )
error_count += errors
# print(' ----Teach ',data_teach)
# print(' ----BP_output ',bp_out3)
UpperCAmelCase_ : int = rp + 1
UpperCAmelCase_ : Any = error_count / patterns
all_mse.append(lowerCamelCase_ )
def draw_error():
UpperCAmelCase_ : Any = [error_accuracy for i in range(int(n_repeat * 1.2 ) )]
plt.plot(lowerCamelCase_ ,"""+-""" )
plt.plot(lowerCamelCase_ ,"""r--""" )
plt.xlabel("""Learning Times""" )
plt.ylabel("""All_mse""" )
plt.grid(lowerCamelCase_ ,alpha=0.5 )
plt.show()
print("""------------------Training Complished---------------------""" )
print((""" - - Training epoch: """, rp, F''' - - Mse: {mse:.6f}''') )
if draw_e:
draw_error()
return mse
def A__ ( self: Optional[int] ,lowerCamelCase_: Any ) -> Tuple:
# model predict
UpperCAmelCase_ : Union[str, Any] = []
print("""-------------------Start Testing-------------------------""" )
print((""" - - Shape: Test_Data """, np.shape(lowerCamelCase_ )) )
for p in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = np.asmatrix(datas_test[p] )
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Optional[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : str = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : str = data_bp_input
UpperCAmelCase_ : Union[str, Any] = bp_outa * self.vji.T - self.thre_bpa
UpperCAmelCase_ : Optional[int] = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = bp_outa * self.wkj.T - self.thre_bpa
UpperCAmelCase_ : List[Any] = self.sig(lowerCamelCase_ )
produce_out.extend(bp_outa.getA().tolist() )
UpperCAmelCase_ : int = [list(map(self.do_round ,lowerCamelCase_ ) ) for each in produce_out]
return np.asarray(lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Tuple:
# return the data of image after convoluting process so we can check it out
UpperCAmelCase_ : Optional[int] = np.asmatrix(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Dict = self.pooling(lowerCamelCase_ ,self.size_poolinga )
return data_conveda, data_pooleda
if __name__ == "__main__":
pass
| 345 | 1 |
import datasets
from .evaluate import evaluate
UpperCamelCase_ = '''\
@article{hendrycks2021cuad,
title={CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review},
author={Dan Hendrycks and Collin Burns and Anya Chen and Spencer Ball},
journal={arXiv preprint arXiv:2103.06268},
year={2021}
}
'''
UpperCamelCase_ = '''
This metric wrap the official scoring script for version 1 of the Contract
Understanding Atticus Dataset (CUAD).
Contract Understanding Atticus Dataset (CUAD) v1 is a corpus of more than 13,000 labels in 510
commercial legal contracts that have been manually labeled to identify 41 categories of important
clauses that lawyers look for when reviewing contracts in connection with corporate transactions.
'''
UpperCamelCase_ = '''
Computes CUAD scores (EM, F1, AUPR, Precision@80%Recall, and Precision@90%Recall).
Args:
predictions: List of question-answers dictionaries with the following key-values:
- \'id\': id of the question-answer pair as given in the references (see below)
- \'prediction_text\': list of possible texts for the answer, as a list of strings
depending on a threshold on the confidence probability of each prediction.
references: List of question-answers dictionaries with the following key-values:
- \'id\': id of the question-answer pair (see above),
- \'answers\': a Dict in the CUAD dataset format
{
\'text\': list of possible texts for the answer, as a list of strings
\'answer_start\': list of start positions for the answer, as a list of ints
}
Note that answer_start values are not taken into account to compute the metric.
Returns:
\'exact_match\': Exact match (the normalized answer exactly match the gold answer)
\'f1\': The F-score of predicted tokens versus the gold answer
\'aupr\': Area Under the Precision-Recall curve
\'prec_at_80_recall\': Precision at 80% recall
\'prec_at_90_recall\': Precision at 90% recall
Examples:
>>> predictions = [{\'prediction_text\': [\'The seller:\', \'The buyer/End-User: Shenzhen LOHAS Supply Chain Management Co., Ltd.\'], \'id\': \'LohaCompanyltd_20191209_F-1_EX-10.16_11917878_EX-10.16_Supply Agreement__Parties\'}]
>>> references = [{\'answers\': {\'answer_start\': [143, 49], \'text\': [\'The seller:\', \'The buyer/End-User: Shenzhen LOHAS Supply Chain Management Co., Ltd.\']}, \'id\': \'LohaCompanyltd_20191209_F-1_EX-10.16_11917878_EX-10.16_Supply Agreement__Parties\'}]
>>> cuad_metric = datasets.load_metric("cuad")
>>> results = cuad_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 100.0, \'f1\': 100.0, \'aupr\': 0.0, \'prec_at_80_recall\': 1.0, \'prec_at_90_recall\': 1.0}
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class _snake_case ( datasets.Metric ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> int:
return datasets.MetricInfo(
description=_DESCRIPTION ,citation=_CITATION ,inputs_description=_KWARGS_DESCRIPTION ,features=datasets.Features(
{
"""predictions""": {
"""id""": datasets.Value("""string""" ),
"""prediction_text""": datasets.features.Sequence(datasets.Value("""string""" ) ),
},
"""references""": {
"""id""": datasets.Value("""string""" ),
"""answers""": datasets.features.Sequence(
{
"""text""": datasets.Value("""string""" ),
"""answer_start""": datasets.Value("""int32""" ),
} ),
},
} ) ,codebase_urls=["""https://www.atticusprojectai.org/cuad"""] ,reference_urls=["""https://www.atticusprojectai.org/cuad"""] ,)
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : str = {prediction["""id"""]: prediction["""prediction_text"""] for prediction in predictions}
UpperCAmelCase_ : Any = [
{
"""paragraphs""": [
{
"""qas""": [
{
"""answers""": [{"""text""": answer_text} for answer_text in ref["""answers"""]["""text"""]],
"""id""": ref["""id"""],
}
for ref in references
]
}
]
}
]
UpperCAmelCase_ : List[Any] = evaluate(dataset=lowerCamelCase_ ,predictions=lowerCamelCase_ )
return score
| 345 |
import json
import os
import unittest
from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = CTRLTokenizer
A__ : Optional[Any] = False
A__ : str = False
def A__ ( self: Optional[int] ) -> List[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
UpperCAmelCase_ : Dict = ["""adapt""", """re@@""", """a@@""", """apt""", """c@@""", """t""", """<unk>"""]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : List[Any] = ["""#version: 0.2""", """a p""", """ap t</w>""", """r e""", """a d""", """ad apt</w>""", """"""]
UpperCAmelCase_ : Optional[Any] = {"""unk_token""": """<unk>"""}
UpperCAmelCase_ : Union[str, Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : Optional[Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase_ ) + """\n""" )
with open(self.merges_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase_ ) )
def A__ ( self: Optional[int] ,**lowerCamelCase_: Any ) -> str:
kwargs.update(self.special_tokens_map )
return CTRLTokenizer.from_pretrained(self.tmpdirname ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: int ) -> str:
UpperCAmelCase_ : List[str] = """adapt react readapt apt"""
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
return input_text, output_text
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = CTRLTokenizer(self.vocab_file ,self.merges_file ,**self.special_tokens_map )
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
UpperCAmelCase_ : Optional[int] = """adapt re@@ a@@ c@@ t re@@ adapt apt""".split()
UpperCAmelCase_ : Tuple = tokenizer.tokenize(lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = tokens + [tokenizer.unk_token]
UpperCAmelCase_ : List[str] = [0, 1, 2, 4, 5, 1, 0, 3, 6]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase_ ) ,lowerCamelCase_ )
| 345 | 1 |
from __future__ import annotations
import random
# Maximum size of the population. Bigger could be faster but is more memory expensive.
UpperCamelCase_ = 200
# Number of elements selected in every generation of evolution. The selection takes
# place from best to worst of that generation and must be smaller than N_POPULATION.
UpperCamelCase_ = 50
# Probability that an element of a generation can mutate, changing one of its genes.
# This will guarantee that all genes will be used during evolution.
UpperCamelCase_ = 0.4
# Just a seed to improve randomness required by the algorithm.
random.seed(random.randint(0, 1000))
def lowerCamelCase_ ( _a : str , _a : str ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = len([g for position, g in enumerate(_a ) if g == main_target[position]] )
return (item, float(_a ))
def lowerCamelCase_ ( _a : str , _a : str ):
'''simple docstring'''
UpperCAmelCase_ : int = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ : Optional[Any] = parent_a[:random_slice] + parent_a[random_slice:]
UpperCAmelCase_ : Tuple = parent_a[:random_slice] + parent_a[random_slice:]
return (child_a, child_a)
def lowerCamelCase_ ( _a : str , _a : list[str] ):
'''simple docstring'''
UpperCAmelCase_ : str = list(_a )
if random.uniform(0 , 1 ) < MUTATION_PROBABILITY:
UpperCAmelCase_ : Tuple = random.choice(_a )
return "".join(_a )
def lowerCamelCase_ ( _a : tuple[str, float] , _a : list[tuple[str, float]] , _a : list[str] , ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = []
# Generate more children proportionally to the fitness score.
UpperCAmelCase_ : int = int(parent_a[1] * 100 ) + 1
UpperCAmelCase_ : Dict = 10 if child_n >= 10 else child_n
for _ in range(_a ):
UpperCAmelCase_ : List[str] = population_score[random.randint(0 , _a )][0]
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = crossover(parent_a[0] , _a )
# Append new string to the population list.
pop.append(mutate(_a , _a ) )
pop.append(mutate(_a , _a ) )
return pop
def lowerCamelCase_ ( _a : str , _a : list[str] , _a : bool = True ):
'''simple docstring'''
if N_POPULATION < N_SELECTED:
UpperCAmelCase_ : Any = F'''{N_POPULATION} must be bigger than {N_SELECTED}'''
raise ValueError(_a )
# Verify that the target contains no genes besides the ones inside genes variable.
UpperCAmelCase_ : Tuple = sorted({c for c in target if c not in genes} )
if not_in_genes_list:
UpperCAmelCase_ : List[Any] = F'''{not_in_genes_list} is not in genes list, evolution cannot converge'''
raise ValueError(_a )
# Generate random starting population.
UpperCAmelCase_ : Any = []
for _ in range(_a ):
population.append("""""".join([random.choice(_a ) for i in range(len(_a ) )] ) )
# Just some logs to know what the algorithms is doing.
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = 0, 0
# This loop will end when we find a perfect match for our target.
while True:
generation += 1
total_population += len(_a )
# Random population created. Now it's time to evaluate.
# Adding a bit of concurrency can make everything faster,
#
# import concurrent.futures
# population_score: list[tuple[str, float]] = []
# with concurrent.futures.ThreadPoolExecutor(
# max_workers=NUM_WORKERS) as executor:
# futures = {executor.submit(evaluate, item) for item in population}
# concurrent.futures.wait(futures)
# population_score = [item.result() for item in futures]
#
# but with a simple algorithm like this, it will probably be slower.
# We just need to call evaluate for every item inside the population.
UpperCAmelCase_ : int = [evaluate(_a , _a ) for item in population]
# Check if there is a matching evolution.
UpperCAmelCase_ : List[str] = sorted(_a , key=lambda _a : x[1] , reverse=_a )
if population_score[0][0] == target:
return (generation, total_population, population_score[0][0])
# Print the best result every 10 generation.
# Just to know that the algorithm is working.
if debug and generation % 10 == 0:
print(
F'''\nGeneration: {generation}'''
F'''\nTotal Population:{total_population}'''
F'''\nBest score: {population_score[0][1]}'''
F'''\nBest string: {population_score[0][0]}''' )
# Flush the old population, keeping some of the best evolutions.
# Keeping this avoid regression of evolution.
UpperCAmelCase_ : Any = population[: int(N_POPULATION / 3 )]
population.clear()
population.extend(_a )
# Normalize population score to be between 0 and 1.
UpperCAmelCase_ : Optional[int] = [
(item, score / len(_a )) for item, score in population_score
]
# This is selection
for i in range(_a ):
population.extend(select(population_score[int(_a )] , _a , _a ) )
# Check if the population has already reached the maximum value and if so,
# break the cycle. If this check is disabled, the algorithm will take
# forever to compute large strings, but will also calculate small strings in
# a far fewer generations.
if len(_a ) > N_POPULATION:
break
if __name__ == "__main__":
UpperCamelCase_ = (
'''This is a genetic algorithm to evaluate, combine, evolve, and mutate a string!'''
)
UpperCamelCase_ = list(
''' ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklm'''
'''nopqrstuvwxyz.,;!?+-*#@^\'èéòà€ù=)(&%$£/\\'''
)
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = basic(target_str, genes_list)
print(
F"\nGeneration: {generation}\nTotal Population: {population}\nTarget: {target}"
)
| 345 |
from __future__ import annotations
from typing import Dict
from ...configuration_utils import PretrainedConfig
UpperCamelCase_ = {
'''susnato/ernie-m-base_pytorch''': '''https://huggingface.co/susnato/ernie-m-base_pytorch/blob/main/config.json''',
'''susnato/ernie-m-large_pytorch''': '''https://huggingface.co/susnato/ernie-m-large_pytorch/blob/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = "ernie_m"
A__ : Dict[str, str] = {"dropout": "classifier_dropout", "num_classes": "num_labels"}
def __init__( self: str ,lowerCamelCase_: int = 250002 ,lowerCamelCase_: int = 768 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 3072 ,lowerCamelCase_: str = "gelu" ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: int = 514 ,lowerCamelCase_: float = 0.0_2 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: float = 1e-05 ,lowerCamelCase_: Any=None ,lowerCamelCase_: List[Any]=False ,lowerCamelCase_: Tuple=0.0 ,**lowerCamelCase_: Optional[int] ,) -> Optional[Any]:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = vocab_size
UpperCAmelCase_ : Any = hidden_size
UpperCAmelCase_ : Optional[Any] = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : List[Any] = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = initializer_range
UpperCAmelCase_ : Union[str, Any] = layer_norm_eps
UpperCAmelCase_ : List[Any] = classifier_dropout
UpperCAmelCase_ : str = is_decoder
UpperCAmelCase_ : List[str] = act_dropout
| 345 | 1 |
import numpy as np
import torch
from torch.nn import CrossEntropyLoss
from transformers import AutoModelForCausalLM, AutoTokenizer
import datasets
from datasets import logging
UpperCamelCase_ = '''\
'''
UpperCamelCase_ = '''
Perplexity (PPL) is one of the most common metrics for evaluating language models.
It is defined as the exponentiated average negative log-likelihood of a sequence.
For more information, see https://huggingface.co/docs/transformers/perplexity
'''
UpperCamelCase_ = '''
Args:
model_id (str): model used for calculating Perplexity
NOTE: Perplexity can only be calculated for causal language models.
This includes models such as gpt2, causal variations of bert,
causal versions of t5, and more (the full list can be found
in the AutoModelForCausalLM documentation here:
https://huggingface.co/docs/transformers/master/en/model_doc/auto#transformers.AutoModelForCausalLM )
input_texts (list of str): input text, each separate text snippet
is one list entry.
batch_size (int): the batch size to run texts through the model. Defaults to 16.
add_start_token (bool): whether to add the start token to the texts,
so the perplexity can include the probability of the first word. Defaults to True.
device (str): device to run on, defaults to \'cuda\' when available
Returns:
perplexity: dictionary containing the perplexity scores for the texts
in the input list, as well as the mean perplexity. If one of the input texts is
longer than the max input length of the model, then it is truncated to the
max length for the perplexity computation.
Examples:
Example 1:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = ["lorem ipsum", "Happy Birthday!", "Bienvenue"]
>>> results = perplexity.compute(model_id=\'gpt2\',
... add_start_token=False,
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
78.22
>>> print(round(results["perplexities"][0], 2))
11.11
Example 2:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = datasets.load_dataset("wikitext",
... "wikitext-2-raw-v1",
... split="test")["text"][:50] # doctest:+ELLIPSIS
[...]
>>> input_texts = [s for s in input_texts if s!=\'\']
>>> results = perplexity.compute(model_id=\'gpt2\',
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
60.35
>>> print(round(results["perplexities"][0], 2))
81.12
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class _snake_case ( datasets.Metric ):
'''simple docstring'''
def A__ ( self: Tuple ) -> List[Any]:
return datasets.MetricInfo(
description=_DESCRIPTION ,citation=_CITATION ,inputs_description=_KWARGS_DESCRIPTION ,features=datasets.Features(
{
"""input_texts""": datasets.Value("""string""" ),
} ) ,reference_urls=["""https://huggingface.co/docs/transformers/perplexity"""] ,)
def A__ ( self: List[Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int = 16 ,lowerCamelCase_: bool = True ,lowerCamelCase_: Any=None ) -> List[str]:
if device is not None:
assert device in ["gpu", "cpu", "cuda"], "device should be either gpu or cpu."
if device == "gpu":
UpperCAmelCase_ : Union[str, Any] = """cuda"""
else:
UpperCAmelCase_ : Union[str, Any] = """cuda""" if torch.cuda.is_available() else """cpu"""
UpperCAmelCase_ : int = AutoModelForCausalLM.from_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model.to(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained(lowerCamelCase_ )
# if batch_size > 1 (which generally leads to padding being required), and
# if there is not an already assigned pad_token, assign an existing
# special token to also be the padding token
if tokenizer.pad_token is None and batch_size > 1:
UpperCAmelCase_ : Optional[int] = list(tokenizer.special_tokens_map_extended.values() )
# check that the model already has at least one special token defined
assert (
len(lowerCamelCase_ ) > 0
), "If batch_size > 1, model must have at least one special token to use for padding. Please use a different model or set batch_size=1."
# assign one of the special tokens to also be the pad token
tokenizer.add_special_tokens({"""pad_token""": existing_special_tokens[0]} )
if add_start_token:
# leave room for <BOS> token to be added:
assert (
tokenizer.bos_token is not None
), "Input model must already have a BOS token if using add_start_token=True. Please use a different model, or set add_start_token=False"
UpperCAmelCase_ : int = model.config.max_length - 1
else:
UpperCAmelCase_ : str = model.config.max_length
UpperCAmelCase_ : Tuple = tokenizer(
lowerCamelCase_ ,add_special_tokens=lowerCamelCase_ ,padding=lowerCamelCase_ ,truncation=lowerCamelCase_ ,max_length=lowerCamelCase_ ,return_tensors="""pt""" ,return_attention_mask=lowerCamelCase_ ,).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = encodings["""input_ids"""]
UpperCAmelCase_ : Any = encodings["""attention_mask"""]
# check that each input is long enough:
if add_start_token:
assert torch.all(torch.ge(attn_masks.sum(1 ) ,1 ) ), "Each input text must be at least one token long."
else:
assert torch.all(
torch.ge(attn_masks.sum(1 ) ,2 ) ), "When add_start_token=False, each input text must be at least two tokens long. Run with add_start_token=True if inputting strings of only one token, and remove all empty input strings."
UpperCAmelCase_ : List[str] = []
UpperCAmelCase_ : List[Any] = CrossEntropyLoss(reduction="""none""" )
for start_index in logging.tqdm(range(0 ,len(lowerCamelCase_ ) ,lowerCamelCase_ ) ):
UpperCAmelCase_ : Tuple = min(start_index + batch_size ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : List[Any] = encoded_texts[start_index:end_index]
UpperCAmelCase_ : int = attn_masks[start_index:end_index]
if add_start_token:
UpperCAmelCase_ : Union[str, Any] = torch.tensor([[tokenizer.bos_token_id]] * encoded_batch.size(dim=0 ) ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.cat([bos_tokens_tensor, encoded_batch] ,dim=1 )
UpperCAmelCase_ : Optional[int] = torch.cat(
[torch.ones(bos_tokens_tensor.size() ,dtype=torch.intaa ).to(lowerCamelCase_ ), attn_mask] ,dim=1 )
UpperCAmelCase_ : int = encoded_batch
with torch.no_grad():
UpperCAmelCase_ : str = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ).logits
UpperCAmelCase_ : Any = out_logits[..., :-1, :].contiguous()
UpperCAmelCase_ : int = labels[..., 1:].contiguous()
UpperCAmelCase_ : Optional[Any] = attn_mask[..., 1:].contiguous()
UpperCAmelCase_ : List[Any] = torch.expa(
(loss_fct(shift_logits.transpose(1 ,2 ) ,lowerCamelCase_ ) * shift_attention_mask_batch).sum(1 )
/ shift_attention_mask_batch.sum(1 ) )
ppls += perplexity_batch.tolist()
return {"perplexities": ppls, "mean_perplexity": np.mean(lowerCamelCase_ )}
| 345 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
UpperCamelCase_ = logging.getLogger(__name__)
torch.set_grad_enabled(False)
UpperCamelCase_ = '''cuda''' if torch.cuda.is_available() else '''cpu'''
def lowerCamelCase_ ( _a : str , _a : Any=100 , _a : int=" " ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = text.split(_a )
return [character.join(text[i : i + n] ).strip() for i in range(0 , len(_a ) , _a )]
def lowerCamelCase_ ( _a : dict ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = [], []
for title, text in zip(documents["""title"""] , documents["""text"""] ):
if text is not None:
for passage in split_text(_a ):
titles.append(title if title is not None else """""" )
texts.append(_a )
return {"title": titles, "text": texts}
def lowerCamelCase_ ( _a : dict , _a : DPRContextEncoder , _a : DPRContextEncoderTokenizerFast ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = ctx_tokenizer(
documents["""title"""] , documents["""text"""] , truncation=_a , padding="""longest""" , return_tensors="""pt""" )["""input_ids"""]
UpperCAmelCase_ : Tuple = ctx_encoder(input_ids.to(device=_a ) , return_dict=_a ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def lowerCamelCase_ ( _a : "RagExampleArguments" , _a : "ProcessingArguments" , _a : "IndexHnswArguments" , ):
'''simple docstring'''
logger.info("""Step 1 - Create the dataset""" )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
UpperCAmelCase_ : Optional[int] = load_dataset(
"""csv""" , data_files=[rag_example_args.csv_path] , split="""train""" , delimiter="""\t""" , column_names=["""title""", """text"""] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
UpperCAmelCase_ : Tuple = dataset.map(_a , batched=_a , num_proc=processing_args.num_proc )
# And compute the embeddings
UpperCAmelCase_ : List[str] = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=_a )
UpperCAmelCase_ : Dict = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
UpperCAmelCase_ : Any = Features(
{"""text""": Value("""string""" ), """title""": Value("""string""" ), """embeddings""": Sequence(Value("""float32""" ) )} ) # optional, save as float32 instead of float64 to save space
UpperCAmelCase_ : List[str] = dataset.map(
partial(_a , ctx_encoder=_a , ctx_tokenizer=_a ) , batched=_a , batch_size=processing_args.batch_size , features=_a , )
# And finally save your dataset
UpperCAmelCase_ : Union[str, Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset""" )
dataset.save_to_disk(_a )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info("""Step 2 - Index the dataset""" )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
UpperCAmelCase_ : Union[str, Any] = faiss.IndexHNSWFlat(index_hnsw_args.d , index_hnsw_args.m , faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index("""embeddings""" , custom_index=_a )
# And save the index
UpperCAmelCase_ : Optional[Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset_hnsw_index.faiss""" )
dataset.get_index("""embeddings""" ).save(_a )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" / "my_knowledge_dataset.csv" ) , metadata={"help": "Path to a tab-separated csv file with columns 'title' and 'text'"} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."} , )
A__ : str = field(
default="facebook/rag-sequence-nq" , metadata={"help": "The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"} , )
A__ : str = field(
default="facebook/dpr-ctx_encoder-multiset-base" , metadata={
"help": (
"The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"
" 'facebook/dpr-ctx_encoder-multiset-base'"
)
} , )
A__ : Optional[str] = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" ) , metadata={"help": "Path to a directory where the dataset passages and the index will be saved"} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": "The number of processes to use to split the documents into passages. Default is single process."
} , )
A__ : int = field(
default=16 , metadata={
"help": "The batch size to use when computing the passages embeddings using the DPR context encoder."
} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : int = field(
default=768 , metadata={"help": "The dimension of the embeddings to pass to the HNSW Faiss index."} , )
A__ : int = field(
default=128 , metadata={
"help": (
"The number of bi-directional links created for every new element during the HNSW index construction."
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
UpperCamelCase_ = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
UpperCamelCase_ = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 345 | 1 |
import multiprocessing
import time
from arguments import PretokenizationArguments
from datasets import load_dataset
from transformers import AutoTokenizer, HfArgumentParser
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = {}
UpperCAmelCase_ : Tuple = tokenizer(example["""content"""] , truncation=_a )["""input_ids"""]
UpperCAmelCase_ : int = len(example["""content"""] ) / len(output["""input_ids"""] )
return output
UpperCamelCase_ = HfArgumentParser(PretokenizationArguments)
UpperCamelCase_ = parser.parse_args()
if args.num_workers is None:
UpperCamelCase_ = multiprocessing.cpu_count()
UpperCamelCase_ = AutoTokenizer.from_pretrained(args.tokenizer_dir)
UpperCamelCase_ = time.time()
UpperCamelCase_ = load_dataset(args.dataset_name, split='''train''')
print(F"Dataset loaded in {time.time()-t_start:.2f}s")
UpperCamelCase_ = time.time()
UpperCamelCase_ = ds.map(
tokenize,
num_proc=args.num_workers,
remove_columns=[
'''repo_name''',
'''path''',
'''copies''',
'''size''',
'''content''',
'''license''',
'''hash''',
'''line_mean''',
'''line_max''',
'''alpha_frac''',
'''autogenerated''',
],
)
print(F"Dataset tokenized in {time.time()-t_start:.2f}s")
UpperCamelCase_ = time.time()
ds.push_to_hub(args.tokenized_data_repo)
print(F"Data pushed to the hub in {time.time()-t_start:.2f}s")
| 345 |
import gc
import unittest
import torch
from parameterized import parameterized
from diffusers import AutoencoderKL
from diffusers.utils import floats_tensor, load_hf_numpy, require_torch_gpu, slow, torch_all_close, torch_device
from diffusers.utils.import_utils import is_xformers_available
from diffusers.utils.testing_utils import enable_full_determinism
from .test_modeling_common import ModelTesterMixin, UNetTesterMixin
enable_full_determinism()
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = AutoencoderKL
A__ : Optional[int] = "sample"
A__ : Tuple = 1E-2
@property
def A__ ( self: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = 4
UpperCAmelCase_ : str = 3
UpperCAmelCase_ : Any = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ).to(lowerCamelCase_ )
return {"sample": image}
@property
def A__ ( self: List[str] ) -> Tuple:
return (3, 32, 32)
@property
def A__ ( self: Optional[Any] ) -> Any:
return (3, 32, 32)
def A__ ( self: Any ) -> Tuple:
UpperCAmelCase_ : List[Any] = {
"""block_out_channels""": [32, 64],
"""in_channels""": 3,
"""out_channels""": 3,
"""down_block_types""": ["""DownEncoderBlock2D""", """DownEncoderBlock2D"""],
"""up_block_types""": ["""UpDecoderBlock2D""", """UpDecoderBlock2D"""],
"""latent_channels""": 4,
}
UpperCAmelCase_ : int = self.dummy_input
return init_dict, inputs_dict
def A__ ( self: Optional[Any] ) -> int:
pass
def A__ ( self: str ) -> Any:
pass
@unittest.skipIf(torch_device == """mps""" ,"""Gradient checkpointing skipped on MPS""" )
def A__ ( self: Union[str, Any] ) -> Dict:
# enable deterministic behavior for gradient checkpointing
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.prepare_init_args_and_inputs_for_common()
UpperCAmelCase_ : List[Any] = self.model_class(**lowerCamelCase_ )
model.to(lowerCamelCase_ )
assert not model.is_gradient_checkpointing and model.training
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model.zero_grad()
UpperCAmelCase_ : Any = torch.randn_like(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = (out - labels).mean()
loss.backward()
# re-instantiate the model now enabling gradient checkpointing
UpperCAmelCase_ : str = self.model_class(**lowerCamelCase_ )
# clone model
model_a.load_state_dict(model.state_dict() )
model_a.to(lowerCamelCase_ )
model_a.enable_gradient_checkpointing()
assert model_a.is_gradient_checkpointing and model_a.training
UpperCAmelCase_ : Optional[int] = model_a(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model_a.zero_grad()
UpperCAmelCase_ : Dict = (out_a - labels).mean()
loss_a.backward()
# compare the output and parameters gradients
self.assertTrue((loss - loss_a).abs() < 1e-5 )
UpperCAmelCase_ : Dict = dict(model.named_parameters() )
UpperCAmelCase_ : Union[str, Any] = dict(model_a.named_parameters() )
for name, param in named_params.items():
self.assertTrue(torch_all_close(param.grad.data ,named_params_a[name].grad.data ,atol=5e-5 ) )
def A__ ( self: Optional[Any] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" ,output_loading_info=lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
self.assertEqual(len(loading_info["""missing_keys"""] ) ,0 )
model.to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model(**self.dummy_input )
assert image is not None, "Make sure output is not None"
def A__ ( self: Optional[int] ) -> int:
UpperCAmelCase_ : Dict = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" )
UpperCAmelCase_ : Tuple = model.to(lowerCamelCase_ )
model.eval()
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.manual_seed(0 )
else:
UpperCAmelCase_ : Optional[int] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = torch.randn(
1 ,model.config.in_channels ,model.config.sample_size ,model.config.sample_size ,generator=torch.manual_seed(0 ) ,)
UpperCAmelCase_ : int = image.to(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ,generator=lowerCamelCase_ ).sample
UpperCAmelCase_ : Optional[int] = output[0, -1, -3:, -3:].flatten().cpu()
# Since the VAE Gaussian prior's generator is seeded on the appropriate device,
# the expected output slices are not the same for CPU and GPU.
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.tensor(
[
-4.0078e-01,
-3.8323e-04,
-1.2681e-01,
-1.1462e-01,
2.0095e-01,
1.0893e-01,
-8.8247e-02,
-3.0361e-01,
-9.8644e-03,
] )
elif torch_device == "cpu":
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.1_3_5_2, 0.0_8_7_8, 0.0_4_1_9, -0.0_8_1_8, -0.1_0_6_9, 0.0_6_8_8, -0.1_4_5_8, -0.4_4_4_6, -0.0_0_2_6] )
else:
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.2_4_2_1, 0.4_6_4_2, 0.2_5_0_7, -0.0_4_3_8, 0.0_6_8_2, 0.3_1_6_0, -0.2_0_1_8, -0.0_7_2_7, 0.2_4_8_5] )
self.assertTrue(torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,rtol=1e-2 ) )
@slow
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
return F'''gaussian_noise_s={seed}_shape={'_'.join([str(lowerCamelCase_ ) for s in shape] )}.npy'''
def A__ ( self: Union[str, Any] ) -> Optional[int]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: List[Any]=(4, 3, 512, 512) ,lowerCamelCase_: Optional[Any]=False ) -> Optional[int]:
UpperCAmelCase_ : Tuple = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : Tuple = torch.from_numpy(load_hf_numpy(self.get_file_format(lowerCamelCase_ ,lowerCamelCase_ ) ) ).to(lowerCamelCase_ ).to(lowerCamelCase_ )
return image
def A__ ( self: List[Any] ,lowerCamelCase_: List[str]="CompVis/stable-diffusion-v1-4" ,lowerCamelCase_: Union[str, Any]=False ) -> Any:
UpperCAmelCase_ : Optional[Any] = """fp16""" if fpaa else None
UpperCAmelCase_ : str = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : int = AutoencoderKL.from_pretrained(
lowerCamelCase_ ,subfolder="""vae""" ,torch_dtype=lowerCamelCase_ ,revision=lowerCamelCase_ ,)
model.to(lowerCamelCase_ ).eval()
return model
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any]=0 ) -> Optional[int]:
if torch_device == "mps":
return torch.manual_seed(lowerCamelCase_ )
return torch.Generator(device=lowerCamelCase_ ).manual_seed(lowerCamelCase_ )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_3, 0.9_8_7_8, -0.0_4_9_5, -0.0_7_9_0, -0.2_7_0_9, 0.8_3_7_5, -0.2_0_6_0, -0.0_8_2_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_6, 0.1_1_6_8, 0.1_3_3_2, -0.4_8_4_0, -0.2_5_0_8, -0.0_7_9_1, -0.0_4_9_3, -0.4_0_8_9], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ) -> Tuple:
UpperCAmelCase_ : List[Any] = self.get_sd_vae_model()
UpperCAmelCase_ : int = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Optional[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Tuple = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[33, [-0.0_5_1_3, 0.0_2_8_9, 1.3_7_9_9, 0.2_1_6_6, -0.2_5_7_3, -0.0_8_7_1, 0.5_1_0_3, -0.0_9_9_9]],
[47, [-0.4_1_2_8, -0.1_3_2_0, -0.3_7_0_4, 0.1_9_6_5, -0.4_1_1_6, -0.2_3_3_2, -0.3_3_4_0, 0.2_2_4_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Tuple:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Tuple = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : Optional[int] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_9, 0.9_8_6_6, -0.0_4_8_7, -0.0_7_7_7, -0.2_7_1_6, 0.8_3_6_8, -0.2_0_5_5, -0.0_8_1_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_7, 0.1_1_4_7, 0.1_3_3_3, -0.4_8_4_1, -0.2_5_0_6, -0.0_8_0_5, -0.0_4_9_1, -0.4_0_8_5], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[int] = self.get_sd_vae_model()
UpperCAmelCase_ : Dict = self.get_sd_image(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : str = model(lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : List[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Any = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[13, [-0.2_0_5_1, -0.1_8_0_3, -0.2_3_1_1, -0.2_1_1_4, -0.3_2_9_2, -0.3_5_7_4, -0.2_9_5_3, -0.3_3_2_3]],
[37, [-0.2_6_3_2, -0.2_6_2_5, -0.2_1_9_9, -0.2_7_4_1, -0.4_5_3_9, -0.4_9_9_0, -0.3_7_2_0, -0.4_9_2_5]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: str ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[int] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : str = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : Any = sample[-1, -2:, :2, -2:].flatten().cpu()
UpperCAmelCase_ : Union[str, Any] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 )
@parameterized.expand(
[
# fmt: off
[27, [-0.0_3_6_9, 0.0_2_0_7, -0.0_7_7_6, -0.0_6_8_2, -0.1_7_4_7, -0.1_9_3_0, -0.1_4_6_5, -0.2_0_3_9]],
[16, [-0.1_6_2_8, -0.2_1_3_4, -0.2_7_4_7, -0.2_6_4_2, -0.3_7_7_4, -0.4_4_0_4, -0.3_6_8_7, -0.4_2_7_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: str ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : str = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : str = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=5e-3 )
@parameterized.expand([(13,), (16,), (27,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> int:
UpperCAmelCase_ : Optional[Any] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-1 )
@parameterized.expand([(13,), (16,), (37,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = self.get_sd_vae_model()
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.3_0_0_1, 0.0_9_1_8, -2.6_9_8_4, -3.9_7_2_0, -3.2_0_9_9, -5.0_3_5_3, 1.7_3_3_8, -0.2_0_6_5, 3.4_2_6_7]],
[47, [-1.5_0_3_0, -4.3_8_7_1, -6.0_3_5_5, -9.1_1_5_7, -1.6_6_6_1, -2.7_8_5_3, 2.1_6_0_7, -5.0_8_2_3, 2.5_6_3_3]],
# fmt: on
] )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[Any] = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : str = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : int = model.encode(lowerCamelCase_ ).latent_dist
UpperCAmelCase_ : Optional[Any] = dist.sample(generator=lowerCamelCase_ )
assert list(sample.shape ) == [image.shape[0], 4] + [i // 8 for i in image.shape[2:]]
UpperCAmelCase_ : Tuple = sample[0, -1, -3:, -3:].flatten().cpu()
UpperCAmelCase_ : Optional[Any] = torch.tensor(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = 3e-3 if torch_device != """mps""" else 1e-2
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=lowerCamelCase_ )
| 345 | 1 |
import math
import flax.linen as nn
import jax.numpy as jnp
def lowerCamelCase_ ( _a : jnp.ndarray , _a : int , _a : float = 1 , _a : float = 1 , _a : float = 1.0E4 , _a : bool = False , _a : float = 1.0 , ):
'''simple docstring'''
assert timesteps.ndim == 1, "Timesteps should be a 1d-array"
assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even'''
UpperCAmelCase_ : Optional[Any] = float(embedding_dim // 2 )
UpperCAmelCase_ : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift)
UpperCAmelCase_ : Tuple = min_timescale * jnp.exp(jnp.arange(_a , dtype=jnp.floataa ) * -log_timescale_increment )
UpperCAmelCase_ : Any = jnp.expand_dims(_a , 1 ) * jnp.expand_dims(_a , 0 )
# scale embeddings
UpperCAmelCase_ : List[str] = scale * emb
if flip_sin_to_cos:
UpperCAmelCase_ : int = jnp.concatenate([jnp.cos(_a ), jnp.sin(_a )] , axis=1 )
else:
UpperCAmelCase_ : Optional[Any] = jnp.concatenate([jnp.sin(_a ), jnp.cos(_a )] , axis=1 )
UpperCAmelCase_ : Union[str, Any] = jnp.reshape(_a , [jnp.shape(_a )[0], embedding_dim] )
return signal
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int = 32
A__ : jnp.dtype = jnp.floataa
@nn.compact
def __call__( self: Any ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Optional[int] = nn.Dense(self.time_embed_dim ,dtype=self.dtype ,name="""linear_1""" )(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = nn.silu(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = nn.Dense(self.time_embed_dim ,dtype=self.dtype ,name="""linear_2""" )(lowerCamelCase_ )
return temb
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int = 32
A__ : bool = False
A__ : float = 1
@nn.compact
def __call__( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
return get_sinusoidal_embeddings(
lowerCamelCase_ ,embedding_dim=self.dim ,flip_sin_to_cos=self.flip_sin_to_cos ,freq_shift=self.freq_shift )
| 345 |
import copy
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Audio, Features, Value
from .base import TaskTemplate
@dataclass(frozen=__snake_case )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = field(default="automatic-speech-recognition" , metadata={"include_in_asdict_even_if_is_default": True} )
A__ : ClassVar[Features] = Features({"audio": Audio()} )
A__ : ClassVar[Features] = Features({"transcription": Value("string" )} )
A__ : str = "audio"
A__ : str = "transcription"
def A__ ( self: int ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
if self.audio_column not in features:
raise ValueError(F'''Column {self.audio_column} is not present in features.''' )
if not isinstance(features[self.audio_column] ,lowerCamelCase_ ):
raise ValueError(F'''Column {self.audio_column} is not an Audio type.''' )
UpperCAmelCase_ : Any = copy.deepcopy(self )
UpperCAmelCase_ : Union[str, Any] = self.input_schema.copy()
UpperCAmelCase_ : Any = features[self.audio_column]
UpperCAmelCase_ : Union[str, Any] = input_schema
return task_template
@property
def A__ ( self: List[str] ) -> Dict[str, str]:
return {self.audio_column: "audio", self.transcription_column: "transcription"}
| 345 | 1 |
from typing import TYPE_CHECKING
# rely on isort to merge the imports
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
UpperCamelCase_ = {
'''configuration_autoformer''': [
'''AUTOFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''AutoformerConfig''',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''AUTOFORMER_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''AutoformerForPrediction''',
'''AutoformerModel''',
'''AutoformerPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_autoformer import (
AUTOFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
AutoformerConfig,
)
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_autoformer import (
AUTOFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
AutoformerForPrediction,
AutoformerModel,
AutoformerPreTrainedModel,
)
else:
import sys
UpperCamelCase_ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 345 |
from collections import OrderedDict
from typing import TYPE_CHECKING, Any, Mapping, Optional
from packaging import version
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...onnx.utils import compute_effective_axis_dimension
from ...utils import logging
if TYPE_CHECKING:
from ...processing_utils import ProcessorMixin
from ...utils import TensorType
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/layoutlmv3-base''': '''https://huggingface.co/microsoft/layoutlmv3-base/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "layoutlmv3"
def __init__( self: str ,lowerCamelCase_: Any=50265 ,lowerCamelCase_: int=768 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=3072 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Any=0.1 ,lowerCamelCase_: Tuple=512 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: List[str]=1e-5 ,lowerCamelCase_: int=1 ,lowerCamelCase_: int=0 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Dict=1024 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=32 ,lowerCamelCase_: Union[str, Any]=128 ,lowerCamelCase_: Tuple=64 ,lowerCamelCase_: Tuple=256 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Any=True ,lowerCamelCase_: Dict=224 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Optional[int]=16 ,lowerCamelCase_: Dict=None ,**lowerCamelCase_: str ,) -> List[Any]:
super().__init__(
vocab_size=lowerCamelCase_ ,hidden_size=lowerCamelCase_ ,num_hidden_layers=lowerCamelCase_ ,num_attention_heads=lowerCamelCase_ ,intermediate_size=lowerCamelCase_ ,hidden_act=lowerCamelCase_ ,hidden_dropout_prob=lowerCamelCase_ ,attention_probs_dropout_prob=lowerCamelCase_ ,max_position_embeddings=lowerCamelCase_ ,type_vocab_size=lowerCamelCase_ ,initializer_range=lowerCamelCase_ ,layer_norm_eps=lowerCamelCase_ ,pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = max_ad_position_embeddings
UpperCAmelCase_ : Optional[int] = coordinate_size
UpperCAmelCase_ : Optional[int] = shape_size
UpperCAmelCase_ : Optional[Any] = has_relative_attention_bias
UpperCAmelCase_ : Optional[int] = rel_pos_bins
UpperCAmelCase_ : Union[str, Any] = max_rel_pos
UpperCAmelCase_ : Dict = has_spatial_attention_bias
UpperCAmelCase_ : Optional[int] = rel_ad_pos_bins
UpperCAmelCase_ : Tuple = max_rel_ad_pos
UpperCAmelCase_ : Union[str, Any] = text_embed
UpperCAmelCase_ : Optional[Any] = visual_embed
UpperCAmelCase_ : List[str] = input_size
UpperCAmelCase_ : str = num_channels
UpperCAmelCase_ : Optional[int] = patch_size
UpperCAmelCase_ : Tuple = classifier_dropout
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = version.parse("1.12" )
@property
def A__ ( self: Dict ) -> Mapping[str, Mapping[int, str]]:
# The order of inputs is different for question answering and sequence classification
if self.task in ["question-answering", "sequence-classification"]:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels""", 2: """height""", 3: """width"""}),
] )
else:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels"""}),
] )
@property
def A__ ( self: Any ) -> float:
return 1e-5
@property
def A__ ( self: int ) -> int:
return 12
def A__ ( self: List[str] ,lowerCamelCase_: "ProcessorMixin" ,lowerCamelCase_: int = -1 ,lowerCamelCase_: int = -1 ,lowerCamelCase_: bool = False ,lowerCamelCase_: Optional["TensorType"] = None ,lowerCamelCase_: int = 3 ,lowerCamelCase_: int = 40 ,lowerCamelCase_: int = 40 ,) -> Mapping[str, Any]:
setattr(processor.image_processor ,"""apply_ocr""" ,lowerCamelCase_ )
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
UpperCAmelCase_ : List[str] = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_batch ,num_token_to_add=0 )
# If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX
UpperCAmelCase_ : int = processor.tokenizer.num_special_tokens_to_add(lowerCamelCase_ )
UpperCAmelCase_ : int = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_sequence ,num_token_to_add=lowerCamelCase_ )
# Generate dummy inputs according to compute batch and sequence
UpperCAmelCase_ : Optional[int] = [[""" """.join([processor.tokenizer.unk_token] ) * seq_length]] * batch_size
# Generate dummy bounding boxes
UpperCAmelCase_ : List[Any] = [[[48, 84, 73, 128]]] * batch_size
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
# batch_size = compute_effective_axis_dimension(batch_size, fixed_dimension=OnnxConfig.default_fixed_batch)
UpperCAmelCase_ : Any = self._generate_dummy_images(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = dict(
processor(
lowerCamelCase_ ,text=lowerCamelCase_ ,boxes=lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,) )
return inputs
| 345 | 1 |
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Optional[Any] = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[Any] ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : int = UNetaDConditionModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=4 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,)
return model
@property
def A__ ( self: str ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = AutoencoderKL(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Tuple:
def extract(*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: str ):
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = torch.ones([0] )
def A__ ( self: List[Any] ,lowerCamelCase_: str ) -> int:
self.pixel_values.to(lowerCamelCase_ )
return self
return Out()
return extract
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : int = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : int = self.dummy_cond_unet
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(
beta_start=0.0_0_0_8_5 ,beta_end=0.0_1_2 ,beta_schedule="""scaled_linear""" ,clip_sample=lowerCamelCase_ ,set_alpha_to_one=lowerCamelCase_ ,)
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : int = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : List[Any] = output.images
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Dict = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : int = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Dict = self.dummy_cond_unet
UpperCAmelCase_ : List[Any] = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : Union[str, Any] = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : str = output.images
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(
"""hf-internal-testing/tiny-stable-diffusion-lms-pipe""" ,safety_checker=lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(pipe.scheduler ,lowerCamelCase_ )
assert pipe.safety_checker is None
UpperCAmelCase_ : List[Any] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(lowerCamelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
UpperCAmelCase_ : Optional[int] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Tuple = self.dummy_cond_unet
UpperCAmelCase_ : Dict = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# put models in fp16
UpperCAmelCase_ : Optional[Any] = unet.half()
UpperCAmelCase_ : Optional[int] = vae.half()
UpperCAmelCase_ : int = bert.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = sd_pipe([prompt] ,num_inference_steps=2 ,output_type="""np""" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Tuple = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : str = (
"""portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"""
""" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"""
""" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"""
""" children from bahnhof zoo, detailed """
)
UpperCAmelCase_ : Optional[int] = 4003660346
UpperCAmelCase_ : int = 7
# without safety guidance (sld_guidance_scale = 0)
UpperCAmelCase_ : Dict = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Optional[int] = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Dict = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
# without safety guidance (strong configuration)
UpperCAmelCase_ : Union[str, Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Any = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Tuple = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : str = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Any = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = """padme amidala taking a bath artwork, safe for work, no nudity"""
UpperCAmelCase_ : List[Any] = 2734971755
UpperCAmelCase_ : Optional[Any] = 7
UpperCAmelCase_ : int = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : Tuple = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
UpperCAmelCase_ : Any = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : List[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" )
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = (
"""the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."""
""" leyendecker"""
)
UpperCAmelCase_ : Optional[Any] = 1044355234
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : List[Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Any = output.images
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-7
UpperCAmelCase_ : Optional[int] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Any = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Any = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 345 |
import argparse
from argparse import Namespace
import torch
from torch import nn
from transformers import XGLMConfig, XGLMForCausalLM
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = [
"""decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = emb.weight.shape
UpperCAmelCase_ : Tuple = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : List[Any] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = torch.load(_a , map_location="""cpu""" )
UpperCAmelCase_ : Dict = Namespace(**checkpoint["""cfg"""]["""model"""] )
UpperCAmelCase_ : Optional[int] = checkpoint["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = state_dict["""decoder.embed_tokens.weight"""].shape[0]
UpperCAmelCase_ : List[str] = {key.replace("""decoder""" , """model""" ): val for key, val in state_dict.items()}
UpperCAmelCase_ : int = XGLMConfig(
vocab_size=_a , max_position_embeddings=args.max_target_positions , num_layers=args.decoder_layers , attention_heads=args.decoder_attention_heads , ffn_dim=args.decoder_ffn_embed_dim , d_model=args.decoder_embed_dim , layerdrop=args.decoder_layerdrop , dropout=args.dropout , attention_dropout=args.attention_dropout , activation_dropout=args.activation_dropout , activation_function="""gelu""" , scale_embedding=not args.no_scale_embedding , tie_word_embeddings=args.share_decoder_input_output_embed , )
UpperCAmelCase_ : List[str] = XGLMForCausalLM(_a )
UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
print(_a )
UpperCAmelCase_ : Optional[Any] = make_linear_from_emb(model.model.embed_tokens )
return model
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument('''fairseq_path''', type=str, help='''path to a model.pt on local filesystem.''')
parser.add_argument('''pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = convert_fairseq_xglm_checkpoint_from_disk(args.fairseq_path)
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 | 1 |
import logging
import os
import quant_trainer
import torch
from torch.utils.data import DataLoader
from transformers import Trainer, is_torch_tpu_available
from transformers.trainer_utils import PredictionOutput
UpperCamelCase_ = logging.getLogger(__name__)
if is_torch_tpu_available(check_device=False):
import torch_xla.core.xla_model as xm
import torch_xla.debug.metrics as met
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: Tuple ,*lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: int=None ,lowerCamelCase_: str=None ,lowerCamelCase_: Any=None ,**lowerCamelCase_: Optional[int] ) -> List[str]:
super().__init__(*lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = eval_examples
UpperCAmelCase_ : str = post_process_function
UpperCAmelCase_ : str = quant_trainer_args
UpperCAmelCase_ : List[Any] = 128 # default number of calibration samples
def A__ ( self: str ,lowerCamelCase_: Optional[int]=None ) -> Union[str, Any]:
if calib_dataset is None and self.calib_dataset is None:
raise ValueError("""Trainer: calibration requires an calib_dataset.""" )
UpperCAmelCase_ : Union[str, Any] = calib_dataset if calib_dataset is not None else self.calib_dataset
UpperCAmelCase_ : Optional[int] = self._remove_unused_columns(lowerCamelCase_ ,description="""Calibration""" )
return DataLoader(
lowerCamelCase_ ,batch_size=self.args.eval_batch_size ,collate_fn=self.data_collator ,drop_last=self.args.dataloader_drop_last ,num_workers=self.args.dataloader_num_workers ,pin_memory=self.args.dataloader_pin_memory ,shuffle=lowerCamelCase_ ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any]=None ) -> str:
UpperCAmelCase_ : Tuple = self.train_dataset if calib_dataset is None else calib_dataset
UpperCAmelCase_ : Tuple = self.get_calib_dataloader(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.model
quant_trainer.configure_model(lowerCamelCase_ ,self.quant_trainer_args ,calib=lowerCamelCase_ )
model.eval()
quant_trainer.enable_calibration(lowerCamelCase_ )
logger.info("""***** Running calibration *****""" )
logger.info(F''' Num examples = {self.calib_num}''' )
logger.info(F''' Batch size = {calib_dataloader.batch_size}''' )
for step, inputs in enumerate(lowerCamelCase_ ):
# Prediction step
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.prediction_step(lowerCamelCase_ ,lowerCamelCase_ ,prediction_loss_only=lowerCamelCase_ )
if (step + 1) * calib_dataloader.batch_size >= self.calib_num:
break
quant_trainer.finish_calibration(lowerCamelCase_ ,self.quant_trainer_args )
UpperCAmelCase_ : int = model
def A__ ( self: str ,lowerCamelCase_: Optional[int]=None ,lowerCamelCase_: Optional[Any]=None ,lowerCamelCase_: int=None ,lowerCamelCase_: str = "eval" ) -> int:
UpperCAmelCase_ : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset
UpperCAmelCase_ : Optional[Any] = self.get_eval_dataloader(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = self.eval_examples if eval_examples is None else eval_examples
# Temporarily disable metric computation, we will do it in the loop here.
UpperCAmelCase_ : int = self.compute_metrics
UpperCAmelCase_ : Optional[Any] = None
UpperCAmelCase_ : Any = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop
try:
UpperCAmelCase_ : Dict = eval_loop(
lowerCamelCase_ ,description="""Evaluation""" ,prediction_loss_only=True if compute_metrics is None else None ,ignore_keys=lowerCamelCase_ ,)
finally:
UpperCAmelCase_ : Optional[Any] = compute_metrics
if self.post_process_function is not None and self.compute_metrics is not None:
UpperCAmelCase_ : List[str] = self.post_process_function(lowerCamelCase_ ,lowerCamelCase_ ,output.predictions )
UpperCAmelCase_ : Dict = self.compute_metrics(lowerCamelCase_ )
# Prefix all keys with metric_key_prefix + '_'
for key in list(metrics.keys() ):
if not key.startswith(F'''{metric_key_prefix}_''' ):
UpperCAmelCase_ : int = metrics.pop(lowerCamelCase_ )
self.log(lowerCamelCase_ )
else:
UpperCAmelCase_ : Optional[int] = {}
if self.args.tpu_metrics_debug or self.args.debug:
# tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.)
xm.master_print(met.metrics_report() )
UpperCAmelCase_ : List[Any] = self.callback_handler.on_evaluate(self.args ,self.state ,self.control ,lowerCamelCase_ )
return metrics
def A__ ( self: str ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any]=None ,lowerCamelCase_: str = "test" ) -> Tuple:
UpperCAmelCase_ : List[str] = self.get_test_dataloader(lowerCamelCase_ )
# Temporarily disable metric computation, we will do it in the loop here.
UpperCAmelCase_ : List[Any] = self.compute_metrics
UpperCAmelCase_ : Any = None
UpperCAmelCase_ : str = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop
try:
UpperCAmelCase_ : Dict = eval_loop(
lowerCamelCase_ ,description="""Prediction""" ,prediction_loss_only=True if compute_metrics is None else None ,ignore_keys=lowerCamelCase_ ,)
finally:
UpperCAmelCase_ : str = compute_metrics
if self.post_process_function is None or self.compute_metrics is None:
return output
UpperCAmelCase_ : List[str] = self.post_process_function(lowerCamelCase_ ,lowerCamelCase_ ,output.predictions ,"""predict""" )
UpperCAmelCase_ : List[Any] = self.compute_metrics(lowerCamelCase_ )
# Prefix all keys with metric_key_prefix + '_'
for key in list(metrics.keys() ):
if not key.startswith(F'''{metric_key_prefix}_''' ):
UpperCAmelCase_ : Optional[Any] = metrics.pop(lowerCamelCase_ )
return PredictionOutput(predictions=predictions.predictions ,label_ids=predictions.label_ids ,metrics=lowerCamelCase_ )
def A__ ( self: List[Any] ,lowerCamelCase_: Any="./" ) -> str:
UpperCAmelCase_ : Optional[Any] = self.eval_dataset
UpperCAmelCase_ : Dict = self.get_eval_dataloader(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = next(iter(lowerCamelCase_ ) )
# saving device - to make it consistent
UpperCAmelCase_ : Optional[Any] = torch.device("""cuda""" if torch.cuda.is_available() else """cpu""" )
# convert to tuple
UpperCAmelCase_ : Optional[Any] = tuple(v.to(lowerCamelCase_ ) for k, v in batch.items() )
logger.info("""Converting model to be onnx compatible""" )
from pytorch_quantization.nn import TensorQuantizer
UpperCAmelCase_ : str = True
UpperCAmelCase_ : Dict = self.model.to(lowerCamelCase_ )
model.eval()
model.float()
UpperCAmelCase_ : List[str] = model.module if hasattr(lowerCamelCase_ ,"""module""" ) else model
quant_trainer.configure_model(lowerCamelCase_ ,self.quant_trainer_args )
UpperCAmelCase_ : List[str] = os.path.join(lowerCamelCase_ ,"""model.onnx""" )
logger.info(F'''exporting model to {output_model_file}''' )
UpperCAmelCase_ : Any = {0: """batch_size""", 1: """seq_len"""}
torch.onnx.export(
lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,export_params=lowerCamelCase_ ,opset_version=13 ,do_constant_folding=lowerCamelCase_ ,input_names=["""input_ids""", """attention_mask""", """token_type_ids"""] ,output_names=["""output_start_logits""", """output_end_logits"""] ,dynamic_axes={
"""input_ids""": axes,
"""attention_mask""": axes,
"""token_type_ids""": axes,
"""output_start_logits""": axes,
"""output_end_logits""": axes,
} ,verbose=lowerCamelCase_ ,)
logger.info("""onnx export finished""" )
| 345 |
import collections
import inspect
import unittest
from transformers import FocalNetConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
FocalNetBackbone,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetModel,
)
from transformers.models.focalnet.modeling_focalnet import FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=13 ,lowerCamelCase_: Optional[int]=32 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Optional[Any]=[32, 64, 128] ,lowerCamelCase_: Optional[int]=[1, 2, 1] ,lowerCamelCase_: Union[str, Any]=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[str]=2.0 ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: int=True ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: str=True ,lowerCamelCase_: Dict=10 ,lowerCamelCase_: str=8 ,lowerCamelCase_: Union[str, Any]=["stage1", "stage2"] ,lowerCamelCase_: Optional[Any]=[1, 2] ,) -> str:
UpperCAmelCase_ : List[Any] = parent
UpperCAmelCase_ : Tuple = batch_size
UpperCAmelCase_ : Any = image_size
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : List[str] = num_channels
UpperCAmelCase_ : Dict = embed_dim
UpperCAmelCase_ : Dict = hidden_sizes
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = num_heads
UpperCAmelCase_ : List[Any] = window_size
UpperCAmelCase_ : Union[str, Any] = mlp_ratio
UpperCAmelCase_ : int = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Union[str, Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[int] = drop_path_rate
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : List[Any] = use_absolute_embeddings
UpperCAmelCase_ : List[Any] = patch_norm
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Optional[Any] = scope
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = encoder_stride
UpperCAmelCase_ : Optional[int] = out_features
UpperCAmelCase_ : Optional[int] = out_indices
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : int = None
if self.use_labels:
UpperCAmelCase_ : str = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Any = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Tuple:
return FocalNetConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,hidden_sizes=self.hidden_sizes ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,out_features=self.out_features ,out_indices=self.out_indices ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : Dict = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : Optional[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[str]:
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size, 8, 8] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,len(config.out_features ) )
self.parent.assertListEqual(model.channels ,config.hidden_sizes[:-1] )
# verify backbone works with out_features=None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size * 2, 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,1 )
self.parent.assertListEqual(model.channels ,[config.hidden_sizes[-1]] )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Any = FocalNetForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.reconstruction.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : int = 1
UpperCAmelCase_ : List[str] = FocalNetForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.reconstruction.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ) -> int:
UpperCAmelCase_ : List[Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
# test greyscale images
UpperCAmelCase_ : List[Any] = 1
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = config_and_inputs
UpperCAmelCase_ : int = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (
(
FocalNetModel,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetBackbone,
)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (
{"feature-extraction": FocalNetModel, "image-classification": FocalNetForImageClassification}
if is_torch_available()
else {}
)
A__ : Optional[Any] = False
A__ : Any = False
A__ : List[str] = False
A__ : Any = False
A__ : Any = False
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : Dict = FocalNetModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: List[str] ) -> Union[str, Any]:
return
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCamelCase_ )
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@unittest.skip(reason="""FocalNet does not use inputs_embeds""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""FocalNet does not use feedforward chunking""" )
def A__ ( self: Optional[Any] ) -> Optional[Any]:
pass
def A__ ( self: Optional[Any] ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : List[Any] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Dict = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Any = [*signature.parameters.keys()]
UpperCAmelCase_ : List[str] = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.hidden_states
UpperCAmelCase_ : List[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# FocalNet has a different seq_length
UpperCAmelCase_ : int = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Optional[int] = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Union[str, Any] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = reshaped_hidden_states[0].shape
UpperCAmelCase_ : List[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Union[str, Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Tuple = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : Union[str, Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Union[str, Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Any = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Optional[int] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
@slow
def A__ ( self: Optional[int] ) -> Optional[Any]:
for model_name in FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Tuple = FocalNetModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[Any] = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Optional[int] ) -> str:
# TODO update organization
return AutoImageProcessor.from_pretrained("""microsoft/focalnet-tiny""" ) if is_vision_available() else None
@slow
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification.from_pretrained("""microsoft/focalnet-tiny""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.default_image_processor
UpperCAmelCase_ : Union[str, Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Dict = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : str = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_1_6_6, -0.4_3_6_8, 0.2_1_9_1] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
self.assertTrue(outputs.logits.argmax(dim=-1 ).item() ,281 )
@require_torch
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (FocalNetBackbone,) if is_torch_available() else ()
A__ : int = FocalNetConfig
A__ : List[str] = False
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : str = FocalNetModelTester(self )
| 345 | 1 |
from __future__ import annotations
class _snake_case :
'''simple docstring'''
def __init__( self: int ,lowerCamelCase_: int ) -> None:
UpperCAmelCase_ : Union[str, Any] = data
UpperCAmelCase_ : Node | None = None
UpperCAmelCase_ : Node | None = None
def lowerCamelCase_ ( _a : Node | None ): # In Order traversal of the tree
'''simple docstring'''
if tree:
display(tree.left )
print(tree.data )
display(tree.right )
def lowerCamelCase_ ( _a : Node | None ):
'''simple docstring'''
return 1 + max(depth_of_tree(tree.left ) , depth_of_tree(tree.right ) ) if tree else 0
def lowerCamelCase_ ( _a : Node ):
'''simple docstring'''
if not tree:
return True
if tree.left and tree.right:
return is_full_binary_tree(tree.left ) and is_full_binary_tree(tree.right )
else:
return not tree.left and not tree.right
def lowerCamelCase_ ( ): # Main function for testing.
'''simple docstring'''
UpperCAmelCase_ : List[str] = Node(1 )
UpperCAmelCase_ : Union[str, Any] = Node(2 )
UpperCAmelCase_ : List[str] = Node(3 )
UpperCAmelCase_ : Tuple = Node(4 )
UpperCAmelCase_ : Optional[int] = Node(5 )
UpperCAmelCase_ : Optional[int] = Node(6 )
UpperCAmelCase_ : Optional[int] = Node(7 )
UpperCAmelCase_ : List[str] = Node(8 )
UpperCAmelCase_ : Optional[Any] = Node(9 )
print(is_full_binary_tree(_a ) )
print(depth_of_tree(_a ) )
print("""Tree is: """ )
display(_a )
if __name__ == "__main__":
main()
| 345 |
import collections
import inspect
import unittest
from transformers import SwinvaConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import SwinvaForImageClassification, SwinvaForMaskedImageModeling, SwinvaModel
from transformers.models.swinva.modeling_swinva import SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: int=13 ,lowerCamelCase_: int=32 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: Any=3 ,lowerCamelCase_: str=16 ,lowerCamelCase_: Optional[Any]=[1, 2, 1] ,lowerCamelCase_: Tuple=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[Any]=2.0 ,lowerCamelCase_: str=True ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[Any]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Union[str, Any]=False ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=10 ,lowerCamelCase_: Tuple=8 ,) -> List[Any]:
UpperCAmelCase_ : List[str] = parent
UpperCAmelCase_ : int = batch_size
UpperCAmelCase_ : int = image_size
UpperCAmelCase_ : Union[str, Any] = patch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : int = embed_dim
UpperCAmelCase_ : Union[str, Any] = depths
UpperCAmelCase_ : List[str] = num_heads
UpperCAmelCase_ : int = window_size
UpperCAmelCase_ : List[str] = mlp_ratio
UpperCAmelCase_ : Tuple = qkv_bias
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : str = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = drop_path_rate
UpperCAmelCase_ : List[str] = hidden_act
UpperCAmelCase_ : int = use_absolute_embeddings
UpperCAmelCase_ : Any = patch_norm
UpperCAmelCase_ : Optional[int] = layer_norm_eps
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Dict = scope
UpperCAmelCase_ : int = use_labels
UpperCAmelCase_ : Optional[Any] = type_sequence_label_size
UpperCAmelCase_ : List[str] = encoder_stride
def A__ ( self: Any ) -> int:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : List[Any] = None
if self.use_labels:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : str = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Union[str, Any]:
return SwinvaConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,)
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ) -> str:
UpperCAmelCase_ : str = SwinvaModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : List[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Any = SwinvaForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.logits.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : str = 1
UpperCAmelCase_ : Optional[Any] = SwinvaForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : int = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ) -> int:
UpperCAmelCase_ : Union[str, Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = SwinvaForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: str ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = config_and_inputs
UpperCAmelCase_ : Optional[int] = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Tuple = (
(SwinvaModel, SwinvaForImageClassification, SwinvaForMaskedImageModeling) if is_torch_available() else ()
)
A__ : Optional[Any] = (
{"feature-extraction": SwinvaModel, "image-classification": SwinvaForImageClassification}
if is_torch_available()
else {}
)
A__ : List[Any] = False
A__ : Tuple = False
A__ : int = False
A__ : Union[str, Any] = False
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = SwinvaModelTester(self )
UpperCAmelCase_ : str = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 )
def A__ ( self: Optional[int] ) -> List[Any]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: Any ) -> Dict:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
@unittest.skip(reason="""Got `CUDA error: misaligned address` with PyTorch 2.0.0.""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""Swinv2 does not use inputs_embeds""" )
def A__ ( self: Tuple ) -> List[str]:
pass
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : Tuple = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Any = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : int = [*signature.parameters.keys()]
UpperCAmelCase_ : Tuple = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Any = True
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
UpperCAmelCase_ : Union[str, Any] = False
UpperCAmelCase_ : str = True
UpperCAmelCase_ : List[Any] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[Any] = outputs.attentions
UpperCAmelCase_ : List[str] = len(self.model_tester.depths )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
UpperCAmelCase_ : str = True
UpperCAmelCase_ : Optional[Any] = config.window_size**2
UpperCAmelCase_ : Optional[int] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[Any] = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
UpperCAmelCase_ : Optional[Any] = len(lowerCamelCase_ )
# Check attention is always last and order is fine
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = True
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
if hasattr(self.model_tester ,"""num_hidden_states_types""" ):
UpperCAmelCase_ : List[Any] = self.model_tester.num_hidden_states_types
else:
# also another +1 for reshaped_hidden_states
UpperCAmelCase_ : List[str] = 2
self.assertEqual(out_len + added_hidden_states ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
def A__ ( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : int = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[str] = outputs.hidden_states
UpperCAmelCase_ : Optional[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# Swinv2 has a different seq_length
UpperCAmelCase_ : Optional[Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : int = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Optional[int] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = reshaped_hidden_states[0].shape
UpperCAmelCase_ : Optional[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Dict = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes:
UpperCAmelCase_ : Any = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Union[str, Any] = 3
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : List[str] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : List[Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Optional[Any] = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : List[str] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@slow
def A__ ( self: str ) -> Tuple:
for model_name in SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Dict = SwinvaModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : List[str] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and "logit_scale" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Dict ) -> Optional[Any]:
return (
AutoImageProcessor.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" )
if is_vision_available()
else None
)
@slow
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ : Tuple = SwinvaForImageClassification.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" ).to(
lowerCamelCase_ )
UpperCAmelCase_ : Any = self.default_image_processor
UpperCAmelCase_ : List[str] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Optional[int] = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : Dict = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.tensor([-0.3_9_4_7, -0.4_3_0_6, 0.0_0_2_6] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
| 345 | 1 |
import warnings
from ...utils import logging
from .image_processing_glpn import GLPNImageProcessor
UpperCamelCase_ = logging.get_logger(__name__)
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: int ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Tuple ) -> None:
warnings.warn(
"""The class GLPNFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please"""
""" use GLPNImageProcessor instead.""" ,lowerCamelCase_ ,)
super().__init__(*lowerCamelCase_ ,**lowerCamelCase_ )
| 345 |
import json
import os
from functools import lru_cache
from typing import Dict, List, Optional, Tuple, Union
import regex as re
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...tokenization_utils_base import BatchEncoding, EncodedInput
from ...utils import PaddingStrategy, logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {'''vocab_file''': '''vocab.json''', '''merges_file''': '''merges.txt'''}
# See all LED models at https://huggingface.co/models?filter=LED
UpperCamelCase_ = {
'''vocab_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/vocab.json''',
},
'''merges_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/merges.txt''',
},
'''tokenizer_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/tokenizer.json''',
},
}
UpperCamelCase_ = {
'''allenai/led-base-16384''': 16384,
}
@lru_cache()
# Copied from transformers.models.bart.tokenization_bart.bytes_to_unicode
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : int = (
list(range(ord("""!""" ) , ord("""~""" ) + 1 ) ) + list(range(ord("""¡""" ) , ord("""¬""" ) + 1 ) ) + list(range(ord("""®""" ) , ord("""ÿ""" ) + 1 ) )
)
UpperCAmelCase_ : Dict = bs[:]
UpperCAmelCase_ : Any = 0
for b in range(2**8 ):
if b not in bs:
bs.append(_a )
cs.append(2**8 + n )
n += 1
UpperCAmelCase_ : Any = [chr(_a ) for n in cs]
return dict(zip(_a , _a ) )
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = set()
UpperCAmelCase_ : List[Any] = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
UpperCAmelCase_ : Optional[int] = char
return pairs
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = VOCAB_FILES_NAMES
A__ : List[str] = PRETRAINED_VOCAB_FILES_MAP
A__ : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Optional[int] = ["input_ids", "attention_mask"]
def __init__( self: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any]="replace" ,lowerCamelCase_: Optional[Any]="<s>" ,lowerCamelCase_: List[Any]="</s>" ,lowerCamelCase_: List[str]="</s>" ,lowerCamelCase_: int="<s>" ,lowerCamelCase_: int="<unk>" ,lowerCamelCase_: str="<pad>" ,lowerCamelCase_: Optional[Any]="<mask>" ,lowerCamelCase_: List[str]=False ,**lowerCamelCase_: Tuple ,) -> Any:
UpperCAmelCase_ : Union[str, Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else bos_token
UpperCAmelCase_ : int = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else eos_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else sep_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else cls_token
UpperCAmelCase_ : Optional[Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else unk_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else pad_token
# Mask token behave like a normal word, i.e. include the space before it
UpperCAmelCase_ : str = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else mask_token
super().__init__(
errors=lowerCamelCase_ ,bos_token=lowerCamelCase_ ,eos_token=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,add_prefix_space=lowerCamelCase_ ,**lowerCamelCase_ ,)
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as vocab_handle:
UpperCAmelCase_ : Union[str, Any] = json.load(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {v: k for k, v in self.encoder.items()}
UpperCAmelCase_ : Any = errors # how to handle errors in decoding
UpperCAmelCase_ : int = bytes_to_unicode()
UpperCAmelCase_ : Dict = {v: k for k, v in self.byte_encoder.items()}
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as merges_handle:
UpperCAmelCase_ : Any = merges_handle.read().split("""\n""" )[1:-1]
UpperCAmelCase_ : int = [tuple(merge.split() ) for merge in bpe_merges]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : Tuple = {}
UpperCAmelCase_ : Optional[int] = add_prefix_space
# Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions
UpperCAmelCase_ : int = re.compile(R"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""" )
@property
# Copied from transformers.models.bart.tokenization_bart.BartTokenizer.vocab_size
def A__ ( self: List[str] ) -> List[str]:
return len(self.encoder )
def A__ ( self: Any ) -> Union[str, Any]:
return dict(self.encoder ,**self.added_tokens_encoder )
def A__ ( self: Tuple ,lowerCamelCase_: Dict ) -> Optional[Any]:
if token in self.cache:
return self.cache[token]
UpperCAmelCase_ : Union[str, Any] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = get_pairs(lowerCamelCase_ )
if not pairs:
return token
while True:
UpperCAmelCase_ : Union[str, Any] = min(lowerCamelCase_ ,key=lambda lowerCamelCase_ : self.bpe_ranks.get(lowerCamelCase_ ,float("""inf""" ) ) )
if bigram not in self.bpe_ranks:
break
UpperCAmelCase_ , UpperCAmelCase_ : Any = bigram
UpperCAmelCase_ : Optional[Any] = []
UpperCAmelCase_ : List[str] = 0
while i < len(lowerCamelCase_ ):
try:
UpperCAmelCase_ : str = word.index(lowerCamelCase_ ,lowerCamelCase_ )
except ValueError:
new_word.extend(word[i:] )
break
else:
new_word.extend(word[i:j] )
UpperCAmelCase_ : Union[str, Any] = j
if word[i] == first and i < len(lowerCamelCase_ ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
UpperCAmelCase_ : List[str] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = new_word
if len(lowerCamelCase_ ) == 1:
break
else:
UpperCAmelCase_ : List[str] = get_pairs(lowerCamelCase_ )
UpperCAmelCase_ : int = """ """.join(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = word
return word
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> List[str]:
UpperCAmelCase_ : str = []
for token in re.findall(self.pat ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = """""".join(
self.byte_encoder[b] for b in token.encode("""utf-8""" ) ) # Maps all our bytes to unicode strings, avoiding control tokens of the BPE (spaces in our case)
bpe_tokens.extend(bpe_token for bpe_token in self.bpe(lowerCamelCase_ ).split(""" """ ) )
return bpe_tokens
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ) -> Optional[int]:
return self.encoder.get(lowerCamelCase_ ,self.encoder.get(self.unk_token ) )
def A__ ( self: List[str] ,lowerCamelCase_: str ) -> Optional[Any]:
return self.decoder.get(lowerCamelCase_ )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ) -> List[Any]:
UpperCAmelCase_ : str = """""".join(lowerCamelCase_ )
UpperCAmelCase_ : int = bytearray([self.byte_decoder[c] for c in text] ).decode("""utf-8""" ,errors=self.errors )
return text
def A__ ( self: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCamelCase_ ):
logger.error(F'''Vocabulary path ({save_directory}) should be a directory''' )
return
UpperCAmelCase_ : List[Any] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : List[str] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""merges_file"""] )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as f:
f.write(json.dumps(self.encoder ,indent=2 ,sort_keys=lowerCamelCase_ ,ensure_ascii=lowerCamelCase_ ) + """\n""" )
UpperCAmelCase_ : str = 0
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as writer:
writer.write("""#version: 0.2\n""" )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() ,key=lambda lowerCamelCase_ : kv[1] ):
if index != token_index:
logger.warning(
F'''Saving vocabulary to {merge_file}: BPE merge indices are not consecutive.'''
""" Please check that the tokenizer is not corrupted!""" )
UpperCAmelCase_ : Tuple = token_index
writer.write(""" """.join(lowerCamelCase_ ) + """\n""" )
index += 1
return vocab_file, merge_file
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
UpperCAmelCase_ : int = [self.cls_token_id]
UpperCAmelCase_ : Optional[int] = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ,lowerCamelCase_: bool = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=lowerCamelCase_ ,token_ids_a=lowerCamelCase_ ,already_has_special_tokens=lowerCamelCase_ )
if token_ids_a is None:
return [1] + ([0] * len(lowerCamelCase_ )) + [1]
return [1] + ([0] * len(lowerCamelCase_ )) + [1, 1] + ([0] * len(lowerCamelCase_ )) + [1]
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Optional[Any] = [self.sep_token_id]
UpperCAmelCase_ : Optional[Any] = [self.cls_token_id]
if token_ids_a is None:
return len(cls + token_ids_a + sep ) * [0]
return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0]
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str=False ,**lowerCamelCase_: List[str] ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = kwargs.pop("""add_prefix_space""" ,self.add_prefix_space )
if (is_split_into_words or add_prefix_space) and (len(lowerCamelCase_ ) > 0 and not text[0].isspace()):
UpperCAmelCase_ : Dict = """ """ + text
return (text, kwargs)
def A__ ( self: List[str] ,lowerCamelCase_: Union[Dict[str, EncodedInput], BatchEncoding] ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: PaddingStrategy = PaddingStrategy.DO_NOT_PAD ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: Optional[bool] = None ,) -> dict:
UpperCAmelCase_ : Optional[int] = super()._pad(
encoded_inputs=lowerCamelCase_ ,max_length=lowerCamelCase_ ,padding_strategy=lowerCamelCase_ ,pad_to_multiple_of=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,)
# Load from model defaults
if return_attention_mask is None:
UpperCAmelCase_ : str = """attention_mask""" in self.model_input_names
if return_attention_mask and "global_attention_mask" in encoded_inputs:
UpperCAmelCase_ : str = encoded_inputs[self.model_input_names[0]]
# `global_attention_mask` need to have the same length as other (sequential) inputs.
UpperCAmelCase_ : List[Any] = len(encoded_inputs["""global_attention_mask"""] ) != len(lowerCamelCase_ )
if needs_to_be_padded:
UpperCAmelCase_ : Dict = len(lowerCamelCase_ ) - len(encoded_inputs["""global_attention_mask"""] )
if self.padding_side == "right":
# Use `-1` since `0` in `global_attention_mask` means `local attention` instead of `not to attend`
UpperCAmelCase_ : str = (
encoded_inputs["""global_attention_mask"""] + [-1] * difference
)
elif self.padding_side == "left":
UpperCAmelCase_ : List[str] = [-1] * difference + encoded_inputs[
"""global_attention_mask"""
]
else:
raise ValueError("""Invalid padding strategy:""" + str(self.padding_side ) )
return encoded_inputs
| 345 | 1 |
import re
import string
import numpy as np
import datasets
UpperCamelCase_ = '''
Returns the rate at which the input predicted strings exactly match their references, ignoring any strings input as part of the regexes_to_ignore list.
'''
UpperCamelCase_ = '''
Args:
predictions: List of predicted texts.
references: List of reference texts.
regexes_to_ignore: List, defaults to None. Regex expressions of characters to
ignore when calculating the exact matches. Note: these regexes are removed
from the input data before the changes based on the options below (e.g. ignore_case,
ignore_punctuation, ignore_numbers) are applied.
ignore_case: Boolean, defaults to False. If true, turns everything
to lowercase so that capitalization differences are ignored.
ignore_punctuation: Boolean, defaults to False. If true, removes all punctuation before
comparing predictions and references.
ignore_numbers: Boolean, defaults to False. If true, removes all punctuation before
comparing predictions and references.
Returns:
exact_match: Dictionary containing exact_match rate. Possible values are between 0.0 and 100.0, inclusive.
Examples:
>>> exact_match = datasets.load_metric("exact_match")
>>> refs = ["the cat", "theater", "YELLING", "agent007"]
>>> preds = ["cat?", "theater", "yelling", "agent"]
>>> results = exact_match.compute(references=refs, predictions=preds)
>>> print(round(results["exact_match"], 1))
25.0
>>> exact_match = datasets.load_metric("exact_match")
>>> refs = ["the cat", "theater", "YELLING", "agent007"]
>>> preds = ["cat?", "theater", "yelling", "agent"]
>>> results = exact_match.compute(references=refs, predictions=preds, regexes_to_ignore=["the ", "yell"], ignore_case=True, ignore_punctuation=True)
>>> print(round(results["exact_match"], 1))
50.0
>>> exact_match = datasets.load_metric("exact_match")
>>> refs = ["the cat", "theater", "YELLING", "agent007"]
>>> preds = ["cat?", "theater", "yelling", "agent"]
>>> results = exact_match.compute(references=refs, predictions=preds, regexes_to_ignore=["the ", "yell", "YELL"], ignore_case=True, ignore_punctuation=True)
>>> print(round(results["exact_match"], 1))
75.0
>>> exact_match = datasets.load_metric("exact_match")
>>> refs = ["the cat", "theater", "YELLING", "agent007"]
>>> preds = ["cat?", "theater", "yelling", "agent"]
>>> results = exact_match.compute(references=refs, predictions=preds, regexes_to_ignore=["the ", "yell", "YELL"], ignore_case=True, ignore_punctuation=True, ignore_numbers=True)
>>> print(round(results["exact_match"], 1))
100.0
>>> exact_match = datasets.load_metric("exact_match")
>>> refs = ["The cat sat on the mat.", "Theaters are great.", "It\'s like comparing oranges and apples."]
>>> preds = ["The cat sat on the mat?", "Theaters are great.", "It\'s like comparing apples and oranges."]
>>> results = exact_match.compute(references=refs, predictions=preds)
>>> print(round(results["exact_match"], 1))
33.3
'''
UpperCamelCase_ = '''
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class _snake_case ( datasets.Metric ):
'''simple docstring'''
def A__ ( self: str ) -> List[str]:
return datasets.MetricInfo(
description=_DESCRIPTION ,citation=_CITATION ,inputs_description=_KWARGS_DESCRIPTION ,features=datasets.Features(
{
"""predictions""": datasets.Value("""string""" ,id="""sequence""" ),
"""references""": datasets.Value("""string""" ,id="""sequence""" ),
} ) ,reference_urls=[] ,)
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: Union[str, Any]=False ,lowerCamelCase_: Tuple=False ,lowerCamelCase_: Tuple=False ,) -> str:
if regexes_to_ignore is not None:
for s in regexes_to_ignore:
UpperCAmelCase_ : int = np.array([re.sub(lowerCamelCase_ ,"""""" ,lowerCamelCase_ ) for x in predictions] )
UpperCAmelCase_ : Optional[Any] = np.array([re.sub(lowerCamelCase_ ,"""""" ,lowerCamelCase_ ) for x in references] )
else:
UpperCAmelCase_ : Optional[int] = np.asarray(lowerCamelCase_ )
UpperCAmelCase_ : Any = np.asarray(lowerCamelCase_ )
if ignore_case:
UpperCAmelCase_ : Dict = np.char.lower(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = np.char.lower(lowerCamelCase_ )
if ignore_punctuation:
UpperCAmelCase_ : Union[str, Any] = string.punctuation.maketrans("""""" ,"""""" ,string.punctuation )
UpperCAmelCase_ : Optional[int] = np.char.translate(lowerCamelCase_ ,table=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = np.char.translate(lowerCamelCase_ ,table=lowerCamelCase_ )
if ignore_numbers:
UpperCAmelCase_ : List[Any] = string.digits.maketrans("""""" ,"""""" ,string.digits )
UpperCAmelCase_ : Optional[int] = np.char.translate(lowerCamelCase_ ,table=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = np.char.translate(lowerCamelCase_ ,table=lowerCamelCase_ )
UpperCAmelCase_ : Dict = predictions == references
return {"exact_match": np.mean(lowerCamelCase_ ) * 100}
| 345 |
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Optional[Any] = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[Any] ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : int = UNetaDConditionModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=4 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,)
return model
@property
def A__ ( self: str ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = AutoencoderKL(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Tuple:
def extract(*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: str ):
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = torch.ones([0] )
def A__ ( self: List[Any] ,lowerCamelCase_: str ) -> int:
self.pixel_values.to(lowerCamelCase_ )
return self
return Out()
return extract
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : int = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : int = self.dummy_cond_unet
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(
beta_start=0.0_0_0_8_5 ,beta_end=0.0_1_2 ,beta_schedule="""scaled_linear""" ,clip_sample=lowerCamelCase_ ,set_alpha_to_one=lowerCamelCase_ ,)
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : int = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : List[Any] = output.images
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Dict = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : int = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Dict = self.dummy_cond_unet
UpperCAmelCase_ : List[Any] = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : Union[str, Any] = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : str = output.images
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(
"""hf-internal-testing/tiny-stable-diffusion-lms-pipe""" ,safety_checker=lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(pipe.scheduler ,lowerCamelCase_ )
assert pipe.safety_checker is None
UpperCAmelCase_ : List[Any] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(lowerCamelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
UpperCAmelCase_ : Optional[int] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Tuple = self.dummy_cond_unet
UpperCAmelCase_ : Dict = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# put models in fp16
UpperCAmelCase_ : Optional[Any] = unet.half()
UpperCAmelCase_ : Optional[int] = vae.half()
UpperCAmelCase_ : int = bert.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = sd_pipe([prompt] ,num_inference_steps=2 ,output_type="""np""" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Tuple = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : str = (
"""portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"""
""" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"""
""" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"""
""" children from bahnhof zoo, detailed """
)
UpperCAmelCase_ : Optional[int] = 4003660346
UpperCAmelCase_ : int = 7
# without safety guidance (sld_guidance_scale = 0)
UpperCAmelCase_ : Dict = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Optional[int] = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Dict = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
# without safety guidance (strong configuration)
UpperCAmelCase_ : Union[str, Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Any = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Tuple = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : str = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Any = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = """padme amidala taking a bath artwork, safe for work, no nudity"""
UpperCAmelCase_ : List[Any] = 2734971755
UpperCAmelCase_ : Optional[Any] = 7
UpperCAmelCase_ : int = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : Tuple = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
UpperCAmelCase_ : Any = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : List[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" )
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = (
"""the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."""
""" leyendecker"""
)
UpperCAmelCase_ : Optional[Any] = 1044355234
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : List[Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Any = output.images
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-7
UpperCAmelCase_ : Optional[int] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Any = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Any = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 345 | 1 |
import warnings
from typing import List, Optional, Tuple, Union
import numpy as np
import PIL
import torch
from ...models import UNetaDModel
from ...schedulers import RePaintScheduler
from ...utils import PIL_INTERPOLATION, logging, randn_tensor
from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
UpperCamelCase_ = logging.get_logger(__name__) # pylint: disable=invalid-name
def lowerCamelCase_ ( _a : Union[List, PIL.Image.Image, torch.Tensor] ):
'''simple docstring'''
warnings.warn(
"""The preprocess method is deprecated and will be removed in a future version. Please"""
""" use VaeImageProcessor.preprocess instead""" , _a , )
if isinstance(_a , torch.Tensor ):
return image
elif isinstance(_a , PIL.Image.Image ):
UpperCAmelCase_ : List[str] = [image]
if isinstance(image[0] , PIL.Image.Image ):
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = image[0].size
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = (x - x % 8 for x in (w, h)) # resize to integer multiple of 8
UpperCAmelCase_ : Optional[Any] = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION["""lanczos"""] ) )[None, :] for i in image]
UpperCAmelCase_ : Union[str, Any] = np.concatenate(_a , axis=0 )
UpperCAmelCase_ : str = np.array(_a ).astype(np.floataa ) / 2_5_5.0
UpperCAmelCase_ : Any = image.transpose(0 , 3 , 1 , 2 )
UpperCAmelCase_ : Any = 2.0 * image - 1.0
UpperCAmelCase_ : Tuple = torch.from_numpy(_a )
elif isinstance(image[0] , torch.Tensor ):
UpperCAmelCase_ : Tuple = torch.cat(_a , dim=0 )
return image
def lowerCamelCase_ ( _a : Union[List, PIL.Image.Image, torch.Tensor] ):
'''simple docstring'''
if isinstance(_a , torch.Tensor ):
return mask
elif isinstance(_a , PIL.Image.Image ):
UpperCAmelCase_ : Dict = [mask]
if isinstance(mask[0] , PIL.Image.Image ):
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = mask[0].size
UpperCAmelCase_ , UpperCAmelCase_ : int = (x - x % 32 for x in (w, h)) # resize to integer multiple of 32
UpperCAmelCase_ : Optional[Any] = [np.array(m.convert("""L""" ).resize((w, h) , resample=PIL_INTERPOLATION["""nearest"""] ) )[None, :] for m in mask]
UpperCAmelCase_ : Optional[Any] = np.concatenate(_a , axis=0 )
UpperCAmelCase_ : Tuple = mask.astype(np.floataa ) / 2_5_5.0
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : List[str] = 1
UpperCAmelCase_ : Optional[int] = torch.from_numpy(_a )
elif isinstance(mask[0] , torch.Tensor ):
UpperCAmelCase_ : List[Any] = torch.cat(_a , dim=0 )
return mask
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : UNetaDModel
A__ : RePaintScheduler
def __init__( self: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: str ) -> List[str]:
super().__init__()
self.register_modules(unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
@torch.no_grad()
def __call__( self: int ,lowerCamelCase_: Union[torch.Tensor, PIL.Image.Image] ,lowerCamelCase_: Union[torch.Tensor, PIL.Image.Image] ,lowerCamelCase_: int = 250 ,lowerCamelCase_: float = 0.0 ,lowerCamelCase_: int = 10 ,lowerCamelCase_: int = 10 ,lowerCamelCase_: Optional[Union[torch.Generator, List[torch.Generator]]] = None ,lowerCamelCase_: Optional[str] = "pil" ,lowerCamelCase_: bool = True ,) -> Union[ImagePipelineOutput, Tuple]:
UpperCAmelCase_ : Any = image
UpperCAmelCase_ : Any = _preprocess_image(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = original_image.to(device=self.device ,dtype=self.unet.dtype )
UpperCAmelCase_ : Dict = _preprocess_mask(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = mask_image.to(device=self.device ,dtype=self.unet.dtype )
UpperCAmelCase_ : List[Any] = original_image.shape[0]
# sample gaussian noise to begin the loop
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) and len(lowerCamelCase_ ) != batch_size:
raise ValueError(
F'''You have passed a list of generators of length {len(lowerCamelCase_ )}, but requested an effective batch'''
F''' size of {batch_size}. Make sure the batch size matches the length of the generators.''' )
UpperCAmelCase_ : str = original_image.shape
UpperCAmelCase_ : Optional[int] = randn_tensor(lowerCamelCase_ ,generator=lowerCamelCase_ ,device=self.device ,dtype=self.unet.dtype )
# set step values
self.scheduler.set_timesteps(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,self.device )
UpperCAmelCase_ : Union[str, Any] = eta
UpperCAmelCase_ : List[str] = self.scheduler.timesteps[0] + 1
UpperCAmelCase_ : List[str] = generator[0] if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else generator
for i, t in enumerate(self.progress_bar(self.scheduler.timesteps ) ):
if t < t_last:
# predict the noise residual
UpperCAmelCase_ : List[str] = self.unet(lowerCamelCase_ ,lowerCamelCase_ ).sample
# compute previous image: x_t -> x_t-1
UpperCAmelCase_ : Optional[Any] = self.scheduler.step(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ).prev_sample
else:
# compute the reverse: x_t-1 -> x_t
UpperCAmelCase_ : List[Any] = self.scheduler.undo_step(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Any = t
UpperCAmelCase_ : Dict = (image / 2 + 0.5).clamp(0 ,1 )
UpperCAmelCase_ : Dict = image.cpu().permute(0 ,2 ,3 ,1 ).numpy()
if output_type == "pil":
UpperCAmelCase_ : Optional[int] = self.numpy_to_pil(lowerCamelCase_ )
if not return_dict:
return (image,)
return ImagePipelineOutput(images=lowerCamelCase_ )
| 345 |
import unittest
from transformers import MobileBertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
MobileBertModel,
)
class _snake_case :
'''simple docstring'''
def __init__( self: Optional[int] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Tuple=13 ,lowerCamelCase_: int=7 ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: str=True ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: int=99 ,lowerCamelCase_: List[str]=64 ,lowerCamelCase_: Tuple=32 ,lowerCamelCase_: List[str]=5 ,lowerCamelCase_: str=4 ,lowerCamelCase_: str=37 ,lowerCamelCase_: Union[str, Any]="gelu" ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: str=0.1 ,lowerCamelCase_: List[str]=512 ,lowerCamelCase_: Dict=16 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: List[str]=0.0_2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: Union[str, Any]=4 ,lowerCamelCase_: str=None ,) -> List[str]:
UpperCAmelCase_ : Any = parent
UpperCAmelCase_ : List[Any] = batch_size
UpperCAmelCase_ : Union[str, Any] = seq_length
UpperCAmelCase_ : Optional[int] = is_training
UpperCAmelCase_ : Dict = use_input_mask
UpperCAmelCase_ : Any = use_token_type_ids
UpperCAmelCase_ : Tuple = use_labels
UpperCAmelCase_ : List[Any] = vocab_size
UpperCAmelCase_ : str = hidden_size
UpperCAmelCase_ : List[str] = embedding_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : List[Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : str = hidden_dropout_prob
UpperCAmelCase_ : List[str] = attention_probs_dropout_prob
UpperCAmelCase_ : Any = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : Any = type_sequence_label_size
UpperCAmelCase_ : Optional[Any] = initializer_range
UpperCAmelCase_ : Optional[int] = num_labels
UpperCAmelCase_ : Optional[int] = num_choices
UpperCAmelCase_ : List[str] = scope
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : List[str] = None
if self.use_input_mask:
UpperCAmelCase_ : Tuple = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : Dict = None
if self.use_token_type_ids:
UpperCAmelCase_ : str = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : int = None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : Union[str, Any] = None
if self.use_labels:
UpperCAmelCase_ : List[str] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : int = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : Tuple = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: Any ) -> Dict:
return MobileBertConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,embedding_size=self.embedding_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> int:
UpperCAmelCase_ : Any = MobileBertModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Dict ) -> int:
UpperCAmelCase_ : Union[str, Any] = MobileBertForMaskedLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: str ,lowerCamelCase_: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : List[Any] = MobileBertForNextSentencePrediction(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 2) )
def A__ ( self: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Tuple = MobileBertForPreTraining(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,next_sentence_label=lowerCamelCase_ ,)
self.parent.assertEqual(result.prediction_logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape ,(self.batch_size, 2) )
def A__ ( self: Any ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = MobileBertForQuestionAnswering(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,start_positions=lowerCamelCase_ ,end_positions=lowerCamelCase_ ,)
self.parent.assertEqual(result.start_logits.shape ,(self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape ,(self.batch_size, self.seq_length) )
def A__ ( self: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> str:
UpperCAmelCase_ : Optional[Any] = self.num_labels
UpperCAmelCase_ : Union[str, Any] = MobileBertForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_labels) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Any:
UpperCAmelCase_ : str = self.num_labels
UpperCAmelCase_ : Optional[int] = MobileBertForTokenClassification(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Tuple ,lowerCamelCase_: str ,lowerCamelCase_: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Union[str, Any] = self.num_choices
UpperCAmelCase_ : Tuple = MobileBertForMultipleChoice(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = input_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Union[str, Any] = token_type_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : str = input_mask.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_choices) )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ : str = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Union[str, Any] = config_and_inputs
UpperCAmelCase_ : Dict = {"""input_ids""": input_ids, """token_type_ids""": token_type_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = (
(
MobileBertModel,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
)
if is_torch_available()
else ()
)
A__ : List[str] = (
{
"feature-extraction": MobileBertModel,
"fill-mask": MobileBertForMaskedLM,
"question-answering": MobileBertForQuestionAnswering,
"text-classification": MobileBertForSequenceClassification,
"token-classification": MobileBertForTokenClassification,
"zero-shot": MobileBertForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : List[str] = True
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: int=False ) -> Union[str, Any]:
UpperCAmelCase_ : List[Any] = super()._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ,return_labels=lowerCamelCase_ )
if return_labels:
if model_class in get_values(lowerCamelCase_ ):
UpperCAmelCase_ : Any = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = torch.zeros(
self.model_tester.batch_size ,dtype=torch.long ,device=lowerCamelCase_ )
return inputs_dict
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[str] = MobileBertModelTester(self )
UpperCAmelCase_ : Union[str, Any] = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Optional[Any] ) -> List[Any]:
self.config_tester.run_common_tests()
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_model(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_masked_lm(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Tuple:
UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_multiple_choice(*lowerCamelCase_ )
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_next_sequence_prediction(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_pretraining(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_question_answering(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_sequence_classification(*lowerCamelCase_ )
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_token_classification(*lowerCamelCase_ )
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return torch.tensor(
_a , dtype=torch.long , device=_a , )
UpperCamelCase_ = 1E-3
@require_torch
@require_sentencepiece
@require_tokenizers
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: List[Any] ) -> str:
UpperCAmelCase_ : Any = MobileBertModel.from_pretrained("""google/mobilebert-uncased""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = _long_tensor([[101, 7110, 1005, 1056, 2023, 11333, 17413, 1029, 102]] )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )[0]
UpperCAmelCase_ : Union[str, Any] = torch.Size((1, 9, 512) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = torch.tensor(
[
[
[-2.473_6526e07, 8.269_1656e04, 1.652_1838e05],
[-5.754_1704e-01, 3.905_6022e00, 4.401_1507e00],
[2.604_7359e00, 1.567_7652e00, -1.732_4188e-01],
]
] ,device=lowerCamelCase_ ,)
# MobileBERT results range from 10e0 to 10e8. Even a 0.0000001% difference with a value of 10e8 results in a
# ~1 difference, it's therefore not a good idea to measure using addition.
# Here, we instead divide the expected result with the result in order to obtain ~1. We then check that the
# result is held between bounds: 1 - TOLERANCE < expected_result / result < 1 + TOLERANCE
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) >= 1 - TOLERANCE )
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) <= 1 + TOLERANCE )
self.assertTrue(lower_bound and upper_bound )
| 345 | 1 |
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
from accelerate.test_utils import execute_subprocess_async
def lowerCamelCase_ ( _a : str=None ):
'''simple docstring'''
if subparsers is not None:
UpperCAmelCase_ : Union[str, Any] = subparsers.add_parser("""test""" )
else:
UpperCAmelCase_ : Union[str, Any] = argparse.ArgumentParser("""Accelerate test command""" )
parser.add_argument(
"""--config_file""" , default=_a , help=(
"""The path to use to store the config file. Will default to a file named default_config.yaml in the cache """
"""location, which is the content of the environment `HF_HOME` suffixed with 'accelerate', or if you don't have """
"""such an environment variable, your cache directory ('~/.cache' or the content of `XDG_CACHE_HOME`) suffixed """
"""with 'huggingface'."""
) , )
if subparsers is not None:
parser.set_defaults(func=_a )
return parser
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Any = os.path.sep.join(__file__.split(os.path.sep )[:-2] + ["""test_utils""", """scripts""", """test_script.py"""] )
if args.config_file is None:
UpperCAmelCase_ : Union[str, Any] = script_name
else:
UpperCAmelCase_ : Optional[int] = F'''--config_file={args.config_file} {script_name}'''
UpperCAmelCase_ : Optional[int] = ["""accelerate-launch"""] + test_args.split()
UpperCAmelCase_ : Union[str, Any] = execute_subprocess_async(_a , env=os.environ.copy() )
if result.returncode == 0:
print("""Test is a success! You are ready for your distributed training!""" )
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = test_command_parser()
UpperCAmelCase_ : Union[str, Any] = parser.parse_args()
test_command(_a )
if __name__ == "__main__":
main()
| 345 |
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import AutoTokenizer, BarkProcessor
from transformers.testing_utils import require_torch, slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: str ) -> int:
UpperCAmelCase_ : List[Any] = """ylacombe/bark-small"""
UpperCAmelCase_ : Tuple = tempfile.mkdtemp()
UpperCAmelCase_ : Union[str, Any] = """en_speaker_1"""
UpperCAmelCase_ : Optional[Any] = """This is a test string"""
UpperCAmelCase_ : int = """speaker_embeddings_path.json"""
UpperCAmelCase_ : Any = """speaker_embeddings"""
def A__ ( self: Tuple ,**lowerCamelCase_: List[str] ) -> List[Any]:
return AutoTokenizer.from_pretrained(self.checkpoint ,**lowerCamelCase_ )
def A__ ( self: str ) -> Union[str, Any]:
shutil.rmtree(self.tmpdirname )
def A__ ( self: List[Any] ) -> int:
UpperCAmelCase_ : int = self.get_tokenizer()
UpperCAmelCase_ : Tuple = BarkProcessor(tokenizer=lowerCamelCase_ )
processor.save_pretrained(self.tmpdirname )
UpperCAmelCase_ : Optional[int] = BarkProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer.get_vocab() )
@slow
def A__ ( self: List[Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
processor.save_pretrained(
self.tmpdirname ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,speaker_embeddings_directory=self.speaker_embeddings_directory ,)
UpperCAmelCase_ : Optional[Any] = self.get_tokenizer(bos_token="""(BOS)""" ,eos_token="""(EOS)""" )
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
self.tmpdirname ,self.speaker_embeddings_dict_path ,bos_token="""(BOS)""" ,eos_token="""(EOS)""" ,)
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer_add_kwargs.get_vocab() )
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
UpperCAmelCase_ : Optional[int] = 35
UpperCAmelCase_ : Optional[int] = 2
UpperCAmelCase_ : Dict = 8
UpperCAmelCase_ : Optional[int] = {
"""semantic_prompt""": np.ones(lowerCamelCase_ ),
"""coarse_prompt""": np.ones((nb_codebooks_coarse, seq_len) ),
"""fine_prompt""": np.ones((nb_codebooks_total, seq_len) ),
}
# test providing already loaded voice_preset
UpperCAmelCase_ : str = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from npz file
UpperCAmelCase_ : List[Any] = os.path.join(self.tmpdirname ,"""file.npz""" )
np.savez(lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : int = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from the hub
UpperCAmelCase_ : Union[str, Any] = processor(text=self.input_string ,voice_preset=self.voice_preset )
def A__ ( self: Dict ) -> Tuple:
UpperCAmelCase_ : Any = self.get_tokenizer()
UpperCAmelCase_ : Dict = BarkProcessor(tokenizer=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string )
UpperCAmelCase_ : str = tokenizer(
self.input_string ,padding="""max_length""" ,max_length=256 ,add_special_tokens=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,return_token_type_ids=lowerCamelCase_ ,)
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] ,encoded_processor[key].squeeze().tolist() )
| 345 | 1 |
import argparse
import io
import requests
import torch
from omegaconf import OmegaConf
from diffusers import AutoencoderKL
from diffusers.pipelines.stable_diffusion.convert_from_ckpt import (
assign_to_checkpoint,
conv_attn_to_linear,
create_vae_diffusers_config,
renew_vae_attention_paths,
renew_vae_resnet_paths,
)
def lowerCamelCase_ ( _a : str , _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : str = checkpoint
UpperCAmelCase_ : Any = {}
UpperCAmelCase_ : Optional[Any] = vae_state_dict["""encoder.conv_in.weight"""]
UpperCAmelCase_ : Optional[Any] = vae_state_dict["""encoder.conv_in.bias"""]
UpperCAmelCase_ : Union[str, Any] = vae_state_dict["""encoder.conv_out.weight"""]
UpperCAmelCase_ : List[Any] = vae_state_dict["""encoder.conv_out.bias"""]
UpperCAmelCase_ : Tuple = vae_state_dict["""encoder.norm_out.weight"""]
UpperCAmelCase_ : Tuple = vae_state_dict["""encoder.norm_out.bias"""]
UpperCAmelCase_ : Optional[int] = vae_state_dict["""decoder.conv_in.weight"""]
UpperCAmelCase_ : Optional[Any] = vae_state_dict["""decoder.conv_in.bias"""]
UpperCAmelCase_ : Tuple = vae_state_dict["""decoder.conv_out.weight"""]
UpperCAmelCase_ : Tuple = vae_state_dict["""decoder.conv_out.bias"""]
UpperCAmelCase_ : Optional[Any] = vae_state_dict["""decoder.norm_out.weight"""]
UpperCAmelCase_ : str = vae_state_dict["""decoder.norm_out.bias"""]
UpperCAmelCase_ : Dict = vae_state_dict["""quant_conv.weight"""]
UpperCAmelCase_ : Union[str, Any] = vae_state_dict["""quant_conv.bias"""]
UpperCAmelCase_ : Tuple = vae_state_dict["""post_quant_conv.weight"""]
UpperCAmelCase_ : List[str] = vae_state_dict["""post_quant_conv.bias"""]
# Retrieves the keys for the encoder down blocks only
UpperCAmelCase_ : str = len({""".""".join(layer.split(""".""" )[:3] ) for layer in vae_state_dict if """encoder.down""" in layer} )
UpperCAmelCase_ : List[str] = {
layer_id: [key for key in vae_state_dict if F'''down.{layer_id}''' in key] for layer_id in range(_a )
}
# Retrieves the keys for the decoder up blocks only
UpperCAmelCase_ : str = len({""".""".join(layer.split(""".""" )[:3] ) for layer in vae_state_dict if """decoder.up""" in layer} )
UpperCAmelCase_ : List[str] = {
layer_id: [key for key in vae_state_dict if F'''up.{layer_id}''' in key] for layer_id in range(_a )
}
for i in range(_a ):
UpperCAmelCase_ : Optional[int] = [key for key in down_blocks[i] if F'''down.{i}''' in key and F'''down.{i}.downsample''' not in key]
if F'''encoder.down.{i}.downsample.conv.weight''' in vae_state_dict:
UpperCAmelCase_ : Optional[Any] = vae_state_dict.pop(
F'''encoder.down.{i}.downsample.conv.weight''' )
UpperCAmelCase_ : Any = vae_state_dict.pop(
F'''encoder.down.{i}.downsample.conv.bias''' )
UpperCAmelCase_ : Optional[Any] = renew_vae_resnet_paths(_a )
UpperCAmelCase_ : Tuple = {"""old""": F'''down.{i}.block''', """new""": F'''down_blocks.{i}.resnets'''}
assign_to_checkpoint(_a , _a , _a , additional_replacements=[meta_path] , config=_a )
UpperCAmelCase_ : List[Any] = [key for key in vae_state_dict if """encoder.mid.block""" in key]
UpperCAmelCase_ : Tuple = 2
for i in range(1 , num_mid_res_blocks + 1 ):
UpperCAmelCase_ : Any = [key for key in mid_resnets if F'''encoder.mid.block_{i}''' in key]
UpperCAmelCase_ : Union[str, Any] = renew_vae_resnet_paths(_a )
UpperCAmelCase_ : Any = {"""old""": F'''mid.block_{i}''', """new""": F'''mid_block.resnets.{i - 1}'''}
assign_to_checkpoint(_a , _a , _a , additional_replacements=[meta_path] , config=_a )
UpperCAmelCase_ : int = [key for key in vae_state_dict if """encoder.mid.attn""" in key]
UpperCAmelCase_ : str = renew_vae_attention_paths(_a )
UpperCAmelCase_ : str = {"""old""": """mid.attn_1""", """new""": """mid_block.attentions.0"""}
assign_to_checkpoint(_a , _a , _a , additional_replacements=[meta_path] , config=_a )
conv_attn_to_linear(_a )
for i in range(_a ):
UpperCAmelCase_ : Optional[Any] = num_up_blocks - 1 - i
UpperCAmelCase_ : List[Any] = [
key for key in up_blocks[block_id] if F'''up.{block_id}''' in key and F'''up.{block_id}.upsample''' not in key
]
if F'''decoder.up.{block_id}.upsample.conv.weight''' in vae_state_dict:
UpperCAmelCase_ : Optional[int] = vae_state_dict[
F'''decoder.up.{block_id}.upsample.conv.weight'''
]
UpperCAmelCase_ : Optional[Any] = vae_state_dict[
F'''decoder.up.{block_id}.upsample.conv.bias'''
]
UpperCAmelCase_ : str = renew_vae_resnet_paths(_a )
UpperCAmelCase_ : Optional[int] = {"""old""": F'''up.{block_id}.block''', """new""": F'''up_blocks.{i}.resnets'''}
assign_to_checkpoint(_a , _a , _a , additional_replacements=[meta_path] , config=_a )
UpperCAmelCase_ : Dict = [key for key in vae_state_dict if """decoder.mid.block""" in key]
UpperCAmelCase_ : Dict = 2
for i in range(1 , num_mid_res_blocks + 1 ):
UpperCAmelCase_ : int = [key for key in mid_resnets if F'''decoder.mid.block_{i}''' in key]
UpperCAmelCase_ : Union[str, Any] = renew_vae_resnet_paths(_a )
UpperCAmelCase_ : int = {"""old""": F'''mid.block_{i}''', """new""": F'''mid_block.resnets.{i - 1}'''}
assign_to_checkpoint(_a , _a , _a , additional_replacements=[meta_path] , config=_a )
UpperCAmelCase_ : Union[str, Any] = [key for key in vae_state_dict if """decoder.mid.attn""" in key]
UpperCAmelCase_ : Dict = renew_vae_attention_paths(_a )
UpperCAmelCase_ : str = {"""old""": """mid.attn_1""", """new""": """mid_block.attentions.0"""}
assign_to_checkpoint(_a , _a , _a , additional_replacements=[meta_path] , config=_a )
conv_attn_to_linear(_a )
return new_checkpoint
def lowerCamelCase_ ( _a : str , _a : str , ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = requests.get(
""" https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml""" )
UpperCAmelCase_ : Union[str, Any] = io.BytesIO(r.content )
UpperCAmelCase_ : str = OmegaConf.load(_a )
UpperCAmelCase_ : Union[str, Any] = 512
UpperCAmelCase_ : List[str] = """cuda""" if torch.cuda.is_available() else """cpu"""
if checkpoint_path.endswith("""safetensors""" ):
from safetensors import safe_open
UpperCAmelCase_ : int = {}
with safe_open(_a , framework="""pt""" , device="""cpu""" ) as f:
for key in f.keys():
UpperCAmelCase_ : Tuple = f.get_tensor(_a )
else:
UpperCAmelCase_ : Dict = torch.load(_a , map_location=_a )["""state_dict"""]
# Convert the VAE model.
UpperCAmelCase_ : Optional[int] = create_vae_diffusers_config(_a , image_size=_a )
UpperCAmelCase_ : List[Any] = custom_convert_ldm_vae_checkpoint(_a , _a )
UpperCAmelCase_ : Any = AutoencoderKL(**_a )
vae.load_state_dict(_a )
vae.save_pretrained(_a )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
parser.add_argument('''--vae_pt_path''', default=None, type=str, required=True, help='''Path to the VAE.pt to convert.''')
parser.add_argument('''--dump_path''', default=None, type=str, required=True, help='''Path to the VAE.pt to convert.''')
UpperCamelCase_ = parser.parse_args()
vae_pt_to_vae_diffuser(args.vae_pt_path, args.dump_path)
| 345 |
import unittest
from queue import Empty
from threading import Thread
from transformers import AutoTokenizer, TextIteratorStreamer, TextStreamer, is_torch_available
from transformers.testing_utils import CaptureStdout, require_torch, torch_device
from ..test_modeling_common import ids_tensor
if is_torch_available():
import torch
from transformers import AutoModelForCausalLM
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Union[str, Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = -1
UpperCAmelCase_ : Dict = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Any = tokenizer.decode(greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : Optional[int] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Dict ) -> Optional[Any]:
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : List[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Dict = tokenizer.decode(greedy_ids[0] )
UpperCAmelCase_ : str = TextIteratorStreamer(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : str = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
UpperCAmelCase_ : int = """"""
for new_text in streamer:
streamer_text += new_text
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : Tuple = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : str = greedy_ids[:, input_ids.shape[1] :]
UpperCAmelCase_ : Dict = tokenizer.decode(new_greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ ,skip_prompt=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : List[str] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: str ) -> str:
# Tests that we can pass `decode_kwargs` to the streamer to control how the tokens are decoded. Must be tested
# with actual models -- the dummy models' tokenizers are not aligned with their models, and
# `skip_special_tokens=True` has no effect on them
UpperCAmelCase_ : Union[str, Any] = AutoTokenizer.from_pretrained("""distilgpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""distilgpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = -1
UpperCAmelCase_ : Union[str, Any] = torch.ones((1, 5) ,device=lowerCamelCase_ ).long() * model.config.bos_token_id
with CaptureStdout() as cs:
UpperCAmelCase_ : Union[str, Any] = TextStreamer(lowerCamelCase_ ,skip_special_tokens=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=1 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The prompt contains a special token, so the streamer should not print it. As such, the output text, when
# re-tokenized, must only contain one token
UpperCAmelCase_ : List[str] = cs.out[:-1] # Remove the final "\n"
UpperCAmelCase_ : Dict = tokenizer(lowerCamelCase_ ,return_tensors="""pt""" )
self.assertEqual(streamer_text_tokenized.input_ids.shape ,(1, 1) )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Any = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = -1
UpperCAmelCase_ : Optional[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = TextIteratorStreamer(lowerCamelCase_ ,timeout=0.0_0_1 )
UpperCAmelCase_ : Any = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : Dict = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
# The streamer will timeout after 0.001 seconds, so an exception will be raised
with self.assertRaises(lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = """"""
for new_text in streamer:
streamer_text += new_text
| 345 | 1 |
from argparse import ArgumentParser, Namespace
from ..utils import logging
from . import BaseTransformersCLICommand
def lowerCamelCase_ ( _a : Namespace ):
'''simple docstring'''
return ConvertCommand(
args.model_type , args.tf_checkpoint , args.pytorch_dump_output , args.config , args.finetuning_task_name )
UpperCamelCase_ = '''
transformers can only be used from the commandline to convert TensorFlow models in PyTorch, In that case, it requires
TensorFlow to be installed. Please see https://www.tensorflow.org/install/ for installation instructions.
'''
class _snake_case ( __snake_case ):
'''simple docstring'''
@staticmethod
def A__ ( lowerCamelCase_: ArgumentParser ) -> Optional[Any]:
UpperCAmelCase_ : str = parser.add_parser(
"""convert""" ,help="""CLI tool to run convert model from original author checkpoints to Transformers PyTorch checkpoints.""" ,)
train_parser.add_argument("""--model_type""" ,type=lowerCamelCase_ ,required=lowerCamelCase_ ,help="""Model's type.""" )
train_parser.add_argument(
"""--tf_checkpoint""" ,type=lowerCamelCase_ ,required=lowerCamelCase_ ,help="""TensorFlow checkpoint path or folder.""" )
train_parser.add_argument(
"""--pytorch_dump_output""" ,type=lowerCamelCase_ ,required=lowerCamelCase_ ,help="""Path to the PyTorch saved model output.""" )
train_parser.add_argument("""--config""" ,type=lowerCamelCase_ ,default="""""" ,help="""Configuration file path or folder.""" )
train_parser.add_argument(
"""--finetuning_task_name""" ,type=lowerCamelCase_ ,default=lowerCamelCase_ ,help="""Optional fine-tuning task name if the TF model was a finetuned model.""" ,)
train_parser.set_defaults(func=lowerCamelCase_ )
def __init__( self: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: str ,lowerCamelCase_: str ,lowerCamelCase_: str ,lowerCamelCase_: str ,*lowerCamelCase_: Dict ,) -> List[Any]:
UpperCAmelCase_ : Any = logging.get_logger("""transformers-cli/converting""" )
self._logger.info(F'''Loading model {model_type}''' )
UpperCAmelCase_ : Tuple = model_type
UpperCAmelCase_ : str = tf_checkpoint
UpperCAmelCase_ : Optional[Any] = pytorch_dump_output
UpperCAmelCase_ : Union[str, Any] = config
UpperCAmelCase_ : int = finetuning_task_name
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
if self._model_type == "albert":
try:
from ..models.albert.convert_albert_original_tf_checkpoint_to_pytorch import (
convert_tf_checkpoint_to_pytorch,
)
except ImportError:
raise ImportError(lowerCamelCase_ )
convert_tf_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
elif self._model_type == "bert":
try:
from ..models.bert.convert_bert_original_tf_checkpoint_to_pytorch import (
convert_tf_checkpoint_to_pytorch,
)
except ImportError:
raise ImportError(lowerCamelCase_ )
convert_tf_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
elif self._model_type == "funnel":
try:
from ..models.funnel.convert_funnel_original_tf_checkpoint_to_pytorch import (
convert_tf_checkpoint_to_pytorch,
)
except ImportError:
raise ImportError(lowerCamelCase_ )
convert_tf_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
elif self._model_type == "t5":
try:
from ..models.ta.convert_ta_original_tf_checkpoint_to_pytorch import convert_tf_checkpoint_to_pytorch
except ImportError:
raise ImportError(lowerCamelCase_ )
convert_tf_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
elif self._model_type == "gpt":
from ..models.openai.convert_openai_original_tf_checkpoint_to_pytorch import (
convert_openai_checkpoint_to_pytorch,
)
convert_openai_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
elif self._model_type == "transfo_xl":
try:
from ..models.transfo_xl.convert_transfo_xl_original_tf_checkpoint_to_pytorch import (
convert_transfo_xl_checkpoint_to_pytorch,
)
except ImportError:
raise ImportError(lowerCamelCase_ )
if "ckpt" in self._tf_checkpoint.lower():
UpperCAmelCase_ : Tuple = self._tf_checkpoint
UpperCAmelCase_ : List[str] = """"""
else:
UpperCAmelCase_ : Dict = self._tf_checkpoint
UpperCAmelCase_ : int = """"""
convert_transfo_xl_checkpoint_to_pytorch(
lowerCamelCase_ ,self._config ,self._pytorch_dump_output ,lowerCamelCase_ )
elif self._model_type == "gpt2":
try:
from ..models.gpta.convert_gpta_original_tf_checkpoint_to_pytorch import (
convert_gpta_checkpoint_to_pytorch,
)
except ImportError:
raise ImportError(lowerCamelCase_ )
convert_gpta_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
elif self._model_type == "xlnet":
try:
from ..models.xlnet.convert_xlnet_original_tf_checkpoint_to_pytorch import (
convert_xlnet_checkpoint_to_pytorch,
)
except ImportError:
raise ImportError(lowerCamelCase_ )
convert_xlnet_checkpoint_to_pytorch(
self._tf_checkpoint ,self._config ,self._pytorch_dump_output ,self._finetuning_task_name )
elif self._model_type == "xlm":
from ..models.xlm.convert_xlm_original_pytorch_checkpoint_to_pytorch import (
convert_xlm_checkpoint_to_pytorch,
)
convert_xlm_checkpoint_to_pytorch(self._tf_checkpoint ,self._pytorch_dump_output )
elif self._model_type == "lxmert":
from ..models.lxmert.convert_lxmert_original_tf_checkpoint_to_pytorch import (
convert_lxmert_checkpoint_to_pytorch,
)
convert_lxmert_checkpoint_to_pytorch(self._tf_checkpoint ,self._pytorch_dump_output )
elif self._model_type == "rembert":
from ..models.rembert.convert_rembert_tf_checkpoint_to_pytorch import (
convert_rembert_tf_checkpoint_to_pytorch,
)
convert_rembert_tf_checkpoint_to_pytorch(self._tf_checkpoint ,self._config ,self._pytorch_dump_output )
else:
raise ValueError(
"""--model_type should be selected in the list [bert, gpt, gpt2, t5, transfo_xl, xlnet, xlm, lxmert]""" )
| 345 |
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel
from diffusers import DDIMScheduler, LDMPipeline, UNetaDModel, VQModel
from diffusers.utils.testing_utils import enable_full_determinism, require_torch, slow, torch_device
enable_full_determinism()
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = UNetaDModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=3 ,out_channels=3 ,down_block_types=("""DownBlock2D""", """AttnDownBlock2D""") ,up_block_types=("""AttnUpBlock2D""", """UpBlock2D""") ,)
return model
@property
def A__ ( self: Tuple ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : List[str] = VQModel(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=3 ,)
return model
@property
def A__ ( self: Tuple ) -> Any:
torch.manual_seed(0 )
UpperCAmelCase_ : int = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
def A__ ( self: str ) -> Optional[Any]:
UpperCAmelCase_ : str = self.dummy_uncond_unet
UpperCAmelCase_ : List[Any] = DDIMScheduler()
UpperCAmelCase_ : List[Any] = self.dummy_vq_model
UpperCAmelCase_ : Optional[int] = LDMPipeline(unet=lowerCamelCase_ ,vqvae=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.manual_seed(0 )
UpperCAmelCase_ : int = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ).images
UpperCAmelCase_ : List[str] = torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ,return_dict=lowerCamelCase_ )[0]
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : str = np.array([0.8_5_1_2, 0.8_1_8, 0.6_4_1_1, 0.6_8_0_8, 0.4_4_6_5, 0.5_6_1_8, 0.4_6, 0.6_2_3_1, 0.5_1_7_2] )
UpperCAmelCase_ : Tuple = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < tolerance
@slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = LDMPipeline.from_pretrained("""CompVis/ldm-celebahq-256""" )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = ldm(generator=lowerCamelCase_ ,num_inference_steps=5 ,output_type="""numpy""" ).images
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
assert image.shape == (1, 256, 256, 3)
UpperCAmelCase_ : int = np.array([0.4_3_9_9, 0.4_4_9_7_5, 0.4_6_8_2_5, 0.4_7_4, 0.4_3_5_9, 0.4_5_8_1, 0.4_5_0_9_5, 0.4_3_4_1, 0.4_4_4_7] )
UpperCAmelCase_ : Union[str, Any] = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
| 345 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''google/canine-s''': '''https://huggingface.co/google/canine-s/resolve/main/config.json''',
# See all CANINE models at https://huggingface.co/models?filter=canine
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "canine"
def __init__( self: Union[str, Any] ,lowerCamelCase_: str=768 ,lowerCamelCase_: int=12 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=3072 ,lowerCamelCase_: int="gelu" ,lowerCamelCase_: List[Any]=0.1 ,lowerCamelCase_: List[Any]=0.1 ,lowerCamelCase_: Optional[int]=16384 ,lowerCamelCase_: Optional[Any]=16 ,lowerCamelCase_: Optional[Any]=0.0_2 ,lowerCamelCase_: Tuple=1e-12 ,lowerCamelCase_: Union[str, Any]=0 ,lowerCamelCase_: List[Any]=0xE_0_0_0 ,lowerCamelCase_: Optional[int]=0xE_0_0_1 ,lowerCamelCase_: Optional[Any]=4 ,lowerCamelCase_: Dict=4 ,lowerCamelCase_: Dict=8 ,lowerCamelCase_: int=16384 ,lowerCamelCase_: Union[str, Any]=128 ,**lowerCamelCase_: str ,) -> Tuple:
super().__init__(pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = max_position_embeddings
UpperCAmelCase_ : List[str] = hidden_size
UpperCAmelCase_ : Optional[int] = num_hidden_layers
UpperCAmelCase_ : Dict = num_attention_heads
UpperCAmelCase_ : str = intermediate_size
UpperCAmelCase_ : List[str] = hidden_act
UpperCAmelCase_ : Union[str, Any] = hidden_dropout_prob
UpperCAmelCase_ : Tuple = attention_probs_dropout_prob
UpperCAmelCase_ : Dict = initializer_range
UpperCAmelCase_ : Dict = type_vocab_size
UpperCAmelCase_ : Union[str, Any] = layer_norm_eps
# Character config:
UpperCAmelCase_ : Tuple = downsampling_rate
UpperCAmelCase_ : Union[str, Any] = upsampling_kernel_size
UpperCAmelCase_ : List[Any] = num_hash_functions
UpperCAmelCase_ : Dict = num_hash_buckets
UpperCAmelCase_ : List[str] = local_transformer_stride
| 345 |
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = [0] * len(_a )
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Dict = 0
for values in graph.values():
for i in values:
indegree[i] += 1
for i in range(len(_a ) ):
if indegree[i] == 0:
queue.append(_a )
while queue:
UpperCAmelCase_ : List[str] = queue.pop(0 )
cnt += 1
topo.append(_a )
for x in graph[vertex]:
indegree[x] -= 1
if indegree[x] == 0:
queue.append(_a )
if cnt != len(_a ):
print("""Cycle exists""" )
else:
print(_a )
# Adjacency List of Graph
UpperCamelCase_ = {0: [1, 2], 1: [3], 2: [3], 3: [4, 5], 4: [], 5: []}
topological_sort(graph)
| 345 | 1 |
from ..utils import DummyObject, requires_backends
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: int ,**lowerCamelCase_: List[str] ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: List[str] ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: int ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Any = ["flax"]
def __init__( self: int ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[str] ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Dict = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[Any] ) -> Any:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: List[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[str] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : int = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[int] ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Optional[Any] ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[Any] = ["flax"]
def __init__( self: Union[str, Any] ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> List[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Dict:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Any ,**lowerCamelCase_: int ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: Union[str, Any] ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Dict ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : str = ["flax"]
def __init__( self: Optional[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: int ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: int ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Optional[int] ) -> List[str]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Dict ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: List[str] ) -> int:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[Any] = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Optional[int] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Dict ) -> int:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Optional[Any]:
requires_backends(cls ,["""flax"""] )
| 345 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/swinv2-tiny-patch4-window8-256''': (
'''https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "swinv2"
A__ : int = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self: List[str] ,lowerCamelCase_: List[str]=224 ,lowerCamelCase_: List[str]=4 ,lowerCamelCase_: List[Any]=3 ,lowerCamelCase_: Optional[Any]=96 ,lowerCamelCase_: Any=[2, 2, 6, 2] ,lowerCamelCase_: Dict=[3, 6, 12, 24] ,lowerCamelCase_: str=7 ,lowerCamelCase_: Optional[Any]=4.0 ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: str=False ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: Union[str, Any]=1e-5 ,lowerCamelCase_: str=32 ,**lowerCamelCase_: List[str] ,) -> Tuple:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = image_size
UpperCAmelCase_ : Tuple = patch_size
UpperCAmelCase_ : Dict = num_channels
UpperCAmelCase_ : List[Any] = embed_dim
UpperCAmelCase_ : Dict = depths
UpperCAmelCase_ : Dict = len(lowerCamelCase_ )
UpperCAmelCase_ : str = num_heads
UpperCAmelCase_ : Tuple = window_size
UpperCAmelCase_ : int = mlp_ratio
UpperCAmelCase_ : str = qkv_bias
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Tuple = attention_probs_dropout_prob
UpperCAmelCase_ : int = drop_path_rate
UpperCAmelCase_ : Optional[Any] = hidden_act
UpperCAmelCase_ : List[str] = use_absolute_embeddings
UpperCAmelCase_ : Dict = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Union[str, Any] = encoder_stride
# we set the hidden_size attribute in order to make Swinv2 work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
UpperCAmelCase_ : List[str] = int(embed_dim * 2 ** (len(lowerCamelCase_ ) - 1) )
UpperCAmelCase_ : Any = (0, 0, 0, 0)
| 345 | 1 |
import warnings
from ...utils import logging
from .image_processing_layoutlmva import LayoutLMvaImageProcessor
UpperCamelCase_ = logging.get_logger(__name__)
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: List[Any] ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Tuple ) -> None:
warnings.warn(
"""The class LayoutLMv2FeatureExtractor is deprecated and will be removed in version 5 of Transformers."""
""" Please use LayoutLMv2ImageProcessor instead.""" ,lowerCamelCase_ ,)
super().__init__(*lowerCamelCase_ ,**lowerCamelCase_ )
| 345 |
import os
import sys
import tempfile
import unittest
import unittest.mock as mock
from pathlib import Path
from huggingface_hub import HfFolder, delete_repo
from huggingface_hub.file_download import http_get
from requests.exceptions import HTTPError
from transformers import (
AlbertTokenizer,
AutoTokenizer,
BertTokenizer,
BertTokenizerFast,
GPTaTokenizerFast,
is_tokenizers_available,
)
from transformers.testing_utils import TOKEN, USER, is_staging_test, require_tokenizers
from transformers.tokenization_utils import Trie
sys.path.append(str(Path(__file__).parent.parent / '''utils'''))
from test_module.custom_tokenization import CustomTokenizer # noqa E402
if is_tokenizers_available():
from test_module.custom_tokenization_fast import CustomTokenizerFast
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: int ) -> str:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : List[str] = mock.Mock()
UpperCAmelCase_ : List[Any] = 500
UpperCAmelCase_ : Union[str, Any] = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : Any = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Union[str, Any] = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# This check we did call the fake head request
mock_head.assert_called()
@require_tokenizers
def A__ ( self: str ) -> int:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : str = mock.Mock()
UpperCAmelCase_ : Optional[int] = 500
UpperCAmelCase_ : int = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : List[Any] = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Optional[int] = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# This check we did call the fake head request
mock_head.assert_called()
def A__ ( self: str ) -> Dict:
# This test is for deprecated behavior and can be removed in v5
try:
UpperCAmelCase_ : Any = tempfile.mktemp()
with open(lowerCamelCase_ ,"""wb""" ) as f:
http_get("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = AlbertTokenizer.from_pretrained(lowerCamelCase_ )
finally:
os.remove(lowerCamelCase_ )
# Supporting this legacy load introduced a weird bug where the tokenizer would load local files if they are in
# the current folder and have the right name.
if os.path.isfile("""tokenizer.json""" ):
# We skip the test if the user has a `tokenizer.json` in this folder to avoid deleting it.
return
try:
with open("""tokenizer.json""" ,"""wb""" ) as f:
http_get("""https://huggingface.co/hf-internal-testing/tiny-random-bert/blob/main/tokenizer.json""" ,lowerCamelCase_ )
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
# The tiny random BERT has a vocab size of 1024, tiny gpt2 as a vocab size of 1000
self.assertEqual(tokenizer.vocab_size ,1000 )
# Tokenizer should depend on the remote checkpoint, not the local tokenizer.json file.
finally:
os.remove("""tokenizer.json""" )
def A__ ( self: List[str] ) -> Tuple:
# This test is for deprecated behavior and can be removed in v5
UpperCAmelCase_ : str = AlbertTokenizer.from_pretrained("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" )
@is_staging_test
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
A__ : str = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "bla", "blou"]
@classmethod
def A__ ( cls: Dict ) -> Optional[int]:
UpperCAmelCase_ : List[str] = TOKEN
HfFolder.save_token(lowerCamelCase_ )
@classmethod
def A__ ( cls: Optional[Any] ) -> List[str]:
try:
delete_repo(token=cls._token ,repo_id="""test-tokenizer""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""valid_org/test-tokenizer-org""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""test-dynamic-tokenizer""" )
except HTTPError:
pass
def A__ ( self: Any ) -> Optional[int]:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Tuple = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : List[Any] = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""test-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""test-tokenizer""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(lowerCamelCase_ ,repo_id="""test-tokenizer""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
def A__ ( self: Optional[int] ) -> Any:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[Any] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Dict = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""valid_org/test-tokenizer-org""" ,use_auth_token=self._token )
UpperCAmelCase_ : Dict = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""valid_org/test-tokenizer-org""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(
lowerCamelCase_ ,repo_id="""valid_org/test-tokenizer-org""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
@require_tokenizers
def A__ ( self: Optional[int] ) -> Optional[Any]:
CustomTokenizer.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Any = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Optional[Any] = CustomTokenizer(lowerCamelCase_ )
# No fast custom tokenizer
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : Optional[Any] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the CustomTokenizer class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
# Fast and slow custom tokenizer
CustomTokenizerFast.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[str] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : str = BertTokenizerFast.from_pretrained(lowerCamelCase_ )
bert_tokenizer.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = CustomTokenizerFast.from_pretrained(lowerCamelCase_ )
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizerFast""" )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(
F'''{USER}/test-dynamic-tokenizer''' ,use_fast=lowerCamelCase_ ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Any = Trie()
trie.add("""Hello 友達""" )
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {""" """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
trie.add("""Hello""" )
trie.data
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {"""""": 1, """ """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
def A__ ( self: Tuple ) -> Optional[int]:
UpperCAmelCase_ : str = Trie()
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS] This is a extra_id_100"""] )
trie.add("""[CLS]""" )
trie.add("""extra_id_1""" )
trie.add("""extra_id_100""" )
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS]""", """ This is a """, """extra_id_100"""] )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ : Dict = Trie()
trie.add("""A""" )
self.assertEqual(trie.split("""ABC""" ) ,["""A""", """BC"""] )
self.assertEqual(trie.split("""BCA""" ) ,["""BC""", """A"""] )
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""TOKEN]""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> Union[str, Any]:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""A""" )
trie.add("""P""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> List[str]:
UpperCAmelCase_ : int = Trie()
trie.add("""AB""" )
trie.add("""B""" )
trie.add("""C""" )
self.assertEqual(trie.split("""ABC""" ) ,["""AB""", """C"""] )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = Trie()
trie.add("""ABC""" )
trie.add("""B""" )
trie.add("""CD""" )
self.assertEqual(trie.split("""ABCD""" ) ,["""ABC""", """D"""] )
def A__ ( self: List[Any] ) -> Any:
# Even if the offsets are wrong, we necessarily output correct string
# parts.
UpperCAmelCase_ : Tuple = Trie()
UpperCAmelCase_ : Optional[Any] = trie.cut_text("""ABC""" ,[0, 0, 2, 1, 2, 3] )
self.assertEqual(lowerCamelCase_ ,["""AB""", """C"""] )
| 345 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/swinv2-tiny-patch4-window8-256''': (
'''https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "swinv2"
A__ : int = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self: List[str] ,lowerCamelCase_: List[str]=224 ,lowerCamelCase_: List[str]=4 ,lowerCamelCase_: List[Any]=3 ,lowerCamelCase_: Optional[Any]=96 ,lowerCamelCase_: Any=[2, 2, 6, 2] ,lowerCamelCase_: Dict=[3, 6, 12, 24] ,lowerCamelCase_: str=7 ,lowerCamelCase_: Optional[Any]=4.0 ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: str=False ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: Union[str, Any]=1e-5 ,lowerCamelCase_: str=32 ,**lowerCamelCase_: List[str] ,) -> Tuple:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = image_size
UpperCAmelCase_ : Tuple = patch_size
UpperCAmelCase_ : Dict = num_channels
UpperCAmelCase_ : List[Any] = embed_dim
UpperCAmelCase_ : Dict = depths
UpperCAmelCase_ : Dict = len(lowerCamelCase_ )
UpperCAmelCase_ : str = num_heads
UpperCAmelCase_ : Tuple = window_size
UpperCAmelCase_ : int = mlp_ratio
UpperCAmelCase_ : str = qkv_bias
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Tuple = attention_probs_dropout_prob
UpperCAmelCase_ : int = drop_path_rate
UpperCAmelCase_ : Optional[Any] = hidden_act
UpperCAmelCase_ : List[str] = use_absolute_embeddings
UpperCAmelCase_ : Dict = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Union[str, Any] = encoder_stride
# we set the hidden_size attribute in order to make Swinv2 work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
UpperCAmelCase_ : List[str] = int(embed_dim * 2 ** (len(lowerCamelCase_ ) - 1) )
UpperCAmelCase_ : Any = (0, 0, 0, 0)
| 345 |
from ..utils import DummyObject, requires_backends
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: int ,**lowerCamelCase_: List[str] ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: List[str] ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: int ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Any = ["flax"]
def __init__( self: int ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[str] ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Dict = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[Any] ) -> Any:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: List[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[str] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : int = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[int] ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Optional[Any] ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[Any] = ["flax"]
def __init__( self: Union[str, Any] ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> List[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Dict:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Any ,**lowerCamelCase_: int ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: Union[str, Any] ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Dict ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : str = ["flax"]
def __init__( self: Optional[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: int ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: int ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Optional[int] ) -> List[str]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Dict ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: List[str] ) -> int:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[Any] = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Optional[int] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Dict ) -> int:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Optional[Any]:
requires_backends(cls ,["""flax"""] )
| 345 | 1 |
from typing import Mapping
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxSeqaSeqConfigWithPast
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''t5-small''': '''https://huggingface.co/t5-small/resolve/main/config.json''',
'''t5-base''': '''https://huggingface.co/t5-base/resolve/main/config.json''',
'''t5-large''': '''https://huggingface.co/t5-large/resolve/main/config.json''',
'''t5-3b''': '''https://huggingface.co/t5-3b/resolve/main/config.json''',
'''t5-11b''': '''https://huggingface.co/t5-11b/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Dict = "t5"
A__ : List[str] = ["past_key_values"]
A__ : Optional[int] = {"hidden_size": "d_model", "num_attention_heads": "num_heads", "num_hidden_layers": "num_layers"}
def __init__( self: List[str] ,lowerCamelCase_: Dict=32128 ,lowerCamelCase_: Any=512 ,lowerCamelCase_: Union[str, Any]=64 ,lowerCamelCase_: str=2048 ,lowerCamelCase_: Optional[int]=6 ,lowerCamelCase_: List[str]=None ,lowerCamelCase_: int=8 ,lowerCamelCase_: str=32 ,lowerCamelCase_: Optional[int]=128 ,lowerCamelCase_: Tuple=0.1 ,lowerCamelCase_: List[Any]=1e-6 ,lowerCamelCase_: Dict=1.0 ,lowerCamelCase_: Tuple="relu" ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: str=1 ,**lowerCamelCase_: Tuple ,) -> List[str]:
UpperCAmelCase_ : Any = vocab_size
UpperCAmelCase_ : Dict = d_model
UpperCAmelCase_ : Optional[int] = d_kv
UpperCAmelCase_ : Dict = d_ff
UpperCAmelCase_ : Dict = num_layers
UpperCAmelCase_ : Union[str, Any] = (
num_decoder_layers if num_decoder_layers is not None else self.num_layers
) # default = symmetry
UpperCAmelCase_ : Union[str, Any] = num_heads
UpperCAmelCase_ : List[Any] = relative_attention_num_buckets
UpperCAmelCase_ : List[Any] = relative_attention_max_distance
UpperCAmelCase_ : Optional[Any] = dropout_rate
UpperCAmelCase_ : Any = layer_norm_epsilon
UpperCAmelCase_ : Optional[Any] = initializer_factor
UpperCAmelCase_ : int = feed_forward_proj
UpperCAmelCase_ : Optional[Any] = use_cache
UpperCAmelCase_ : str = self.feed_forward_proj.split("""-""" )
UpperCAmelCase_ : int = act_info[-1]
UpperCAmelCase_ : Tuple = act_info[0] == """gated"""
if len(lowerCamelCase_ ) > 1 and act_info[0] != "gated" or len(lowerCamelCase_ ) > 2:
raise ValueError(
F'''`feed_forward_proj`: {feed_forward_proj} is not a valid activation function of the dense layer.'''
"""Please make sure `feed_forward_proj` is of the format `gated-{ACT_FN}` or `{ACT_FN}`, e.g. """
"""'gated-gelu' or 'relu'""" )
# for backwards compatibility
if feed_forward_proj == "gated-gelu":
UpperCAmelCase_ : List[str] = """gelu_new"""
super().__init__(
pad_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,is_encoder_decoder=lowerCamelCase_ ,**lowerCamelCase_ ,)
class _snake_case ( __snake_case ):
'''simple docstring'''
@property
def A__ ( self: str ) -> Mapping[str, Mapping[int, str]]:
UpperCAmelCase_ : int = {
"""input_ids""": {0: """batch""", 1: """encoder_sequence"""},
"""attention_mask""": {0: """batch""", 1: """encoder_sequence"""},
}
if self.use_past:
UpperCAmelCase_ : Any = """past_encoder_sequence + sequence"""
UpperCAmelCase_ : Optional[int] = {0: """batch"""}
UpperCAmelCase_ : List[str] = {0: """batch""", 1: """past_decoder_sequence + sequence"""}
else:
UpperCAmelCase_ : Tuple = {0: """batch""", 1: """decoder_sequence"""}
UpperCAmelCase_ : Optional[Any] = {0: """batch""", 1: """decoder_sequence"""}
if self.use_past:
self.fill_with_past_key_values_(lowerCamelCase_ ,direction="""inputs""" )
return common_inputs
@property
def A__ ( self: Optional[Any] ) -> int:
return 13
| 345 |
import random
from typing import Any
def lowerCamelCase_ ( _a : list ):
'''simple docstring'''
for _ in range(len(_a ) ):
UpperCAmelCase_ : Tuple = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ : List[Any] = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ , UpperCAmelCase_ : int = data[b], data[a]
return data
if __name__ == "__main__":
UpperCamelCase_ = [0, 1, 2, 3, 4, 5, 6, 7]
UpperCamelCase_ = ['''python''', '''says''', '''hello''', '''!''']
print('''Fisher-Yates Shuffle:''')
print('''List''', integers, strings)
print('''FY Shuffle''', fisher_yates_shuffle(integers), fisher_yates_shuffle(strings))
| 345 | 1 |
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
if b == 0:
return 1
if (b % 2) == 0:
return actual_power(_a , int(b / 2 ) ) * actual_power(_a , int(b / 2 ) )
else:
return a * actual_power(_a , int(b / 2 ) ) * actual_power(_a , int(b / 2 ) )
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
if b < 0:
return 1 / actual_power(_a , _a )
return actual_power(_a , _a )
if __name__ == "__main__":
print(power(-2, -3))
| 345 |
import flax.linen as nn
import jax.numpy as jnp
from .attention_flax import FlaxTransformeraDModel
from .resnet_flax import FlaxDownsampleaD, FlaxResnetBlockaD, FlaxUpsampleaD
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Optional[int] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : List[Any] = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : List[Any] = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : int = resnets
UpperCAmelCase_ : Tuple = attentions
if self.add_downsample:
UpperCAmelCase_ : List[Any] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int=True ) -> int:
UpperCAmelCase_ : List[Any] = ()
for resnet, attn in zip(self.resnets ,self.attentions ):
UpperCAmelCase_ : str = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[Any] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> int:
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : Dict = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnets
if self.add_downsample:
UpperCAmelCase_ : List[str] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any]=True ) -> Any:
UpperCAmelCase_ : Union[str, Any] = ()
for resnet in self.resnets:
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[str] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: str ) -> Any:
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : int = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Optional[Any] = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : int = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = resnets
UpperCAmelCase_ : Dict = attentions
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: Any ,lowerCamelCase_: str ,lowerCamelCase_: List[str]=True ) -> List[str]:
for resnet, attn in zip(self.resnets ,self.attentions ):
# pop res hidden states
UpperCAmelCase_ : List[str] = res_hidden_states_tuple[-1]
UpperCAmelCase_ : Union[str, Any] = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : Optional[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Dict = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> Dict:
UpperCAmelCase_ : Any = []
for i in range(self.num_layers ):
UpperCAmelCase_ : str = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : Optional[int] = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : str = resnets
if self.add_upsample:
UpperCAmelCase_ : Union[str, Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Dict ,lowerCamelCase_: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any=True ) -> List[str]:
for resnet in self.resnets:
# pop res hidden states
UpperCAmelCase_ : Dict = res_hidden_states_tuple[-1]
UpperCAmelCase_ : str = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : List[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : List[str] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
# there is always at least one resnet
UpperCAmelCase_ : List[Any] = [
FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
]
UpperCAmelCase_ : Any = []
for _ in range(self.num_layers ):
UpperCAmelCase_ : Optional[Any] = FlaxTransformeraDModel(
in_channels=self.in_channels ,n_heads=self.num_attention_heads ,d_head=self.in_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Dict = resnets
UpperCAmelCase_ : Any = attentions
def __call__( self: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any]=True ) -> List[Any]:
UpperCAmelCase_ : List[Any] = self.resnets[0](lowerCamelCase_ ,lowerCamelCase_ )
for attn, resnet in zip(self.attentions ,self.resnets[1:] ):
UpperCAmelCase_ : Optional[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
return hidden_states
| 345 | 1 |
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_flax_available, is_torch_available
UpperCamelCase_ = {
'''configuration_gpt_neo''': ['''GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''GPTNeoConfig''', '''GPTNeoOnnxConfig'''],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''GPTNeoForCausalLM''',
'''GPTNeoForQuestionAnswering''',
'''GPTNeoForSequenceClassification''',
'''GPTNeoForTokenClassification''',
'''GPTNeoModel''',
'''GPTNeoPreTrainedModel''',
'''load_tf_weights_in_gpt_neo''',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''FlaxGPTNeoForCausalLM''',
'''FlaxGPTNeoModel''',
'''FlaxGPTNeoPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_gpt_neo import GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoConfig, GPTNeoOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_gpt_neo import (
GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoForCausalLM,
GPTNeoForQuestionAnswering,
GPTNeoForSequenceClassification,
GPTNeoForTokenClassification,
GPTNeoModel,
GPTNeoPreTrainedModel,
load_tf_weights_in_gpt_neo,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_gpt_neo import FlaxGPTNeoForCausalLM, FlaxGPTNeoModel, FlaxGPTNeoPreTrainedModel
else:
import sys
UpperCamelCase_ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 345 |
import pickle
import numpy as np
from matplotlib import pyplot as plt
class _snake_case :
'''simple docstring'''
def __init__( self: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple=0.2 ,lowerCamelCase_: Union[str, Any]=0.2 ) -> List[str]:
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : str = bp_numa
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : Optional[int] = conva_get[:2]
UpperCAmelCase_ : List[Any] = conva_get[2]
UpperCAmelCase_ : str = size_pa
UpperCAmelCase_ : Optional[int] = rate_w
UpperCAmelCase_ : Dict = rate_t
UpperCAmelCase_ : List[Any] = [
np.mat(-1 * np.random.rand(self.conva[0] ,self.conva[0] ) + 0.5 )
for i in range(self.conva[1] )
]
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : Dict = -2 * np.random.rand(self.conva[1] ) + 1
UpperCAmelCase_ : str = -2 * np.random.rand(self.num_bpa ) + 1
UpperCAmelCase_ : Union[str, Any] = -2 * np.random.rand(self.num_bpa ) + 1
def A__ ( self: str ,lowerCamelCase_: Optional[Any] ) -> Tuple:
# save model dict with pickle
UpperCAmelCase_ : Dict = {
"""num_bp1""": self.num_bpa,
"""num_bp2""": self.num_bpa,
"""num_bp3""": self.num_bpa,
"""conv1""": self.conva,
"""step_conv1""": self.step_conva,
"""size_pooling1""": self.size_poolinga,
"""rate_weight""": self.rate_weight,
"""rate_thre""": self.rate_thre,
"""w_conv1""": self.w_conva,
"""wkj""": self.wkj,
"""vji""": self.vji,
"""thre_conv1""": self.thre_conva,
"""thre_bp2""": self.thre_bpa,
"""thre_bp3""": self.thre_bpa,
}
with open(lowerCamelCase_ ,"""wb""" ) as f:
pickle.dump(lowerCamelCase_ ,lowerCamelCase_ )
print(F'''Model saved： {save_path}''' )
@classmethod
def A__ ( cls: List[str] ,lowerCamelCase_: str ) -> List[str]:
# read saved model
with open(lowerCamelCase_ ,"""rb""" ) as f:
UpperCAmelCase_ : Any = pickle.load(lowerCamelCase_ ) # noqa: S301
UpperCAmelCase_ : Union[str, Any] = model_dic.get("""conv1""" )
conv_get.append(model_dic.get("""step_conv1""" ) )
UpperCAmelCase_ : List[str] = model_dic.get("""size_pooling1""" )
UpperCAmelCase_ : Tuple = model_dic.get("""num_bp1""" )
UpperCAmelCase_ : Optional[Any] = model_dic.get("""num_bp2""" )
UpperCAmelCase_ : List[str] = model_dic.get("""num_bp3""" )
UpperCAmelCase_ : List[Any] = model_dic.get("""rate_weight""" )
UpperCAmelCase_ : Dict = model_dic.get("""rate_thre""" )
# create model instance
UpperCAmelCase_ : List[Any] = CNN(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# modify model parameter
UpperCAmelCase_ : Any = model_dic.get("""w_conv1""" )
UpperCAmelCase_ : int = model_dic.get("""wkj""" )
UpperCAmelCase_ : int = model_dic.get("""vji""" )
UpperCAmelCase_ : Optional[int] = model_dic.get("""thre_conv1""" )
UpperCAmelCase_ : List[str] = model_dic.get("""thre_bp2""" )
UpperCAmelCase_ : Dict = model_dic.get("""thre_bp3""" )
return conv_ins
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> Tuple:
return 1 / (1 + np.exp(-1 * x ))
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
return round(lowerCamelCase_ ,3 )
def A__ ( self: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Any:
# convolution process
UpperCAmelCase_ : Optional[Any] = convs[0]
UpperCAmelCase_ : int = convs[1]
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )[0]
# get the data slice of original image data, data_focus
UpperCAmelCase_ : Dict = []
for i_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
for j_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = data[
i_focus : i_focus + size_conv, j_focus : j_focus + size_conv
]
data_focus.append(lowerCamelCase_ )
# calculate the feature map of every single kernel, and saved as list of matrix
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = int((size_data - size_conv) / conv_step + 1 )
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[int] = []
for i_focus in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = (
np.sum(np.multiply(data_focus[i_focus] ,w_convs[i_map] ) )
- thre_convs[i_map]
)
featuremap.append(self.sig(lowerCamelCase_ ) )
UpperCAmelCase_ : Union[str, Any] = np.asmatrix(lowerCamelCase_ ).reshape(
lowerCamelCase_ ,lowerCamelCase_ )
data_featuremap.append(lowerCamelCase_ )
# expanding the data slice to One dimenssion
UpperCAmelCase_ : Optional[Any] = []
for each_focus in data_focus:
focusa_list.extend(self.Expand_Mat(lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[int] = np.asarray(lowerCamelCase_ )
return focus_list, data_featuremap
def A__ ( self: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any]="average_pool" ) -> List[Any]:
# pooling process
UpperCAmelCase_ : Optional[Any] = len(featuremaps[0] )
UpperCAmelCase_ : Any = int(size_map / size_pooling )
UpperCAmelCase_ : Optional[int] = []
for i_map in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Any = featuremaps[i_map]
UpperCAmelCase_ : Tuple = []
for i_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : str = feature_map[
i_focus : i_focus + size_pooling,
j_focus : j_focus + size_pooling,
]
if pooling_type == "average_pool":
# average pooling
map_pooled.append(np.average(lowerCamelCase_ ) )
elif pooling_type == "max_pooling":
# max pooling
map_pooled.append(np.max(lowerCamelCase_ ) )
UpperCAmelCase_ : int = np.asmatrix(lowerCamelCase_ ).reshape(lowerCamelCase_ ,lowerCamelCase_ )
featuremap_pooled.append(lowerCamelCase_ )
return featuremap_pooled
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> Optional[int]:
# expanding three dimension data to one dimension list
UpperCAmelCase_ : List[Any] = []
for i in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Tuple = np.shape(data[i] )
UpperCAmelCase_ : Optional[int] = data[i].reshape(1 ,shapes[0] * shapes[1] )
UpperCAmelCase_ : Optional[int] = data_listed.getA().tolist()[0]
data_expanded.extend(lowerCamelCase_ )
UpperCAmelCase_ : int = np.asarray(lowerCamelCase_ )
return data_expanded
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
# expanding matrix to one dimension list
UpperCAmelCase_ : List[Any] = np.asarray(lowerCamelCase_ )
UpperCAmelCase_ : str = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = data_mat.reshape(1 ,shapes[0] * shapes[1] )
return data_expanded
def A__ ( self: str ,lowerCamelCase_: Dict ,lowerCamelCase_: int ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = 0
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = np.ones((size_map, size_map) )
for i in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Any = pd_pool[
i_pool
]
UpperCAmelCase_ : List[str] = i_pool + 1
UpperCAmelCase_ : Optional[Any] = np.multiply(
lowerCamelCase_ ,np.multiply(out_map[i_map] ,(1 - out_map[i_map]) ) )
pd_all.append(lowerCamelCase_ )
return pd_all
def A__ ( self: str ,lowerCamelCase_: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any=bool ) -> Optional[int]:
# model traning
print("""----------------------Start Training-------------------------""" )
print((""" - - Shape: Train_Data """, np.shape(lowerCamelCase_ )) )
print((""" - - Shape: Teach_Data """, np.shape(lowerCamelCase_ )) )
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : Any = 10000
while rp < n_repeat and mse >= error_accuracy:
UpperCAmelCase_ : List[str] = 0
print(F'''-------------Learning Time {rp}--------------''' )
for p in range(len(lowerCamelCase_ ) ):
# print('------------Learning Image: %d--------------'%p)
UpperCAmelCase_ : str = np.asmatrix(datas_train[p] )
UpperCAmelCase_ : Optional[Any] = np.asarray(datas_teach[p] )
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : List[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = data_bp_input
UpperCAmelCase_ : Optional[Any] = np.dot(lowerCamelCase_ ,self.vji.T ) - self.thre_bpa
UpperCAmelCase_ : int = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = np.dot(lowerCamelCase_ ,self.wkj.T ) - self.thre_bpa
UpperCAmelCase_ : Optional[Any] = self.sig(lowerCamelCase_ )
# --------------Model Leaning ------------------------
# calculate error and gradient---------------
UpperCAmelCase_ : List[str] = np.multiply(
(data_teach - bp_outa) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : List[Any] = np.multiply(
np.dot(lowerCamelCase_ ,self.wkj ) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : Any = np.dot(lowerCamelCase_ ,self.vji )
UpperCAmelCase_ : Tuple = pd_i_all / (self.size_poolinga * self.size_poolinga)
UpperCAmelCase_ : List[str] = pd_conva_pooled.T.getA().tolist()
UpperCAmelCase_ : str = self._calculate_gradient_from_pool(
lowerCamelCase_ ,lowerCamelCase_ ,shape_featuremapa[0] ,shape_featuremapa[1] ,self.size_poolinga ,)
# weight and threshold learning process---------
# convolution layer
for k_conv in range(self.conva[1] ):
UpperCAmelCase_ : List[str] = self._expand_mat(pd_conva_all[k_conv] )
UpperCAmelCase_ : Optional[Any] = self.rate_weight * np.dot(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : int = self.w_conva[k_conv] + delta_w.reshape(
(self.conva[0], self.conva[0]) )
UpperCAmelCase_ : str = (
self.thre_conva[k_conv]
- np.sum(pd_conva_all[k_conv] ) * self.rate_thre
)
# all connected layer
UpperCAmelCase_ : int = self.wkj + pd_k_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : Tuple = self.vji + pd_j_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : int = self.thre_bpa - pd_k_all * self.rate_thre
UpperCAmelCase_ : str = self.thre_bpa - pd_j_all * self.rate_thre
# calculate the sum error of all single image
UpperCAmelCase_ : int = np.sum(abs(data_teach - bp_outa ) )
error_count += errors
# print(' ----Teach ',data_teach)
# print(' ----BP_output ',bp_out3)
UpperCAmelCase_ : int = rp + 1
UpperCAmelCase_ : Any = error_count / patterns
all_mse.append(lowerCamelCase_ )
def draw_error():
UpperCAmelCase_ : Any = [error_accuracy for i in range(int(n_repeat * 1.2 ) )]
plt.plot(lowerCamelCase_ ,"""+-""" )
plt.plot(lowerCamelCase_ ,"""r--""" )
plt.xlabel("""Learning Times""" )
plt.ylabel("""All_mse""" )
plt.grid(lowerCamelCase_ ,alpha=0.5 )
plt.show()
print("""------------------Training Complished---------------------""" )
print((""" - - Training epoch: """, rp, F''' - - Mse: {mse:.6f}''') )
if draw_e:
draw_error()
return mse
def A__ ( self: Optional[int] ,lowerCamelCase_: Any ) -> Tuple:
# model predict
UpperCAmelCase_ : Union[str, Any] = []
print("""-------------------Start Testing-------------------------""" )
print((""" - - Shape: Test_Data """, np.shape(lowerCamelCase_ )) )
for p in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = np.asmatrix(datas_test[p] )
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Optional[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : str = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : str = data_bp_input
UpperCAmelCase_ : Union[str, Any] = bp_outa * self.vji.T - self.thre_bpa
UpperCAmelCase_ : Optional[int] = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = bp_outa * self.wkj.T - self.thre_bpa
UpperCAmelCase_ : List[Any] = self.sig(lowerCamelCase_ )
produce_out.extend(bp_outa.getA().tolist() )
UpperCAmelCase_ : int = [list(map(self.do_round ,lowerCamelCase_ ) ) for each in produce_out]
return np.asarray(lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Tuple:
# return the data of image after convoluting process so we can check it out
UpperCAmelCase_ : Optional[int] = np.asmatrix(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Dict = self.pooling(lowerCamelCase_ ,self.size_poolinga )
return data_conveda, data_pooleda
if __name__ == "__main__":
pass
| 345 | 1 |
import unittest
from queue import Empty
from threading import Thread
from transformers import AutoTokenizer, TextIteratorStreamer, TextStreamer, is_torch_available
from transformers.testing_utils import CaptureStdout, require_torch, torch_device
from ..test_modeling_common import ids_tensor
if is_torch_available():
import torch
from transformers import AutoModelForCausalLM
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Union[str, Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = -1
UpperCAmelCase_ : Dict = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Any = tokenizer.decode(greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : Optional[int] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Dict ) -> Optional[Any]:
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : List[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Dict = tokenizer.decode(greedy_ids[0] )
UpperCAmelCase_ : str = TextIteratorStreamer(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : str = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
UpperCAmelCase_ : int = """"""
for new_text in streamer:
streamer_text += new_text
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : Tuple = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : str = greedy_ids[:, input_ids.shape[1] :]
UpperCAmelCase_ : Dict = tokenizer.decode(new_greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ ,skip_prompt=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : List[str] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: str ) -> str:
# Tests that we can pass `decode_kwargs` to the streamer to control how the tokens are decoded. Must be tested
# with actual models -- the dummy models' tokenizers are not aligned with their models, and
# `skip_special_tokens=True` has no effect on them
UpperCAmelCase_ : Union[str, Any] = AutoTokenizer.from_pretrained("""distilgpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""distilgpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = -1
UpperCAmelCase_ : Union[str, Any] = torch.ones((1, 5) ,device=lowerCamelCase_ ).long() * model.config.bos_token_id
with CaptureStdout() as cs:
UpperCAmelCase_ : Union[str, Any] = TextStreamer(lowerCamelCase_ ,skip_special_tokens=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=1 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The prompt contains a special token, so the streamer should not print it. As such, the output text, when
# re-tokenized, must only contain one token
UpperCAmelCase_ : List[str] = cs.out[:-1] # Remove the final "\n"
UpperCAmelCase_ : Dict = tokenizer(lowerCamelCase_ ,return_tensors="""pt""" )
self.assertEqual(streamer_text_tokenized.input_ids.shape ,(1, 1) )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Any = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = -1
UpperCAmelCase_ : Optional[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = TextIteratorStreamer(lowerCamelCase_ ,timeout=0.0_0_1 )
UpperCAmelCase_ : Any = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : Dict = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
# The streamer will timeout after 0.001 seconds, so an exception will be raised
with self.assertRaises(lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = """"""
for new_text in streamer:
streamer_text += new_text
| 345 |
import json
import os
import unittest
from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = CTRLTokenizer
A__ : Optional[Any] = False
A__ : str = False
def A__ ( self: Optional[int] ) -> List[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
UpperCAmelCase_ : Dict = ["""adapt""", """re@@""", """a@@""", """apt""", """c@@""", """t""", """<unk>"""]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : List[Any] = ["""#version: 0.2""", """a p""", """ap t</w>""", """r e""", """a d""", """ad apt</w>""", """"""]
UpperCAmelCase_ : Optional[Any] = {"""unk_token""": """<unk>"""}
UpperCAmelCase_ : Union[str, Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : Optional[Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase_ ) + """\n""" )
with open(self.merges_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase_ ) )
def A__ ( self: Optional[int] ,**lowerCamelCase_: Any ) -> str:
kwargs.update(self.special_tokens_map )
return CTRLTokenizer.from_pretrained(self.tmpdirname ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: int ) -> str:
UpperCAmelCase_ : List[str] = """adapt react readapt apt"""
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
return input_text, output_text
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = CTRLTokenizer(self.vocab_file ,self.merges_file ,**self.special_tokens_map )
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
UpperCAmelCase_ : Optional[int] = """adapt re@@ a@@ c@@ t re@@ adapt apt""".split()
UpperCAmelCase_ : Tuple = tokenizer.tokenize(lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = tokens + [tokenizer.unk_token]
UpperCAmelCase_ : List[str] = [0, 1, 2, 4, 5, 1, 0, 3, 6]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase_ ) ,lowerCamelCase_ )
| 345 | 1 |
from typing import List, Optional, Union
import torch
from ...models import UNetaDConditionModel, VQModel
from ...pipelines import DiffusionPipeline
from ...pipelines.pipeline_utils import ImagePipelineOutput
from ...schedulers import DDPMScheduler
from ...utils import (
is_accelerate_available,
is_accelerate_version,
logging,
randn_tensor,
replace_example_docstring,
)
UpperCamelCase_ = logging.get_logger(__name__) # pylint: disable=invalid-name
UpperCamelCase_ = '''
Examples:
```py
>>> from diffusers import KandinskyV22Pipeline, KandinskyV22PriorPipeline
>>> import torch
>>> pipe_prior = KandinskyV22PriorPipeline.from_pretrained("kandinsky-community/kandinsky-2-2-prior")
>>> pipe_prior.to("cuda")
>>> prompt = "red cat, 4k photo"
>>> out = pipe_prior(prompt)
>>> image_emb = out.image_embeds
>>> zero_image_emb = out.negative_image_embeds
>>> pipe = KandinskyV22Pipeline.from_pretrained("kandinsky-community/kandinsky-2-2-decoder")
>>> pipe.to("cuda")
>>> image = pipe(
... image_embeds=image_emb,
... negative_image_embeds=zero_image_emb,
... height=768,
... width=768,
... num_inference_steps=50,
... ).images
>>> image[0].save("cat.png")
```
'''
def lowerCamelCase_ ( _a : List[Any] , _a : Tuple , _a : Any=8 ):
'''simple docstring'''
UpperCAmelCase_ : List[Any] = height // scale_factor**2
if height % scale_factor**2 != 0:
new_height += 1
UpperCAmelCase_ : Optional[Any] = width // scale_factor**2
if width % scale_factor**2 != 0:
new_width += 1
return new_height * scale_factor, new_width * scale_factor
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: UNetaDConditionModel ,lowerCamelCase_: DDPMScheduler ,lowerCamelCase_: VQModel ,) -> Dict:
super().__init__()
self.register_modules(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,movq=lowerCamelCase_ ,)
UpperCAmelCase_ : Dict = 2 ** (len(self.movq.config.block_out_channels ) - 1)
def A__ ( self: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[Any] ) -> Optional[int]:
if latents is None:
UpperCAmelCase_ : int = randn_tensor(lowerCamelCase_ ,generator=lowerCamelCase_ ,device=lowerCamelCase_ ,dtype=lowerCamelCase_ )
else:
if latents.shape != shape:
raise ValueError(F'''Unexpected latents shape, got {latents.shape}, expected {shape}''' )
UpperCAmelCase_ : Tuple = latents.to(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = latents * scheduler.init_noise_sigma
return latents
def A__ ( self: Any ,lowerCamelCase_: List[Any]=0 ) -> Tuple:
if is_accelerate_available():
from accelerate import cpu_offload
else:
raise ImportError("""Please install accelerate via `pip install accelerate`""" )
UpperCAmelCase_ : Optional[Any] = torch.device(F'''cuda:{gpu_id}''' )
UpperCAmelCase_ : List[str] = [
self.unet,
self.movq,
]
for cpu_offloaded_model in models:
if cpu_offloaded_model is not None:
cpu_offload(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Any ,lowerCamelCase_: Optional[int]=0 ) -> int:
if is_accelerate_available() and is_accelerate_version(""">=""" ,"""0.17.0.dev0""" ):
from accelerate import cpu_offload_with_hook
else:
raise ImportError("""`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher.""" )
UpperCAmelCase_ : List[str] = torch.device(F'''cuda:{gpu_id}''' )
if self.device.type != "cpu":
self.to("""cpu""" ,silence_dtype_warnings=lowerCamelCase_ )
torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist)
UpperCAmelCase_ : int = None
for cpu_offloaded_model in [self.unet, self.movq]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = cpu_offload_with_hook(lowerCamelCase_ ,lowerCamelCase_ ,prev_module_hook=lowerCamelCase_ )
# We'll offload the last model manually.
UpperCAmelCase_ : Tuple = hook
@property
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._execution_device
def A__ ( self: Optional[int] ) -> Union[str, Any]:
if not hasattr(self.unet ,"""_hf_hook""" ):
return self.device
for module in self.unet.modules():
if (
hasattr(lowerCamelCase_ ,"""_hf_hook""" )
and hasattr(module._hf_hook ,"""execution_device""" )
and module._hf_hook.execution_device is not None
):
return torch.device(module._hf_hook.execution_device )
return self.device
@torch.no_grad()
@replace_example_docstring(lowerCamelCase_ )
def __call__( self: Dict ,lowerCamelCase_: Union[torch.FloatTensor, List[torch.FloatTensor]] ,lowerCamelCase_: Union[torch.FloatTensor, List[torch.FloatTensor]] ,lowerCamelCase_: int = 512 ,lowerCamelCase_: int = 512 ,lowerCamelCase_: int = 100 ,lowerCamelCase_: float = 4.0 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: Optional[Union[torch.Generator, List[torch.Generator]]] = None ,lowerCamelCase_: Optional[torch.FloatTensor] = None ,lowerCamelCase_: Optional[str] = "pil" ,lowerCamelCase_: bool = True ,) -> Optional[Any]:
UpperCAmelCase_ : Any = self._execution_device
UpperCAmelCase_ : List[str] = guidance_scale > 1.0
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = torch.cat(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : Union[str, Any] = image_embeds.shape[0] * num_images_per_prompt
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Dict = torch.cat(lowerCamelCase_ ,dim=0 )
if do_classifier_free_guidance:
UpperCAmelCase_ : Union[str, Any] = image_embeds.repeat_interleave(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : Any = negative_image_embeds.repeat_interleave(lowerCamelCase_ ,dim=0 )
UpperCAmelCase_ : List[Any] = torch.cat([negative_image_embeds, image_embeds] ,dim=0 ).to(dtype=self.unet.dtype ,device=lowerCamelCase_ )
self.scheduler.set_timesteps(lowerCamelCase_ ,device=lowerCamelCase_ )
UpperCAmelCase_ : str = self.scheduler.timesteps
UpperCAmelCase_ : List[str] = self.unet.config.in_channels
UpperCAmelCase_ , UpperCAmelCase_ : Dict = downscale_height_and_width(lowerCamelCase_ ,lowerCamelCase_ ,self.movq_scale_factor )
# create initial latent
UpperCAmelCase_ : Optional[int] = self.prepare_latents(
(batch_size, num_channels_latents, height, width) ,image_embeds.dtype ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,self.scheduler ,)
for i, t in enumerate(self.progress_bar(lowerCamelCase_ ) ):
# expand the latents if we are doing classifier free guidance
UpperCAmelCase_ : List[str] = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents
UpperCAmelCase_ : Union[str, Any] = {"""image_embeds""": image_embeds}
UpperCAmelCase_ : Union[str, Any] = self.unet(
sample=lowerCamelCase_ ,timestep=lowerCamelCase_ ,encoder_hidden_states=lowerCamelCase_ ,added_cond_kwargs=lowerCamelCase_ ,return_dict=lowerCamelCase_ ,)[0]
if do_classifier_free_guidance:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = noise_pred.split(latents.shape[1] ,dim=1 )
UpperCAmelCase_ , UpperCAmelCase_ : Dict = noise_pred.chunk(2 )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = variance_pred.chunk(2 )
UpperCAmelCase_ : List[Any] = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
UpperCAmelCase_ : List[str] = torch.cat([noise_pred, variance_pred_text] ,dim=1 )
if not (
hasattr(self.scheduler.config ,"""variance_type""" )
and self.scheduler.config.variance_type in ["learned", "learned_range"]
):
UpperCAmelCase_ , UpperCAmelCase_ : Dict = noise_pred.split(latents.shape[1] ,dim=1 )
# compute the previous noisy sample x_t -> x_t-1
UpperCAmelCase_ : Any = self.scheduler.step(
lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,generator=lowerCamelCase_ ,)[0]
# post-processing
UpperCAmelCase_ : Optional[int] = self.movq.decode(lowerCamelCase_ ,force_not_quantize=lowerCamelCase_ )["""sample"""]
if output_type not in ["pt", "np", "pil"]:
raise ValueError(F'''Only the output types `pt`, `pil` and `np` are supported not output_type={output_type}''' )
if output_type in ["np", "pil"]:
UpperCAmelCase_ : Optional[int] = image * 0.5 + 0.5
UpperCAmelCase_ : Dict = image.clamp(0 ,1 )
UpperCAmelCase_ : int = image.cpu().permute(0 ,2 ,3 ,1 ).float().numpy()
if output_type == "pil":
UpperCAmelCase_ : List[str] = self.numpy_to_pil(lowerCamelCase_ )
if not return_dict:
return (image,)
return ImagePipelineOutput(images=lowerCamelCase_ )
| 345 |
from __future__ import annotations
from typing import Dict
from ...configuration_utils import PretrainedConfig
UpperCamelCase_ = {
'''susnato/ernie-m-base_pytorch''': '''https://huggingface.co/susnato/ernie-m-base_pytorch/blob/main/config.json''',
'''susnato/ernie-m-large_pytorch''': '''https://huggingface.co/susnato/ernie-m-large_pytorch/blob/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = "ernie_m"
A__ : Dict[str, str] = {"dropout": "classifier_dropout", "num_classes": "num_labels"}
def __init__( self: str ,lowerCamelCase_: int = 250002 ,lowerCamelCase_: int = 768 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 3072 ,lowerCamelCase_: str = "gelu" ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: int = 514 ,lowerCamelCase_: float = 0.0_2 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: float = 1e-05 ,lowerCamelCase_: Any=None ,lowerCamelCase_: List[Any]=False ,lowerCamelCase_: Tuple=0.0 ,**lowerCamelCase_: Optional[int] ,) -> Optional[Any]:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = vocab_size
UpperCAmelCase_ : Any = hidden_size
UpperCAmelCase_ : Optional[Any] = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : List[Any] = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = initializer_range
UpperCAmelCase_ : Union[str, Any] = layer_norm_eps
UpperCAmelCase_ : List[Any] = classifier_dropout
UpperCAmelCase_ : str = is_decoder
UpperCAmelCase_ : List[str] = act_dropout
| 345 | 1 |
import argparse
from argparse import Namespace
import torch
from torch import nn
from transformers import XGLMConfig, XGLMForCausalLM
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = [
"""decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = emb.weight.shape
UpperCAmelCase_ : Tuple = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : List[Any] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = torch.load(_a , map_location="""cpu""" )
UpperCAmelCase_ : Dict = Namespace(**checkpoint["""cfg"""]["""model"""] )
UpperCAmelCase_ : Optional[int] = checkpoint["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = state_dict["""decoder.embed_tokens.weight"""].shape[0]
UpperCAmelCase_ : List[str] = {key.replace("""decoder""" , """model""" ): val for key, val in state_dict.items()}
UpperCAmelCase_ : int = XGLMConfig(
vocab_size=_a , max_position_embeddings=args.max_target_positions , num_layers=args.decoder_layers , attention_heads=args.decoder_attention_heads , ffn_dim=args.decoder_ffn_embed_dim , d_model=args.decoder_embed_dim , layerdrop=args.decoder_layerdrop , dropout=args.dropout , attention_dropout=args.attention_dropout , activation_dropout=args.activation_dropout , activation_function="""gelu""" , scale_embedding=not args.no_scale_embedding , tie_word_embeddings=args.share_decoder_input_output_embed , )
UpperCAmelCase_ : List[str] = XGLMForCausalLM(_a )
UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
print(_a )
UpperCAmelCase_ : Optional[Any] = make_linear_from_emb(model.model.embed_tokens )
return model
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument('''fairseq_path''', type=str, help='''path to a model.pt on local filesystem.''')
parser.add_argument('''pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = convert_fairseq_xglm_checkpoint_from_disk(args.fairseq_path)
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
UpperCamelCase_ = logging.getLogger(__name__)
torch.set_grad_enabled(False)
UpperCamelCase_ = '''cuda''' if torch.cuda.is_available() else '''cpu'''
def lowerCamelCase_ ( _a : str , _a : Any=100 , _a : int=" " ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = text.split(_a )
return [character.join(text[i : i + n] ).strip() for i in range(0 , len(_a ) , _a )]
def lowerCamelCase_ ( _a : dict ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = [], []
for title, text in zip(documents["""title"""] , documents["""text"""] ):
if text is not None:
for passage in split_text(_a ):
titles.append(title if title is not None else """""" )
texts.append(_a )
return {"title": titles, "text": texts}
def lowerCamelCase_ ( _a : dict , _a : DPRContextEncoder , _a : DPRContextEncoderTokenizerFast ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = ctx_tokenizer(
documents["""title"""] , documents["""text"""] , truncation=_a , padding="""longest""" , return_tensors="""pt""" )["""input_ids"""]
UpperCAmelCase_ : Tuple = ctx_encoder(input_ids.to(device=_a ) , return_dict=_a ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def lowerCamelCase_ ( _a : "RagExampleArguments" , _a : "ProcessingArguments" , _a : "IndexHnswArguments" , ):
'''simple docstring'''
logger.info("""Step 1 - Create the dataset""" )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
UpperCAmelCase_ : Optional[int] = load_dataset(
"""csv""" , data_files=[rag_example_args.csv_path] , split="""train""" , delimiter="""\t""" , column_names=["""title""", """text"""] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
UpperCAmelCase_ : Tuple = dataset.map(_a , batched=_a , num_proc=processing_args.num_proc )
# And compute the embeddings
UpperCAmelCase_ : List[str] = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=_a )
UpperCAmelCase_ : Dict = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
UpperCAmelCase_ : Any = Features(
{"""text""": Value("""string""" ), """title""": Value("""string""" ), """embeddings""": Sequence(Value("""float32""" ) )} ) # optional, save as float32 instead of float64 to save space
UpperCAmelCase_ : List[str] = dataset.map(
partial(_a , ctx_encoder=_a , ctx_tokenizer=_a ) , batched=_a , batch_size=processing_args.batch_size , features=_a , )
# And finally save your dataset
UpperCAmelCase_ : Union[str, Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset""" )
dataset.save_to_disk(_a )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info("""Step 2 - Index the dataset""" )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
UpperCAmelCase_ : Union[str, Any] = faiss.IndexHNSWFlat(index_hnsw_args.d , index_hnsw_args.m , faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index("""embeddings""" , custom_index=_a )
# And save the index
UpperCAmelCase_ : Optional[Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset_hnsw_index.faiss""" )
dataset.get_index("""embeddings""" ).save(_a )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" / "my_knowledge_dataset.csv" ) , metadata={"help": "Path to a tab-separated csv file with columns 'title' and 'text'"} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."} , )
A__ : str = field(
default="facebook/rag-sequence-nq" , metadata={"help": "The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"} , )
A__ : str = field(
default="facebook/dpr-ctx_encoder-multiset-base" , metadata={
"help": (
"The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"
" 'facebook/dpr-ctx_encoder-multiset-base'"
)
} , )
A__ : Optional[str] = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" ) , metadata={"help": "Path to a directory where the dataset passages and the index will be saved"} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": "The number of processes to use to split the documents into passages. Default is single process."
} , )
A__ : int = field(
default=16 , metadata={
"help": "The batch size to use when computing the passages embeddings using the DPR context encoder."
} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : int = field(
default=768 , metadata={"help": "The dimension of the embeddings to pass to the HNSW Faiss index."} , )
A__ : int = field(
default=128 , metadata={
"help": (
"The number of bi-directional links created for every new element during the HNSW index construction."
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
UpperCamelCase_ = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
UpperCamelCase_ = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 345 | 1 |
import argparse
import json
import requests
import timm
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import AutoImageProcessor, SwinConfig, SwinForImageClassification
def lowerCamelCase_ ( _a : Optional[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = SwinConfig()
UpperCAmelCase_ : Union[str, Any] = swin_name.split("""_""" )
UpperCAmelCase_ : Optional[int] = name_split[1]
UpperCAmelCase_ : Union[str, Any] = int(name_split[4] )
UpperCAmelCase_ : str = int(name_split[3][-1] )
if model_size == "tiny":
UpperCAmelCase_ : Dict = 96
UpperCAmelCase_ : Tuple = (2, 2, 6, 2)
UpperCAmelCase_ : Any = (3, 6, 12, 24)
elif model_size == "small":
UpperCAmelCase_ : List[Any] = 96
UpperCAmelCase_ : int = (2, 2, 18, 2)
UpperCAmelCase_ : Optional[Any] = (3, 6, 12, 24)
elif model_size == "base":
UpperCAmelCase_ : Optional[Any] = 128
UpperCAmelCase_ : Any = (2, 2, 18, 2)
UpperCAmelCase_ : Any = (4, 8, 16, 32)
else:
UpperCAmelCase_ : Dict = 192
UpperCAmelCase_ : List[str] = (2, 2, 18, 2)
UpperCAmelCase_ : Dict = (6, 12, 24, 48)
if "in22k" in swin_name:
UpperCAmelCase_ : Dict = 2_1841
else:
UpperCAmelCase_ : Optional[int] = 1000
UpperCAmelCase_ : List[str] = """huggingface/label-files"""
UpperCAmelCase_ : Dict = """imagenet-1k-id2label.json"""
UpperCAmelCase_ : Union[str, Any] = json.load(open(hf_hub_download(_a , _a , repo_type="""dataset""" ) , """r""" ) )
UpperCAmelCase_ : List[Any] = {int(_a ): v for k, v in idalabel.items()}
UpperCAmelCase_ : List[str] = idalabel
UpperCAmelCase_ : Tuple = {v: k for k, v in idalabel.items()}
UpperCAmelCase_ : List[Any] = img_size
UpperCAmelCase_ : Tuple = num_classes
UpperCAmelCase_ : Optional[int] = embed_dim
UpperCAmelCase_ : Optional[int] = depths
UpperCAmelCase_ : List[str] = num_heads
UpperCAmelCase_ : Optional[Any] = window_size
return config
def lowerCamelCase_ ( _a : Optional[Any] ):
'''simple docstring'''
if "patch_embed.proj" in name:
UpperCAmelCase_ : List[Any] = name.replace("""patch_embed.proj""" , """embeddings.patch_embeddings.projection""" )
if "patch_embed.norm" in name:
UpperCAmelCase_ : Dict = name.replace("""patch_embed.norm""" , """embeddings.norm""" )
if "layers" in name:
UpperCAmelCase_ : List[str] = """encoder.""" + name
if "attn.proj" in name:
UpperCAmelCase_ : Optional[int] = name.replace("""attn.proj""" , """attention.output.dense""" )
if "attn" in name:
UpperCAmelCase_ : Union[str, Any] = name.replace("""attn""" , """attention.self""" )
if "norm1" in name:
UpperCAmelCase_ : Dict = name.replace("""norm1""" , """layernorm_before""" )
if "norm2" in name:
UpperCAmelCase_ : Optional[Any] = name.replace("""norm2""" , """layernorm_after""" )
if "mlp.fc1" in name:
UpperCAmelCase_ : Union[str, Any] = name.replace("""mlp.fc1""" , """intermediate.dense""" )
if "mlp.fc2" in name:
UpperCAmelCase_ : Any = name.replace("""mlp.fc2""" , """output.dense""" )
if name == "norm.weight":
UpperCAmelCase_ : List[Any] = """layernorm.weight"""
if name == "norm.bias":
UpperCAmelCase_ : int = """layernorm.bias"""
if "head" in name:
UpperCAmelCase_ : Any = name.replace("""head""" , """classifier""" )
else:
UpperCAmelCase_ : List[str] = """swin.""" + name
return name
def lowerCamelCase_ ( _a : Tuple , _a : Any ):
'''simple docstring'''
for key in orig_state_dict.copy().keys():
UpperCAmelCase_ : Optional[Any] = orig_state_dict.pop(_a )
if "mask" in key:
continue
elif "qkv" in key:
UpperCAmelCase_ : Optional[Any] = key.split(""".""" )
UpperCAmelCase_ : Tuple = int(key_split[1] )
UpperCAmelCase_ : Optional[int] = int(key_split[3] )
UpperCAmelCase_ : Optional[Any] = model.swin.encoder.layers[layer_num].blocks[block_num].attention.self.all_head_size
if "weight" in key:
UpperCAmelCase_ : int = val[:dim, :]
UpperCAmelCase_ : Tuple = val[
dim : dim * 2, :
]
UpperCAmelCase_ : int = val[-dim:, :]
else:
UpperCAmelCase_ : List[str] = val[
:dim
]
UpperCAmelCase_ : str = val[
dim : dim * 2
]
UpperCAmelCase_ : Any = val[
-dim:
]
else:
UpperCAmelCase_ : int = val
return orig_state_dict
def lowerCamelCase_ ( _a : Tuple , _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = timm.create_model(_a , pretrained=_a )
timm_model.eval()
UpperCAmelCase_ : Optional[int] = get_swin_config(_a )
UpperCAmelCase_ : Any = SwinForImageClassification(_a )
model.eval()
UpperCAmelCase_ : List[Any] = convert_state_dict(timm_model.state_dict() , _a )
model.load_state_dict(_a )
UpperCAmelCase_ : Optional[Any] = """http://images.cocodataset.org/val2017/000000039769.jpg"""
UpperCAmelCase_ : int = AutoImageProcessor.from_pretrained("""microsoft/{}""".format(swin_name.replace("""_""" , """-""" ) ) )
UpperCAmelCase_ : Optional[Any] = Image.open(requests.get(_a , stream=_a ).raw )
UpperCAmelCase_ : Union[str, Any] = image_processor(images=_a , return_tensors="""pt""" )
UpperCAmelCase_ : int = timm_model(inputs["""pixel_values"""] )
UpperCAmelCase_ : int = model(**_a ).logits
assert torch.allclose(_a , _a , atol=1E-3 )
print(F'''Saving model {swin_name} to {pytorch_dump_folder_path}''' )
model.save_pretrained(_a )
print(F'''Saving image processor to {pytorch_dump_folder_path}''' )
image_processor.save_pretrained(_a )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--swin_name''',
default='''swin_tiny_patch4_window7_224''',
type=str,
help='''Name of the Swin timm model you\'d like to convert.''',
)
parser.add_argument(
'''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model directory.'''
)
UpperCamelCase_ = parser.parse_args()
convert_swin_checkpoint(args.swin_name, args.pytorch_dump_folder_path)
| 345 |
import gc
import unittest
import torch
from parameterized import parameterized
from diffusers import AutoencoderKL
from diffusers.utils import floats_tensor, load_hf_numpy, require_torch_gpu, slow, torch_all_close, torch_device
from diffusers.utils.import_utils import is_xformers_available
from diffusers.utils.testing_utils import enable_full_determinism
from .test_modeling_common import ModelTesterMixin, UNetTesterMixin
enable_full_determinism()
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = AutoencoderKL
A__ : Optional[int] = "sample"
A__ : Tuple = 1E-2
@property
def A__ ( self: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = 4
UpperCAmelCase_ : str = 3
UpperCAmelCase_ : Any = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ).to(lowerCamelCase_ )
return {"sample": image}
@property
def A__ ( self: List[str] ) -> Tuple:
return (3, 32, 32)
@property
def A__ ( self: Optional[Any] ) -> Any:
return (3, 32, 32)
def A__ ( self: Any ) -> Tuple:
UpperCAmelCase_ : List[Any] = {
"""block_out_channels""": [32, 64],
"""in_channels""": 3,
"""out_channels""": 3,
"""down_block_types""": ["""DownEncoderBlock2D""", """DownEncoderBlock2D"""],
"""up_block_types""": ["""UpDecoderBlock2D""", """UpDecoderBlock2D"""],
"""latent_channels""": 4,
}
UpperCAmelCase_ : int = self.dummy_input
return init_dict, inputs_dict
def A__ ( self: Optional[Any] ) -> int:
pass
def A__ ( self: str ) -> Any:
pass
@unittest.skipIf(torch_device == """mps""" ,"""Gradient checkpointing skipped on MPS""" )
def A__ ( self: Union[str, Any] ) -> Dict:
# enable deterministic behavior for gradient checkpointing
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.prepare_init_args_and_inputs_for_common()
UpperCAmelCase_ : List[Any] = self.model_class(**lowerCamelCase_ )
model.to(lowerCamelCase_ )
assert not model.is_gradient_checkpointing and model.training
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model.zero_grad()
UpperCAmelCase_ : Any = torch.randn_like(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = (out - labels).mean()
loss.backward()
# re-instantiate the model now enabling gradient checkpointing
UpperCAmelCase_ : str = self.model_class(**lowerCamelCase_ )
# clone model
model_a.load_state_dict(model.state_dict() )
model_a.to(lowerCamelCase_ )
model_a.enable_gradient_checkpointing()
assert model_a.is_gradient_checkpointing and model_a.training
UpperCAmelCase_ : Optional[int] = model_a(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model_a.zero_grad()
UpperCAmelCase_ : Dict = (out_a - labels).mean()
loss_a.backward()
# compare the output and parameters gradients
self.assertTrue((loss - loss_a).abs() < 1e-5 )
UpperCAmelCase_ : Dict = dict(model.named_parameters() )
UpperCAmelCase_ : Union[str, Any] = dict(model_a.named_parameters() )
for name, param in named_params.items():
self.assertTrue(torch_all_close(param.grad.data ,named_params_a[name].grad.data ,atol=5e-5 ) )
def A__ ( self: Optional[Any] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" ,output_loading_info=lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
self.assertEqual(len(loading_info["""missing_keys"""] ) ,0 )
model.to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model(**self.dummy_input )
assert image is not None, "Make sure output is not None"
def A__ ( self: Optional[int] ) -> int:
UpperCAmelCase_ : Dict = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" )
UpperCAmelCase_ : Tuple = model.to(lowerCamelCase_ )
model.eval()
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.manual_seed(0 )
else:
UpperCAmelCase_ : Optional[int] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = torch.randn(
1 ,model.config.in_channels ,model.config.sample_size ,model.config.sample_size ,generator=torch.manual_seed(0 ) ,)
UpperCAmelCase_ : int = image.to(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ,generator=lowerCamelCase_ ).sample
UpperCAmelCase_ : Optional[int] = output[0, -1, -3:, -3:].flatten().cpu()
# Since the VAE Gaussian prior's generator is seeded on the appropriate device,
# the expected output slices are not the same for CPU and GPU.
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.tensor(
[
-4.0078e-01,
-3.8323e-04,
-1.2681e-01,
-1.1462e-01,
2.0095e-01,
1.0893e-01,
-8.8247e-02,
-3.0361e-01,
-9.8644e-03,
] )
elif torch_device == "cpu":
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.1_3_5_2, 0.0_8_7_8, 0.0_4_1_9, -0.0_8_1_8, -0.1_0_6_9, 0.0_6_8_8, -0.1_4_5_8, -0.4_4_4_6, -0.0_0_2_6] )
else:
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.2_4_2_1, 0.4_6_4_2, 0.2_5_0_7, -0.0_4_3_8, 0.0_6_8_2, 0.3_1_6_0, -0.2_0_1_8, -0.0_7_2_7, 0.2_4_8_5] )
self.assertTrue(torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,rtol=1e-2 ) )
@slow
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
return F'''gaussian_noise_s={seed}_shape={'_'.join([str(lowerCamelCase_ ) for s in shape] )}.npy'''
def A__ ( self: Union[str, Any] ) -> Optional[int]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: List[Any]=(4, 3, 512, 512) ,lowerCamelCase_: Optional[Any]=False ) -> Optional[int]:
UpperCAmelCase_ : Tuple = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : Tuple = torch.from_numpy(load_hf_numpy(self.get_file_format(lowerCamelCase_ ,lowerCamelCase_ ) ) ).to(lowerCamelCase_ ).to(lowerCamelCase_ )
return image
def A__ ( self: List[Any] ,lowerCamelCase_: List[str]="CompVis/stable-diffusion-v1-4" ,lowerCamelCase_: Union[str, Any]=False ) -> Any:
UpperCAmelCase_ : Optional[Any] = """fp16""" if fpaa else None
UpperCAmelCase_ : str = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : int = AutoencoderKL.from_pretrained(
lowerCamelCase_ ,subfolder="""vae""" ,torch_dtype=lowerCamelCase_ ,revision=lowerCamelCase_ ,)
model.to(lowerCamelCase_ ).eval()
return model
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any]=0 ) -> Optional[int]:
if torch_device == "mps":
return torch.manual_seed(lowerCamelCase_ )
return torch.Generator(device=lowerCamelCase_ ).manual_seed(lowerCamelCase_ )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_3, 0.9_8_7_8, -0.0_4_9_5, -0.0_7_9_0, -0.2_7_0_9, 0.8_3_7_5, -0.2_0_6_0, -0.0_8_2_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_6, 0.1_1_6_8, 0.1_3_3_2, -0.4_8_4_0, -0.2_5_0_8, -0.0_7_9_1, -0.0_4_9_3, -0.4_0_8_9], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ) -> Tuple:
UpperCAmelCase_ : List[Any] = self.get_sd_vae_model()
UpperCAmelCase_ : int = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Optional[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Tuple = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[33, [-0.0_5_1_3, 0.0_2_8_9, 1.3_7_9_9, 0.2_1_6_6, -0.2_5_7_3, -0.0_8_7_1, 0.5_1_0_3, -0.0_9_9_9]],
[47, [-0.4_1_2_8, -0.1_3_2_0, -0.3_7_0_4, 0.1_9_6_5, -0.4_1_1_6, -0.2_3_3_2, -0.3_3_4_0, 0.2_2_4_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Tuple:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Tuple = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : Optional[int] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_9, 0.9_8_6_6, -0.0_4_8_7, -0.0_7_7_7, -0.2_7_1_6, 0.8_3_6_8, -0.2_0_5_5, -0.0_8_1_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_7, 0.1_1_4_7, 0.1_3_3_3, -0.4_8_4_1, -0.2_5_0_6, -0.0_8_0_5, -0.0_4_9_1, -0.4_0_8_5], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[int] = self.get_sd_vae_model()
UpperCAmelCase_ : Dict = self.get_sd_image(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : str = model(lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : List[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Any = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[13, [-0.2_0_5_1, -0.1_8_0_3, -0.2_3_1_1, -0.2_1_1_4, -0.3_2_9_2, -0.3_5_7_4, -0.2_9_5_3, -0.3_3_2_3]],
[37, [-0.2_6_3_2, -0.2_6_2_5, -0.2_1_9_9, -0.2_7_4_1, -0.4_5_3_9, -0.4_9_9_0, -0.3_7_2_0, -0.4_9_2_5]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: str ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[int] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : str = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : Any = sample[-1, -2:, :2, -2:].flatten().cpu()
UpperCAmelCase_ : Union[str, Any] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 )
@parameterized.expand(
[
# fmt: off
[27, [-0.0_3_6_9, 0.0_2_0_7, -0.0_7_7_6, -0.0_6_8_2, -0.1_7_4_7, -0.1_9_3_0, -0.1_4_6_5, -0.2_0_3_9]],
[16, [-0.1_6_2_8, -0.2_1_3_4, -0.2_7_4_7, -0.2_6_4_2, -0.3_7_7_4, -0.4_4_0_4, -0.3_6_8_7, -0.4_2_7_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: str ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : str = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : str = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=5e-3 )
@parameterized.expand([(13,), (16,), (27,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> int:
UpperCAmelCase_ : Optional[Any] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-1 )
@parameterized.expand([(13,), (16,), (37,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = self.get_sd_vae_model()
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.3_0_0_1, 0.0_9_1_8, -2.6_9_8_4, -3.9_7_2_0, -3.2_0_9_9, -5.0_3_5_3, 1.7_3_3_8, -0.2_0_6_5, 3.4_2_6_7]],
[47, [-1.5_0_3_0, -4.3_8_7_1, -6.0_3_5_5, -9.1_1_5_7, -1.6_6_6_1, -2.7_8_5_3, 2.1_6_0_7, -5.0_8_2_3, 2.5_6_3_3]],
# fmt: on
] )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[Any] = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : str = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : int = model.encode(lowerCamelCase_ ).latent_dist
UpperCAmelCase_ : Optional[Any] = dist.sample(generator=lowerCamelCase_ )
assert list(sample.shape ) == [image.shape[0], 4] + [i // 8 for i in image.shape[2:]]
UpperCAmelCase_ : Tuple = sample[0, -1, -3:, -3:].flatten().cpu()
UpperCAmelCase_ : Optional[Any] = torch.tensor(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = 3e-3 if torch_device != """mps""" else 1e-2
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=lowerCamelCase_ )
| 345 | 1 |
UpperCamelCase_ = [
'''Audio''',
'''Array2D''',
'''Array3D''',
'''Array4D''',
'''Array5D''',
'''ClassLabel''',
'''Features''',
'''Sequence''',
'''Value''',
'''Image''',
'''Translation''',
'''TranslationVariableLanguages''',
]
from .audio import Audio
from .features import ArrayaD, ArrayaD, ArrayaD, ArrayaD, ClassLabel, Features, Sequence, Value
from .image import Image
from .translation import Translation, TranslationVariableLanguages
| 345 |
import copy
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Audio, Features, Value
from .base import TaskTemplate
@dataclass(frozen=__snake_case )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = field(default="automatic-speech-recognition" , metadata={"include_in_asdict_even_if_is_default": True} )
A__ : ClassVar[Features] = Features({"audio": Audio()} )
A__ : ClassVar[Features] = Features({"transcription": Value("string" )} )
A__ : str = "audio"
A__ : str = "transcription"
def A__ ( self: int ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
if self.audio_column not in features:
raise ValueError(F'''Column {self.audio_column} is not present in features.''' )
if not isinstance(features[self.audio_column] ,lowerCamelCase_ ):
raise ValueError(F'''Column {self.audio_column} is not an Audio type.''' )
UpperCAmelCase_ : Any = copy.deepcopy(self )
UpperCAmelCase_ : Union[str, Any] = self.input_schema.copy()
UpperCAmelCase_ : Any = features[self.audio_column]
UpperCAmelCase_ : Union[str, Any] = input_schema
return task_template
@property
def A__ ( self: List[str] ) -> Dict[str, str]:
return {self.audio_column: "audio", self.transcription_column: "transcription"}
| 345 | 1 |
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = 0
# if input_string is "aba" than new_input_string become "a|b|a"
UpperCAmelCase_ : Union[str, Any] = """"""
UpperCAmelCase_ : str = """"""
# append each character + "|" in new_string for range(0, length-1)
for i in input_string[: len(_a ) - 1]:
new_input_string += i + "|"
# append last character
new_input_string += input_string[-1]
# we will store the starting and ending of previous furthest ending palindromic
# substring
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = 0, 0
# length[i] shows the length of palindromic substring with center i
UpperCAmelCase_ : Tuple = [1 for i in range(len(_a ) )]
# for each character in new_string find corresponding palindromic string
UpperCAmelCase_ : Optional[Any] = 0
for j in range(len(_a ) ):
UpperCAmelCase_ : Tuple = 1 if j > r else min(length[l + r - j] // 2 , r - j + 1 )
while (
j - k >= 0
and j + k < len(_a )
and new_input_string[k + j] == new_input_string[j - k]
):
k += 1
UpperCAmelCase_ : Optional[Any] = 2 * k - 1
# does this string is ending after the previously explored end (that is r) ?
# if yes the update the new r to the last index of this
if j + k - 1 > r:
UpperCAmelCase_ : int = j - k + 1 # noqa: E741
UpperCAmelCase_ : List[str] = j + k - 1
# update max_length and start position
if max_length < length[j]:
UpperCAmelCase_ : List[str] = length[j]
UpperCAmelCase_ : List[str] = j
# create that string
UpperCAmelCase_ : Optional[Any] = new_input_string[start - max_length // 2 : start + max_length // 2 + 1]
for i in s:
if i != "|":
output_string += i
return output_string
if __name__ == "__main__":
import doctest
doctest.testmod()
| 345 |
from collections import OrderedDict
from typing import TYPE_CHECKING, Any, Mapping, Optional
from packaging import version
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...onnx.utils import compute_effective_axis_dimension
from ...utils import logging
if TYPE_CHECKING:
from ...processing_utils import ProcessorMixin
from ...utils import TensorType
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/layoutlmv3-base''': '''https://huggingface.co/microsoft/layoutlmv3-base/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "layoutlmv3"
def __init__( self: str ,lowerCamelCase_: Any=50265 ,lowerCamelCase_: int=768 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=3072 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Any=0.1 ,lowerCamelCase_: Tuple=512 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: List[str]=1e-5 ,lowerCamelCase_: int=1 ,lowerCamelCase_: int=0 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Dict=1024 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=32 ,lowerCamelCase_: Union[str, Any]=128 ,lowerCamelCase_: Tuple=64 ,lowerCamelCase_: Tuple=256 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Any=True ,lowerCamelCase_: Dict=224 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Optional[int]=16 ,lowerCamelCase_: Dict=None ,**lowerCamelCase_: str ,) -> List[Any]:
super().__init__(
vocab_size=lowerCamelCase_ ,hidden_size=lowerCamelCase_ ,num_hidden_layers=lowerCamelCase_ ,num_attention_heads=lowerCamelCase_ ,intermediate_size=lowerCamelCase_ ,hidden_act=lowerCamelCase_ ,hidden_dropout_prob=lowerCamelCase_ ,attention_probs_dropout_prob=lowerCamelCase_ ,max_position_embeddings=lowerCamelCase_ ,type_vocab_size=lowerCamelCase_ ,initializer_range=lowerCamelCase_ ,layer_norm_eps=lowerCamelCase_ ,pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = max_ad_position_embeddings
UpperCAmelCase_ : Optional[int] = coordinate_size
UpperCAmelCase_ : Optional[int] = shape_size
UpperCAmelCase_ : Optional[Any] = has_relative_attention_bias
UpperCAmelCase_ : Optional[int] = rel_pos_bins
UpperCAmelCase_ : Union[str, Any] = max_rel_pos
UpperCAmelCase_ : Dict = has_spatial_attention_bias
UpperCAmelCase_ : Optional[int] = rel_ad_pos_bins
UpperCAmelCase_ : Tuple = max_rel_ad_pos
UpperCAmelCase_ : Union[str, Any] = text_embed
UpperCAmelCase_ : Optional[Any] = visual_embed
UpperCAmelCase_ : List[str] = input_size
UpperCAmelCase_ : str = num_channels
UpperCAmelCase_ : Optional[int] = patch_size
UpperCAmelCase_ : Tuple = classifier_dropout
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = version.parse("1.12" )
@property
def A__ ( self: Dict ) -> Mapping[str, Mapping[int, str]]:
# The order of inputs is different for question answering and sequence classification
if self.task in ["question-answering", "sequence-classification"]:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels""", 2: """height""", 3: """width"""}),
] )
else:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels"""}),
] )
@property
def A__ ( self: Any ) -> float:
return 1e-5
@property
def A__ ( self: int ) -> int:
return 12
def A__ ( self: List[str] ,lowerCamelCase_: "ProcessorMixin" ,lowerCamelCase_: int = -1 ,lowerCamelCase_: int = -1 ,lowerCamelCase_: bool = False ,lowerCamelCase_: Optional["TensorType"] = None ,lowerCamelCase_: int = 3 ,lowerCamelCase_: int = 40 ,lowerCamelCase_: int = 40 ,) -> Mapping[str, Any]:
setattr(processor.image_processor ,"""apply_ocr""" ,lowerCamelCase_ )
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
UpperCAmelCase_ : List[str] = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_batch ,num_token_to_add=0 )
# If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX
UpperCAmelCase_ : int = processor.tokenizer.num_special_tokens_to_add(lowerCamelCase_ )
UpperCAmelCase_ : int = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_sequence ,num_token_to_add=lowerCamelCase_ )
# Generate dummy inputs according to compute batch and sequence
UpperCAmelCase_ : Optional[int] = [[""" """.join([processor.tokenizer.unk_token] ) * seq_length]] * batch_size
# Generate dummy bounding boxes
UpperCAmelCase_ : List[Any] = [[[48, 84, 73, 128]]] * batch_size
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
# batch_size = compute_effective_axis_dimension(batch_size, fixed_dimension=OnnxConfig.default_fixed_batch)
UpperCAmelCase_ : Any = self._generate_dummy_images(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = dict(
processor(
lowerCamelCase_ ,text=lowerCamelCase_ ,boxes=lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,) )
return inputs
| 345 | 1 |
from __future__ import annotations
from fractions import Fraction
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
return (
num != den and num % 10 == den // 10 and (num // 10) / (den % 10) == num / den
)
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Union[str, Any] = 11
UpperCAmelCase_ : Optional[Any] = int("""1""" + """0""" * digit_len )
for num in range(_a , _a ):
while den <= 99:
if (num != den) and (num % 10 == den // 10) and (den % 10 != 0):
if is_digit_cancelling(_a , _a ):
solutions.append(F'''{num}/{den}''' )
den += 1
num += 1
UpperCAmelCase_ : str = 10
return solutions
def lowerCamelCase_ ( _a : int = 2 ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = 1.0
for fraction in fraction_list(_a ):
UpperCAmelCase_ : Tuple = Fraction(_a )
result *= frac.denominator / frac.numerator
return int(_a )
if __name__ == "__main__":
print(solution())
| 345 |
import argparse
from argparse import Namespace
import torch
from torch import nn
from transformers import XGLMConfig, XGLMForCausalLM
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = [
"""decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = emb.weight.shape
UpperCAmelCase_ : Tuple = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : List[Any] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = torch.load(_a , map_location="""cpu""" )
UpperCAmelCase_ : Dict = Namespace(**checkpoint["""cfg"""]["""model"""] )
UpperCAmelCase_ : Optional[int] = checkpoint["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = state_dict["""decoder.embed_tokens.weight"""].shape[0]
UpperCAmelCase_ : List[str] = {key.replace("""decoder""" , """model""" ): val for key, val in state_dict.items()}
UpperCAmelCase_ : int = XGLMConfig(
vocab_size=_a , max_position_embeddings=args.max_target_positions , num_layers=args.decoder_layers , attention_heads=args.decoder_attention_heads , ffn_dim=args.decoder_ffn_embed_dim , d_model=args.decoder_embed_dim , layerdrop=args.decoder_layerdrop , dropout=args.dropout , attention_dropout=args.attention_dropout , activation_dropout=args.activation_dropout , activation_function="""gelu""" , scale_embedding=not args.no_scale_embedding , tie_word_embeddings=args.share_decoder_input_output_embed , )
UpperCAmelCase_ : List[str] = XGLMForCausalLM(_a )
UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
print(_a )
UpperCAmelCase_ : Optional[Any] = make_linear_from_emb(model.model.embed_tokens )
return model
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument('''fairseq_path''', type=str, help='''path to a model.pt on local filesystem.''')
parser.add_argument('''pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = convert_fairseq_xglm_checkpoint_from_disk(args.fairseq_path)
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 | 1 |
import collections
import os
from shutil import copyfile
from typing import Any, Dict, List, Optional, Tuple
from ...tokenization_utils import PreTrainedTokenizer
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = '''▁'''
UpperCamelCase_ = {'''vocab_file''': '''prophetnet.tokenizer'''}
UpperCamelCase_ = {
'''vocab_file''': {
'''microsoft/xprophetnet-large-wiki100-cased''': (
'''https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/prophetnet.tokenizer'''
),
}
}
UpperCamelCase_ = {
'''microsoft/xprophetnet-large-wiki100-cased''': {'''do_lower_case''': False},
}
UpperCamelCase_ = {
'''microsoft/xprophetnet-large-wiki100-cased''': 512,
}
def lowerCamelCase_ ( _a : Optional[int] ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = collections.OrderedDict()
with open(_a , """r""" , encoding="""utf-8""" ) as reader:
UpperCAmelCase_ : Dict = reader.readlines()
for index, token in enumerate(_a ):
UpperCAmelCase_ : str = token.rstrip("""\n""" )
UpperCAmelCase_ : int = index
return vocab
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = VOCAB_FILES_NAMES
A__ : Tuple = PRETRAINED_VOCAB_FILES_MAP
A__ : int = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Optional[Any] = ["input_ids", "attention_mask"]
def __init__( self: Union[str, Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: Union[str, Any]="[SEP]" ,lowerCamelCase_: Any="[SEP]" ,lowerCamelCase_: Any="[SEP]" ,lowerCamelCase_: List[Any]="[UNK]" ,lowerCamelCase_: Union[str, Any]="[PAD]" ,lowerCamelCase_: str="[CLS]" ,lowerCamelCase_: Optional[Any]="[MASK]" ,lowerCamelCase_: Optional[Dict[str, Any]] = None ,**lowerCamelCase_: List[Any] ,) -> None:
UpperCAmelCase_ : Dict = {} if sp_model_kwargs is None else sp_model_kwargs
super().__init__(
bos_token=lowerCamelCase_ ,eos_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,sp_model_kwargs=self.sp_model_kwargs ,**lowerCamelCase_ ,)
try:
import sentencepiece as spm
except ImportError:
logger.warning(
"""You need to install SentencePiece to use XLMRobertaTokenizer: https://github.com/google/sentencepiece"""
""" pip install sentencepiece""" )
raise
UpperCAmelCase_ : List[str] = spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(str(lowerCamelCase_ ) )
UpperCAmelCase_ : Tuple = vocab_file
# Original fairseq vocab and spm vocab must be "aligned":
# Vocab | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
# -------- | ------- | ------- | ------ | ------- | --- | --- | --- | ----- | ----- | ----
# fairseq | '<s>' | '<pad>' | '</s>' | '<unk>' | ',' | '.' | '▁' | 's' | '▁de' | '-'
# spm | '<unk>' | '<s>' | '</s>' | ',' | '.' | '▁' | 's' | '▁de' | '-' | '▁a'
# put special tokens and [unused] tokens into the vocab
UpperCAmelCase_ : str = {"""[PAD]""": 0, """[CLS]""": 1, """[SEP]""": 2, """[UNK]""": 3, """[MASK]""": 4}
for i in range(10 ):
UpperCAmelCase_ : Union[str, Any] = F'''[unused{i}]'''
UpperCAmelCase_ : Optional[int] = 5 + i
# The first "real" token "," has position 15 in the embedding vocab and position 3 in the spm vocab
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : int = {v: k for k, v in self.fairseq_tokens_to_ids.items()}
for k in self.fairseq_tokens_to_ids.keys():
self.unique_no_split_tokens.append(lowerCamelCase_ )
def __getstate__( self: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : str = self.__dict__.copy()
UpperCAmelCase_ : Union[str, Any] = None
return state
def __setstate__( self: int ,lowerCamelCase_: Optional[int] ) -> Any:
UpperCAmelCase_ : List[str] = d
try:
import sentencepiece as spm
except ImportError:
logger.warning(
"""You need to install SentencePiece to use XLMRobertaTokenizer: https://github.com/google/sentencepiece"""
""" pip install sentencepiece""" )
raise
# for backward compatibility
if not hasattr(self ,"""sp_model_kwargs""" ):
UpperCAmelCase_ : Any = {}
UpperCAmelCase_ : Any = spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(self.vocab_file )
def A__ ( self: List[str] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ,lowerCamelCase_: bool = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=lowerCamelCase_ ,token_ids_a=lowerCamelCase_ ,already_has_special_tokens=lowerCamelCase_ )
if token_ids_a is None:
return ([0] * len(lowerCamelCase_ )) + [1]
return ([0] * len(lowerCamelCase_ )) + [1] + ([0] * len(lowerCamelCase_ )) + [1]
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Tuple = [self.sep_token_id]
if token_ids_a is None:
return len(token_ids_a + sep ) * [0]
return len(token_ids_a + sep + sep + token_ids_a + sep ) * [0]
@property
def A__ ( self: Union[str, Any] ) -> Optional[int]:
return len(self.sp_model ) + self.fairseq_offset
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[int] = {self.convert_ids_to_tokens(lowerCamelCase_ ): i for i in range(self.vocab_size )}
vocab.update(self.added_tokens_encoder )
return vocab
def A__ ( self: Optional[Any] ,lowerCamelCase_: str ) -> str:
return self.sp_model.encode(lowerCamelCase_ ,out_type=lowerCamelCase_ )
def A__ ( self: List[Any] ,lowerCamelCase_: Dict ) -> Optional[int]:
if token in self.fairseq_tokens_to_ids:
return self.fairseq_tokens_to_ids[token]
UpperCAmelCase_ : int = self.sp_model.PieceToId(lowerCamelCase_ )
# Need to return unknown token if the SP model returned 0
return spm_id + self.fairseq_offset if spm_id else self.unk_token_id
def A__ ( self: List[str] ,lowerCamelCase_: Dict ) -> Union[str, Any]:
if index in self.fairseq_ids_to_tokens:
return self.fairseq_ids_to_tokens[index]
return self.sp_model.IdToPiece(index - self.fairseq_offset )
def A__ ( self: Tuple ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Union[str, Any] = """""".join(lowerCamelCase_ ).replace(lowerCamelCase_ ,""" """ ).strip()
return out_string
def A__ ( self: Tuple ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCamelCase_ ):
logger.error(F'''Vocabulary path ({save_directory}) should be a directory''' )
return
UpperCAmelCase_ : Optional[int] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""vocab_file"""] )
if os.path.abspath(self.vocab_file ) != os.path.abspath(lowerCamelCase_ ) and os.path.isfile(self.vocab_file ):
copyfile(self.vocab_file ,lowerCamelCase_ )
elif not os.path.isfile(self.vocab_file ):
with open(lowerCamelCase_ ,"""wb""" ) as fi:
UpperCAmelCase_ : List[Any] = self.sp_model.serialized_model_proto()
fi.write(lowerCamelCase_ )
return (out_vocab_file,)
def A__ ( self: Optional[int] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
if token_ids_a is None:
return token_ids_a + [self.sep_token_id]
UpperCAmelCase_ : Optional[Any] = [self.sep_token_id]
return token_ids_a + sep + token_ids_a + sep
| 345 |
import collections
import inspect
import unittest
from transformers import FocalNetConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
FocalNetBackbone,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetModel,
)
from transformers.models.focalnet.modeling_focalnet import FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=13 ,lowerCamelCase_: Optional[int]=32 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Optional[Any]=[32, 64, 128] ,lowerCamelCase_: Optional[int]=[1, 2, 1] ,lowerCamelCase_: Union[str, Any]=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[str]=2.0 ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: int=True ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: str=True ,lowerCamelCase_: Dict=10 ,lowerCamelCase_: str=8 ,lowerCamelCase_: Union[str, Any]=["stage1", "stage2"] ,lowerCamelCase_: Optional[Any]=[1, 2] ,) -> str:
UpperCAmelCase_ : List[Any] = parent
UpperCAmelCase_ : Tuple = batch_size
UpperCAmelCase_ : Any = image_size
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : List[str] = num_channels
UpperCAmelCase_ : Dict = embed_dim
UpperCAmelCase_ : Dict = hidden_sizes
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = num_heads
UpperCAmelCase_ : List[Any] = window_size
UpperCAmelCase_ : Union[str, Any] = mlp_ratio
UpperCAmelCase_ : int = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Union[str, Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[int] = drop_path_rate
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : List[Any] = use_absolute_embeddings
UpperCAmelCase_ : List[Any] = patch_norm
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Optional[Any] = scope
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = encoder_stride
UpperCAmelCase_ : Optional[int] = out_features
UpperCAmelCase_ : Optional[int] = out_indices
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : int = None
if self.use_labels:
UpperCAmelCase_ : str = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Any = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Tuple:
return FocalNetConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,hidden_sizes=self.hidden_sizes ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,out_features=self.out_features ,out_indices=self.out_indices ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : Dict = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : Optional[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[str]:
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size, 8, 8] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,len(config.out_features ) )
self.parent.assertListEqual(model.channels ,config.hidden_sizes[:-1] )
# verify backbone works with out_features=None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size * 2, 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,1 )
self.parent.assertListEqual(model.channels ,[config.hidden_sizes[-1]] )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Any = FocalNetForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.reconstruction.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : int = 1
UpperCAmelCase_ : List[str] = FocalNetForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.reconstruction.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ) -> int:
UpperCAmelCase_ : List[Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
# test greyscale images
UpperCAmelCase_ : List[Any] = 1
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = config_and_inputs
UpperCAmelCase_ : int = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (
(
FocalNetModel,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetBackbone,
)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (
{"feature-extraction": FocalNetModel, "image-classification": FocalNetForImageClassification}
if is_torch_available()
else {}
)
A__ : Optional[Any] = False
A__ : Any = False
A__ : List[str] = False
A__ : Any = False
A__ : Any = False
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : Dict = FocalNetModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: List[str] ) -> Union[str, Any]:
return
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCamelCase_ )
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@unittest.skip(reason="""FocalNet does not use inputs_embeds""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""FocalNet does not use feedforward chunking""" )
def A__ ( self: Optional[Any] ) -> Optional[Any]:
pass
def A__ ( self: Optional[Any] ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : List[Any] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Dict = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Any = [*signature.parameters.keys()]
UpperCAmelCase_ : List[str] = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.hidden_states
UpperCAmelCase_ : List[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# FocalNet has a different seq_length
UpperCAmelCase_ : int = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Optional[int] = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Union[str, Any] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = reshaped_hidden_states[0].shape
UpperCAmelCase_ : List[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Union[str, Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Tuple = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : Union[str, Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Union[str, Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Any = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Optional[int] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
@slow
def A__ ( self: Optional[int] ) -> Optional[Any]:
for model_name in FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Tuple = FocalNetModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[Any] = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Optional[int] ) -> str:
# TODO update organization
return AutoImageProcessor.from_pretrained("""microsoft/focalnet-tiny""" ) if is_vision_available() else None
@slow
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification.from_pretrained("""microsoft/focalnet-tiny""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.default_image_processor
UpperCAmelCase_ : Union[str, Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Dict = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : str = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_1_6_6, -0.4_3_6_8, 0.2_1_9_1] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
self.assertTrue(outputs.logits.argmax(dim=-1 ).item() ,281 )
@require_torch
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (FocalNetBackbone,) if is_torch_available() else ()
A__ : int = FocalNetConfig
A__ : List[str] = False
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : str = FocalNetModelTester(self )
| 345 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
UpperCamelCase_ = {
'''configuration_deberta''': ['''DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''DebertaConfig''', '''DebertaOnnxConfig'''],
'''tokenization_deberta''': ['''DebertaTokenizer'''],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = ['''DebertaTokenizerFast''']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''DebertaForMaskedLM''',
'''DebertaForQuestionAnswering''',
'''DebertaForSequenceClassification''',
'''DebertaForTokenClassification''',
'''DebertaModel''',
'''DebertaPreTrainedModel''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''TF_DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TFDebertaForMaskedLM''',
'''TFDebertaForQuestionAnswering''',
'''TFDebertaForSequenceClassification''',
'''TFDebertaForTokenClassification''',
'''TFDebertaModel''',
'''TFDebertaPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_deberta import DEBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, DebertaConfig, DebertaOnnxConfig
from .tokenization_deberta import DebertaTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_deberta_fast import DebertaTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_deberta import (
DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
DebertaForMaskedLM,
DebertaForQuestionAnswering,
DebertaForSequenceClassification,
DebertaForTokenClassification,
DebertaModel,
DebertaPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_deberta import (
TF_DEBERTA_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDebertaForMaskedLM,
TFDebertaForQuestionAnswering,
TFDebertaForSequenceClassification,
TFDebertaForTokenClassification,
TFDebertaModel,
TFDebertaPreTrainedModel,
)
else:
import sys
UpperCamelCase_ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 345 |
import collections
import inspect
import unittest
from transformers import SwinvaConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import SwinvaForImageClassification, SwinvaForMaskedImageModeling, SwinvaModel
from transformers.models.swinva.modeling_swinva import SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: int=13 ,lowerCamelCase_: int=32 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: Any=3 ,lowerCamelCase_: str=16 ,lowerCamelCase_: Optional[Any]=[1, 2, 1] ,lowerCamelCase_: Tuple=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[Any]=2.0 ,lowerCamelCase_: str=True ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[Any]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Union[str, Any]=False ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=10 ,lowerCamelCase_: Tuple=8 ,) -> List[Any]:
UpperCAmelCase_ : List[str] = parent
UpperCAmelCase_ : int = batch_size
UpperCAmelCase_ : int = image_size
UpperCAmelCase_ : Union[str, Any] = patch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : int = embed_dim
UpperCAmelCase_ : Union[str, Any] = depths
UpperCAmelCase_ : List[str] = num_heads
UpperCAmelCase_ : int = window_size
UpperCAmelCase_ : List[str] = mlp_ratio
UpperCAmelCase_ : Tuple = qkv_bias
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : str = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = drop_path_rate
UpperCAmelCase_ : List[str] = hidden_act
UpperCAmelCase_ : int = use_absolute_embeddings
UpperCAmelCase_ : Any = patch_norm
UpperCAmelCase_ : Optional[int] = layer_norm_eps
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Dict = scope
UpperCAmelCase_ : int = use_labels
UpperCAmelCase_ : Optional[Any] = type_sequence_label_size
UpperCAmelCase_ : List[str] = encoder_stride
def A__ ( self: Any ) -> int:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : List[Any] = None
if self.use_labels:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : str = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Union[str, Any]:
return SwinvaConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,)
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ) -> str:
UpperCAmelCase_ : str = SwinvaModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : List[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Any = SwinvaForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.logits.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : str = 1
UpperCAmelCase_ : Optional[Any] = SwinvaForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : int = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ) -> int:
UpperCAmelCase_ : Union[str, Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = SwinvaForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: str ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = config_and_inputs
UpperCAmelCase_ : Optional[int] = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Tuple = (
(SwinvaModel, SwinvaForImageClassification, SwinvaForMaskedImageModeling) if is_torch_available() else ()
)
A__ : Optional[Any] = (
{"feature-extraction": SwinvaModel, "image-classification": SwinvaForImageClassification}
if is_torch_available()
else {}
)
A__ : List[Any] = False
A__ : Tuple = False
A__ : int = False
A__ : Union[str, Any] = False
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = SwinvaModelTester(self )
UpperCAmelCase_ : str = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 )
def A__ ( self: Optional[int] ) -> List[Any]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: Any ) -> Dict:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
@unittest.skip(reason="""Got `CUDA error: misaligned address` with PyTorch 2.0.0.""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""Swinv2 does not use inputs_embeds""" )
def A__ ( self: Tuple ) -> List[str]:
pass
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : Tuple = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Any = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : int = [*signature.parameters.keys()]
UpperCAmelCase_ : Tuple = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Any = True
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
UpperCAmelCase_ : Union[str, Any] = False
UpperCAmelCase_ : str = True
UpperCAmelCase_ : List[Any] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[Any] = outputs.attentions
UpperCAmelCase_ : List[str] = len(self.model_tester.depths )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
UpperCAmelCase_ : str = True
UpperCAmelCase_ : Optional[Any] = config.window_size**2
UpperCAmelCase_ : Optional[int] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[Any] = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
UpperCAmelCase_ : Optional[Any] = len(lowerCamelCase_ )
# Check attention is always last and order is fine
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = True
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
if hasattr(self.model_tester ,"""num_hidden_states_types""" ):
UpperCAmelCase_ : List[Any] = self.model_tester.num_hidden_states_types
else:
# also another +1 for reshaped_hidden_states
UpperCAmelCase_ : List[str] = 2
self.assertEqual(out_len + added_hidden_states ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
def A__ ( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : int = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[str] = outputs.hidden_states
UpperCAmelCase_ : Optional[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# Swinv2 has a different seq_length
UpperCAmelCase_ : Optional[Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : int = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Optional[int] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = reshaped_hidden_states[0].shape
UpperCAmelCase_ : Optional[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Dict = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes:
UpperCAmelCase_ : Any = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Union[str, Any] = 3
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : List[str] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : List[Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Optional[Any] = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : List[str] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@slow
def A__ ( self: str ) -> Tuple:
for model_name in SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Dict = SwinvaModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : List[str] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and "logit_scale" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Dict ) -> Optional[Any]:
return (
AutoImageProcessor.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" )
if is_vision_available()
else None
)
@slow
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ : Tuple = SwinvaForImageClassification.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" ).to(
lowerCamelCase_ )
UpperCAmelCase_ : Any = self.default_image_processor
UpperCAmelCase_ : List[str] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Optional[int] = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : Dict = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.tensor([-0.3_9_4_7, -0.4_3_0_6, 0.0_0_2_6] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
| 345 | 1 |
import argparse
import requests
import torch
from PIL import Image
from transformers import CLIPProcessor, GroupViTConfig, GroupViTModel
def lowerCamelCase_ ( _a : Optional[Any] ):
'''simple docstring'''
if "img_encoder.pos_embed" in name:
UpperCAmelCase_ : Optional[int] = name.replace("""img_encoder.pos_embed""" , """vision_model.embeddings.position_embeddings""" )
if "img_encoder.patch_embed.proj" in name:
UpperCAmelCase_ : Optional[int] = name.replace("""img_encoder.patch_embed.proj""" , """vision_model.embeddings.patch_embeddings.projection""" )
if "img_encoder.patch_embed.norm" in name:
UpperCAmelCase_ : List[str] = name.replace("""img_encoder.patch_embed.norm""" , """vision_model.embeddings.layernorm""" )
if "img_encoder.layers" in name:
UpperCAmelCase_ : Any = name.replace("""img_encoder.layers""" , """vision_model.encoder.stages""" )
if "blocks" in name and "res" not in name:
UpperCAmelCase_ : Tuple = name.replace("""blocks""" , """layers""" )
if "attn" in name and "pre_assign" not in name:
UpperCAmelCase_ : List[Any] = name.replace("""attn""" , """self_attn""" )
if "proj" in name and "self_attn" in name and "text" not in name:
UpperCAmelCase_ : Optional[Any] = name.replace("""proj""" , """out_proj""" )
if "pre_assign_attn.attn.proj" in name:
UpperCAmelCase_ : List[str] = name.replace("""pre_assign_attn.attn.proj""" , """pre_assign_attn.attn.out_proj""" )
if "norm1" in name:
UpperCAmelCase_ : Tuple = name.replace("""norm1""" , """layer_norm1""" )
if "norm2" in name and "pre_assign" not in name:
UpperCAmelCase_ : Union[str, Any] = name.replace("""norm2""" , """layer_norm2""" )
if "img_encoder.norm" in name:
UpperCAmelCase_ : Optional[Any] = name.replace("""img_encoder.norm""" , """vision_model.layernorm""" )
# text encoder
if "text_encoder.token_embedding" in name:
UpperCAmelCase_ : Any = name.replace("""text_encoder.token_embedding""" , """text_model.embeddings.token_embedding""" )
if "text_encoder.positional_embedding" in name:
UpperCAmelCase_ : int = name.replace("""text_encoder.positional_embedding""" , """text_model.embeddings.position_embedding.weight""" )
if "text_encoder.transformer.resblocks." in name:
UpperCAmelCase_ : List[Any] = name.replace("""text_encoder.transformer.resblocks.""" , """text_model.encoder.layers.""" )
if "ln_1" in name:
UpperCAmelCase_ : str = name.replace("""ln_1""" , """layer_norm1""" )
if "ln_2" in name:
UpperCAmelCase_ : Union[str, Any] = name.replace("""ln_2""" , """layer_norm2""" )
if "c_fc" in name:
UpperCAmelCase_ : List[Any] = name.replace("""c_fc""" , """fc1""" )
if "c_proj" in name:
UpperCAmelCase_ : List[Any] = name.replace("""c_proj""" , """fc2""" )
if "text_encoder" in name:
UpperCAmelCase_ : Union[str, Any] = name.replace("""text_encoder""" , """text_model""" )
if "ln_final" in name:
UpperCAmelCase_ : str = name.replace("""ln_final""" , """final_layer_norm""" )
# projection layers
if "img_projector.linear_hidden." in name:
UpperCAmelCase_ : Optional[Any] = name.replace("""img_projector.linear_hidden.""" , """visual_projection.""" )
if "img_projector.linear_out." in name:
UpperCAmelCase_ : Optional[int] = name.replace("""img_projector.linear_out.""" , """visual_projection.3.""" )
if "text_projector.linear_hidden" in name:
UpperCAmelCase_ : Tuple = name.replace("""text_projector.linear_hidden""" , """text_projection""" )
if "text_projector.linear_out" in name:
UpperCAmelCase_ : str = name.replace("""text_projector.linear_out""" , """text_projection.3""" )
return name
def lowerCamelCase_ ( _a : Optional[Any] , _a : Dict ):
'''simple docstring'''
for key in orig_state_dict.copy().keys():
UpperCAmelCase_ : str = orig_state_dict.pop(_a )
if "qkv" in key:
# weights and biases of the key, value and query projections of vision encoder's attention layers require special treatment:
# we need to split them up into separate matrices/vectors
UpperCAmelCase_ : Tuple = key.split(""".""" )
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = int(key_split[2] ), int(key_split[4] )
UpperCAmelCase_ : int = config.vision_config.hidden_size
if "weight" in key:
UpperCAmelCase_ : Tuple = val[:dim, :]
UpperCAmelCase_ : Dict = val[dim : dim * 2, :]
UpperCAmelCase_ : Optional[Any] = val[-dim:, :]
else:
UpperCAmelCase_ : Tuple = val[:dim]
UpperCAmelCase_ : str = val[dim : dim * 2]
UpperCAmelCase_ : Optional[Any] = val[-dim:]
elif "in_proj" in key:
# weights and biases of the key, value and query projections of text encoder's attention layers require special treatment:
# we need to split them up into separate matrices/vectors
UpperCAmelCase_ : List[Any] = key.split(""".""" )
UpperCAmelCase_ : int = int(key_split[3] )
UpperCAmelCase_ : int = config.text_config.hidden_size
if "weight" in key:
UpperCAmelCase_ : Dict = val[:dim, :]
UpperCAmelCase_ : Optional[int] = val[
dim : dim * 2, :
]
UpperCAmelCase_ : str = val[-dim:, :]
else:
UpperCAmelCase_ : Optional[Any] = val[:dim]
UpperCAmelCase_ : Optional[int] = val[dim : dim * 2]
UpperCAmelCase_ : int = val[-dim:]
else:
UpperCAmelCase_ : List[str] = rename_key(_a )
# squeeze if necessary
if (
"text_projection.0" in new_name
or "text_projection.3" in new_name
or "visual_projection.0" in new_name
or "visual_projection.3" in new_name
):
UpperCAmelCase_ : Tuple = val.squeeze_()
else:
UpperCAmelCase_ : int = val
return orig_state_dict
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = """http://images.cocodataset.org/val2017/000000039769.jpg"""
UpperCAmelCase_ : Optional[Any] = Image.open(requests.get(_a , stream=_a ).raw )
return im
@torch.no_grad()
def lowerCamelCase_ ( _a : Optional[Any] , _a : str , _a : List[str]="groupvit-gcc-yfcc" , _a : Tuple=False ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = GroupViTConfig()
UpperCAmelCase_ : Union[str, Any] = GroupViTModel(_a ).eval()
UpperCAmelCase_ : str = torch.load(_a , map_location="""cpu""" )["""model"""]
UpperCAmelCase_ : List[str] = convert_state_dict(_a , _a )
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
assert missing_keys == ["text_model.embeddings.position_ids"]
assert (unexpected_keys == ["multi_label_logit_scale"]) or (len(_a ) == 0)
# verify result
UpperCAmelCase_ : int = CLIPProcessor.from_pretrained("""openai/clip-vit-base-patch32""" )
UpperCAmelCase_ : Union[str, Any] = prepare_img()
UpperCAmelCase_ : Optional[int] = processor(text=["""a photo of a cat""", """a photo of a dog"""] , images=_a , padding=_a , return_tensors="""pt""" )
with torch.no_grad():
UpperCAmelCase_ : int = model(**_a )
if model_name == "groupvit-gcc-yfcc":
UpperCAmelCase_ : Optional[Any] = torch.tensor([[1_3.3_5_2_3, 6.3_6_2_9]] )
elif model_name == "groupvit-gcc-redcaps":
UpperCAmelCase_ : str = torch.tensor([[1_6.1_8_7_3, 8.6_2_3_0]] )
else:
raise ValueError(F'''Model name {model_name} not supported.''' )
assert torch.allclose(outputs.logits_per_image , _a , atol=1E-3 )
processor.save_pretrained(_a )
model.save_pretrained(_a )
print("""Successfully saved processor and model to""" , _a )
if push_to_hub:
print("""Pushing to the hub...""" )
processor.push_to_hub(_a , organization="""nielsr""" )
model.push_to_hub(_a , organization="""nielsr""" )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
parser.add_argument(
'''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to dump the processor and PyTorch model.'''
)
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to GroupViT checkpoint''')
parser.add_argument(
'''--model_name''',
default='''groupvit-gccy-fcc''',
type=str,
help='''Name of the model. Expecting either \'groupvit-gcc-yfcc\' or \'groupvit-gcc-redcaps\'''',
)
parser.add_argument(
'''--push_to_hub''',
action='''store_true''',
help='''Whether or not to push the converted model and processor to the 🤗 hub using the provided `model_name`.''',
)
UpperCamelCase_ = parser.parse_args()
convert_groupvit_checkpoint(args.checkpoint_path, args.pytorch_dump_folder_path, args.model_name, args.push_to_hub)
| 345 |
import json
import os
from functools import lru_cache
from typing import Dict, List, Optional, Tuple, Union
import regex as re
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...tokenization_utils_base import BatchEncoding, EncodedInput
from ...utils import PaddingStrategy, logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {'''vocab_file''': '''vocab.json''', '''merges_file''': '''merges.txt'''}
# See all LED models at https://huggingface.co/models?filter=LED
UpperCamelCase_ = {
'''vocab_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/vocab.json''',
},
'''merges_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/merges.txt''',
},
'''tokenizer_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/tokenizer.json''',
},
}
UpperCamelCase_ = {
'''allenai/led-base-16384''': 16384,
}
@lru_cache()
# Copied from transformers.models.bart.tokenization_bart.bytes_to_unicode
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : int = (
list(range(ord("""!""" ) , ord("""~""" ) + 1 ) ) + list(range(ord("""¡""" ) , ord("""¬""" ) + 1 ) ) + list(range(ord("""®""" ) , ord("""ÿ""" ) + 1 ) )
)
UpperCAmelCase_ : Dict = bs[:]
UpperCAmelCase_ : Any = 0
for b in range(2**8 ):
if b not in bs:
bs.append(_a )
cs.append(2**8 + n )
n += 1
UpperCAmelCase_ : Any = [chr(_a ) for n in cs]
return dict(zip(_a , _a ) )
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = set()
UpperCAmelCase_ : List[Any] = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
UpperCAmelCase_ : Optional[int] = char
return pairs
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = VOCAB_FILES_NAMES
A__ : List[str] = PRETRAINED_VOCAB_FILES_MAP
A__ : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Optional[int] = ["input_ids", "attention_mask"]
def __init__( self: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any]="replace" ,lowerCamelCase_: Optional[Any]="<s>" ,lowerCamelCase_: List[Any]="</s>" ,lowerCamelCase_: List[str]="</s>" ,lowerCamelCase_: int="<s>" ,lowerCamelCase_: int="<unk>" ,lowerCamelCase_: str="<pad>" ,lowerCamelCase_: Optional[Any]="<mask>" ,lowerCamelCase_: List[str]=False ,**lowerCamelCase_: Tuple ,) -> Any:
UpperCAmelCase_ : Union[str, Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else bos_token
UpperCAmelCase_ : int = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else eos_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else sep_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else cls_token
UpperCAmelCase_ : Optional[Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else unk_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else pad_token
# Mask token behave like a normal word, i.e. include the space before it
UpperCAmelCase_ : str = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else mask_token
super().__init__(
errors=lowerCamelCase_ ,bos_token=lowerCamelCase_ ,eos_token=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,add_prefix_space=lowerCamelCase_ ,**lowerCamelCase_ ,)
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as vocab_handle:
UpperCAmelCase_ : Union[str, Any] = json.load(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {v: k for k, v in self.encoder.items()}
UpperCAmelCase_ : Any = errors # how to handle errors in decoding
UpperCAmelCase_ : int = bytes_to_unicode()
UpperCAmelCase_ : Dict = {v: k for k, v in self.byte_encoder.items()}
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as merges_handle:
UpperCAmelCase_ : Any = merges_handle.read().split("""\n""" )[1:-1]
UpperCAmelCase_ : int = [tuple(merge.split() ) for merge in bpe_merges]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : Tuple = {}
UpperCAmelCase_ : Optional[int] = add_prefix_space
# Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions
UpperCAmelCase_ : int = re.compile(R"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""" )
@property
# Copied from transformers.models.bart.tokenization_bart.BartTokenizer.vocab_size
def A__ ( self: List[str] ) -> List[str]:
return len(self.encoder )
def A__ ( self: Any ) -> Union[str, Any]:
return dict(self.encoder ,**self.added_tokens_encoder )
def A__ ( self: Tuple ,lowerCamelCase_: Dict ) -> Optional[Any]:
if token in self.cache:
return self.cache[token]
UpperCAmelCase_ : Union[str, Any] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = get_pairs(lowerCamelCase_ )
if not pairs:
return token
while True:
UpperCAmelCase_ : Union[str, Any] = min(lowerCamelCase_ ,key=lambda lowerCamelCase_ : self.bpe_ranks.get(lowerCamelCase_ ,float("""inf""" ) ) )
if bigram not in self.bpe_ranks:
break
UpperCAmelCase_ , UpperCAmelCase_ : Any = bigram
UpperCAmelCase_ : Optional[Any] = []
UpperCAmelCase_ : List[str] = 0
while i < len(lowerCamelCase_ ):
try:
UpperCAmelCase_ : str = word.index(lowerCamelCase_ ,lowerCamelCase_ )
except ValueError:
new_word.extend(word[i:] )
break
else:
new_word.extend(word[i:j] )
UpperCAmelCase_ : Union[str, Any] = j
if word[i] == first and i < len(lowerCamelCase_ ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
UpperCAmelCase_ : List[str] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = new_word
if len(lowerCamelCase_ ) == 1:
break
else:
UpperCAmelCase_ : List[str] = get_pairs(lowerCamelCase_ )
UpperCAmelCase_ : int = """ """.join(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = word
return word
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> List[str]:
UpperCAmelCase_ : str = []
for token in re.findall(self.pat ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = """""".join(
self.byte_encoder[b] for b in token.encode("""utf-8""" ) ) # Maps all our bytes to unicode strings, avoiding control tokens of the BPE (spaces in our case)
bpe_tokens.extend(bpe_token for bpe_token in self.bpe(lowerCamelCase_ ).split(""" """ ) )
return bpe_tokens
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ) -> Optional[int]:
return self.encoder.get(lowerCamelCase_ ,self.encoder.get(self.unk_token ) )
def A__ ( self: List[str] ,lowerCamelCase_: str ) -> Optional[Any]:
return self.decoder.get(lowerCamelCase_ )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ) -> List[Any]:
UpperCAmelCase_ : str = """""".join(lowerCamelCase_ )
UpperCAmelCase_ : int = bytearray([self.byte_decoder[c] for c in text] ).decode("""utf-8""" ,errors=self.errors )
return text
def A__ ( self: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCamelCase_ ):
logger.error(F'''Vocabulary path ({save_directory}) should be a directory''' )
return
UpperCAmelCase_ : List[Any] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : List[str] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""merges_file"""] )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as f:
f.write(json.dumps(self.encoder ,indent=2 ,sort_keys=lowerCamelCase_ ,ensure_ascii=lowerCamelCase_ ) + """\n""" )
UpperCAmelCase_ : str = 0
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as writer:
writer.write("""#version: 0.2\n""" )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() ,key=lambda lowerCamelCase_ : kv[1] ):
if index != token_index:
logger.warning(
F'''Saving vocabulary to {merge_file}: BPE merge indices are not consecutive.'''
""" Please check that the tokenizer is not corrupted!""" )
UpperCAmelCase_ : Tuple = token_index
writer.write(""" """.join(lowerCamelCase_ ) + """\n""" )
index += 1
return vocab_file, merge_file
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
UpperCAmelCase_ : int = [self.cls_token_id]
UpperCAmelCase_ : Optional[int] = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ,lowerCamelCase_: bool = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=lowerCamelCase_ ,token_ids_a=lowerCamelCase_ ,already_has_special_tokens=lowerCamelCase_ )
if token_ids_a is None:
return [1] + ([0] * len(lowerCamelCase_ )) + [1]
return [1] + ([0] * len(lowerCamelCase_ )) + [1, 1] + ([0] * len(lowerCamelCase_ )) + [1]
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Optional[Any] = [self.sep_token_id]
UpperCAmelCase_ : Optional[Any] = [self.cls_token_id]
if token_ids_a is None:
return len(cls + token_ids_a + sep ) * [0]
return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0]
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str=False ,**lowerCamelCase_: List[str] ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = kwargs.pop("""add_prefix_space""" ,self.add_prefix_space )
if (is_split_into_words or add_prefix_space) and (len(lowerCamelCase_ ) > 0 and not text[0].isspace()):
UpperCAmelCase_ : Dict = """ """ + text
return (text, kwargs)
def A__ ( self: List[str] ,lowerCamelCase_: Union[Dict[str, EncodedInput], BatchEncoding] ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: PaddingStrategy = PaddingStrategy.DO_NOT_PAD ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: Optional[bool] = None ,) -> dict:
UpperCAmelCase_ : Optional[int] = super()._pad(
encoded_inputs=lowerCamelCase_ ,max_length=lowerCamelCase_ ,padding_strategy=lowerCamelCase_ ,pad_to_multiple_of=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,)
# Load from model defaults
if return_attention_mask is None:
UpperCAmelCase_ : str = """attention_mask""" in self.model_input_names
if return_attention_mask and "global_attention_mask" in encoded_inputs:
UpperCAmelCase_ : str = encoded_inputs[self.model_input_names[0]]
# `global_attention_mask` need to have the same length as other (sequential) inputs.
UpperCAmelCase_ : List[Any] = len(encoded_inputs["""global_attention_mask"""] ) != len(lowerCamelCase_ )
if needs_to_be_padded:
UpperCAmelCase_ : Dict = len(lowerCamelCase_ ) - len(encoded_inputs["""global_attention_mask"""] )
if self.padding_side == "right":
# Use `-1` since `0` in `global_attention_mask` means `local attention` instead of `not to attend`
UpperCAmelCase_ : str = (
encoded_inputs["""global_attention_mask"""] + [-1] * difference
)
elif self.padding_side == "left":
UpperCAmelCase_ : List[str] = [-1] * difference + encoded_inputs[
"""global_attention_mask"""
]
else:
raise ValueError("""Invalid padding strategy:""" + str(self.padding_side ) )
return encoded_inputs
| 345 | 1 |
import logging
import os
import random
import sys
from dataclasses import dataclass, field
from typing import Optional
import datasets
import numpy as np
import pandas as pd
from datasets import load_dataset
import transformers
from transformers import (
AutoConfig,
BartForSequenceClassification,
DataCollatorWithPadding,
EvalPrediction,
HfArgumentParser,
TapexTokenizer,
Trainer,
TrainingArguments,
default_data_collator,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version
from transformers.utils.versions import require_version
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version('''4.17.0.dev0''')
require_version('''datasets>=1.8.0''', '''To fix: pip install -r examples/pytorch/text-classification/requirements.txt''')
UpperCamelCase_ = logging.getLogger(__name__)
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[str] = field(
default="tab_fact" , metadata={"help": "The name of the dataset to use (via the datasets library)."} )
A__ : Optional[str] = field(
default="tab_fact" , metadata={"help": "The configuration name of the dataset to use (via the datasets library)."} , )
A__ : int = field(
default=1_024 , metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
} , )
A__ : bool = field(
default=__snake_case , metadata={"help": "Overwrite the cached preprocessed datasets or not."} )
A__ : bool = field(
default=__snake_case , metadata={
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
} , )
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
} , )
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
} , )
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "A csv or a json file containing the training data."} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "A csv or a json file containing the validation data."} )
A__ : Optional[str] = field(default=__snake_case , metadata={"help": "A csv or a json file containing the test data."} )
def A__ ( self: List[str] ) -> List[Any]:
if self.dataset_name is not None:
pass
elif self.train_file is None or self.validation_file is None:
raise ValueError("""Need either a GLUE task, a training/validation file or a dataset name.""" )
else:
UpperCAmelCase_ : Optional[Any] = self.train_file.split(""".""" )[-1]
assert train_extension in ["csv", "json"], "`train_file` should be a csv or a json file."
UpperCAmelCase_ : str = self.validation_file.split(""".""" )[-1]
assert (
validation_extension == train_extension
), "`validation_file` should have the same extension (csv or json) as `train_file`."
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=__snake_case , metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Pretrained config name or path if not the same as model_name"} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"} , )
A__ : bool = field(
default=__snake_case , metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."} , )
A__ : str = field(
default="main" , metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."} , )
A__ : bool = field(
default=__snake_case , metadata={
"help": (
"Will use the token generated when running `huggingface-cli login` (necessary to use this script "
"with private models)."
)
} , )
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : int = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments) )
if len(sys.argv ) == 2 and sys.argv[1].endswith(""".json""" ):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Dict = parser.parse_args_into_dataclasses()
# Setup logging
logging.basicConfig(
format="""%(asctime)s - %(levelname)s - %(name)s - %(message)s""" , datefmt="""%m/%d/%Y %H:%M:%S""" , handlers=[logging.StreamHandler(sys.stdout )] , )
UpperCAmelCase_ : int = training_args.get_process_log_level()
logger.setLevel(_a )
datasets.utils.logging.set_verbosity(_a )
transformers.utils.logging.set_verbosity(_a )
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Log on each process the small summary:
logger.warning(
F'''Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}'''
+ F'''distributed training: {bool(training_args.local_rank != -1 )}, 16-bits training: {training_args.fpaa}''' )
logger.info(F'''Training/evaluation parameters {training_args}''' )
# Detecting last checkpoint.
UpperCAmelCase_ : str = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
UpperCAmelCase_ : str = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F'''Output directory ({training_args.output_dir}) already exists and is not empty. '''
"""Use --overwrite_output_dir to overcome.""" )
elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
logger.info(
F'''Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change '''
"""the `--output_dir` or add `--overwrite_output_dir` to train from scratch.""" )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON training and evaluation files (see below)
# or specify a GLUE benchmark task (the dataset will be downloaded automatically from the datasets Hub).
#
# For JSON files, this script will use the `question` column for the input question and `table` column for the corresponding table.
#
# If the CSVs/JSONs contain only one non-label column, the script does single sentence classification on this
# single column. You can easily tweak this behavior (see below)
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
UpperCAmelCase_ : List[str] = load_dataset(
data_args.dataset_name , data_args.dataset_config_name , cache_dir=model_args.cache_dir )
else:
# Loading a dataset from your local files.
# CSV/JSON training and evaluation files are needed.
UpperCAmelCase_ : Any = {"""train""": data_args.train_file, """validation""": data_args.validation_file}
# Get the test dataset: you can provide your own CSV/JSON test file (see below)
# when you use `do_predict` without specifying a GLUE benchmark task.
if training_args.do_predict:
if data_args.test_file is not None:
UpperCAmelCase_ : Tuple = data_args.train_file.split(""".""" )[-1]
UpperCAmelCase_ : int = data_args.test_file.split(""".""" )[-1]
assert (
test_extension == train_extension
), "`test_file` should have the same extension (csv or json) as `train_file`."
UpperCAmelCase_ : Dict = data_args.test_file
else:
raise ValueError("""Need either a GLUE task or a test file for `do_predict`.""" )
for key in data_files.keys():
logger.info(F'''load a local file for {key}: {data_files[key]}''' )
if data_args.train_file.endswith(""".csv""" ):
# Loading a dataset from local csv files
UpperCAmelCase_ : Union[str, Any] = load_dataset("""csv""" , data_files=_a , cache_dir=model_args.cache_dir )
else:
# Loading a dataset from local json files
UpperCAmelCase_ : List[str] = load_dataset("""json""" , data_files=_a , cache_dir=model_args.cache_dir )
# See more about loading any type of standard or custom dataset at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Labels
UpperCAmelCase_ : Any = raw_datasets["""train"""].features["""label"""].names
UpperCAmelCase_ : Optional[Any] = len(_a )
# Load pretrained model and tokenizer
#
# In distributed training, the .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
UpperCAmelCase_ : List[str] = AutoConfig.from_pretrained(
model_args.config_name if model_args.config_name else model_args.model_name_or_path , num_labels=_a , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , )
# load tapex tokenizer
UpperCAmelCase_ : Optional[Any] = TapexTokenizer.from_pretrained(
model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path , cache_dir=model_args.cache_dir , use_fast=model_args.use_fast_tokenizer , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , add_prefix_space=_a , )
UpperCAmelCase_ : List[Any] = BartForSequenceClassification.from_pretrained(
model_args.model_name_or_path , from_tf=bool(""".ckpt""" in model_args.model_name_or_path ) , config=_a , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , )
# Padding strategy
if data_args.pad_to_max_length:
UpperCAmelCase_ : Union[str, Any] = """max_length"""
else:
# We will pad later, dynamically at batch creation, to the max sequence length in each batch
UpperCAmelCase_ : int = False
# Some models have set the order of the labels to use, so let's make sure we do use it.
UpperCAmelCase_ : int = {"""Refused""": 0, """Entailed""": 1}
UpperCAmelCase_ : List[Any] = {0: """Refused""", 1: """Entailed"""}
if data_args.max_seq_length > tokenizer.model_max_length:
logger.warning(
F'''The max_seq_length passed ({data_args.max_seq_length}) is larger than the maximum length for the'''
F'''model ({tokenizer.model_max_length}). Using max_seq_length={tokenizer.model_max_length}.''' )
UpperCAmelCase_ : Optional[Any] = min(data_args.max_seq_length , tokenizer.model_max_length )
def preprocess_tabfact_function(_a : Optional[Any] ):
# Tokenize the texts
def _convert_table_text_to_pandas(_a : Optional[Any] ):
UpperCAmelCase_ : int = [_table_row.split("""#""" ) for _table_row in _table_text.strip("""\n""" ).split("""\n""" )]
UpperCAmelCase_ : Union[str, Any] = pd.DataFrame.from_records(_table_content[1:] , columns=_table_content[0] )
return _table_pd
UpperCAmelCase_ : List[str] = examples["""statement"""]
UpperCAmelCase_ : int = list(map(_convert_table_text_to_pandas , examples["""table_text"""] ) )
UpperCAmelCase_ : List[Any] = tokenizer(_a , _a , padding=_a , max_length=_a , truncation=_a )
UpperCAmelCase_ : Union[str, Any] = examples["""label"""]
return result
with training_args.main_process_first(desc="""dataset map pre-processing""" ):
UpperCAmelCase_ : List[Any] = raw_datasets.map(
_a , batched=_a , load_from_cache_file=not data_args.overwrite_cache , desc="""Running tokenizer on dataset""" , )
if training_args.do_train:
if "train" not in raw_datasets:
raise ValueError("""--do_train requires a train dataset""" )
UpperCAmelCase_ : str = raw_datasets["""train"""]
if data_args.max_train_samples is not None:
UpperCAmelCase_ : List[Any] = train_dataset.select(range(data_args.max_train_samples ) )
if training_args.do_eval:
if "validation" not in raw_datasets and "validation_matched" not in raw_datasets:
raise ValueError("""--do_eval requires a validation dataset""" )
UpperCAmelCase_ : str = raw_datasets["""validation"""]
if data_args.max_eval_samples is not None:
UpperCAmelCase_ : Tuple = eval_dataset.select(range(data_args.max_eval_samples ) )
if training_args.do_predict or data_args.test_file is not None:
if "test" not in raw_datasets and "test_matched" not in raw_datasets:
raise ValueError("""--do_predict requires a test dataset""" )
UpperCAmelCase_ : Optional[Any] = raw_datasets["""test"""]
if data_args.max_predict_samples is not None:
UpperCAmelCase_ : Optional[int] = predict_dataset.select(range(data_args.max_predict_samples ) )
# Log a few random samples from the training set:
if training_args.do_train:
for index in random.sample(range(len(_a ) ) , 3 ):
logger.info(F'''Sample {index} of the training set: {train_dataset[index]}.''' )
# You can define your custom compute_metrics function. It takes an `EvalPrediction` object (a namedtuple with a
# predictions and label_ids field) and has to return a dictionary string to float.
def compute_metrics(_a : EvalPrediction ):
UpperCAmelCase_ : Any = p.predictions[0] if isinstance(p.predictions , _a ) else p.predictions
UpperCAmelCase_ : List[str] = np.argmax(_a , axis=1 )
return {"accuracy": (preds == p.label_ids).astype(np.floataa ).mean().item()}
# Data collator will default to DataCollatorWithPadding, so we change it if we already did the padding.
if data_args.pad_to_max_length:
UpperCAmelCase_ : Union[str, Any] = default_data_collator
elif training_args.fpaa:
UpperCAmelCase_ : int = DataCollatorWithPadding(_a , pad_to_multiple_of=8 )
else:
UpperCAmelCase_ : str = None
# Initialize our Trainer
UpperCAmelCase_ : List[Any] = Trainer(
model=_a , args=_a , train_dataset=train_dataset if training_args.do_train else None , eval_dataset=eval_dataset if training_args.do_eval else None , compute_metrics=_a , tokenizer=_a , data_collator=_a , )
# Training
if training_args.do_train:
UpperCAmelCase_ : List[str] = None
if training_args.resume_from_checkpoint is not None:
UpperCAmelCase_ : Optional[Any] = training_args.resume_from_checkpoint
elif last_checkpoint is not None:
UpperCAmelCase_ : Optional[Any] = last_checkpoint
UpperCAmelCase_ : Any = trainer.train(resume_from_checkpoint=_a )
UpperCAmelCase_ : str = train_result.metrics
UpperCAmelCase_ : List[str] = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(_a )
)
UpperCAmelCase_ : str = min(_a , len(_a ) )
trainer.save_model() # Saves the tokenizer too for easy upload
trainer.log_metrics("""train""" , _a )
trainer.save_metrics("""train""" , _a )
trainer.save_state()
# Evaluation
if training_args.do_eval:
logger.info("""*** Evaluate ***""" )
UpperCAmelCase_ : Dict = trainer.evaluate(eval_dataset=_a )
UpperCAmelCase_ : Union[str, Any] = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(_a )
UpperCAmelCase_ : List[Any] = min(_a , len(_a ) )
trainer.log_metrics("""eval""" , _a )
trainer.save_metrics("""eval""" , _a )
if training_args.do_predict:
logger.info("""*** Predict ***""" )
# Removing the `label` columns because it contains -1 and Trainer won't like that.
UpperCAmelCase_ : Dict = predict_dataset.remove_columns("""label""" )
UpperCAmelCase_ : Any = trainer.predict(_a , metric_key_prefix="""predict""" ).predictions
UpperCAmelCase_ : Optional[int] = np.argmax(_a , axis=1 )
UpperCAmelCase_ : Union[str, Any] = os.path.join(training_args.output_dir , """predict_results_tabfact.txt""" )
if trainer.is_world_process_zero():
with open(_a , """w""" ) as writer:
logger.info("""***** Predict Results *****""" )
writer.write("""index\tprediction\n""" )
for index, item in enumerate(_a ):
UpperCAmelCase_ : List[Any] = label_list[item]
writer.write(F'''{index}\t{item}\n''' )
UpperCAmelCase_ : Tuple = {"""finetuned_from""": model_args.model_name_or_path, """tasks""": """text-classification"""}
if training_args.push_to_hub:
trainer.push_to_hub(**_a )
else:
trainer.create_model_card(**_a )
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
main()
if __name__ == "__main__":
main()
| 345 |
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Optional[Any] = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[Any] ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : int = UNetaDConditionModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=4 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,)
return model
@property
def A__ ( self: str ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = AutoencoderKL(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Tuple:
def extract(*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: str ):
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = torch.ones([0] )
def A__ ( self: List[Any] ,lowerCamelCase_: str ) -> int:
self.pixel_values.to(lowerCamelCase_ )
return self
return Out()
return extract
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : int = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : int = self.dummy_cond_unet
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(
beta_start=0.0_0_0_8_5 ,beta_end=0.0_1_2 ,beta_schedule="""scaled_linear""" ,clip_sample=lowerCamelCase_ ,set_alpha_to_one=lowerCamelCase_ ,)
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : int = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : List[Any] = output.images
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Dict = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : int = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Dict = self.dummy_cond_unet
UpperCAmelCase_ : List[Any] = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : Union[str, Any] = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : str = output.images
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(
"""hf-internal-testing/tiny-stable-diffusion-lms-pipe""" ,safety_checker=lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(pipe.scheduler ,lowerCamelCase_ )
assert pipe.safety_checker is None
UpperCAmelCase_ : List[Any] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(lowerCamelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
UpperCAmelCase_ : Optional[int] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Tuple = self.dummy_cond_unet
UpperCAmelCase_ : Dict = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# put models in fp16
UpperCAmelCase_ : Optional[Any] = unet.half()
UpperCAmelCase_ : Optional[int] = vae.half()
UpperCAmelCase_ : int = bert.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = sd_pipe([prompt] ,num_inference_steps=2 ,output_type="""np""" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Tuple = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : str = (
"""portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"""
""" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"""
""" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"""
""" children from bahnhof zoo, detailed """
)
UpperCAmelCase_ : Optional[int] = 4003660346
UpperCAmelCase_ : int = 7
# without safety guidance (sld_guidance_scale = 0)
UpperCAmelCase_ : Dict = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Optional[int] = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Dict = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
# without safety guidance (strong configuration)
UpperCAmelCase_ : Union[str, Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Any = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Tuple = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : str = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Any = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = """padme amidala taking a bath artwork, safe for work, no nudity"""
UpperCAmelCase_ : List[Any] = 2734971755
UpperCAmelCase_ : Optional[Any] = 7
UpperCAmelCase_ : int = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : Tuple = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
UpperCAmelCase_ : Any = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : List[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" )
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = (
"""the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."""
""" leyendecker"""
)
UpperCAmelCase_ : Optional[Any] = 1044355234
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : List[Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Any = output.images
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-7
UpperCAmelCase_ : Optional[int] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Any = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Any = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 345 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''facebook/dpr-ctx_encoder-single-nq-base''': (
'''https://huggingface.co/facebook/dpr-ctx_encoder-single-nq-base/resolve/main/config.json'''
),
'''facebook/dpr-question_encoder-single-nq-base''': (
'''https://huggingface.co/facebook/dpr-question_encoder-single-nq-base/resolve/main/config.json'''
),
'''facebook/dpr-reader-single-nq-base''': (
'''https://huggingface.co/facebook/dpr-reader-single-nq-base/resolve/main/config.json'''
),
'''facebook/dpr-ctx_encoder-multiset-base''': (
'''https://huggingface.co/facebook/dpr-ctx_encoder-multiset-base/resolve/main/config.json'''
),
'''facebook/dpr-question_encoder-multiset-base''': (
'''https://huggingface.co/facebook/dpr-question_encoder-multiset-base/resolve/main/config.json'''
),
'''facebook/dpr-reader-multiset-base''': (
'''https://huggingface.co/facebook/dpr-reader-multiset-base/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Tuple = "dpr"
def __init__( self: List[str] ,lowerCamelCase_: Tuple=30522 ,lowerCamelCase_: Union[str, Any]=768 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: Optional[Any]=12 ,lowerCamelCase_: Optional[Any]=3072 ,lowerCamelCase_: Optional[Any]="gelu" ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: str=512 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: int=0.0_2 ,lowerCamelCase_: List[Any]=1e-12 ,lowerCamelCase_: Tuple=0 ,lowerCamelCase_: Dict="absolute" ,lowerCamelCase_: int = 0 ,**lowerCamelCase_: List[str] ,) -> Optional[int]:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Dict = vocab_size
UpperCAmelCase_ : List[str] = hidden_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : Dict = num_attention_heads
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : Optional[Any] = intermediate_size
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : List[Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Any = max_position_embeddings
UpperCAmelCase_ : str = type_vocab_size
UpperCAmelCase_ : Any = initializer_range
UpperCAmelCase_ : Optional[Any] = layer_norm_eps
UpperCAmelCase_ : Tuple = projection_dim
UpperCAmelCase_ : str = position_embedding_type
| 345 |
import unittest
from transformers import MobileBertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
MobileBertModel,
)
class _snake_case :
'''simple docstring'''
def __init__( self: Optional[int] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Tuple=13 ,lowerCamelCase_: int=7 ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: str=True ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: int=99 ,lowerCamelCase_: List[str]=64 ,lowerCamelCase_: Tuple=32 ,lowerCamelCase_: List[str]=5 ,lowerCamelCase_: str=4 ,lowerCamelCase_: str=37 ,lowerCamelCase_: Union[str, Any]="gelu" ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: str=0.1 ,lowerCamelCase_: List[str]=512 ,lowerCamelCase_: Dict=16 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: List[str]=0.0_2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: Union[str, Any]=4 ,lowerCamelCase_: str=None ,) -> List[str]:
UpperCAmelCase_ : Any = parent
UpperCAmelCase_ : List[Any] = batch_size
UpperCAmelCase_ : Union[str, Any] = seq_length
UpperCAmelCase_ : Optional[int] = is_training
UpperCAmelCase_ : Dict = use_input_mask
UpperCAmelCase_ : Any = use_token_type_ids
UpperCAmelCase_ : Tuple = use_labels
UpperCAmelCase_ : List[Any] = vocab_size
UpperCAmelCase_ : str = hidden_size
UpperCAmelCase_ : List[str] = embedding_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : List[Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : str = hidden_dropout_prob
UpperCAmelCase_ : List[str] = attention_probs_dropout_prob
UpperCAmelCase_ : Any = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : Any = type_sequence_label_size
UpperCAmelCase_ : Optional[Any] = initializer_range
UpperCAmelCase_ : Optional[int] = num_labels
UpperCAmelCase_ : Optional[int] = num_choices
UpperCAmelCase_ : List[str] = scope
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : List[str] = None
if self.use_input_mask:
UpperCAmelCase_ : Tuple = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : Dict = None
if self.use_token_type_ids:
UpperCAmelCase_ : str = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : int = None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : Union[str, Any] = None
if self.use_labels:
UpperCAmelCase_ : List[str] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : int = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : Tuple = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: Any ) -> Dict:
return MobileBertConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,embedding_size=self.embedding_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> int:
UpperCAmelCase_ : Any = MobileBertModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Dict ) -> int:
UpperCAmelCase_ : Union[str, Any] = MobileBertForMaskedLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: str ,lowerCamelCase_: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : List[Any] = MobileBertForNextSentencePrediction(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 2) )
def A__ ( self: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Tuple = MobileBertForPreTraining(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,next_sentence_label=lowerCamelCase_ ,)
self.parent.assertEqual(result.prediction_logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape ,(self.batch_size, 2) )
def A__ ( self: Any ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = MobileBertForQuestionAnswering(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,start_positions=lowerCamelCase_ ,end_positions=lowerCamelCase_ ,)
self.parent.assertEqual(result.start_logits.shape ,(self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape ,(self.batch_size, self.seq_length) )
def A__ ( self: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> str:
UpperCAmelCase_ : Optional[Any] = self.num_labels
UpperCAmelCase_ : Union[str, Any] = MobileBertForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_labels) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Any:
UpperCAmelCase_ : str = self.num_labels
UpperCAmelCase_ : Optional[int] = MobileBertForTokenClassification(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Tuple ,lowerCamelCase_: str ,lowerCamelCase_: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Union[str, Any] = self.num_choices
UpperCAmelCase_ : Tuple = MobileBertForMultipleChoice(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = input_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Union[str, Any] = token_type_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : str = input_mask.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_choices) )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ : str = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Union[str, Any] = config_and_inputs
UpperCAmelCase_ : Dict = {"""input_ids""": input_ids, """token_type_ids""": token_type_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = (
(
MobileBertModel,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
)
if is_torch_available()
else ()
)
A__ : List[str] = (
{
"feature-extraction": MobileBertModel,
"fill-mask": MobileBertForMaskedLM,
"question-answering": MobileBertForQuestionAnswering,
"text-classification": MobileBertForSequenceClassification,
"token-classification": MobileBertForTokenClassification,
"zero-shot": MobileBertForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : List[str] = True
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: int=False ) -> Union[str, Any]:
UpperCAmelCase_ : List[Any] = super()._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ,return_labels=lowerCamelCase_ )
if return_labels:
if model_class in get_values(lowerCamelCase_ ):
UpperCAmelCase_ : Any = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = torch.zeros(
self.model_tester.batch_size ,dtype=torch.long ,device=lowerCamelCase_ )
return inputs_dict
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[str] = MobileBertModelTester(self )
UpperCAmelCase_ : Union[str, Any] = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Optional[Any] ) -> List[Any]:
self.config_tester.run_common_tests()
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_model(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_masked_lm(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Tuple:
UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_multiple_choice(*lowerCamelCase_ )
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_next_sequence_prediction(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_pretraining(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_question_answering(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_sequence_classification(*lowerCamelCase_ )
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_token_classification(*lowerCamelCase_ )
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return torch.tensor(
_a , dtype=torch.long , device=_a , )
UpperCamelCase_ = 1E-3
@require_torch
@require_sentencepiece
@require_tokenizers
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: List[Any] ) -> str:
UpperCAmelCase_ : Any = MobileBertModel.from_pretrained("""google/mobilebert-uncased""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = _long_tensor([[101, 7110, 1005, 1056, 2023, 11333, 17413, 1029, 102]] )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )[0]
UpperCAmelCase_ : Union[str, Any] = torch.Size((1, 9, 512) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = torch.tensor(
[
[
[-2.473_6526e07, 8.269_1656e04, 1.652_1838e05],
[-5.754_1704e-01, 3.905_6022e00, 4.401_1507e00],
[2.604_7359e00, 1.567_7652e00, -1.732_4188e-01],
]
] ,device=lowerCamelCase_ ,)
# MobileBERT results range from 10e0 to 10e8. Even a 0.0000001% difference with a value of 10e8 results in a
# ~1 difference, it's therefore not a good idea to measure using addition.
# Here, we instead divide the expected result with the result in order to obtain ~1. We then check that the
# result is held between bounds: 1 - TOLERANCE < expected_result / result < 1 + TOLERANCE
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) >= 1 - TOLERANCE )
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) <= 1 + TOLERANCE )
self.assertTrue(lower_bound and upper_bound )
| 345 | 1 |
import gc
import random
import unittest
import numpy as np
import torch
from PIL import Image
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, DDPMScheduler, StableDiffusionUpscalePipeline, UNetaDConditionModel
from diffusers.utils import floats_tensor, load_image, load_numpy, slow, torch_device
from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu
enable_full_determinism()
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Dict ) -> str:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: Optional[int] ) -> Optional[Any]:
UpperCAmelCase_ : Optional[int] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : List[Any] = (32, 32)
UpperCAmelCase_ : str = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[str] ) -> Tuple:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = UNetaDConditionModel(
block_out_channels=(32, 32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=7 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,attention_head_dim=8 ,use_linear_projection=lowerCamelCase_ ,only_cross_attention=(True, True, False) ,num_class_embeds=100 ,)
return model
@property
def A__ ( self: str ) -> List[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : List[str] = AutoencoderKL(
block_out_channels=[32, 32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Tuple ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,hidden_act="""gelu""" ,projection_dim=512 ,)
return CLIPTextModel(lowerCamelCase_ )
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Optional[Any] = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Any = self.dummy_cond_unet_upscale
UpperCAmelCase_ : int = DDPMScheduler()
UpperCAmelCase_ : List[Any] = DDIMScheduler(prediction_type="""v_prediction""" )
UpperCAmelCase_ : List[str] = self.dummy_vae
UpperCAmelCase_ : str = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
UpperCAmelCase_ : Any = self.dummy_image.cpu().permute(0 ,2 ,3 ,1 )[0]
UpperCAmelCase_ : Optional[int] = Image.fromarray(np.uinta(lowerCamelCase_ ) ).convert("""RGB""" ).resize((64, 64) )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionUpscalePipeline(
unet=lowerCamelCase_ ,low_res_scheduler=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,max_noise_level=350 ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,noise_level=20 ,num_inference_steps=2 ,output_type="""np""" ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Optional[int] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = sd_pipe(
[prompt] ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,noise_level=20 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[int] = image_from_tuple[0, -3:, -3:, -1]
UpperCAmelCase_ : str = low_res_image.size[0] * 4
assert image.shape == (1, expected_height_width, expected_height_width, 3)
UpperCAmelCase_ : Union[str, Any] = np.array([0.3_1_1_3, 0.3_9_1_0, 0.4_2_7_2, 0.4_8_5_9, 0.5_0_6_1, 0.4_6_5_2, 0.5_3_6_2, 0.5_7_1_5, 0.5_6_6_1] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Optional[Any]:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Tuple = self.dummy_cond_unet_upscale
UpperCAmelCase_ : Tuple = DDPMScheduler()
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(prediction_type="""v_prediction""" )
UpperCAmelCase_ : Any = self.dummy_vae
UpperCAmelCase_ : Optional[int] = self.dummy_text_encoder
UpperCAmelCase_ : Optional[Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
UpperCAmelCase_ : int = self.dummy_image.cpu().permute(0 ,2 ,3 ,1 )[0]
UpperCAmelCase_ : List[str] = Image.fromarray(np.uinta(lowerCamelCase_ ) ).convert("""RGB""" ).resize((64, 64) )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Optional[Any] = StableDiffusionUpscalePipeline(
unet=lowerCamelCase_ ,low_res_scheduler=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,max_noise_level=350 ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : int = sd_pipe(
2 * [prompt] ,image=2 * [low_res_image] ,guidance_scale=6.0 ,noise_level=20 ,num_inference_steps=2 ,output_type="""np""" ,)
UpperCAmelCase_ : Union[str, Any] = output.images
assert image.shape[0] == 2
UpperCAmelCase_ : Any = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = sd_pipe(
[prompt] ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,num_images_per_prompt=2 ,guidance_scale=6.0 ,noise_level=20 ,num_inference_steps=2 ,output_type="""np""" ,)
UpperCAmelCase_ : Union[str, Any] = output.images
assert image.shape[0] == 2
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : Optional[int] = self.dummy_cond_unet_upscale
UpperCAmelCase_ : Optional[Any] = DDPMScheduler()
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(prediction_type="""v_prediction""" )
UpperCAmelCase_ : List[str] = self.dummy_vae
UpperCAmelCase_ : Any = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
UpperCAmelCase_ : List[Any] = self.dummy_image.cpu().permute(0 ,2 ,3 ,1 )[0]
UpperCAmelCase_ : Tuple = Image.fromarray(np.uinta(lowerCamelCase_ ) ).convert("""RGB""" ).resize((64, 64) )
# put models in fp16, except vae as it overflows in fp16
UpperCAmelCase_ : int = unet.half()
UpperCAmelCase_ : Optional[int] = text_encoder.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : List[str] = StableDiffusionUpscalePipeline(
unet=lowerCamelCase_ ,low_res_scheduler=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,max_noise_level=350 ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = torch.manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""np""" ,).images
UpperCAmelCase_ : List[str] = low_res_image.size[0] * 4
assert image.shape == (1, expected_height_width, expected_height_width, 3)
@slow
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: int ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Tuple = load_image(
"""https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"""
"""/sd2-upscale/low_res_cat.png""" )
UpperCAmelCase_ : Optional[Any] = load_numpy(
"""https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd2-upscale"""
"""/upsampled_cat.npy""" )
UpperCAmelCase_ : int = """stabilityai/stable-diffusion-x4-upscaler"""
UpperCAmelCase_ : Any = StableDiffusionUpscalePipeline.from_pretrained(lowerCamelCase_ )
pipe.to(lowerCamelCase_ )
pipe.set_progress_bar_config(disable=lowerCamelCase_ )
pipe.enable_attention_slicing()
UpperCAmelCase_ : int = """a cat sitting on a park bench"""
UpperCAmelCase_ : Optional[int] = torch.manual_seed(0 )
UpperCAmelCase_ : Dict = pipe(
prompt=lowerCamelCase_ ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,output_type="""np""" ,)
UpperCAmelCase_ : Optional[Any] = output.images[0]
assert image.shape == (512, 512, 3)
assert np.abs(expected_image - image ).max() < 1e-3
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : List[str] = load_image(
"""https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"""
"""/sd2-upscale/low_res_cat.png""" )
UpperCAmelCase_ : int = load_numpy(
"""https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd2-upscale"""
"""/upsampled_cat_fp16.npy""" )
UpperCAmelCase_ : Optional[int] = """stabilityai/stable-diffusion-x4-upscaler"""
UpperCAmelCase_ : Tuple = StableDiffusionUpscalePipeline.from_pretrained(
lowerCamelCase_ ,torch_dtype=torch.floataa ,)
pipe.to(lowerCamelCase_ )
pipe.set_progress_bar_config(disable=lowerCamelCase_ )
pipe.enable_attention_slicing()
UpperCAmelCase_ : List[str] = """a cat sitting on a park bench"""
UpperCAmelCase_ : List[Any] = torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = pipe(
prompt=lowerCamelCase_ ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,output_type="""np""" ,)
UpperCAmelCase_ : Optional[Any] = output.images[0]
assert image.shape == (512, 512, 3)
assert np.abs(expected_image - image ).max() < 5e-1
def A__ ( self: int ) -> Tuple:
torch.cuda.empty_cache()
torch.cuda.reset_max_memory_allocated()
torch.cuda.reset_peak_memory_stats()
UpperCAmelCase_ : Dict = load_image(
"""https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"""
"""/sd2-upscale/low_res_cat.png""" )
UpperCAmelCase_ : Tuple = """stabilityai/stable-diffusion-x4-upscaler"""
UpperCAmelCase_ : Optional[int] = StableDiffusionUpscalePipeline.from_pretrained(
lowerCamelCase_ ,torch_dtype=torch.floataa ,)
pipe.to(lowerCamelCase_ )
pipe.set_progress_bar_config(disable=lowerCamelCase_ )
pipe.enable_attention_slicing(1 )
pipe.enable_sequential_cpu_offload()
UpperCAmelCase_ : List[Any] = """a cat sitting on a park bench"""
UpperCAmelCase_ : Optional[Any] = torch.manual_seed(0 )
UpperCAmelCase_ : Any = pipe(
prompt=lowerCamelCase_ ,image=lowerCamelCase_ ,generator=lowerCamelCase_ ,num_inference_steps=5 ,output_type="""np""" ,)
UpperCAmelCase_ : Dict = torch.cuda.max_memory_allocated()
# make sure that less than 2.9 GB is allocated
assert mem_bytes < 2.9 * 10**9
| 345 |
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import AutoTokenizer, BarkProcessor
from transformers.testing_utils import require_torch, slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: str ) -> int:
UpperCAmelCase_ : List[Any] = """ylacombe/bark-small"""
UpperCAmelCase_ : Tuple = tempfile.mkdtemp()
UpperCAmelCase_ : Union[str, Any] = """en_speaker_1"""
UpperCAmelCase_ : Optional[Any] = """This is a test string"""
UpperCAmelCase_ : int = """speaker_embeddings_path.json"""
UpperCAmelCase_ : Any = """speaker_embeddings"""
def A__ ( self: Tuple ,**lowerCamelCase_: List[str] ) -> List[Any]:
return AutoTokenizer.from_pretrained(self.checkpoint ,**lowerCamelCase_ )
def A__ ( self: str ) -> Union[str, Any]:
shutil.rmtree(self.tmpdirname )
def A__ ( self: List[Any] ) -> int:
UpperCAmelCase_ : int = self.get_tokenizer()
UpperCAmelCase_ : Tuple = BarkProcessor(tokenizer=lowerCamelCase_ )
processor.save_pretrained(self.tmpdirname )
UpperCAmelCase_ : Optional[int] = BarkProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer.get_vocab() )
@slow
def A__ ( self: List[Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
processor.save_pretrained(
self.tmpdirname ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,speaker_embeddings_directory=self.speaker_embeddings_directory ,)
UpperCAmelCase_ : Optional[Any] = self.get_tokenizer(bos_token="""(BOS)""" ,eos_token="""(EOS)""" )
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
self.tmpdirname ,self.speaker_embeddings_dict_path ,bos_token="""(BOS)""" ,eos_token="""(EOS)""" ,)
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer_add_kwargs.get_vocab() )
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
UpperCAmelCase_ : Optional[int] = 35
UpperCAmelCase_ : Optional[int] = 2
UpperCAmelCase_ : Dict = 8
UpperCAmelCase_ : Optional[int] = {
"""semantic_prompt""": np.ones(lowerCamelCase_ ),
"""coarse_prompt""": np.ones((nb_codebooks_coarse, seq_len) ),
"""fine_prompt""": np.ones((nb_codebooks_total, seq_len) ),
}
# test providing already loaded voice_preset
UpperCAmelCase_ : str = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from npz file
UpperCAmelCase_ : List[Any] = os.path.join(self.tmpdirname ,"""file.npz""" )
np.savez(lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : int = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from the hub
UpperCAmelCase_ : Union[str, Any] = processor(text=self.input_string ,voice_preset=self.voice_preset )
def A__ ( self: Dict ) -> Tuple:
UpperCAmelCase_ : Any = self.get_tokenizer()
UpperCAmelCase_ : Dict = BarkProcessor(tokenizer=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string )
UpperCAmelCase_ : str = tokenizer(
self.input_string ,padding="""max_length""" ,max_length=256 ,add_special_tokens=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,return_token_type_ids=lowerCamelCase_ ,)
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] ,encoded_processor[key].squeeze().tolist() )
| 345 | 1 |
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Features, Value
from .base import TaskTemplate
@dataclass(frozen=__snake_case )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = field(default="language-modeling" , metadata={"include_in_asdict_even_if_is_default": True} )
A__ : ClassVar[Features] = Features({"text": Value("string" )} )
A__ : ClassVar[Features] = Features({} )
A__ : str = "text"
@property
def A__ ( self: Any ) -> Dict[str, str]:
return {self.text_column: "text"}
| 345 |
import unittest
from queue import Empty
from threading import Thread
from transformers import AutoTokenizer, TextIteratorStreamer, TextStreamer, is_torch_available
from transformers.testing_utils import CaptureStdout, require_torch, torch_device
from ..test_modeling_common import ids_tensor
if is_torch_available():
import torch
from transformers import AutoModelForCausalLM
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Union[str, Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = -1
UpperCAmelCase_ : Dict = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Any = tokenizer.decode(greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : Optional[int] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Dict ) -> Optional[Any]:
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : List[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Dict = tokenizer.decode(greedy_ids[0] )
UpperCAmelCase_ : str = TextIteratorStreamer(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : str = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
UpperCAmelCase_ : int = """"""
for new_text in streamer:
streamer_text += new_text
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : Tuple = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : str = greedy_ids[:, input_ids.shape[1] :]
UpperCAmelCase_ : Dict = tokenizer.decode(new_greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ ,skip_prompt=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : List[str] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: str ) -> str:
# Tests that we can pass `decode_kwargs` to the streamer to control how the tokens are decoded. Must be tested
# with actual models -- the dummy models' tokenizers are not aligned with their models, and
# `skip_special_tokens=True` has no effect on them
UpperCAmelCase_ : Union[str, Any] = AutoTokenizer.from_pretrained("""distilgpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""distilgpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = -1
UpperCAmelCase_ : Union[str, Any] = torch.ones((1, 5) ,device=lowerCamelCase_ ).long() * model.config.bos_token_id
with CaptureStdout() as cs:
UpperCAmelCase_ : Union[str, Any] = TextStreamer(lowerCamelCase_ ,skip_special_tokens=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=1 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The prompt contains a special token, so the streamer should not print it. As such, the output text, when
# re-tokenized, must only contain one token
UpperCAmelCase_ : List[str] = cs.out[:-1] # Remove the final "\n"
UpperCAmelCase_ : Dict = tokenizer(lowerCamelCase_ ,return_tensors="""pt""" )
self.assertEqual(streamer_text_tokenized.input_ids.shape ,(1, 1) )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Any = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = -1
UpperCAmelCase_ : Optional[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = TextIteratorStreamer(lowerCamelCase_ ,timeout=0.0_0_1 )
UpperCAmelCase_ : Any = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : Dict = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
# The streamer will timeout after 0.001 seconds, so an exception will be raised
with self.assertRaises(lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = """"""
for new_text in streamer:
streamer_text += new_text
| 345 | 1 |
import pandas as pd
from matplotlib import pyplot as plt
from sklearn.linear_model import LinearRegression
# Splitting the dataset into the Training set and Test set
from sklearn.model_selection import train_test_split
# Fitting Polynomial Regression to the dataset
from sklearn.preprocessing import PolynomialFeatures
# Importing the dataset
UpperCamelCase_ = pd.read_csv(
'''https://s3.us-west-2.amazonaws.com/public.gamelab.fun/dataset/'''
'''position_salaries.csv'''
)
UpperCamelCase_ = dataset.iloc[:, 1:2].values
UpperCamelCase_ = dataset.iloc[:, 2].values
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = train_test_split(X, y, test_size=0.2, random_state=0)
UpperCamelCase_ = PolynomialFeatures(degree=4)
UpperCamelCase_ = poly_reg.fit_transform(X)
UpperCamelCase_ = LinearRegression()
pol_reg.fit(X_poly, y)
def lowerCamelCase_ ( ):
'''simple docstring'''
plt.scatter(_a , _a , color="""red""" )
plt.plot(_a , pol_reg.predict(poly_reg.fit_transform(_a ) ) , color="""blue""" )
plt.title("""Truth or Bluff (Linear Regression)""" )
plt.xlabel("""Position level""" )
plt.ylabel("""Salary""" )
plt.show()
if __name__ == "__main__":
viz_polymonial()
# Predicting a new result with Polymonial Regression
pol_reg.predict(poly_reg.fit_transform([[5.5]]))
# output should be 132148.43750003
| 345 |
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel
from diffusers import DDIMScheduler, LDMPipeline, UNetaDModel, VQModel
from diffusers.utils.testing_utils import enable_full_determinism, require_torch, slow, torch_device
enable_full_determinism()
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = UNetaDModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=3 ,out_channels=3 ,down_block_types=("""DownBlock2D""", """AttnDownBlock2D""") ,up_block_types=("""AttnUpBlock2D""", """UpBlock2D""") ,)
return model
@property
def A__ ( self: Tuple ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : List[str] = VQModel(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=3 ,)
return model
@property
def A__ ( self: Tuple ) -> Any:
torch.manual_seed(0 )
UpperCAmelCase_ : int = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
def A__ ( self: str ) -> Optional[Any]:
UpperCAmelCase_ : str = self.dummy_uncond_unet
UpperCAmelCase_ : List[Any] = DDIMScheduler()
UpperCAmelCase_ : List[Any] = self.dummy_vq_model
UpperCAmelCase_ : Optional[int] = LDMPipeline(unet=lowerCamelCase_ ,vqvae=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.manual_seed(0 )
UpperCAmelCase_ : int = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ).images
UpperCAmelCase_ : List[str] = torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ,return_dict=lowerCamelCase_ )[0]
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : str = np.array([0.8_5_1_2, 0.8_1_8, 0.6_4_1_1, 0.6_8_0_8, 0.4_4_6_5, 0.5_6_1_8, 0.4_6, 0.6_2_3_1, 0.5_1_7_2] )
UpperCAmelCase_ : Tuple = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < tolerance
@slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = LDMPipeline.from_pretrained("""CompVis/ldm-celebahq-256""" )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = ldm(generator=lowerCamelCase_ ,num_inference_steps=5 ,output_type="""numpy""" ).images
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
assert image.shape == (1, 256, 256, 3)
UpperCAmelCase_ : int = np.array([0.4_3_9_9, 0.4_4_9_7_5, 0.4_6_8_2_5, 0.4_7_4, 0.4_3_5_9, 0.4_5_8_1, 0.4_5_0_9_5, 0.4_3_4_1, 0.4_4_4_7] )
UpperCAmelCase_ : Union[str, Any] = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
| 345 | 1 |
import json
import os
import unittest
from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = CTRLTokenizer
A__ : Optional[Any] = False
A__ : str = False
def A__ ( self: Optional[int] ) -> List[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
UpperCAmelCase_ : Dict = ["""adapt""", """re@@""", """a@@""", """apt""", """c@@""", """t""", """<unk>"""]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : List[Any] = ["""#version: 0.2""", """a p""", """ap t</w>""", """r e""", """a d""", """ad apt</w>""", """"""]
UpperCAmelCase_ : Optional[Any] = {"""unk_token""": """<unk>"""}
UpperCAmelCase_ : Union[str, Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : Optional[Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase_ ) + """\n""" )
with open(self.merges_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase_ ) )
def A__ ( self: Optional[int] ,**lowerCamelCase_: Any ) -> str:
kwargs.update(self.special_tokens_map )
return CTRLTokenizer.from_pretrained(self.tmpdirname ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: int ) -> str:
UpperCAmelCase_ : List[str] = """adapt react readapt apt"""
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
return input_text, output_text
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = CTRLTokenizer(self.vocab_file ,self.merges_file ,**self.special_tokens_map )
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
UpperCAmelCase_ : Optional[int] = """adapt re@@ a@@ c@@ t re@@ adapt apt""".split()
UpperCAmelCase_ : Tuple = tokenizer.tokenize(lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = tokens + [tokenizer.unk_token]
UpperCAmelCase_ : List[str] = [0, 1, 2, 4, 5, 1, 0, 3, 6]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase_ ) ,lowerCamelCase_ )
| 345 |
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = [0] * len(_a )
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Dict = 0
for values in graph.values():
for i in values:
indegree[i] += 1
for i in range(len(_a ) ):
if indegree[i] == 0:
queue.append(_a )
while queue:
UpperCAmelCase_ : List[str] = queue.pop(0 )
cnt += 1
topo.append(_a )
for x in graph[vertex]:
indegree[x] -= 1
if indegree[x] == 0:
queue.append(_a )
if cnt != len(_a ):
print("""Cycle exists""" )
else:
print(_a )
# Adjacency List of Graph
UpperCamelCase_ = {0: [1, 2], 1: [3], 2: [3], 3: [4, 5], 4: [], 5: []}
topological_sort(graph)
| 345 | 1 |
from collections.abc import Generator
from math import sin
def lowerCamelCase_ ( _a : bytes ):
'''simple docstring'''
if len(_a ) != 32:
raise ValueError("""Input must be of length 32""" )
UpperCAmelCase_ : Dict = B""""""
for i in [3, 2, 1, 0]:
little_endian += string_aa[8 * i : 8 * i + 8]
return little_endian
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
if i < 0:
raise ValueError("""Input must be non-negative""" )
UpperCAmelCase_ : Optional[int] = format(_a , """08x""" )[-8:]
UpperCAmelCase_ : int = B""""""
for i in [3, 2, 1, 0]:
little_endian_hex += hex_rep[2 * i : 2 * i + 2].encode("""utf-8""" )
return little_endian_hex
def lowerCamelCase_ ( _a : bytes ):
'''simple docstring'''
UpperCAmelCase_ : str = B""""""
for char in message:
bit_string += format(_a , """08b""" ).encode("""utf-8""" )
UpperCAmelCase_ : Tuple = format(len(_a ) , """064b""" ).encode("""utf-8""" )
# Pad bit_string to a multiple of 512 chars
bit_string += b"1"
while len(_a ) % 512 != 448:
bit_string += b"0"
bit_string += to_little_endian(start_len[32:] ) + to_little_endian(start_len[:32] )
return bit_string
def lowerCamelCase_ ( _a : bytes ):
'''simple docstring'''
if len(_a ) % 512 != 0:
raise ValueError("""Input must have length that's a multiple of 512""" )
for pos in range(0 , len(_a ) , 512 ):
UpperCAmelCase_ : str = bit_string[pos : pos + 512]
UpperCAmelCase_ : Any = []
for i in range(0 , 512 , 32 ):
block_words.append(int(to_little_endian(block[i : i + 32] ) , 2 ) )
yield block_words
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
if i < 0:
raise ValueError("""Input must be non-negative""" )
UpperCAmelCase_ : List[str] = format(_a , """032b""" )
UpperCAmelCase_ : Optional[int] = """"""
for c in i_str:
new_str += "1" if c == "0" else "0"
return int(_a , 2 )
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
return (a + b) % 2**32
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
if i < 0:
raise ValueError("""Input must be non-negative""" )
if shift < 0:
raise ValueError("""Shift must be non-negative""" )
return ((i << shift) ^ (i >> (32 - shift))) % 2**32
def lowerCamelCase_ ( _a : bytes ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = preprocess(_a )
UpperCAmelCase_ : List[Any] = [int(2**32 * abs(sin(i + 1 ) ) ) for i in range(64 )]
# Starting states
UpperCAmelCase_ : Any = 0x6745_2301
UpperCAmelCase_ : int = 0xEFCD_AB89
UpperCAmelCase_ : int = 0x98BA_DCFE
UpperCAmelCase_ : Tuple = 0x1032_5476
UpperCAmelCase_ : List[Any] = [
7,
12,
17,
22,
7,
12,
17,
22,
7,
12,
17,
22,
7,
12,
17,
22,
5,
9,
14,
20,
5,
9,
14,
20,
5,
9,
14,
20,
5,
9,
14,
20,
4,
11,
16,
23,
4,
11,
16,
23,
4,
11,
16,
23,
4,
11,
16,
23,
6,
10,
15,
21,
6,
10,
15,
21,
6,
10,
15,
21,
6,
10,
15,
21,
]
# Process bit string in chunks, each with 16 32-char words
for block_words in get_block_words(_a ):
UpperCAmelCase_ : int = aa
UpperCAmelCase_ : Union[str, Any] = ba
UpperCAmelCase_ : Optional[int] = ca
UpperCAmelCase_ : str = da
# Hash current chunk
for i in range(64 ):
if i <= 15:
# f = (b & c) | (not_32(b) & d) # Alternate definition for f
UpperCAmelCase_ : Tuple = d ^ (b & (c ^ d))
UpperCAmelCase_ : List[Any] = i
elif i <= 31:
# f = (d & b) | (not_32(d) & c) # Alternate definition for f
UpperCAmelCase_ : Any = c ^ (d & (b ^ c))
UpperCAmelCase_ : Any = (5 * i + 1) % 16
elif i <= 47:
UpperCAmelCase_ : Any = b ^ c ^ d
UpperCAmelCase_ : Union[str, Any] = (3 * i + 5) % 16
else:
UpperCAmelCase_ : int = c ^ (b | not_aa(_a ))
UpperCAmelCase_ : Optional[int] = (7 * i) % 16
UpperCAmelCase_ : Tuple = (f + a + added_consts[i] + block_words[g]) % 2**32
UpperCAmelCase_ : Optional[int] = d
UpperCAmelCase_ : Any = c
UpperCAmelCase_ : List[Any] = b
UpperCAmelCase_ : Union[str, Any] = sum_aa(_a , left_rotate_aa(_a , shift_amounts[i] ) )
# Add hashed chunk to running total
UpperCAmelCase_ : Optional[int] = sum_aa(_a , _a )
UpperCAmelCase_ : Optional[int] = sum_aa(_a , _a )
UpperCAmelCase_ : Any = sum_aa(_a , _a )
UpperCAmelCase_ : Optional[int] = sum_aa(_a , _a )
UpperCAmelCase_ : Any = reformat_hex(_a ) + reformat_hex(_a ) + reformat_hex(_a ) + reformat_hex(_a )
return digest
if __name__ == "__main__":
import doctest
doctest.testmod()
| 345 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/swinv2-tiny-patch4-window8-256''': (
'''https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "swinv2"
A__ : int = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self: List[str] ,lowerCamelCase_: List[str]=224 ,lowerCamelCase_: List[str]=4 ,lowerCamelCase_: List[Any]=3 ,lowerCamelCase_: Optional[Any]=96 ,lowerCamelCase_: Any=[2, 2, 6, 2] ,lowerCamelCase_: Dict=[3, 6, 12, 24] ,lowerCamelCase_: str=7 ,lowerCamelCase_: Optional[Any]=4.0 ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: str=False ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: Union[str, Any]=1e-5 ,lowerCamelCase_: str=32 ,**lowerCamelCase_: List[str] ,) -> Tuple:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = image_size
UpperCAmelCase_ : Tuple = patch_size
UpperCAmelCase_ : Dict = num_channels
UpperCAmelCase_ : List[Any] = embed_dim
UpperCAmelCase_ : Dict = depths
UpperCAmelCase_ : Dict = len(lowerCamelCase_ )
UpperCAmelCase_ : str = num_heads
UpperCAmelCase_ : Tuple = window_size
UpperCAmelCase_ : int = mlp_ratio
UpperCAmelCase_ : str = qkv_bias
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Tuple = attention_probs_dropout_prob
UpperCAmelCase_ : int = drop_path_rate
UpperCAmelCase_ : Optional[Any] = hidden_act
UpperCAmelCase_ : List[str] = use_absolute_embeddings
UpperCAmelCase_ : Dict = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Union[str, Any] = encoder_stride
# we set the hidden_size attribute in order to make Swinv2 work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
UpperCAmelCase_ : List[str] = int(embed_dim * 2 ** (len(lowerCamelCase_ ) - 1) )
UpperCAmelCase_ : Any = (0, 0, 0, 0)
| 345 | 1 |
import math
import os
import unittest
from transformers import MegatronBertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
MegatronBertForCausalLM,
MegatronBertForMaskedLM,
MegatronBertForMultipleChoice,
MegatronBertForNextSentencePrediction,
MegatronBertForPreTraining,
MegatronBertForQuestionAnswering,
MegatronBertForSequenceClassification,
MegatronBertForTokenClassification,
MegatronBertModel,
)
class _snake_case :
'''simple docstring'''
def __init__( self: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any]=13 ,lowerCamelCase_: Optional[Any]=7 ,lowerCamelCase_: int=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Dict=99 ,lowerCamelCase_: Optional[Any]=64 ,lowerCamelCase_: Dict=32 ,lowerCamelCase_: str=5 ,lowerCamelCase_: Any=4 ,lowerCamelCase_: List[str]=37 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: Dict=0.1 ,lowerCamelCase_: List[Any]=0.1 ,lowerCamelCase_: Tuple=512 ,lowerCamelCase_: Optional[int]=16 ,lowerCamelCase_: Tuple=2 ,lowerCamelCase_: Tuple=0.0_2 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Any=4 ,lowerCamelCase_: Dict=None ,) -> Optional[Any]:
UpperCAmelCase_ : Any = parent
UpperCAmelCase_ : str = batch_size
UpperCAmelCase_ : List[Any] = seq_length
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : str = use_input_mask
UpperCAmelCase_ : int = use_token_type_ids
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : List[str] = vocab_size
UpperCAmelCase_ : Dict = hidden_size
UpperCAmelCase_ : str = embedding_size
UpperCAmelCase_ : str = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : Optional[int] = intermediate_size
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : int = attention_probs_dropout_prob
UpperCAmelCase_ : Dict = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = type_vocab_size
UpperCAmelCase_ : Dict = type_sequence_label_size
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Tuple = num_labels
UpperCAmelCase_ : str = num_choices
UpperCAmelCase_ : Optional[int] = scope
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : List[Any] = None
if self.use_input_mask:
UpperCAmelCase_ : Dict = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : int = None
if self.use_token_type_ids:
UpperCAmelCase_ : Dict = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : List[str] = None
UpperCAmelCase_ : Dict = None
UpperCAmelCase_ : Tuple = None
if self.use_labels:
UpperCAmelCase_ : Union[str, Any] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : int = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : Optional[Any] = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: Dict ) -> Tuple:
return MegatronBertConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,embedding_size=self.embedding_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: int ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ) -> int:
UpperCAmelCase_ : int = MegatronBertModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : str = model(lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : int = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ) -> Tuple:
UpperCAmelCase_ : Dict = MegatronBertForMaskedLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: str ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Any ) -> Any:
UpperCAmelCase_ : Union[str, Any] = MegatronBertForCausalLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : str = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: str ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = MegatronBertForNextSentencePrediction(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 2) )
def A__ ( self: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[int] = MegatronBertForPreTraining(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,next_sentence_label=lowerCamelCase_ ,)
self.parent.assertEqual(result.prediction_logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape ,(self.batch_size, 2) )
def A__ ( self: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ) -> int:
UpperCAmelCase_ : List[str] = MegatronBertForQuestionAnswering(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,start_positions=lowerCamelCase_ ,end_positions=lowerCamelCase_ ,)
self.parent.assertEqual(result.start_logits.shape ,(self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape ,(self.batch_size, self.seq_length) )
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ) -> Optional[int]:
UpperCAmelCase_ : int = self.num_labels
UpperCAmelCase_ : Union[str, Any] = MegatronBertForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_labels) )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: Dict ) -> List[str]:
UpperCAmelCase_ : Dict = self.num_labels
UpperCAmelCase_ : List[str] = MegatronBertForTokenClassification(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: str ) -> Optional[int]:
UpperCAmelCase_ : Dict = self.num_choices
UpperCAmelCase_ : Union[str, Any] = MegatronBertForMultipleChoice(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[str] = input_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : List[Any] = token_type_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : List[str] = input_mask.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Dict = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_choices) )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Optional[Any] = config_and_inputs
UpperCAmelCase_ : List[str] = {"""input_ids""": input_ids, """token_type_ids""": token_type_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = (
(
MegatronBertModel,
MegatronBertForMaskedLM,
MegatronBertForCausalLM,
MegatronBertForMultipleChoice,
MegatronBertForNextSentencePrediction,
MegatronBertForPreTraining,
MegatronBertForQuestionAnswering,
MegatronBertForSequenceClassification,
MegatronBertForTokenClassification,
)
if is_torch_available()
else ()
)
A__ : Optional[int] = (
{
"feature-extraction": MegatronBertModel,
"fill-mask": MegatronBertForMaskedLM,
"question-answering": MegatronBertForQuestionAnswering,
"text-classification": MegatronBertForSequenceClassification,
"text-generation": MegatronBertForCausalLM,
"token-classification": MegatronBertForTokenClassification,
"zero-shot": MegatronBertForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : List[str] = True
# test_resize_embeddings = False
A__ : int = False
def A__ ( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Union[str, Any]=False ) -> Any:
UpperCAmelCase_ : Any = super()._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ,return_labels=lowerCamelCase_ )
if return_labels:
if model_class in get_values(lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.zeros(
self.model_tester.batch_size ,dtype=torch.long ,device=lowerCamelCase_ )
return inputs_dict
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Optional[Any] = MegatronBertModelTester(self )
UpperCAmelCase_ : Tuple = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Optional[int] ) -> int:
self.config_tester.run_common_tests()
def A__ ( self: List[str] ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_model(*lowerCamelCase_ )
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : Any = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_masked_lm(*lowerCamelCase_ )
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_multiple_choice(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_next_sequence_prediction(*lowerCamelCase_ )
def A__ ( self: Any ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_pretraining(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_question_answering(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_sequence_classification(*lowerCamelCase_ )
def A__ ( self: List[Any] ) -> int:
UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_megatron_bert_for_token_classification(*lowerCamelCase_ )
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return torch.tensor(
_a , dtype=torch.long , device=_a , )
UpperCamelCase_ = 1E-4
@require_torch
@require_sentencepiece
@require_tokenizers
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
@unittest.skip("""Model is not available.""" )
def A__ ( self: Dict ) -> Any:
UpperCAmelCase_ : Any = """nvidia/megatron-bert-uncased-345m"""
if "MYDIR" in os.environ:
UpperCAmelCase_ : Union[str, Any] = os.path.join(os.environ["""MYDIR"""] ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = MegatronBertModel.from_pretrained(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.half()
UpperCAmelCase_ : Tuple = _long_tensor([[101, 7110, 1005, 1056, 2023, 11333, 17413, 1029, 102]] )
with torch.no_grad():
UpperCAmelCase_ : int = model(lowerCamelCase_ )[0]
UpperCAmelCase_ : str = torch.Size((1, 9, 1024) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = [-0.6_0_4_0, -0.2_5_1_7, -0.1_0_2_5, 0.3_4_2_0, -0.6_7_5_8, -0.0_0_1_7, -0.1_0_8_9, -0.1_9_9_0, 0.5_7_2_8]
for ii in range(3 ):
for jj in range(3 ):
UpperCAmelCase_ : Optional[Any] = output[0, ii, jj]
UpperCAmelCase_ : List[str] = expected[3 * ii + jj]
UpperCAmelCase_ : Optional[int] = """ii={} jj={} a={} b={}""".format(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
self.assertTrue(math.isclose(lowerCamelCase_ ,lowerCamelCase_ ,rel_tol=lowerCamelCase_ ,abs_tol=lowerCamelCase_ ) ,msg=lowerCamelCase_ )
| 345 |
import os
import sys
import tempfile
import unittest
import unittest.mock as mock
from pathlib import Path
from huggingface_hub import HfFolder, delete_repo
from huggingface_hub.file_download import http_get
from requests.exceptions import HTTPError
from transformers import (
AlbertTokenizer,
AutoTokenizer,
BertTokenizer,
BertTokenizerFast,
GPTaTokenizerFast,
is_tokenizers_available,
)
from transformers.testing_utils import TOKEN, USER, is_staging_test, require_tokenizers
from transformers.tokenization_utils import Trie
sys.path.append(str(Path(__file__).parent.parent / '''utils'''))
from test_module.custom_tokenization import CustomTokenizer # noqa E402
if is_tokenizers_available():
from test_module.custom_tokenization_fast import CustomTokenizerFast
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: int ) -> str:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : List[str] = mock.Mock()
UpperCAmelCase_ : List[Any] = 500
UpperCAmelCase_ : Union[str, Any] = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : Any = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Union[str, Any] = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# This check we did call the fake head request
mock_head.assert_called()
@require_tokenizers
def A__ ( self: str ) -> int:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : str = mock.Mock()
UpperCAmelCase_ : Optional[int] = 500
UpperCAmelCase_ : int = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : List[Any] = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Optional[int] = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# This check we did call the fake head request
mock_head.assert_called()
def A__ ( self: str ) -> Dict:
# This test is for deprecated behavior and can be removed in v5
try:
UpperCAmelCase_ : Any = tempfile.mktemp()
with open(lowerCamelCase_ ,"""wb""" ) as f:
http_get("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = AlbertTokenizer.from_pretrained(lowerCamelCase_ )
finally:
os.remove(lowerCamelCase_ )
# Supporting this legacy load introduced a weird bug where the tokenizer would load local files if they are in
# the current folder and have the right name.
if os.path.isfile("""tokenizer.json""" ):
# We skip the test if the user has a `tokenizer.json` in this folder to avoid deleting it.
return
try:
with open("""tokenizer.json""" ,"""wb""" ) as f:
http_get("""https://huggingface.co/hf-internal-testing/tiny-random-bert/blob/main/tokenizer.json""" ,lowerCamelCase_ )
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
# The tiny random BERT has a vocab size of 1024, tiny gpt2 as a vocab size of 1000
self.assertEqual(tokenizer.vocab_size ,1000 )
# Tokenizer should depend on the remote checkpoint, not the local tokenizer.json file.
finally:
os.remove("""tokenizer.json""" )
def A__ ( self: List[str] ) -> Tuple:
# This test is for deprecated behavior and can be removed in v5
UpperCAmelCase_ : str = AlbertTokenizer.from_pretrained("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" )
@is_staging_test
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
A__ : str = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "bla", "blou"]
@classmethod
def A__ ( cls: Dict ) -> Optional[int]:
UpperCAmelCase_ : List[str] = TOKEN
HfFolder.save_token(lowerCamelCase_ )
@classmethod
def A__ ( cls: Optional[Any] ) -> List[str]:
try:
delete_repo(token=cls._token ,repo_id="""test-tokenizer""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""valid_org/test-tokenizer-org""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""test-dynamic-tokenizer""" )
except HTTPError:
pass
def A__ ( self: Any ) -> Optional[int]:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Tuple = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : List[Any] = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""test-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""test-tokenizer""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(lowerCamelCase_ ,repo_id="""test-tokenizer""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
def A__ ( self: Optional[int] ) -> Any:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[Any] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Dict = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""valid_org/test-tokenizer-org""" ,use_auth_token=self._token )
UpperCAmelCase_ : Dict = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""valid_org/test-tokenizer-org""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(
lowerCamelCase_ ,repo_id="""valid_org/test-tokenizer-org""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
@require_tokenizers
def A__ ( self: Optional[int] ) -> Optional[Any]:
CustomTokenizer.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Any = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Optional[Any] = CustomTokenizer(lowerCamelCase_ )
# No fast custom tokenizer
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : Optional[Any] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the CustomTokenizer class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
# Fast and slow custom tokenizer
CustomTokenizerFast.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[str] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : str = BertTokenizerFast.from_pretrained(lowerCamelCase_ )
bert_tokenizer.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = CustomTokenizerFast.from_pretrained(lowerCamelCase_ )
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizerFast""" )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(
F'''{USER}/test-dynamic-tokenizer''' ,use_fast=lowerCamelCase_ ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Any = Trie()
trie.add("""Hello 友達""" )
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {""" """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
trie.add("""Hello""" )
trie.data
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {"""""": 1, """ """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
def A__ ( self: Tuple ) -> Optional[int]:
UpperCAmelCase_ : str = Trie()
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS] This is a extra_id_100"""] )
trie.add("""[CLS]""" )
trie.add("""extra_id_1""" )
trie.add("""extra_id_100""" )
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS]""", """ This is a """, """extra_id_100"""] )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ : Dict = Trie()
trie.add("""A""" )
self.assertEqual(trie.split("""ABC""" ) ,["""A""", """BC"""] )
self.assertEqual(trie.split("""BCA""" ) ,["""BC""", """A"""] )
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""TOKEN]""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> Union[str, Any]:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""A""" )
trie.add("""P""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> List[str]:
UpperCAmelCase_ : int = Trie()
trie.add("""AB""" )
trie.add("""B""" )
trie.add("""C""" )
self.assertEqual(trie.split("""ABC""" ) ,["""AB""", """C"""] )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = Trie()
trie.add("""ABC""" )
trie.add("""B""" )
trie.add("""CD""" )
self.assertEqual(trie.split("""ABCD""" ) ,["""ABC""", """D"""] )
def A__ ( self: List[Any] ) -> Any:
# Even if the offsets are wrong, we necessarily output correct string
# parts.
UpperCAmelCase_ : Tuple = Trie()
UpperCAmelCase_ : Optional[Any] = trie.cut_text("""ABC""" ,[0, 0, 2, 1, 2, 3] )
self.assertEqual(lowerCamelCase_ ,["""AB""", """C"""] )
| 345 | 1 |
import json
import os
import subprocess
import unittest
from ast import literal_eval
import pytest
from parameterized import parameterized_class
from . import is_sagemaker_available
if is_sagemaker_available():
from sagemaker import Session, TrainingJobAnalytics
from sagemaker.huggingface import HuggingFace
@pytest.mark.skipif(
literal_eval(os.getenv("TEST_SAGEMAKER" , "False" ) ) is not True , reason="Skipping test because should only be run when releasing minor transformers version" , )
@pytest.mark.usefixtures("sm_env" )
@parameterized_class(
[
{
"framework": "pytorch",
"script": "run_glue.py",
"model_name_or_path": "distilbert-base-cased",
"instance_type": "ml.g4dn.xlarge",
"results": {"train_runtime": 650, "eval_accuracy": 0.6, "eval_loss": 0.9},
},
{
"framework": "tensorflow",
"script": "run_tf.py",
"model_name_or_path": "distilbert-base-cased",
"instance_type": "ml.g4dn.xlarge",
"results": {"train_runtime": 600, "eval_accuracy": 0.3, "eval_loss": 0.9},
},
] )
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Any ) -> Union[str, Any]:
if self.framework == "pytorch":
subprocess.run(
F'''cp ./examples/pytorch/text-classification/run_glue.py {self.env.test_path}/run_glue.py'''.split() ,encoding="""utf-8""" ,check=lowerCamelCase_ ,)
assert hasattr(self ,"""env""" )
def A__ ( self: Optional[int] ,lowerCamelCase_: str=1 ) -> Dict:
# creates estimator
return HuggingFace(
entry_point=self.script ,source_dir=self.env.test_path ,role=self.env.role ,image_uri=self.env.image_uri ,base_job_name=F'''{self.env.base_job_name}-single''' ,instance_count=lowerCamelCase_ ,instance_type=self.instance_type ,debugger_hook_config=lowerCamelCase_ ,hyperparameters={**self.env.hyperparameters, """model_name_or_path""": self.model_name_or_path} ,metric_definitions=self.env.metric_definitions ,py_version="""py36""" ,)
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ) -> Optional[Any]:
TrainingJobAnalytics(lowerCamelCase_ ).export_csv(F'''{self.env.test_path}/{job_name}_metrics.csv''' )
def A__ ( self: Union[str, Any] ) -> Dict:
# create estimator
UpperCAmelCase_ : int = self.create_estimator()
# run training
estimator.fit()
# result dataframe
UpperCAmelCase_ : Tuple = TrainingJobAnalytics(estimator.latest_training_job.name ).dataframe()
# extract kpis
UpperCAmelCase_ : Optional[int] = list(result_metrics_df[result_metrics_df.metric_name == """eval_accuracy"""]["""value"""] )
UpperCAmelCase_ : Union[str, Any] = list(result_metrics_df[result_metrics_df.metric_name == """eval_loss"""]["""value"""] )
# get train time from SageMaker job, this includes starting, preprocessing, stopping
UpperCAmelCase_ : Optional[int] = (
Session().describe_training_job(estimator.latest_training_job.name ).get("""TrainingTimeInSeconds""" ,999999 )
)
# assert kpis
assert train_runtime <= self.results["train_runtime"]
assert all(t >= self.results["""eval_accuracy"""] for t in eval_accuracy )
assert all(t <= self.results["""eval_loss"""] for t in eval_loss )
# dump tests result into json file to share in PR
with open(F'''{estimator.latest_training_job.name}.json''' ,"""w""" ) as outfile:
json.dump({"""train_time""": train_runtime, """eval_accuracy""": eval_accuracy, """eval_loss""": eval_loss} ,lowerCamelCase_ )
| 345 |
from ..utils import DummyObject, requires_backends
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: int ,**lowerCamelCase_: List[str] ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: List[str] ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: int ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Any = ["flax"]
def __init__( self: int ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[str] ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Dict = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[Any] ) -> Any:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: List[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[str] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : int = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[int] ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Optional[Any] ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[Any] = ["flax"]
def __init__( self: Union[str, Any] ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> List[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Dict:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Any ,**lowerCamelCase_: int ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: Union[str, Any] ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Dict ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : str = ["flax"]
def __init__( self: Optional[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: int ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: int ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Optional[int] ) -> List[str]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Dict ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: List[str] ) -> int:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[Any] = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Optional[int] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Dict ) -> int:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Optional[Any]:
requires_backends(cls ,["""flax"""] )
| 345 | 1 |
from __future__ import annotations
from math import gcd
def lowerCamelCase_ ( _a : int , _a : int = 2 , _a : int = 1 , _a : int = 3 , ):
'''simple docstring'''
if num < 2:
raise ValueError("""The input value cannot be less than 2""" )
# Because of the relationship between ``f(f(x))`` and ``f(x)``, this
# algorithm struggles to find factors that are divisible by two.
# As a workaround, we specifically check for two and even inputs.
# See: https://math.stackexchange.com/a/2856214/165820
if num > 2 and num % 2 == 0:
return 2
# Pollard's Rho algorithm requires a function that returns pseudorandom
# values between 0 <= X < ``num``. It doesn't need to be random in the
# sense that the output value is cryptographically secure or difficult
# to calculate, it only needs to be random in the sense that all output
# values should be equally likely to appear.
# For this reason, Pollard suggested using ``f(x) = (x**2 - 1) % num``
# However, the success of Pollard's algorithm isn't guaranteed and is
# determined in part by the initial seed and the chosen random function.
# To make retries easier, we will instead use ``f(x) = (x**2 + C) % num``
# where ``C`` is a value that we can modify between each attempt.
def rand_fn(_a : int , _a : int , _a : int ) -> int:
return (pow(_a , 2 ) + step) % modulus
for _ in range(_a ):
# These track the position within the cycle detection logic.
UpperCAmelCase_ : Optional[Any] = seed
UpperCAmelCase_ : int = seed
while True:
# At each iteration, the tortoise moves one step and the hare moves two.
UpperCAmelCase_ : Tuple = rand_fn(_a , _a , _a )
UpperCAmelCase_ : str = rand_fn(_a , _a , _a )
UpperCAmelCase_ : Tuple = rand_fn(_a , _a , _a )
# At some point both the tortoise and the hare will enter a cycle whose
# length ``p`` is a divisor of ``num``. Once in that cycle, at some point
# the tortoise and hare will end up on the same value modulo ``p``.
# We can detect when this happens because the position difference between
# the tortoise and the hare will share a common divisor with ``num``.
UpperCAmelCase_ : Optional[Any] = gcd(hare - tortoise , _a )
if divisor == 1:
# No common divisor yet, just keep searching.
continue
else:
# We found a common divisor!
if divisor == num:
# Unfortunately, the divisor is ``num`` itself and is useless.
break
else:
# The divisor is a nontrivial factor of ``num``!
return divisor
# If we made it here, then this attempt failed.
# We need to pick a new starting seed for the tortoise and hare
# in addition to a new step value for the random function.
# To keep this example implementation deterministic, the
# new values will be generated based on currently available
# values instead of using something like ``random.randint``.
# We can use the hare's position as the new seed.
# This is actually what Richard Brent's the "optimized" variant does.
UpperCAmelCase_ : Union[str, Any] = hare
# The new step value for the random function can just be incremented.
# At first the results will be similar to what the old function would
# have produced, but the value will quickly diverge after a bit.
step += 1
# We haven't found a divisor within the requested number of attempts.
# We were unlucky or ``num`` itself is actually prime.
return None
if __name__ == "__main__":
import argparse
UpperCamelCase_ = argparse.ArgumentParser()
parser.add_argument(
'''num''',
type=int,
help='''The value to find a divisor of''',
)
parser.add_argument(
'''--attempts''',
type=int,
default=3,
help='''The number of attempts before giving up''',
)
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = pollard_rho(args.num, attempts=args.attempts)
if divisor is None:
print(F"{args.num} is probably prime")
else:
UpperCamelCase_ = args.num // divisor
print(F"{args.num} = {divisor} * {quotient}")
| 345 |
import random
from typing import Any
def lowerCamelCase_ ( _a : list ):
'''simple docstring'''
for _ in range(len(_a ) ):
UpperCAmelCase_ : Tuple = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ : List[Any] = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ , UpperCAmelCase_ : int = data[b], data[a]
return data
if __name__ == "__main__":
UpperCamelCase_ = [0, 1, 2, 3, 4, 5, 6, 7]
UpperCamelCase_ = ['''python''', '''says''', '''hello''', '''!''']
print('''Fisher-Yates Shuffle:''')
print('''List''', integers, strings)
print('''FY Shuffle''', fisher_yates_shuffle(integers), fisher_yates_shuffle(strings))
| 345 | 1 |
import argparse
import json
import os
import torch
from torch import nn
from transformers import NllbMoeConfig, NllbMoeModel
from transformers.modeling_utils import dtype_byte_size
from transformers.utils import WEIGHTS_INDEX_NAME, WEIGHTS_NAME
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = [
"""encoder.version""",
"""decoder.version""",
"""model.encoder.version""",
"""model.decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""encoder.embed_positions._float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Optional[int] ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = emb.weight.shape
UpperCAmelCase_ : List[Any] = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : Optional[int] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Union[str, Any] , _a : Dict=None ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = {}
for old_key in state_dict.keys():
UpperCAmelCase_ : Union[str, Any] = old_key
if "moe_layer.experts." in key:
if expert_idx is not None:
UpperCAmelCase_ : Any = key.replace("""moe_layer.experts.0""" , F'''ffn.experts.expert_{expert_idx}''' )
else:
UpperCAmelCase_ : List[str] = key.replace("""moe_layer.experts.""" , """ffn.experts.expert_""" )
if "gate" in key:
UpperCAmelCase_ : List[str] = key.replace(""".moe_layer.gate.wg""" , """.ffn.router.classifier""" )
if "fc2" and "experts" not in key:
UpperCAmelCase_ : List[Any] = key.replace(""".fc2.""" , """.ffn.fc2.""" )
if "fc1" and "experts" not in key:
UpperCAmelCase_ : Optional[int] = key.replace(""".fc1.""" , """.ffn.fc1.""" )
if ".encoder_attn." in key:
UpperCAmelCase_ : Union[str, Any] = key.replace(""".encoder_attn.""" , """.cross_attention.""" )
if "encoder_attn_layer_norm" in key:
UpperCAmelCase_ : List[str] = key.replace("""encoder_attn_layer_norm""" , """cross_attention_layer_norm""" )
if "final_layer_norm" in key:
UpperCAmelCase_ : Optional[int] = key.replace("""final_layer_norm""" , """ff_layer_norm""" )
UpperCAmelCase_ : Dict = state_dict[old_key]
return new_dict
def lowerCamelCase_ ( _a : Any , _a : str , _a : Dict , _a : Any , _a : str = WEIGHTS_NAME ):
'''simple docstring'''
UpperCAmelCase_ : List[Any] = []
UpperCAmelCase_ : Optional[Any] = 0
os.makedirs(_a , exist_ok=_a )
for expert in range(_a ):
UpperCAmelCase_ : List[Any] = switch_checkpoint_path + F'''-rank-{expert}.pt'''
if os.path.isfile(_a ):
UpperCAmelCase_ : Union[str, Any] = torch.load(_a )["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : Any = rename_fairseq_keys(_a , _a )
UpperCAmelCase_ : int = os.path.join(
_a , weights_name.replace(""".bin""" , F'''-{len(_a )+1:05d}-of-???.bin''' ) )
torch.save(_a , _a )
sharded_state_dicts.append(expert_state.keys() )
total_size += sum([value.numel() for key, value in expert_state.items()] ) * dtype_byte_size(
expert_state[list(_a )[0]].dtype )
# Add the last block
UpperCAmelCase_ : Any = os.path.join(_a , weights_name.replace(""".bin""" , F'''-{len(_a )+1:05d}-of-???.bin''' ) )
UpperCAmelCase_ : Any = torch.load(switch_checkpoint_path + """-shared.pt""" )["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = rename_fairseq_keys(_a , _a )
UpperCAmelCase_ : Optional[int] = shared_weights["""decoder.embed_tokens.weight"""]
sharded_state_dicts.append(shared_weights.keys() )
# If we only have the shared weights (dummy model/experts saved on the same file)
if len(_a ) == 1:
UpperCAmelCase_ : Optional[int] = os.path.join(_a , _a )
torch.save(_a , _a )
return {weights_name: sharded_state_dicts[0]}, None
else:
torch.save(_a , _a )
# Otherwise, let's build the index
UpperCAmelCase_ : Union[str, Any] = {}
for idx, shard in enumerate(_a ):
UpperCAmelCase_ : int = weights_name.replace(""".bin""" , F'''-{idx+1:05d}-of-{len(_a ):05d}.bin''' )
UpperCAmelCase_ : Optional[int] = os.path.join(_a , weights_name.replace(""".bin""" , F'''-{idx+1:05d}-of-???.bin''' ) )
os.rename(_a , os.path.join(_a , _a ) )
for key in shard:
UpperCAmelCase_ : Dict = shard_file
# Add the metadata
UpperCAmelCase_ : Optional[int] = {"""total_size""": total_size}
UpperCAmelCase_ : List[str] = {"""metadata""": metadata, """weight_map""": weight_map}
with open(os.path.join(_a , _a ) , """w""" , encoding="""utf-8""" ) as f:
UpperCAmelCase_ : Any = json.dumps(_a , indent=2 , sort_keys=_a ) + """\n"""
f.write(_a )
return metadata, index
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--nllb_moe_checkpoint_path''',
default='''/home/arthur_huggingface_co/fairseq/weights/checkpoints/model_moe_54b/checkpoint_2_300000''',
type=str,
required=False,
help='''Path to a directory containing a folder per layer. Follows the original Google format.''',
)
parser.add_argument('''--dtype''', default='''float32''', type=str, required=False, help='''dtype of the saved model''')
parser.add_argument(
'''--pytorch_dump_folder_path''',
default='''/home/arthur_huggingface_co/fairseq/weights/checkpoints/hf-converted-moe-54b''',
type=str,
required=False,
help='''Path to the output pytorch model.''',
)
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ ,UpperCamelCase_ = shard_on_the_fly(
args.nllb_moe_checkpoint_path,
args.pytorch_dump_folder_path,
128,
args.dtype,
)
UpperCamelCase_ = NllbMoeConfig.from_pretrained(
'''facebook/nllb-200-3.3B''', encoder_sparse_step=4, decoder_sparse_step=4, num_experts=128
)
config.save_pretrained(args.pytorch_dump_folder_path)
UpperCamelCase_ = NllbMoeModel.from_pretrained(args.pytorch_dump_folder_path)
print('''Done''')
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 |
import flax.linen as nn
import jax.numpy as jnp
from .attention_flax import FlaxTransformeraDModel
from .resnet_flax import FlaxDownsampleaD, FlaxResnetBlockaD, FlaxUpsampleaD
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Optional[int] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : List[Any] = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : List[Any] = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : int = resnets
UpperCAmelCase_ : Tuple = attentions
if self.add_downsample:
UpperCAmelCase_ : List[Any] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int=True ) -> int:
UpperCAmelCase_ : List[Any] = ()
for resnet, attn in zip(self.resnets ,self.attentions ):
UpperCAmelCase_ : str = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[Any] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> int:
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : Dict = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnets
if self.add_downsample:
UpperCAmelCase_ : List[str] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any]=True ) -> Any:
UpperCAmelCase_ : Union[str, Any] = ()
for resnet in self.resnets:
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[str] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: str ) -> Any:
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : int = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Optional[Any] = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : int = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = resnets
UpperCAmelCase_ : Dict = attentions
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: Any ,lowerCamelCase_: str ,lowerCamelCase_: List[str]=True ) -> List[str]:
for resnet, attn in zip(self.resnets ,self.attentions ):
# pop res hidden states
UpperCAmelCase_ : List[str] = res_hidden_states_tuple[-1]
UpperCAmelCase_ : Union[str, Any] = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : Optional[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Dict = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> Dict:
UpperCAmelCase_ : Any = []
for i in range(self.num_layers ):
UpperCAmelCase_ : str = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : Optional[int] = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : str = resnets
if self.add_upsample:
UpperCAmelCase_ : Union[str, Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Dict ,lowerCamelCase_: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any=True ) -> List[str]:
for resnet in self.resnets:
# pop res hidden states
UpperCAmelCase_ : Dict = res_hidden_states_tuple[-1]
UpperCAmelCase_ : str = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : List[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : List[str] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
# there is always at least one resnet
UpperCAmelCase_ : List[Any] = [
FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
]
UpperCAmelCase_ : Any = []
for _ in range(self.num_layers ):
UpperCAmelCase_ : Optional[Any] = FlaxTransformeraDModel(
in_channels=self.in_channels ,n_heads=self.num_attention_heads ,d_head=self.in_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Dict = resnets
UpperCAmelCase_ : Any = attentions
def __call__( self: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any]=True ) -> List[Any]:
UpperCAmelCase_ : List[Any] = self.resnets[0](lowerCamelCase_ ,lowerCamelCase_ )
for attn, resnet in zip(self.attentions ,self.resnets[1:] ):
UpperCAmelCase_ : Optional[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
return hidden_states
| 345 | 1 |
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = [1]
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : int = 0, 0, 0
UpperCAmelCase_ : str = ugly_nums[ia] * 2
UpperCAmelCase_ : List[Any] = ugly_nums[ia] * 3
UpperCAmelCase_ : List[str] = ugly_nums[ia] * 5
for _ in range(1 , _a ):
UpperCAmelCase_ : List[str] = min(_a , _a , _a )
ugly_nums.append(_a )
if next_num == next_a:
ia += 1
UpperCAmelCase_ : Optional[Any] = ugly_nums[ia] * 2
if next_num == next_a:
ia += 1
UpperCAmelCase_ : List[str] = ugly_nums[ia] * 3
if next_num == next_a:
ia += 1
UpperCAmelCase_ : Optional[Any] = ugly_nums[ia] * 5
return ugly_nums[-1]
if __name__ == "__main__":
from doctest import testmod
testmod(verbose=True)
print(F"{ugly_numbers(200) = }")
| 345 |
import pickle
import numpy as np
from matplotlib import pyplot as plt
class _snake_case :
'''simple docstring'''
def __init__( self: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple=0.2 ,lowerCamelCase_: Union[str, Any]=0.2 ) -> List[str]:
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : str = bp_numa
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : Optional[int] = conva_get[:2]
UpperCAmelCase_ : List[Any] = conva_get[2]
UpperCAmelCase_ : str = size_pa
UpperCAmelCase_ : Optional[int] = rate_w
UpperCAmelCase_ : Dict = rate_t
UpperCAmelCase_ : List[Any] = [
np.mat(-1 * np.random.rand(self.conva[0] ,self.conva[0] ) + 0.5 )
for i in range(self.conva[1] )
]
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : Dict = -2 * np.random.rand(self.conva[1] ) + 1
UpperCAmelCase_ : str = -2 * np.random.rand(self.num_bpa ) + 1
UpperCAmelCase_ : Union[str, Any] = -2 * np.random.rand(self.num_bpa ) + 1
def A__ ( self: str ,lowerCamelCase_: Optional[Any] ) -> Tuple:
# save model dict with pickle
UpperCAmelCase_ : Dict = {
"""num_bp1""": self.num_bpa,
"""num_bp2""": self.num_bpa,
"""num_bp3""": self.num_bpa,
"""conv1""": self.conva,
"""step_conv1""": self.step_conva,
"""size_pooling1""": self.size_poolinga,
"""rate_weight""": self.rate_weight,
"""rate_thre""": self.rate_thre,
"""w_conv1""": self.w_conva,
"""wkj""": self.wkj,
"""vji""": self.vji,
"""thre_conv1""": self.thre_conva,
"""thre_bp2""": self.thre_bpa,
"""thre_bp3""": self.thre_bpa,
}
with open(lowerCamelCase_ ,"""wb""" ) as f:
pickle.dump(lowerCamelCase_ ,lowerCamelCase_ )
print(F'''Model saved： {save_path}''' )
@classmethod
def A__ ( cls: List[str] ,lowerCamelCase_: str ) -> List[str]:
# read saved model
with open(lowerCamelCase_ ,"""rb""" ) as f:
UpperCAmelCase_ : Any = pickle.load(lowerCamelCase_ ) # noqa: S301
UpperCAmelCase_ : Union[str, Any] = model_dic.get("""conv1""" )
conv_get.append(model_dic.get("""step_conv1""" ) )
UpperCAmelCase_ : List[str] = model_dic.get("""size_pooling1""" )
UpperCAmelCase_ : Tuple = model_dic.get("""num_bp1""" )
UpperCAmelCase_ : Optional[Any] = model_dic.get("""num_bp2""" )
UpperCAmelCase_ : List[str] = model_dic.get("""num_bp3""" )
UpperCAmelCase_ : List[Any] = model_dic.get("""rate_weight""" )
UpperCAmelCase_ : Dict = model_dic.get("""rate_thre""" )
# create model instance
UpperCAmelCase_ : List[Any] = CNN(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# modify model parameter
UpperCAmelCase_ : Any = model_dic.get("""w_conv1""" )
UpperCAmelCase_ : int = model_dic.get("""wkj""" )
UpperCAmelCase_ : int = model_dic.get("""vji""" )
UpperCAmelCase_ : Optional[int] = model_dic.get("""thre_conv1""" )
UpperCAmelCase_ : List[str] = model_dic.get("""thre_bp2""" )
UpperCAmelCase_ : Dict = model_dic.get("""thre_bp3""" )
return conv_ins
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> Tuple:
return 1 / (1 + np.exp(-1 * x ))
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
return round(lowerCamelCase_ ,3 )
def A__ ( self: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Any:
# convolution process
UpperCAmelCase_ : Optional[Any] = convs[0]
UpperCAmelCase_ : int = convs[1]
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )[0]
# get the data slice of original image data, data_focus
UpperCAmelCase_ : Dict = []
for i_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
for j_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = data[
i_focus : i_focus + size_conv, j_focus : j_focus + size_conv
]
data_focus.append(lowerCamelCase_ )
# calculate the feature map of every single kernel, and saved as list of matrix
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = int((size_data - size_conv) / conv_step + 1 )
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[int] = []
for i_focus in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = (
np.sum(np.multiply(data_focus[i_focus] ,w_convs[i_map] ) )
- thre_convs[i_map]
)
featuremap.append(self.sig(lowerCamelCase_ ) )
UpperCAmelCase_ : Union[str, Any] = np.asmatrix(lowerCamelCase_ ).reshape(
lowerCamelCase_ ,lowerCamelCase_ )
data_featuremap.append(lowerCamelCase_ )
# expanding the data slice to One dimenssion
UpperCAmelCase_ : Optional[Any] = []
for each_focus in data_focus:
focusa_list.extend(self.Expand_Mat(lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[int] = np.asarray(lowerCamelCase_ )
return focus_list, data_featuremap
def A__ ( self: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any]="average_pool" ) -> List[Any]:
# pooling process
UpperCAmelCase_ : Optional[Any] = len(featuremaps[0] )
UpperCAmelCase_ : Any = int(size_map / size_pooling )
UpperCAmelCase_ : Optional[int] = []
for i_map in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Any = featuremaps[i_map]
UpperCAmelCase_ : Tuple = []
for i_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : str = feature_map[
i_focus : i_focus + size_pooling,
j_focus : j_focus + size_pooling,
]
if pooling_type == "average_pool":
# average pooling
map_pooled.append(np.average(lowerCamelCase_ ) )
elif pooling_type == "max_pooling":
# max pooling
map_pooled.append(np.max(lowerCamelCase_ ) )
UpperCAmelCase_ : int = np.asmatrix(lowerCamelCase_ ).reshape(lowerCamelCase_ ,lowerCamelCase_ )
featuremap_pooled.append(lowerCamelCase_ )
return featuremap_pooled
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> Optional[int]:
# expanding three dimension data to one dimension list
UpperCAmelCase_ : List[Any] = []
for i in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Tuple = np.shape(data[i] )
UpperCAmelCase_ : Optional[int] = data[i].reshape(1 ,shapes[0] * shapes[1] )
UpperCAmelCase_ : Optional[int] = data_listed.getA().tolist()[0]
data_expanded.extend(lowerCamelCase_ )
UpperCAmelCase_ : int = np.asarray(lowerCamelCase_ )
return data_expanded
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
# expanding matrix to one dimension list
UpperCAmelCase_ : List[Any] = np.asarray(lowerCamelCase_ )
UpperCAmelCase_ : str = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = data_mat.reshape(1 ,shapes[0] * shapes[1] )
return data_expanded
def A__ ( self: str ,lowerCamelCase_: Dict ,lowerCamelCase_: int ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = 0
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = np.ones((size_map, size_map) )
for i in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Any = pd_pool[
i_pool
]
UpperCAmelCase_ : List[str] = i_pool + 1
UpperCAmelCase_ : Optional[Any] = np.multiply(
lowerCamelCase_ ,np.multiply(out_map[i_map] ,(1 - out_map[i_map]) ) )
pd_all.append(lowerCamelCase_ )
return pd_all
def A__ ( self: str ,lowerCamelCase_: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any=bool ) -> Optional[int]:
# model traning
print("""----------------------Start Training-------------------------""" )
print((""" - - Shape: Train_Data """, np.shape(lowerCamelCase_ )) )
print((""" - - Shape: Teach_Data """, np.shape(lowerCamelCase_ )) )
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : Any = 10000
while rp < n_repeat and mse >= error_accuracy:
UpperCAmelCase_ : List[str] = 0
print(F'''-------------Learning Time {rp}--------------''' )
for p in range(len(lowerCamelCase_ ) ):
# print('------------Learning Image: %d--------------'%p)
UpperCAmelCase_ : str = np.asmatrix(datas_train[p] )
UpperCAmelCase_ : Optional[Any] = np.asarray(datas_teach[p] )
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : List[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = data_bp_input
UpperCAmelCase_ : Optional[Any] = np.dot(lowerCamelCase_ ,self.vji.T ) - self.thre_bpa
UpperCAmelCase_ : int = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = np.dot(lowerCamelCase_ ,self.wkj.T ) - self.thre_bpa
UpperCAmelCase_ : Optional[Any] = self.sig(lowerCamelCase_ )
# --------------Model Leaning ------------------------
# calculate error and gradient---------------
UpperCAmelCase_ : List[str] = np.multiply(
(data_teach - bp_outa) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : List[Any] = np.multiply(
np.dot(lowerCamelCase_ ,self.wkj ) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : Any = np.dot(lowerCamelCase_ ,self.vji )
UpperCAmelCase_ : Tuple = pd_i_all / (self.size_poolinga * self.size_poolinga)
UpperCAmelCase_ : List[str] = pd_conva_pooled.T.getA().tolist()
UpperCAmelCase_ : str = self._calculate_gradient_from_pool(
lowerCamelCase_ ,lowerCamelCase_ ,shape_featuremapa[0] ,shape_featuremapa[1] ,self.size_poolinga ,)
# weight and threshold learning process---------
# convolution layer
for k_conv in range(self.conva[1] ):
UpperCAmelCase_ : List[str] = self._expand_mat(pd_conva_all[k_conv] )
UpperCAmelCase_ : Optional[Any] = self.rate_weight * np.dot(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : int = self.w_conva[k_conv] + delta_w.reshape(
(self.conva[0], self.conva[0]) )
UpperCAmelCase_ : str = (
self.thre_conva[k_conv]
- np.sum(pd_conva_all[k_conv] ) * self.rate_thre
)
# all connected layer
UpperCAmelCase_ : int = self.wkj + pd_k_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : Tuple = self.vji + pd_j_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : int = self.thre_bpa - pd_k_all * self.rate_thre
UpperCAmelCase_ : str = self.thre_bpa - pd_j_all * self.rate_thre
# calculate the sum error of all single image
UpperCAmelCase_ : int = np.sum(abs(data_teach - bp_outa ) )
error_count += errors
# print(' ----Teach ',data_teach)
# print(' ----BP_output ',bp_out3)
UpperCAmelCase_ : int = rp + 1
UpperCAmelCase_ : Any = error_count / patterns
all_mse.append(lowerCamelCase_ )
def draw_error():
UpperCAmelCase_ : Any = [error_accuracy for i in range(int(n_repeat * 1.2 ) )]
plt.plot(lowerCamelCase_ ,"""+-""" )
plt.plot(lowerCamelCase_ ,"""r--""" )
plt.xlabel("""Learning Times""" )
plt.ylabel("""All_mse""" )
plt.grid(lowerCamelCase_ ,alpha=0.5 )
plt.show()
print("""------------------Training Complished---------------------""" )
print((""" - - Training epoch: """, rp, F''' - - Mse: {mse:.6f}''') )
if draw_e:
draw_error()
return mse
def A__ ( self: Optional[int] ,lowerCamelCase_: Any ) -> Tuple:
# model predict
UpperCAmelCase_ : Union[str, Any] = []
print("""-------------------Start Testing-------------------------""" )
print((""" - - Shape: Test_Data """, np.shape(lowerCamelCase_ )) )
for p in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = np.asmatrix(datas_test[p] )
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Optional[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : str = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : str = data_bp_input
UpperCAmelCase_ : Union[str, Any] = bp_outa * self.vji.T - self.thre_bpa
UpperCAmelCase_ : Optional[int] = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = bp_outa * self.wkj.T - self.thre_bpa
UpperCAmelCase_ : List[Any] = self.sig(lowerCamelCase_ )
produce_out.extend(bp_outa.getA().tolist() )
UpperCAmelCase_ : int = [list(map(self.do_round ,lowerCamelCase_ ) ) for each in produce_out]
return np.asarray(lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Tuple:
# return the data of image after convoluting process so we can check it out
UpperCAmelCase_ : Optional[int] = np.asmatrix(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Dict = self.pooling(lowerCamelCase_ ,self.size_poolinga )
return data_conveda, data_pooleda
if __name__ == "__main__":
pass
| 345 | 1 |
import math
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
if 1 < number < 4:
# 2 and 3 are primes
return True
elif number < 2 or number % 2 == 0 or number % 3 == 0:
# Negatives, 0, 1, all even numbers, all multiples of 3 are not primes
return False
# All primes number are in format of 6k +/- 1
for i in range(5 , int(math.sqrt(_a ) + 1 ) , 6 ):
if number % i == 0 or number % (i + 2) == 0:
return False
return True
def lowerCamelCase_ ( _a : int = 1_0001 ):
'''simple docstring'''
try:
UpperCAmelCase_ : Dict = int(_a )
except (TypeError, ValueError):
raise TypeError("""Parameter nth must be int or castable to int.""" ) from None
if nth <= 0:
raise ValueError("""Parameter nth must be greater than or equal to one.""" )
UpperCAmelCase_ : list[int] = []
UpperCAmelCase_ : Union[str, Any] = 2
while len(_a ) < nth:
if is_prime(_a ):
primes.append(_a )
num += 1
else:
num += 1
return primes[len(_a ) - 1]
if __name__ == "__main__":
print(F"{solution() = }")
| 345 |
import json
import os
import unittest
from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = CTRLTokenizer
A__ : Optional[Any] = False
A__ : str = False
def A__ ( self: Optional[int] ) -> List[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
UpperCAmelCase_ : Dict = ["""adapt""", """re@@""", """a@@""", """apt""", """c@@""", """t""", """<unk>"""]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : List[Any] = ["""#version: 0.2""", """a p""", """ap t</w>""", """r e""", """a d""", """ad apt</w>""", """"""]
UpperCAmelCase_ : Optional[Any] = {"""unk_token""": """<unk>"""}
UpperCAmelCase_ : Union[str, Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : Optional[Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase_ ) + """\n""" )
with open(self.merges_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase_ ) )
def A__ ( self: Optional[int] ,**lowerCamelCase_: Any ) -> str:
kwargs.update(self.special_tokens_map )
return CTRLTokenizer.from_pretrained(self.tmpdirname ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: int ) -> str:
UpperCAmelCase_ : List[str] = """adapt react readapt apt"""
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
return input_text, output_text
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = CTRLTokenizer(self.vocab_file ,self.merges_file ,**self.special_tokens_map )
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
UpperCAmelCase_ : Optional[int] = """adapt re@@ a@@ c@@ t re@@ adapt apt""".split()
UpperCAmelCase_ : Tuple = tokenizer.tokenize(lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = tokens + [tokenizer.unk_token]
UpperCAmelCase_ : List[str] = [0, 1, 2, 4, 5, 1, 0, 3, 6]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase_ ) ,lowerCamelCase_ )
| 345 | 1 |
import argparse
import torch
from transformers import LxmertConfig, LxmertForPreTraining, load_tf_weights_in_lxmert
from transformers.utils import logging
logging.set_verbosity_info()
def lowerCamelCase_ ( _a : Union[str, Any] , _a : Any , _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Dict = LxmertConfig.from_json_file(_a )
print(F'''Building PyTorch model from configuration: {config}''' )
UpperCAmelCase_ : List[Any] = LxmertForPreTraining(_a )
# Load weights from tf checkpoint
load_tf_weights_in_lxmert(_a , _a , _a )
# Save pytorch-model
print(F'''Save PyTorch model to {pytorch_dump_path}''' )
torch.save(model.state_dict() , _a )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--tf_checkpoint_path''', default=None, type=str, required=True, help='''Path to the TensorFlow checkpoint path.'''
)
parser.add_argument(
'''--config_file''',
default=None,
type=str,
required=True,
help='''The config json file corresponding to the pre-trained model. \nThis specifies the model architecture.''',
)
parser.add_argument(
'''--pytorch_dump_path''', default=None, type=str, required=True, help='''Path to the output PyTorch model.'''
)
UpperCamelCase_ = parser.parse_args()
convert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.config_file, args.pytorch_dump_path)
| 345 |
from __future__ import annotations
from typing import Dict
from ...configuration_utils import PretrainedConfig
UpperCamelCase_ = {
'''susnato/ernie-m-base_pytorch''': '''https://huggingface.co/susnato/ernie-m-base_pytorch/blob/main/config.json''',
'''susnato/ernie-m-large_pytorch''': '''https://huggingface.co/susnato/ernie-m-large_pytorch/blob/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = "ernie_m"
A__ : Dict[str, str] = {"dropout": "classifier_dropout", "num_classes": "num_labels"}
def __init__( self: str ,lowerCamelCase_: int = 250002 ,lowerCamelCase_: int = 768 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 3072 ,lowerCamelCase_: str = "gelu" ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: int = 514 ,lowerCamelCase_: float = 0.0_2 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: float = 1e-05 ,lowerCamelCase_: Any=None ,lowerCamelCase_: List[Any]=False ,lowerCamelCase_: Tuple=0.0 ,**lowerCamelCase_: Optional[int] ,) -> Optional[Any]:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = vocab_size
UpperCAmelCase_ : Any = hidden_size
UpperCAmelCase_ : Optional[Any] = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : List[Any] = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = initializer_range
UpperCAmelCase_ : Union[str, Any] = layer_norm_eps
UpperCAmelCase_ : List[Any] = classifier_dropout
UpperCAmelCase_ : str = is_decoder
UpperCAmelCase_ : List[str] = act_dropout
| 345 | 1 |
from itertools import product
def lowerCamelCase_ ( _a : int , _a : int ):
'''simple docstring'''
UpperCAmelCase_ : List[Any] = sides_number
UpperCAmelCase_ : int = max_face_number * dice_number
UpperCAmelCase_ : str = [0] * (max_total + 1)
UpperCAmelCase_ : Tuple = 1
UpperCAmelCase_ : Dict = range(_a , max_face_number + 1 )
for dice_numbers in product(_a , repeat=_a ):
UpperCAmelCase_ : str = sum(_a )
totals_frequencies[total] += 1
return totals_frequencies
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : Any = total_frequency_distribution(
sides_number=4 , dice_number=9 )
UpperCAmelCase_ : List[Any] = total_frequency_distribution(
sides_number=6 , dice_number=6 )
UpperCAmelCase_ : Union[str, Any] = 0
UpperCAmelCase_ : str = 9
UpperCAmelCase_ : List[str] = 4 * 9
UpperCAmelCase_ : List[str] = 6
for peter_total in range(_a , max_peter_total + 1 ):
peter_wins_count += peter_totals_frequencies[peter_total] * sum(
colin_totals_frequencies[min_colin_total:peter_total] )
UpperCAmelCase_ : Union[str, Any] = (4**9) * (6**6)
UpperCAmelCase_ : Optional[Any] = peter_wins_count / total_games_number
UpperCAmelCase_ : Optional[int] = round(_a , ndigits=7 )
return rounded_peter_win_probability
if __name__ == "__main__":
print(F"{solution() = }")
| 345 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
UpperCamelCase_ = logging.getLogger(__name__)
torch.set_grad_enabled(False)
UpperCamelCase_ = '''cuda''' if torch.cuda.is_available() else '''cpu'''
def lowerCamelCase_ ( _a : str , _a : Any=100 , _a : int=" " ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = text.split(_a )
return [character.join(text[i : i + n] ).strip() for i in range(0 , len(_a ) , _a )]
def lowerCamelCase_ ( _a : dict ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = [], []
for title, text in zip(documents["""title"""] , documents["""text"""] ):
if text is not None:
for passage in split_text(_a ):
titles.append(title if title is not None else """""" )
texts.append(_a )
return {"title": titles, "text": texts}
def lowerCamelCase_ ( _a : dict , _a : DPRContextEncoder , _a : DPRContextEncoderTokenizerFast ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = ctx_tokenizer(
documents["""title"""] , documents["""text"""] , truncation=_a , padding="""longest""" , return_tensors="""pt""" )["""input_ids"""]
UpperCAmelCase_ : Tuple = ctx_encoder(input_ids.to(device=_a ) , return_dict=_a ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def lowerCamelCase_ ( _a : "RagExampleArguments" , _a : "ProcessingArguments" , _a : "IndexHnswArguments" , ):
'''simple docstring'''
logger.info("""Step 1 - Create the dataset""" )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
UpperCAmelCase_ : Optional[int] = load_dataset(
"""csv""" , data_files=[rag_example_args.csv_path] , split="""train""" , delimiter="""\t""" , column_names=["""title""", """text"""] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
UpperCAmelCase_ : Tuple = dataset.map(_a , batched=_a , num_proc=processing_args.num_proc )
# And compute the embeddings
UpperCAmelCase_ : List[str] = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=_a )
UpperCAmelCase_ : Dict = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
UpperCAmelCase_ : Any = Features(
{"""text""": Value("""string""" ), """title""": Value("""string""" ), """embeddings""": Sequence(Value("""float32""" ) )} ) # optional, save as float32 instead of float64 to save space
UpperCAmelCase_ : List[str] = dataset.map(
partial(_a , ctx_encoder=_a , ctx_tokenizer=_a ) , batched=_a , batch_size=processing_args.batch_size , features=_a , )
# And finally save your dataset
UpperCAmelCase_ : Union[str, Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset""" )
dataset.save_to_disk(_a )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info("""Step 2 - Index the dataset""" )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
UpperCAmelCase_ : Union[str, Any] = faiss.IndexHNSWFlat(index_hnsw_args.d , index_hnsw_args.m , faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index("""embeddings""" , custom_index=_a )
# And save the index
UpperCAmelCase_ : Optional[Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset_hnsw_index.faiss""" )
dataset.get_index("""embeddings""" ).save(_a )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" / "my_knowledge_dataset.csv" ) , metadata={"help": "Path to a tab-separated csv file with columns 'title' and 'text'"} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."} , )
A__ : str = field(
default="facebook/rag-sequence-nq" , metadata={"help": "The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"} , )
A__ : str = field(
default="facebook/dpr-ctx_encoder-multiset-base" , metadata={
"help": (
"The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"
" 'facebook/dpr-ctx_encoder-multiset-base'"
)
} , )
A__ : Optional[str] = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" ) , metadata={"help": "Path to a directory where the dataset passages and the index will be saved"} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": "The number of processes to use to split the documents into passages. Default is single process."
} , )
A__ : int = field(
default=16 , metadata={
"help": "The batch size to use when computing the passages embeddings using the DPR context encoder."
} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : int = field(
default=768 , metadata={"help": "The dimension of the embeddings to pass to the HNSW Faiss index."} , )
A__ : int = field(
default=128 , metadata={
"help": (
"The number of bi-directional links created for every new element during the HNSW index construction."
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
UpperCamelCase_ = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
UpperCamelCase_ = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 345 | 1 |
import io
import json
import fsspec
import pytest
from datasets import Dataset, DatasetDict, Features, NamedSplit, Value
from datasets.io.json import JsonDatasetReader, JsonDatasetWriter
from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases
def lowerCamelCase_ ( _a : Union[str, Any] , _a : int ):
'''simple docstring'''
assert isinstance(_a , _a )
assert dataset.num_rows == 4
assert dataset.num_columns == 3
assert dataset.column_names == ["col_1", "col_2", "col_3"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize("""keep_in_memory""" , [False, True] )
def lowerCamelCase_ ( _a : Optional[int] , _a : Union[str, Any] , _a : Optional[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = tmp_path / """cache"""
UpperCAmelCase_ : List[Any] = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
UpperCAmelCase_ : List[Any] = JsonDatasetReader(_a , cache_dir=_a , keep_in_memory=_a ).read()
_check_json_dataset(_a , _a )
@pytest.mark.parametrize(
"""features""" , [
None,
{"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""},
{"""col_1""": """string""", """col_2""": """string""", """col_3""": """string"""},
{"""col_1""": """int32""", """col_2""": """int32""", """col_3""": """int32"""},
{"""col_1""": """float32""", """col_2""": """float32""", """col_3""": """float32"""},
] , )
def lowerCamelCase_ ( _a : str , _a : int , _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = tmp_path / """cache"""
UpperCAmelCase_ : Tuple = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
UpperCAmelCase_ : int = features.copy() if features else default_expected_features
UpperCAmelCase_ : Dict = (
Features({feature: Value(_a ) for feature, dtype in features.items()} ) if features is not None else None
)
UpperCAmelCase_ : int = JsonDatasetReader(_a , features=_a , cache_dir=_a ).read()
_check_json_dataset(_a , _a )
@pytest.mark.parametrize(
"""features""" , [
None,
{"""col_3""": """float64""", """col_1""": """string""", """col_2""": """int64"""},
] , )
def lowerCamelCase_ ( _a : Any , _a : Dict , _a : int ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = tmp_path / """cache"""
UpperCAmelCase_ : List[Any] = {"""col_3""": """float64""", """col_1""": """string""", """col_2""": """int64"""}
UpperCAmelCase_ : Tuple = features.copy() if features else default_expected_features
UpperCAmelCase_ : Optional[int] = (
Features({feature: Value(_a ) for feature, dtype in features.items()} ) if features is not None else None
)
UpperCAmelCase_ : List[str] = JsonDatasetReader(_a , features=_a , cache_dir=_a ).read()
assert isinstance(_a , _a )
assert dataset.num_rows == 2
assert dataset.num_columns == 3
assert dataset.column_names == ["col_3", "col_1", "col_2"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
def lowerCamelCase_ ( _a : Optional[Any] , _a : Union[str, Any] ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = {"""col_2""": """int64""", """col_3""": """float64""", """col_1""": """string"""}
UpperCAmelCase_ : Optional[int] = features.copy()
UpperCAmelCase_ : Optional[Any] = (
Features({feature: Value(_a ) for feature, dtype in features.items()} ) if features is not None else None
)
UpperCAmelCase_ : List[str] = tmp_path / """cache"""
UpperCAmelCase_ : List[str] = JsonDatasetReader(_a , features=_a , cache_dir=_a ).read()
assert isinstance(_a , _a )
assert dataset.num_rows == 2
assert dataset.num_columns == 3
assert dataset.column_names == ["col_2", "col_3", "col_1"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize("""split""" , [None, NamedSplit("""train""" ), """train""", """test"""] )
def lowerCamelCase_ ( _a : Optional[Any] , _a : List[str] , _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = tmp_path / """cache"""
UpperCAmelCase_ : str = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
UpperCAmelCase_ : Optional[Any] = JsonDatasetReader(_a , cache_dir=_a , split=_a ).read()
_check_json_dataset(_a , _a )
assert dataset.split == split if split else "train"
@pytest.mark.parametrize("""path_type""" , [str, list] )
def lowerCamelCase_ ( _a : Dict , _a : Optional[int] , _a : Optional[Any] ):
'''simple docstring'''
if issubclass(_a , _a ):
UpperCAmelCase_ : List[str] = jsonl_path
elif issubclass(_a , _a ):
UpperCAmelCase_ : Tuple = [jsonl_path]
UpperCAmelCase_ : int = tmp_path / """cache"""
UpperCAmelCase_ : Any = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
UpperCAmelCase_ : Tuple = JsonDatasetReader(_a , cache_dir=_a ).read()
_check_json_dataset(_a , _a )
def lowerCamelCase_ ( _a : List[str] , _a : Optional[int] , _a : List[Any]=("train",) ):
'''simple docstring'''
assert isinstance(_a , _a )
for split in splits:
UpperCAmelCase_ : Tuple = dataset_dict[split]
assert dataset.num_rows == 4
assert dataset.num_columns == 3
assert dataset.column_names == ["col_1", "col_2", "col_3"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize("""keep_in_memory""" , [False, True] )
def lowerCamelCase_ ( _a : List[Any] , _a : Optional[int] , _a : str ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = tmp_path / """cache"""
UpperCAmelCase_ : Dict = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
UpperCAmelCase_ : int = JsonDatasetReader({"""train""": jsonl_path} , cache_dir=_a , keep_in_memory=_a ).read()
_check_json_datasetdict(_a , _a )
@pytest.mark.parametrize(
"""features""" , [
None,
{"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""},
{"""col_1""": """string""", """col_2""": """string""", """col_3""": """string"""},
{"""col_1""": """int32""", """col_2""": """int32""", """col_3""": """int32"""},
{"""col_1""": """float32""", """col_2""": """float32""", """col_3""": """float32"""},
] , )
def lowerCamelCase_ ( _a : Tuple , _a : Union[str, Any] , _a : Tuple ):
'''simple docstring'''
UpperCAmelCase_ : Any = tmp_path / """cache"""
UpperCAmelCase_ : List[str] = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
UpperCAmelCase_ : Any = features.copy() if features else default_expected_features
UpperCAmelCase_ : List[Any] = (
Features({feature: Value(_a ) for feature, dtype in features.items()} ) if features is not None else None
)
UpperCAmelCase_ : Any = JsonDatasetReader({"""train""": jsonl_path} , features=_a , cache_dir=_a ).read()
_check_json_datasetdict(_a , _a )
@pytest.mark.parametrize("""split""" , [None, NamedSplit("""train""" ), """train""", """test"""] )
def lowerCamelCase_ ( _a : Any , _a : str , _a : str ):
'''simple docstring'''
if split:
UpperCAmelCase_ : Optional[Any] = {split: jsonl_path}
else:
UpperCAmelCase_ : Tuple = """train"""
UpperCAmelCase_ : Union[str, Any] = {"""train""": jsonl_path, """test""": jsonl_path}
UpperCAmelCase_ : Tuple = tmp_path / """cache"""
UpperCAmelCase_ : Dict = {"""col_1""": """string""", """col_2""": """int64""", """col_3""": """float64"""}
UpperCAmelCase_ : Union[str, Any] = JsonDatasetReader(_a , cache_dir=_a ).read()
_check_json_datasetdict(_a , _a , splits=list(path.keys() ) )
assert all(dataset[split].split == split for split in path.keys() )
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
return json.load(_a )
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return [json.loads(_a ) for line in buffer]
class _snake_case :
'''simple docstring'''
@pytest.mark.parametrize("""lines, load_json_function""" ,[(True, load_json_lines), (False, load_json)] )
def A__ ( self: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Optional[int] ) -> Any:
with io.BytesIO() as buffer:
JsonDatasetWriter(lowerCamelCase_ ,lowerCamelCase_ ,lines=lowerCamelCase_ ).write()
buffer.seek(0 )
UpperCAmelCase_ : List[str] = load_json_function(lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(exported_content[0] ,lowerCamelCase_ )
assert len(lowerCamelCase_ ) == 10
@pytest.mark.parametrize(
"""orient, container, keys, len_at""" ,[
("""records""", list, {"""tokens""", """labels""", """answers""", """id"""}, None),
("""split""", dict, {"""columns""", """data"""}, """data"""),
("""index""", dict, set("""0123456789""" ), None),
("""columns""", dict, {"""tokens""", """labels""", """answers""", """id"""}, """tokens"""),
("""values""", list, None, None),
("""table""", dict, {"""schema""", """data"""}, """data"""),
] ,)
def A__ ( self: List[str] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[int] ) -> Tuple:
with io.BytesIO() as buffer:
JsonDatasetWriter(lowerCamelCase_ ,lowerCamelCase_ ,lines=lowerCamelCase_ ,orient=lowerCamelCase_ ).write()
buffer.seek(0 )
UpperCAmelCase_ : Dict = load_json(lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
if keys:
if container is dict:
assert exported_content.keys() == keys
else:
assert exported_content[0].keys() == keys
else:
assert not hasattr(lowerCamelCase_ ,"""keys""" ) and not hasattr(exported_content[0] ,"""keys""" )
if len_at:
assert len(exported_content[len_at] ) == 10
else:
assert len(lowerCamelCase_ ) == 10
@pytest.mark.parametrize("""lines, load_json_function""" ,[(True, load_json_lines), (False, load_json)] )
def A__ ( self: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ) -> int:
with io.BytesIO() as buffer:
JsonDatasetWriter(lowerCamelCase_ ,lowerCamelCase_ ,lines=lowerCamelCase_ ,num_proc=2 ).write()
buffer.seek(0 )
UpperCAmelCase_ : List[Any] = load_json_function(lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(exported_content[0] ,lowerCamelCase_ )
assert len(lowerCamelCase_ ) == 10
@pytest.mark.parametrize(
"""orient, container, keys, len_at""" ,[
("""records""", list, {"""tokens""", """labels""", """answers""", """id"""}, None),
("""split""", dict, {"""columns""", """data"""}, """data"""),
("""index""", dict, set("""0123456789""" ), None),
("""columns""", dict, {"""tokens""", """labels""", """answers""", """id"""}, """tokens"""),
("""values""", list, None, None),
("""table""", dict, {"""schema""", """data"""}, """data"""),
] ,)
def A__ ( self: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: int ,lowerCamelCase_: str ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ) -> Any:
with io.BytesIO() as buffer:
JsonDatasetWriter(lowerCamelCase_ ,lowerCamelCase_ ,lines=lowerCamelCase_ ,orient=lowerCamelCase_ ,num_proc=2 ).write()
buffer.seek(0 )
UpperCAmelCase_ : str = load_json(lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
if keys:
if container is dict:
assert exported_content.keys() == keys
else:
assert exported_content[0].keys() == keys
else:
assert not hasattr(lowerCamelCase_ ,"""keys""" ) and not hasattr(exported_content[0] ,"""keys""" )
if len_at:
assert len(exported_content[len_at] ) == 10
else:
assert len(lowerCamelCase_ ) == 10
def A__ ( self: Tuple ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
with pytest.raises(lowerCamelCase_ ):
with io.BytesIO() as buffer:
JsonDatasetWriter(lowerCamelCase_ ,lowerCamelCase_ ,num_proc=0 )
@pytest.mark.parametrize("""compression, extension""" ,[("""gzip""", """gz"""), ("""bz2""", """bz2"""), ("""xz""", """xz""")] )
def A__ ( self: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int ) -> Any:
UpperCAmelCase_ : Any = tmp_path_factory.mktemp("""data""" ) / F'''test.json.{extension}'''
UpperCAmelCase_ : Optional[int] = str(shared_datadir / F'''test_file.json.{extension}''' )
JsonDatasetWriter(lowerCamelCase_ ,lowerCamelCase_ ,compression=lowerCamelCase_ ).write()
with fsspec.open(lowerCamelCase_ ,"""rb""" ,compression="""infer""" ) as f:
UpperCAmelCase_ : Dict = f.read()
with fsspec.open(lowerCamelCase_ ,"""rb""" ,compression="""infer""" ) as f:
UpperCAmelCase_ : Any = f.read()
assert exported_content == original_content
| 345 |
import gc
import unittest
import torch
from parameterized import parameterized
from diffusers import AutoencoderKL
from diffusers.utils import floats_tensor, load_hf_numpy, require_torch_gpu, slow, torch_all_close, torch_device
from diffusers.utils.import_utils import is_xformers_available
from diffusers.utils.testing_utils import enable_full_determinism
from .test_modeling_common import ModelTesterMixin, UNetTesterMixin
enable_full_determinism()
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = AutoencoderKL
A__ : Optional[int] = "sample"
A__ : Tuple = 1E-2
@property
def A__ ( self: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = 4
UpperCAmelCase_ : str = 3
UpperCAmelCase_ : Any = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ).to(lowerCamelCase_ )
return {"sample": image}
@property
def A__ ( self: List[str] ) -> Tuple:
return (3, 32, 32)
@property
def A__ ( self: Optional[Any] ) -> Any:
return (3, 32, 32)
def A__ ( self: Any ) -> Tuple:
UpperCAmelCase_ : List[Any] = {
"""block_out_channels""": [32, 64],
"""in_channels""": 3,
"""out_channels""": 3,
"""down_block_types""": ["""DownEncoderBlock2D""", """DownEncoderBlock2D"""],
"""up_block_types""": ["""UpDecoderBlock2D""", """UpDecoderBlock2D"""],
"""latent_channels""": 4,
}
UpperCAmelCase_ : int = self.dummy_input
return init_dict, inputs_dict
def A__ ( self: Optional[Any] ) -> int:
pass
def A__ ( self: str ) -> Any:
pass
@unittest.skipIf(torch_device == """mps""" ,"""Gradient checkpointing skipped on MPS""" )
def A__ ( self: Union[str, Any] ) -> Dict:
# enable deterministic behavior for gradient checkpointing
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.prepare_init_args_and_inputs_for_common()
UpperCAmelCase_ : List[Any] = self.model_class(**lowerCamelCase_ )
model.to(lowerCamelCase_ )
assert not model.is_gradient_checkpointing and model.training
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model.zero_grad()
UpperCAmelCase_ : Any = torch.randn_like(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = (out - labels).mean()
loss.backward()
# re-instantiate the model now enabling gradient checkpointing
UpperCAmelCase_ : str = self.model_class(**lowerCamelCase_ )
# clone model
model_a.load_state_dict(model.state_dict() )
model_a.to(lowerCamelCase_ )
model_a.enable_gradient_checkpointing()
assert model_a.is_gradient_checkpointing and model_a.training
UpperCAmelCase_ : Optional[int] = model_a(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model_a.zero_grad()
UpperCAmelCase_ : Dict = (out_a - labels).mean()
loss_a.backward()
# compare the output and parameters gradients
self.assertTrue((loss - loss_a).abs() < 1e-5 )
UpperCAmelCase_ : Dict = dict(model.named_parameters() )
UpperCAmelCase_ : Union[str, Any] = dict(model_a.named_parameters() )
for name, param in named_params.items():
self.assertTrue(torch_all_close(param.grad.data ,named_params_a[name].grad.data ,atol=5e-5 ) )
def A__ ( self: Optional[Any] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" ,output_loading_info=lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
self.assertEqual(len(loading_info["""missing_keys"""] ) ,0 )
model.to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model(**self.dummy_input )
assert image is not None, "Make sure output is not None"
def A__ ( self: Optional[int] ) -> int:
UpperCAmelCase_ : Dict = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" )
UpperCAmelCase_ : Tuple = model.to(lowerCamelCase_ )
model.eval()
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.manual_seed(0 )
else:
UpperCAmelCase_ : Optional[int] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = torch.randn(
1 ,model.config.in_channels ,model.config.sample_size ,model.config.sample_size ,generator=torch.manual_seed(0 ) ,)
UpperCAmelCase_ : int = image.to(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ,generator=lowerCamelCase_ ).sample
UpperCAmelCase_ : Optional[int] = output[0, -1, -3:, -3:].flatten().cpu()
# Since the VAE Gaussian prior's generator is seeded on the appropriate device,
# the expected output slices are not the same for CPU and GPU.
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.tensor(
[
-4.0078e-01,
-3.8323e-04,
-1.2681e-01,
-1.1462e-01,
2.0095e-01,
1.0893e-01,
-8.8247e-02,
-3.0361e-01,
-9.8644e-03,
] )
elif torch_device == "cpu":
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.1_3_5_2, 0.0_8_7_8, 0.0_4_1_9, -0.0_8_1_8, -0.1_0_6_9, 0.0_6_8_8, -0.1_4_5_8, -0.4_4_4_6, -0.0_0_2_6] )
else:
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.2_4_2_1, 0.4_6_4_2, 0.2_5_0_7, -0.0_4_3_8, 0.0_6_8_2, 0.3_1_6_0, -0.2_0_1_8, -0.0_7_2_7, 0.2_4_8_5] )
self.assertTrue(torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,rtol=1e-2 ) )
@slow
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
return F'''gaussian_noise_s={seed}_shape={'_'.join([str(lowerCamelCase_ ) for s in shape] )}.npy'''
def A__ ( self: Union[str, Any] ) -> Optional[int]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: List[Any]=(4, 3, 512, 512) ,lowerCamelCase_: Optional[Any]=False ) -> Optional[int]:
UpperCAmelCase_ : Tuple = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : Tuple = torch.from_numpy(load_hf_numpy(self.get_file_format(lowerCamelCase_ ,lowerCamelCase_ ) ) ).to(lowerCamelCase_ ).to(lowerCamelCase_ )
return image
def A__ ( self: List[Any] ,lowerCamelCase_: List[str]="CompVis/stable-diffusion-v1-4" ,lowerCamelCase_: Union[str, Any]=False ) -> Any:
UpperCAmelCase_ : Optional[Any] = """fp16""" if fpaa else None
UpperCAmelCase_ : str = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : int = AutoencoderKL.from_pretrained(
lowerCamelCase_ ,subfolder="""vae""" ,torch_dtype=lowerCamelCase_ ,revision=lowerCamelCase_ ,)
model.to(lowerCamelCase_ ).eval()
return model
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any]=0 ) -> Optional[int]:
if torch_device == "mps":
return torch.manual_seed(lowerCamelCase_ )
return torch.Generator(device=lowerCamelCase_ ).manual_seed(lowerCamelCase_ )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_3, 0.9_8_7_8, -0.0_4_9_5, -0.0_7_9_0, -0.2_7_0_9, 0.8_3_7_5, -0.2_0_6_0, -0.0_8_2_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_6, 0.1_1_6_8, 0.1_3_3_2, -0.4_8_4_0, -0.2_5_0_8, -0.0_7_9_1, -0.0_4_9_3, -0.4_0_8_9], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ) -> Tuple:
UpperCAmelCase_ : List[Any] = self.get_sd_vae_model()
UpperCAmelCase_ : int = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Optional[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Tuple = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[33, [-0.0_5_1_3, 0.0_2_8_9, 1.3_7_9_9, 0.2_1_6_6, -0.2_5_7_3, -0.0_8_7_1, 0.5_1_0_3, -0.0_9_9_9]],
[47, [-0.4_1_2_8, -0.1_3_2_0, -0.3_7_0_4, 0.1_9_6_5, -0.4_1_1_6, -0.2_3_3_2, -0.3_3_4_0, 0.2_2_4_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Tuple:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Tuple = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : Optional[int] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_9, 0.9_8_6_6, -0.0_4_8_7, -0.0_7_7_7, -0.2_7_1_6, 0.8_3_6_8, -0.2_0_5_5, -0.0_8_1_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_7, 0.1_1_4_7, 0.1_3_3_3, -0.4_8_4_1, -0.2_5_0_6, -0.0_8_0_5, -0.0_4_9_1, -0.4_0_8_5], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[int] = self.get_sd_vae_model()
UpperCAmelCase_ : Dict = self.get_sd_image(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : str = model(lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : List[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Any = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[13, [-0.2_0_5_1, -0.1_8_0_3, -0.2_3_1_1, -0.2_1_1_4, -0.3_2_9_2, -0.3_5_7_4, -0.2_9_5_3, -0.3_3_2_3]],
[37, [-0.2_6_3_2, -0.2_6_2_5, -0.2_1_9_9, -0.2_7_4_1, -0.4_5_3_9, -0.4_9_9_0, -0.3_7_2_0, -0.4_9_2_5]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: str ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[int] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : str = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : Any = sample[-1, -2:, :2, -2:].flatten().cpu()
UpperCAmelCase_ : Union[str, Any] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 )
@parameterized.expand(
[
# fmt: off
[27, [-0.0_3_6_9, 0.0_2_0_7, -0.0_7_7_6, -0.0_6_8_2, -0.1_7_4_7, -0.1_9_3_0, -0.1_4_6_5, -0.2_0_3_9]],
[16, [-0.1_6_2_8, -0.2_1_3_4, -0.2_7_4_7, -0.2_6_4_2, -0.3_7_7_4, -0.4_4_0_4, -0.3_6_8_7, -0.4_2_7_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: str ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : str = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : str = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=5e-3 )
@parameterized.expand([(13,), (16,), (27,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> int:
UpperCAmelCase_ : Optional[Any] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-1 )
@parameterized.expand([(13,), (16,), (37,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = self.get_sd_vae_model()
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.3_0_0_1, 0.0_9_1_8, -2.6_9_8_4, -3.9_7_2_0, -3.2_0_9_9, -5.0_3_5_3, 1.7_3_3_8, -0.2_0_6_5, 3.4_2_6_7]],
[47, [-1.5_0_3_0, -4.3_8_7_1, -6.0_3_5_5, -9.1_1_5_7, -1.6_6_6_1, -2.7_8_5_3, 2.1_6_0_7, -5.0_8_2_3, 2.5_6_3_3]],
# fmt: on
] )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[Any] = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : str = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : int = model.encode(lowerCamelCase_ ).latent_dist
UpperCAmelCase_ : Optional[Any] = dist.sample(generator=lowerCamelCase_ )
assert list(sample.shape ) == [image.shape[0], 4] + [i // 8 for i in image.shape[2:]]
UpperCAmelCase_ : Tuple = sample[0, -1, -3:, -3:].flatten().cpu()
UpperCAmelCase_ : Optional[Any] = torch.tensor(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = 3e-3 if torch_device != """mps""" else 1e-2
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=lowerCamelCase_ )
| 345 | 1 |
import copy
import os
from typing import Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''BridgeTower/bridgetower-base''': '''https://huggingface.co/BridgeTower/bridgetower-base/blob/main/config.json''',
'''BridgeTower/bridgetower-base-itm-mlm''': (
'''https://huggingface.co/BridgeTower/bridgetower-base-itm-mlm/blob/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : List[str] = "bridgetower_vision_model"
def __init__( self: List[str] ,lowerCamelCase_: Union[str, Any]=768 ,lowerCamelCase_: Optional[Any]=12 ,lowerCamelCase_: Union[str, Any]=3 ,lowerCamelCase_: Tuple=16 ,lowerCamelCase_: Any=288 ,lowerCamelCase_: List[str]=1 ,lowerCamelCase_: int=1e-05 ,lowerCamelCase_: Optional[Any]=False ,lowerCamelCase_: Optional[Any]=True ,lowerCamelCase_: Dict=False ,**lowerCamelCase_: Dict ,) -> Union[str, Any]:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = hidden_size
UpperCAmelCase_ : Optional[int] = num_hidden_layers
UpperCAmelCase_ : int = num_channels
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : int = image_size
UpperCAmelCase_ : Tuple = initializer_factor
UpperCAmelCase_ : Tuple = layer_norm_eps
UpperCAmelCase_ : List[str] = stop_gradient
UpperCAmelCase_ : Any = share_layernorm
UpperCAmelCase_ : Optional[Any] = remove_last_layer
@classmethod
def A__ ( cls: Optional[Any] ,lowerCamelCase_: Union[str, os.PathLike] ,**lowerCamelCase_: Tuple ) -> "PretrainedConfig":
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = cls.get_config_dict(lowerCamelCase_ ,**lowerCamelCase_ )
if config_dict.get("""model_type""" ) == "bridgetower":
UpperCAmelCase_ : List[str] = config_dict["""text_config"""]
if "model_type" in config_dict and hasattr(cls ,"""model_type""" ) and config_dict["model_type"] != cls.model_type:
logger.warning(
F'''You are using a model of type {config_dict['model_type']} to instantiate a model of type '''
F'''{cls.model_type}. This is not supported for all configurations of models and can yield errors.''' )
return cls.from_dict(lowerCamelCase_ ,**lowerCamelCase_ )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : List[str] = "bridgetower_text_model"
def __init__( self: Optional[int] ,lowerCamelCase_: int=50265 ,lowerCamelCase_: List[Any]=768 ,lowerCamelCase_: Tuple=12 ,lowerCamelCase_: Dict=12 ,lowerCamelCase_: int=1 ,lowerCamelCase_: str=3072 ,lowerCamelCase_: List[str]="gelu" ,lowerCamelCase_: Optional[Any]=0.1 ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: Tuple=514 ,lowerCamelCase_: Union[str, Any]=1 ,lowerCamelCase_: Tuple=1e-05 ,lowerCamelCase_: Any=1 ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Tuple="absolute" ,lowerCamelCase_: Tuple=True ,**lowerCamelCase_: Union[str, Any] ,) -> str:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = vocab_size
UpperCAmelCase_ : Optional[Any] = hidden_size
UpperCAmelCase_ : Tuple = num_hidden_layers
UpperCAmelCase_ : Optional[int] = num_attention_heads
UpperCAmelCase_ : str = hidden_act
UpperCAmelCase_ : List[str] = initializer_factor
UpperCAmelCase_ : List[str] = intermediate_size
UpperCAmelCase_ : Optional[int] = hidden_dropout_prob
UpperCAmelCase_ : Any = attention_probs_dropout_prob
UpperCAmelCase_ : Union[str, Any] = max_position_embeddings
UpperCAmelCase_ : Any = type_vocab_size
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : Tuple = position_embedding_type
UpperCAmelCase_ : List[Any] = use_cache
UpperCAmelCase_ : int = pad_token_id
UpperCAmelCase_ : int = bos_token_id
UpperCAmelCase_ : List[Any] = eos_token_id
@classmethod
def A__ ( cls: Optional[int] ,lowerCamelCase_: Union[str, os.PathLike] ,**lowerCamelCase_: List[str] ) -> "PretrainedConfig":
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = cls.get_config_dict(lowerCamelCase_ ,**lowerCamelCase_ )
if config_dict.get("""model_type""" ) == "bridgetower":
UpperCAmelCase_ : Tuple = config_dict["""text_config"""]
if "model_type" in config_dict and hasattr(cls ,"""model_type""" ) and config_dict["model_type"] != cls.model_type:
logger.warning(
F'''You are using a model of type {config_dict['model_type']} to instantiate a model of type '''
F'''{cls.model_type}. This is not supported for all configurations of models and can yield errors.''' )
return cls.from_dict(lowerCamelCase_ ,**lowerCamelCase_ )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = "bridgetower"
def __init__( self: Dict ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[Any]="gelu" ,lowerCamelCase_: int=768 ,lowerCamelCase_: List[str]=1 ,lowerCamelCase_: Tuple=1e-05 ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict="add" ,lowerCamelCase_: Optional[int]=12 ,lowerCamelCase_: Optional[int]=6 ,lowerCamelCase_: List[Any]=False ,lowerCamelCase_: Any=False ,lowerCamelCase_: List[Any]=None ,lowerCamelCase_: List[Any]=None ,**lowerCamelCase_: int ,) -> Optional[Any]:
# TODO: remove this once the Hub files are updated.
UpperCAmelCase_ : Union[str, Any] = kwargs.pop("""text_config_dict""" ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = kwargs.pop("""vision_config_dict""" ,lowerCamelCase_ )
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = share_cross_modal_transformer_layers
UpperCAmelCase_ : Dict = hidden_act
UpperCAmelCase_ : List[Any] = hidden_size
UpperCAmelCase_ : Dict = initializer_factor
UpperCAmelCase_ : List[Any] = layer_norm_eps
UpperCAmelCase_ : int = share_link_tower_layers
UpperCAmelCase_ : List[Any] = link_tower_type
UpperCAmelCase_ : int = num_attention_heads
UpperCAmelCase_ : List[str] = num_hidden_layers
UpperCAmelCase_ : str = tie_word_embeddings
UpperCAmelCase_ : int = init_layernorm_from_vision_encoder
if text_config is None:
UpperCAmelCase_ : Tuple = {}
logger.info("""`text_config` is `None`. Initializing the `BridgeTowerTextConfig` with default values.""" )
if vision_config is None:
UpperCAmelCase_ : List[Any] = {}
logger.info("""`vision_config` is `None`. Initializing the `BridgeTowerVisionConfig` with default values.""" )
UpperCAmelCase_ : str = BridgeTowerTextConfig(**lowerCamelCase_ )
UpperCAmelCase_ : List[str] = BridgeTowerVisionConfig(**lowerCamelCase_ )
@classmethod
def A__ ( cls: List[str] ,lowerCamelCase_: BridgeTowerTextConfig ,lowerCamelCase_: BridgeTowerVisionConfig ,**lowerCamelCase_: Dict ) -> Optional[Any]:
return cls(text_config=text_config.to_dict() ,vision_config=vision_config.to_dict() ,**lowerCamelCase_ )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Union[str, Any] = copy.deepcopy(self.__dict__ )
UpperCAmelCase_ : str = self.text_config.to_dict()
UpperCAmelCase_ : List[Any] = self.vision_config.to_dict()
UpperCAmelCase_ : Optional[int] = self.__class__.model_type
return output
| 345 |
import copy
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Audio, Features, Value
from .base import TaskTemplate
@dataclass(frozen=__snake_case )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = field(default="automatic-speech-recognition" , metadata={"include_in_asdict_even_if_is_default": True} )
A__ : ClassVar[Features] = Features({"audio": Audio()} )
A__ : ClassVar[Features] = Features({"transcription": Value("string" )} )
A__ : str = "audio"
A__ : str = "transcription"
def A__ ( self: int ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
if self.audio_column not in features:
raise ValueError(F'''Column {self.audio_column} is not present in features.''' )
if not isinstance(features[self.audio_column] ,lowerCamelCase_ ):
raise ValueError(F'''Column {self.audio_column} is not an Audio type.''' )
UpperCAmelCase_ : Any = copy.deepcopy(self )
UpperCAmelCase_ : Union[str, Any] = self.input_schema.copy()
UpperCAmelCase_ : Any = features[self.audio_column]
UpperCAmelCase_ : Union[str, Any] = input_schema
return task_template
@property
def A__ ( self: List[str] ) -> Dict[str, str]:
return {self.audio_column: "audio", self.transcription_column: "transcription"}
| 345 | 1 |
from typing import List, Optional, Tuple, Union
import PIL
import torch
from torchvision import transforms
from diffusers.pipeline_utils import DiffusionPipeline, ImagePipelineOutput
from diffusers.schedulers import DDIMScheduler
from diffusers.utils import randn_tensor
UpperCamelCase_ = transforms.Compose(
[
transforms.Resize((256, 256)),
transforms.ToTensor(),
transforms.Normalize([0.5], [0.5]),
]
)
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
if isinstance(_a , torch.Tensor ):
return image
elif isinstance(_a , PIL.Image.Image ):
UpperCAmelCase_ : List[str] = [image]
UpperCAmelCase_ : Union[str, Any] = [trans(img.convert("""RGB""" ) ) for img in image]
UpperCAmelCase_ : Any = torch.stack(_a )
return image
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ) -> Tuple:
super().__init__()
# make sure scheduler can always be converted to DDIM
UpperCAmelCase_ : Any = DDIMScheduler.from_config(scheduler.config )
self.register_modules(unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
def A__ ( self: List[Any] ,lowerCamelCase_: List[Any] ) -> Optional[int]:
if strength < 0 or strength > 1:
raise ValueError(F'''The value of strength should in [0.0, 1.0] but is {strength}''' )
def A__ ( self: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ) -> Any:
# get the original timestep using init_timestep
UpperCAmelCase_ : Tuple = min(int(num_inference_steps * strength ) ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = max(num_inference_steps - init_timestep ,0 )
UpperCAmelCase_ : Tuple = self.scheduler.timesteps[t_start:]
return timesteps, num_inference_steps - t_start
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=None ) -> List[str]:
if not isinstance(lowerCamelCase_ ,(torch.Tensor, PIL.Image.Image, list) ):
raise ValueError(
F'''`image` has to be of type `torch.Tensor`, `PIL.Image.Image` or list but is {type(lowerCamelCase_ )}''' )
UpperCAmelCase_ : Any = image.to(device=lowerCamelCase_ ,dtype=lowerCamelCase_ )
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) and len(lowerCamelCase_ ) != batch_size:
raise ValueError(
F'''You have passed a list of generators of length {len(lowerCamelCase_ )}, but requested an effective batch'''
F''' size of {batch_size}. Make sure the batch size matches the length of the generators.''' )
UpperCAmelCase_ : Tuple = init_latents.shape
UpperCAmelCase_ : Any = randn_tensor(lowerCamelCase_ ,generator=lowerCamelCase_ ,device=lowerCamelCase_ ,dtype=lowerCamelCase_ )
# get latents
print("""add noise to latents at timestep""" ,lowerCamelCase_ )
UpperCAmelCase_ : Any = self.scheduler.add_noise(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Dict = init_latents
return latents
@torch.no_grad()
def __call__( self: List[str] ,lowerCamelCase_: Union[torch.FloatTensor, PIL.Image.Image] = None ,lowerCamelCase_: float = 0.8 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: Optional[Union[torch.Generator, List[torch.Generator]]] = None ,lowerCamelCase_: float = 0.0 ,lowerCamelCase_: int = 50 ,lowerCamelCase_: Optional[bool] = None ,lowerCamelCase_: Optional[str] = "pil" ,lowerCamelCase_: bool = True ,) -> Union[ImagePipelineOutput, Tuple]:
self.check_inputs(lowerCamelCase_ )
# 2. Preprocess image
UpperCAmelCase_ : List[str] = preprocess(lowerCamelCase_ )
# 3. set timesteps
self.scheduler.set_timesteps(lowerCamelCase_ ,device=self.device )
UpperCAmelCase_ , UpperCAmelCase_ : str = self.get_timesteps(lowerCamelCase_ ,lowerCamelCase_ ,self.device )
UpperCAmelCase_ : int = timesteps[:1].repeat(lowerCamelCase_ )
# 4. Prepare latent variables
UpperCAmelCase_ : Dict = self.prepare_latents(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,self.unet.dtype ,self.device ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = latents
# 5. Denoising loop
for t in self.progress_bar(lowerCamelCase_ ):
# 1. predict noise model_output
UpperCAmelCase_ : Tuple = self.unet(lowerCamelCase_ ,lowerCamelCase_ ).sample
# 2. predict previous mean of image x_t-1 and add variance depending on eta
# eta corresponds to η in paper and should be between [0, 1]
# do x_t -> x_t-1
UpperCAmelCase_ : int = self.scheduler.step(
lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,eta=lowerCamelCase_ ,use_clipped_model_output=lowerCamelCase_ ,generator=lowerCamelCase_ ,).prev_sample
UpperCAmelCase_ : Tuple = (image / 2 + 0.5).clamp(0 ,1 )
UpperCAmelCase_ : Any = image.cpu().permute(0 ,2 ,3 ,1 ).numpy()
if output_type == "pil":
UpperCAmelCase_ : str = self.numpy_to_pil(lowerCamelCase_ )
if not return_dict:
return (image, latent_timestep.item())
return ImagePipelineOutput(images=lowerCamelCase_ )
| 345 |
from collections import OrderedDict
from typing import TYPE_CHECKING, Any, Mapping, Optional
from packaging import version
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...onnx.utils import compute_effective_axis_dimension
from ...utils import logging
if TYPE_CHECKING:
from ...processing_utils import ProcessorMixin
from ...utils import TensorType
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/layoutlmv3-base''': '''https://huggingface.co/microsoft/layoutlmv3-base/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "layoutlmv3"
def __init__( self: str ,lowerCamelCase_: Any=50265 ,lowerCamelCase_: int=768 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=3072 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Any=0.1 ,lowerCamelCase_: Tuple=512 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: List[str]=1e-5 ,lowerCamelCase_: int=1 ,lowerCamelCase_: int=0 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Dict=1024 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=32 ,lowerCamelCase_: Union[str, Any]=128 ,lowerCamelCase_: Tuple=64 ,lowerCamelCase_: Tuple=256 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Any=True ,lowerCamelCase_: Dict=224 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Optional[int]=16 ,lowerCamelCase_: Dict=None ,**lowerCamelCase_: str ,) -> List[Any]:
super().__init__(
vocab_size=lowerCamelCase_ ,hidden_size=lowerCamelCase_ ,num_hidden_layers=lowerCamelCase_ ,num_attention_heads=lowerCamelCase_ ,intermediate_size=lowerCamelCase_ ,hidden_act=lowerCamelCase_ ,hidden_dropout_prob=lowerCamelCase_ ,attention_probs_dropout_prob=lowerCamelCase_ ,max_position_embeddings=lowerCamelCase_ ,type_vocab_size=lowerCamelCase_ ,initializer_range=lowerCamelCase_ ,layer_norm_eps=lowerCamelCase_ ,pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = max_ad_position_embeddings
UpperCAmelCase_ : Optional[int] = coordinate_size
UpperCAmelCase_ : Optional[int] = shape_size
UpperCAmelCase_ : Optional[Any] = has_relative_attention_bias
UpperCAmelCase_ : Optional[int] = rel_pos_bins
UpperCAmelCase_ : Union[str, Any] = max_rel_pos
UpperCAmelCase_ : Dict = has_spatial_attention_bias
UpperCAmelCase_ : Optional[int] = rel_ad_pos_bins
UpperCAmelCase_ : Tuple = max_rel_ad_pos
UpperCAmelCase_ : Union[str, Any] = text_embed
UpperCAmelCase_ : Optional[Any] = visual_embed
UpperCAmelCase_ : List[str] = input_size
UpperCAmelCase_ : str = num_channels
UpperCAmelCase_ : Optional[int] = patch_size
UpperCAmelCase_ : Tuple = classifier_dropout
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = version.parse("1.12" )
@property
def A__ ( self: Dict ) -> Mapping[str, Mapping[int, str]]:
# The order of inputs is different for question answering and sequence classification
if self.task in ["question-answering", "sequence-classification"]:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels""", 2: """height""", 3: """width"""}),
] )
else:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels"""}),
] )
@property
def A__ ( self: Any ) -> float:
return 1e-5
@property
def A__ ( self: int ) -> int:
return 12
def A__ ( self: List[str] ,lowerCamelCase_: "ProcessorMixin" ,lowerCamelCase_: int = -1 ,lowerCamelCase_: int = -1 ,lowerCamelCase_: bool = False ,lowerCamelCase_: Optional["TensorType"] = None ,lowerCamelCase_: int = 3 ,lowerCamelCase_: int = 40 ,lowerCamelCase_: int = 40 ,) -> Mapping[str, Any]:
setattr(processor.image_processor ,"""apply_ocr""" ,lowerCamelCase_ )
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
UpperCAmelCase_ : List[str] = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_batch ,num_token_to_add=0 )
# If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX
UpperCAmelCase_ : int = processor.tokenizer.num_special_tokens_to_add(lowerCamelCase_ )
UpperCAmelCase_ : int = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_sequence ,num_token_to_add=lowerCamelCase_ )
# Generate dummy inputs according to compute batch and sequence
UpperCAmelCase_ : Optional[int] = [[""" """.join([processor.tokenizer.unk_token] ) * seq_length]] * batch_size
# Generate dummy bounding boxes
UpperCAmelCase_ : List[Any] = [[[48, 84, 73, 128]]] * batch_size
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
# batch_size = compute_effective_axis_dimension(batch_size, fixed_dimension=OnnxConfig.default_fixed_batch)
UpperCAmelCase_ : Any = self._generate_dummy_images(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = dict(
processor(
lowerCamelCase_ ,text=lowerCamelCase_ ,boxes=lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,) )
return inputs
| 345 | 1 |
import unittest
from transformers import (
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
Pipeline,
ZeroShotClassificationPipeline,
pipeline,
)
from transformers.testing_utils import is_pipeline_test, nested_simplify, require_tf, require_torch, slow
from .test_pipelines_common import ANY
# These 2 model types require different inputs than those of the usual text models.
UpperCamelCase_ = {'''LayoutLMv2Config''', '''LayoutLMv3Config'''}
@is_pipeline_test
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
A__ : Optional[int] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
A__ : Union[str, Any] = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
if model_mapping is not None:
A__ : List[str] = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP}
if tf_model_mapping is not None:
A__ : str = {
config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP
}
def A__ ( self: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[str] = ZeroShotClassificationPipeline(
model=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,candidate_labels=["""polics""", """health"""] )
return classifier, ["Who are you voting for in 2020?", "My stomach hurts."]
def A__ ( self: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Any:
UpperCAmelCase_ : Optional[int] = classifier("""Who are you voting for in 2020?""" ,candidate_labels="""politics""" )
self.assertEqual(lowerCamelCase_ ,{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ )]} )
# No kwarg
UpperCAmelCase_ : Dict = classifier("""Who are you voting for in 2020?""" ,["""politics"""] )
self.assertEqual(lowerCamelCase_ ,{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ )]} )
UpperCAmelCase_ : Optional[Any] = classifier("""Who are you voting for in 2020?""" ,candidate_labels=["""politics"""] )
self.assertEqual(lowerCamelCase_ ,{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ )]} )
UpperCAmelCase_ : Union[str, Any] = classifier("""Who are you voting for in 2020?""" ,candidate_labels="""politics, public health""" )
self.assertEqual(
lowerCamelCase_ ,{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )]} )
self.assertAlmostEqual(sum(nested_simplify(outputs["""scores"""] ) ) ,1.0 )
UpperCAmelCase_ : Dict = classifier("""Who are you voting for in 2020?""" ,candidate_labels=["""politics""", """public health"""] )
self.assertEqual(
lowerCamelCase_ ,{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )]} )
self.assertAlmostEqual(sum(nested_simplify(outputs["""scores"""] ) ) ,1.0 )
UpperCAmelCase_ : int = classifier(
"""Who are you voting for in 2020?""" ,candidate_labels="""politics""" ,hypothesis_template="""This text is about {}""" )
self.assertEqual(lowerCamelCase_ ,{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ )]} )
# https://github.com/huggingface/transformers/issues/13846
UpperCAmelCase_ : Any = classifier(["""I am happy"""] ,["""positive""", """negative"""] )
self.assertEqual(
lowerCamelCase_ ,[
{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )]}
for i in range(1 )
] ,)
UpperCAmelCase_ : Optional[int] = classifier(["""I am happy""", """I am sad"""] ,["""positive""", """negative"""] )
self.assertEqual(
lowerCamelCase_ ,[
{"""sequence""": ANY(lowerCamelCase_ ), """labels""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )], """scores""": [ANY(lowerCamelCase_ ), ANY(lowerCamelCase_ )]}
for i in range(2 )
] ,)
with self.assertRaises(lowerCamelCase_ ):
classifier("""""" ,candidate_labels="""politics""" )
with self.assertRaises(lowerCamelCase_ ):
classifier(lowerCamelCase_ ,candidate_labels="""politics""" )
with self.assertRaises(lowerCamelCase_ ):
classifier("""Who are you voting for in 2020?""" ,candidate_labels="""""" )
with self.assertRaises(lowerCamelCase_ ):
classifier("""Who are you voting for in 2020?""" ,candidate_labels=lowerCamelCase_ )
with self.assertRaises(lowerCamelCase_ ):
classifier(
"""Who are you voting for in 2020?""" ,candidate_labels="""politics""" ,hypothesis_template="""Not formatting template""" ,)
with self.assertRaises(lowerCamelCase_ ):
classifier(
"""Who are you voting for in 2020?""" ,candidate_labels="""politics""" ,hypothesis_template=lowerCamelCase_ ,)
self.run_entailment_id(lowerCamelCase_ )
def A__ ( self: str ,lowerCamelCase_: Pipeline ) -> Optional[Any]:
UpperCAmelCase_ : List[Any] = zero_shot_classifier.model.config
UpperCAmelCase_ : List[str] = config.labelaid
UpperCAmelCase_ : Union[str, Any] = zero_shot_classifier.entailment_id
UpperCAmelCase_ : str = {"""LABEL_0""": 0, """LABEL_1""": 1, """LABEL_2""": 2}
self.assertEqual(zero_shot_classifier.entailment_id ,-1 )
UpperCAmelCase_ : Optional[Any] = {"""entailment""": 0, """neutral""": 1, """contradiction""": 2}
self.assertEqual(zero_shot_classifier.entailment_id ,0 )
UpperCAmelCase_ : Tuple = {"""ENTAIL""": 0, """NON-ENTAIL""": 1}
self.assertEqual(zero_shot_classifier.entailment_id ,0 )
UpperCAmelCase_ : str = {"""ENTAIL""": 2, """NEUTRAL""": 1, """CONTR""": 0}
self.assertEqual(zero_shot_classifier.entailment_id ,2 )
UpperCAmelCase_ : Optional[Any] = original_labelaid
self.assertEqual(lowerCamelCase_ ,zero_shot_classifier.entailment_id )
@require_torch
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ : Optional[Any] = pipeline(
"""zero-shot-classification""" ,model="""sshleifer/tiny-distilbert-base-cased-distilled-squad""" ,framework="""pt""" ,)
# There was a regression in 4.10 for this
# Adding a test so we don't make the mistake again.
# https://github.com/huggingface/transformers/issues/13381#issuecomment-912343499
zero_shot_classifier(
"""Who are you voting for in 2020?""" * 100 ,candidate_labels=["""politics""", """public health""", """science"""] )
@require_torch
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[int] = pipeline(
"""zero-shot-classification""" ,model="""sshleifer/tiny-distilbert-base-cased-distilled-squad""" ,framework="""pt""" ,)
UpperCAmelCase_ : Optional[Any] = zero_shot_classifier(
"""Who are you voting for in 2020?""" ,candidate_labels=["""politics""", """public health""", """science"""] )
self.assertEqual(
nested_simplify(lowerCamelCase_ ) ,{
"""sequence""": """Who are you voting for in 2020?""",
"""labels""": ["""science""", """public health""", """politics"""],
"""scores""": [0.3_3_3, 0.3_3_3, 0.3_3_3],
} ,)
@require_tf
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Optional[int] = pipeline(
"""zero-shot-classification""" ,model="""sshleifer/tiny-distilbert-base-cased-distilled-squad""" ,framework="""tf""" ,)
UpperCAmelCase_ : Dict = zero_shot_classifier(
"""Who are you voting for in 2020?""" ,candidate_labels=["""politics""", """public health""", """science"""] )
self.assertEqual(
nested_simplify(lowerCamelCase_ ) ,{
"""sequence""": """Who are you voting for in 2020?""",
"""labels""": ["""science""", """public health""", """politics"""],
"""scores""": [0.3_3_3, 0.3_3_3, 0.3_3_3],
} ,)
@slow
@require_torch
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[Any] = pipeline("""zero-shot-classification""" ,model="""roberta-large-mnli""" ,framework="""pt""" )
UpperCAmelCase_ : List[str] = zero_shot_classifier(
"""Who are you voting for in 2020?""" ,candidate_labels=["""politics""", """public health""", """science"""] )
self.assertEqual(
nested_simplify(lowerCamelCase_ ) ,{
"""sequence""": """Who are you voting for in 2020?""",
"""labels""": ["""politics""", """public health""", """science"""],
"""scores""": [0.9_7_6, 0.0_1_5, 0.0_0_9],
} ,)
UpperCAmelCase_ : Any = zero_shot_classifier(
"""The dominant sequence transduction models are based on complex recurrent or convolutional neural networks"""
""" in an encoder-decoder configuration. The best performing models also connect the encoder and decoder"""
""" through an attention mechanism. We propose a new simple network architecture, the Transformer, based"""
""" solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two"""
""" machine translation tasks show these models to be superior in quality while being more parallelizable"""
""" and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014"""
""" English-to-German translation task, improving over the existing best results, including ensembles by"""
""" over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new"""
""" single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small"""
""" fraction of the training costs of the best models from the literature. We show that the Transformer"""
""" generalizes well to other tasks by applying it successfully to English constituency parsing both with"""
""" large and limited training data.""" ,candidate_labels=["""machine learning""", """statistics""", """translation""", """vision"""] ,multi_label=lowerCamelCase_ ,)
self.assertEqual(
nested_simplify(lowerCamelCase_ ) ,{
"""sequence""": (
"""The dominant sequence transduction models are based on complex recurrent or convolutional neural"""
""" networks in an encoder-decoder configuration. The best performing models also connect the"""
""" encoder and decoder through an attention mechanism. We propose a new simple network"""
""" architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence"""
""" and convolutions entirely. Experiments on two machine translation tasks show these models to be"""
""" superior in quality while being more parallelizable and requiring significantly less time to"""
""" train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task,"""
""" improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014"""
""" English-to-French translation task, our model establishes a new single-model state-of-the-art"""
""" BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training"""
""" costs of the best models from the literature. We show that the Transformer generalizes well to"""
""" other tasks by applying it successfully to English constituency parsing both with large and"""
""" limited training data."""
),
"""labels""": ["""translation""", """machine learning""", """vision""", """statistics"""],
"""scores""": [0.8_1_7, 0.7_1_3, 0.0_1_8, 0.0_1_8],
} ,)
@slow
@require_tf
def A__ ( self: int ) -> List[str]:
UpperCAmelCase_ : List[Any] = pipeline("""zero-shot-classification""" ,model="""roberta-large-mnli""" ,framework="""tf""" )
UpperCAmelCase_ : Union[str, Any] = zero_shot_classifier(
"""Who are you voting for in 2020?""" ,candidate_labels=["""politics""", """public health""", """science"""] )
self.assertEqual(
nested_simplify(lowerCamelCase_ ) ,{
"""sequence""": """Who are you voting for in 2020?""",
"""labels""": ["""politics""", """public health""", """science"""],
"""scores""": [0.9_7_6, 0.0_1_5, 0.0_0_9],
} ,)
UpperCAmelCase_ : Optional[int] = zero_shot_classifier(
"""The dominant sequence transduction models are based on complex recurrent or convolutional neural networks"""
""" in an encoder-decoder configuration. The best performing models also connect the encoder and decoder"""
""" through an attention mechanism. We propose a new simple network architecture, the Transformer, based"""
""" solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two"""
""" machine translation tasks show these models to be superior in quality while being more parallelizable"""
""" and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014"""
""" English-to-German translation task, improving over the existing best results, including ensembles by"""
""" over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new"""
""" single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small"""
""" fraction of the training costs of the best models from the literature. We show that the Transformer"""
""" generalizes well to other tasks by applying it successfully to English constituency parsing both with"""
""" large and limited training data.""" ,candidate_labels=["""machine learning""", """statistics""", """translation""", """vision"""] ,multi_label=lowerCamelCase_ ,)
self.assertEqual(
nested_simplify(lowerCamelCase_ ) ,{
"""sequence""": (
"""The dominant sequence transduction models are based on complex recurrent or convolutional neural"""
""" networks in an encoder-decoder configuration. The best performing models also connect the"""
""" encoder and decoder through an attention mechanism. We propose a new simple network"""
""" architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence"""
""" and convolutions entirely. Experiments on two machine translation tasks show these models to be"""
""" superior in quality while being more parallelizable and requiring significantly less time to"""
""" train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task,"""
""" improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014"""
""" English-to-French translation task, our model establishes a new single-model state-of-the-art"""
""" BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training"""
""" costs of the best models from the literature. We show that the Transformer generalizes well to"""
""" other tasks by applying it successfully to English constituency parsing both with large and"""
""" limited training data."""
),
"""labels""": ["""translation""", """machine learning""", """vision""", """statistics"""],
"""scores""": [0.8_1_7, 0.7_1_3, 0.0_1_8, 0.0_1_8],
} ,)
| 345 |
import argparse
from argparse import Namespace
import torch
from torch import nn
from transformers import XGLMConfig, XGLMForCausalLM
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = [
"""decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = emb.weight.shape
UpperCAmelCase_ : Tuple = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : List[Any] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = torch.load(_a , map_location="""cpu""" )
UpperCAmelCase_ : Dict = Namespace(**checkpoint["""cfg"""]["""model"""] )
UpperCAmelCase_ : Optional[int] = checkpoint["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = state_dict["""decoder.embed_tokens.weight"""].shape[0]
UpperCAmelCase_ : List[str] = {key.replace("""decoder""" , """model""" ): val for key, val in state_dict.items()}
UpperCAmelCase_ : int = XGLMConfig(
vocab_size=_a , max_position_embeddings=args.max_target_positions , num_layers=args.decoder_layers , attention_heads=args.decoder_attention_heads , ffn_dim=args.decoder_ffn_embed_dim , d_model=args.decoder_embed_dim , layerdrop=args.decoder_layerdrop , dropout=args.dropout , attention_dropout=args.attention_dropout , activation_dropout=args.activation_dropout , activation_function="""gelu""" , scale_embedding=not args.no_scale_embedding , tie_word_embeddings=args.share_decoder_input_output_embed , )
UpperCAmelCase_ : List[str] = XGLMForCausalLM(_a )
UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
print(_a )
UpperCAmelCase_ : Optional[Any] = make_linear_from_emb(model.model.embed_tokens )
return model
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument('''fairseq_path''', type=str, help='''path to a model.pt on local filesystem.''')
parser.add_argument('''pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = convert_fairseq_xglm_checkpoint_from_disk(args.fairseq_path)
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 | 1 |
import argparse
import datetime
def lowerCamelCase_ ( _a : str ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = {
"""0""": """Sunday""",
"""1""": """Monday""",
"""2""": """Tuesday""",
"""3""": """Wednesday""",
"""4""": """Thursday""",
"""5""": """Friday""",
"""6""": """Saturday""",
}
UpperCAmelCase_ : Tuple = {0: 1, 1: 2, 2: 3, 3: 4, 4: 5, 5: 6, 6: 0}
# Validate
if not 0 < len(_a ) < 11:
raise ValueError("""Must be 10 characters long""" )
# Get month
UpperCAmelCase_ : int = int(date_input[0] + date_input[1] )
# Validate
if not 0 < m < 13:
raise ValueError("""Month must be between 1 - 12""" )
UpperCAmelCase_ : str = date_input[2]
# Validate
if sep_a not in ["-", "/"]:
raise ValueError("""Date separator must be '-' or '/'""" )
# Get day
UpperCAmelCase_ : int = int(date_input[3] + date_input[4] )
# Validate
if not 0 < d < 32:
raise ValueError("""Date must be between 1 - 31""" )
# Get second separator
UpperCAmelCase_ : str = date_input[5]
# Validate
if sep_a not in ["-", "/"]:
raise ValueError("""Date separator must be '-' or '/'""" )
# Get year
UpperCAmelCase_ : int = int(date_input[6] + date_input[7] + date_input[8] + date_input[9] )
# Arbitrary year range
if not 45 < y < 8500:
raise ValueError(
"""Year out of range. There has to be some sort of limit...right?""" )
# Get datetime obj for validation
UpperCAmelCase_ : int = datetime.date(int(_a ) , int(_a ) , int(_a ) )
# Start math
if m <= 2:
UpperCAmelCase_ : str = y - 1
UpperCAmelCase_ : Optional[int] = m + 12
# maths var
UpperCAmelCase_ : int = int(str(_a )[:2] )
UpperCAmelCase_ : int = int(str(_a )[2:] )
UpperCAmelCase_ : int = int(2.6 * m - 5.3_9 )
UpperCAmelCase_ : int = int(c / 4 )
UpperCAmelCase_ : int = int(k / 4 )
UpperCAmelCase_ : int = int(d + k )
UpperCAmelCase_ : int = int(t + u + v + x )
UpperCAmelCase_ : int = int(z - (2 * c) )
UpperCAmelCase_ : int = round(w % 7 )
# End math
# Validate math
if f != convert_datetime_days[dt_ck.weekday()]:
raise AssertionError("""The date was evaluated incorrectly. Contact developer.""" )
# Response
UpperCAmelCase_ : str = F'''Your date {date_input}, is a {days[str(_a )]}!'''
return response
if __name__ == "__main__":
import doctest
doctest.testmod()
UpperCamelCase_ = argparse.ArgumentParser(
description=(
'''Find out what day of the week nearly any date is or was. Enter '''
'''date as a string in the mm-dd-yyyy or mm/dd/yyyy format'''
)
)
parser.add_argument(
'''date_input''', type=str, help='''Date as a string (mm-dd-yyyy or mm/dd/yyyy)'''
)
UpperCamelCase_ = parser.parse_args()
zeller(args.date_input)
| 345 |
import collections
import inspect
import unittest
from transformers import FocalNetConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
FocalNetBackbone,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetModel,
)
from transformers.models.focalnet.modeling_focalnet import FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=13 ,lowerCamelCase_: Optional[int]=32 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Optional[Any]=[32, 64, 128] ,lowerCamelCase_: Optional[int]=[1, 2, 1] ,lowerCamelCase_: Union[str, Any]=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[str]=2.0 ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: int=True ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: str=True ,lowerCamelCase_: Dict=10 ,lowerCamelCase_: str=8 ,lowerCamelCase_: Union[str, Any]=["stage1", "stage2"] ,lowerCamelCase_: Optional[Any]=[1, 2] ,) -> str:
UpperCAmelCase_ : List[Any] = parent
UpperCAmelCase_ : Tuple = batch_size
UpperCAmelCase_ : Any = image_size
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : List[str] = num_channels
UpperCAmelCase_ : Dict = embed_dim
UpperCAmelCase_ : Dict = hidden_sizes
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = num_heads
UpperCAmelCase_ : List[Any] = window_size
UpperCAmelCase_ : Union[str, Any] = mlp_ratio
UpperCAmelCase_ : int = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Union[str, Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[int] = drop_path_rate
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : List[Any] = use_absolute_embeddings
UpperCAmelCase_ : List[Any] = patch_norm
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Optional[Any] = scope
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = encoder_stride
UpperCAmelCase_ : Optional[int] = out_features
UpperCAmelCase_ : Optional[int] = out_indices
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : int = None
if self.use_labels:
UpperCAmelCase_ : str = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Any = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Tuple:
return FocalNetConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,hidden_sizes=self.hidden_sizes ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,out_features=self.out_features ,out_indices=self.out_indices ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : Dict = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : Optional[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[str]:
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size, 8, 8] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,len(config.out_features ) )
self.parent.assertListEqual(model.channels ,config.hidden_sizes[:-1] )
# verify backbone works with out_features=None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size * 2, 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,1 )
self.parent.assertListEqual(model.channels ,[config.hidden_sizes[-1]] )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Any = FocalNetForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.reconstruction.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : int = 1
UpperCAmelCase_ : List[str] = FocalNetForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.reconstruction.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ) -> int:
UpperCAmelCase_ : List[Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
# test greyscale images
UpperCAmelCase_ : List[Any] = 1
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = config_and_inputs
UpperCAmelCase_ : int = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (
(
FocalNetModel,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetBackbone,
)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (
{"feature-extraction": FocalNetModel, "image-classification": FocalNetForImageClassification}
if is_torch_available()
else {}
)
A__ : Optional[Any] = False
A__ : Any = False
A__ : List[str] = False
A__ : Any = False
A__ : Any = False
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : Dict = FocalNetModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: List[str] ) -> Union[str, Any]:
return
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCamelCase_ )
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@unittest.skip(reason="""FocalNet does not use inputs_embeds""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""FocalNet does not use feedforward chunking""" )
def A__ ( self: Optional[Any] ) -> Optional[Any]:
pass
def A__ ( self: Optional[Any] ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : List[Any] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Dict = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Any = [*signature.parameters.keys()]
UpperCAmelCase_ : List[str] = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.hidden_states
UpperCAmelCase_ : List[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# FocalNet has a different seq_length
UpperCAmelCase_ : int = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Optional[int] = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Union[str, Any] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = reshaped_hidden_states[0].shape
UpperCAmelCase_ : List[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Union[str, Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Tuple = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : Union[str, Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Union[str, Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Any = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Optional[int] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
@slow
def A__ ( self: Optional[int] ) -> Optional[Any]:
for model_name in FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Tuple = FocalNetModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[Any] = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Optional[int] ) -> str:
# TODO update organization
return AutoImageProcessor.from_pretrained("""microsoft/focalnet-tiny""" ) if is_vision_available() else None
@slow
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification.from_pretrained("""microsoft/focalnet-tiny""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.default_image_processor
UpperCAmelCase_ : Union[str, Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Dict = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : str = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_1_6_6, -0.4_3_6_8, 0.2_1_9_1] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
self.assertTrue(outputs.logits.argmax(dim=-1 ).item() ,281 )
@require_torch
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (FocalNetBackbone,) if is_torch_available() else ()
A__ : int = FocalNetConfig
A__ : List[str] = False
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : str = FocalNetModelTester(self )
| 345 | 1 |
import argparse
import torch
from transformers import (
EncodecConfig,
EncodecFeatureExtractor,
EncodecModel,
logging,
)
# checkpoints downloaded from:
# https://dl.fbaipublicfiles.com/encodec/v0/encodec_24khz-d7cc33bc.th
# https://huggingface.co/facebook/musicgen-small/resolve/main/compression_state_dict.bin
# https://dl.fbaipublicfiles.com/encodec/v0/encodec_48khz-7e698e3e.th
logging.set_verbosity_info()
UpperCamelCase_ = logging.get_logger('''transformers.models.encodec''')
UpperCamelCase_ = {
'''quantizer.vq.layers.*._codebook.inited''': '''quantizer.layers.*.codebook.inited''',
'''quantizer.vq.layers.*._codebook.cluster_size''': '''quantizer.layers.*.codebook.cluster_size''',
'''quantizer.vq.layers.*._codebook.embed''': '''quantizer.layers.*.codebook.embed''',
'''quantizer.vq.layers.*._codebook.embed_avg''': '''quantizer.layers.*.codebook.embed_avg''',
}
UpperCamelCase_ = {
'''encoder.model.0.conv.conv''': '''encoder.layers.0.conv''',
'''encoder.model.1.block.1.conv.conv''': '''encoder.layers.1.block.1.conv''',
'''encoder.model.1.block.3.conv.conv''': '''encoder.layers.1.block.3.conv''',
'''encoder.model.1.shortcut.conv.conv''': '''encoder.layers.1.shortcut.conv''',
'''encoder.model.3.conv.conv''': '''encoder.layers.3.conv''',
'''encoder.model.4.block.1.conv.conv''': '''encoder.layers.4.block.1.conv''',
'''encoder.model.4.block.3.conv.conv''': '''encoder.layers.4.block.3.conv''',
'''encoder.model.4.shortcut.conv.conv''': '''encoder.layers.4.shortcut.conv''',
'''encoder.model.6.conv.conv''': '''encoder.layers.6.conv''',
'''encoder.model.7.block.1.conv.conv''': '''encoder.layers.7.block.1.conv''',
'''encoder.model.7.block.3.conv.conv''': '''encoder.layers.7.block.3.conv''',
'''encoder.model.7.shortcut.conv.conv''': '''encoder.layers.7.shortcut.conv''',
'''encoder.model.9.conv.conv''': '''encoder.layers.9.conv''',
'''encoder.model.10.block.1.conv.conv''': '''encoder.layers.10.block.1.conv''',
'''encoder.model.10.block.3.conv.conv''': '''encoder.layers.10.block.3.conv''',
'''encoder.model.10.shortcut.conv.conv''': '''encoder.layers.10.shortcut.conv''',
'''encoder.model.12.conv.conv''': '''encoder.layers.12.conv''',
'''encoder.model.13.lstm''': '''encoder.layers.13.lstm''',
'''encoder.model.15.conv.conv''': '''encoder.layers.15.conv''',
}
UpperCamelCase_ = {
'''encoder.model.0.conv.norm''': '''encoder.layers.0.norm''',
'''encoder.model.1.block.1.conv.norm''': '''encoder.layers.1.block.1.norm''',
'''encoder.model.1.block.3.conv.norm''': '''encoder.layers.1.block.3.norm''',
'''encoder.model.1.shortcut.conv.norm''': '''encoder.layers.1.shortcut.norm''',
'''encoder.model.3.conv.norm''': '''encoder.layers.3.norm''',
'''encoder.model.4.block.1.conv.norm''': '''encoder.layers.4.block.1.norm''',
'''encoder.model.4.block.3.conv.norm''': '''encoder.layers.4.block.3.norm''',
'''encoder.model.4.shortcut.conv.norm''': '''encoder.layers.4.shortcut.norm''',
'''encoder.model.6.conv.norm''': '''encoder.layers.6.norm''',
'''encoder.model.7.block.1.conv.norm''': '''encoder.layers.7.block.1.norm''',
'''encoder.model.7.block.3.conv.norm''': '''encoder.layers.7.block.3.norm''',
'''encoder.model.7.shortcut.conv.norm''': '''encoder.layers.7.shortcut.norm''',
'''encoder.model.9.conv.norm''': '''encoder.layers.9.norm''',
'''encoder.model.10.block.1.conv.norm''': '''encoder.layers.10.block.1.norm''',
'''encoder.model.10.block.3.conv.norm''': '''encoder.layers.10.block.3.norm''',
'''encoder.model.10.shortcut.conv.norm''': '''encoder.layers.10.shortcut.norm''',
'''encoder.model.12.conv.norm''': '''encoder.layers.12.norm''',
'''encoder.model.15.conv.norm''': '''encoder.layers.15.norm''',
}
UpperCamelCase_ = {
'''decoder.model.0.conv.conv''': '''decoder.layers.0.conv''',
'''decoder.model.1.lstm''': '''decoder.layers.1.lstm''',
'''decoder.model.3.convtr.convtr''': '''decoder.layers.3.conv''',
'''decoder.model.4.block.1.conv.conv''': '''decoder.layers.4.block.1.conv''',
'''decoder.model.4.block.3.conv.conv''': '''decoder.layers.4.block.3.conv''',
'''decoder.model.4.shortcut.conv.conv''': '''decoder.layers.4.shortcut.conv''',
'''decoder.model.6.convtr.convtr''': '''decoder.layers.6.conv''',
'''decoder.model.7.block.1.conv.conv''': '''decoder.layers.7.block.1.conv''',
'''decoder.model.7.block.3.conv.conv''': '''decoder.layers.7.block.3.conv''',
'''decoder.model.7.shortcut.conv.conv''': '''decoder.layers.7.shortcut.conv''',
'''decoder.model.9.convtr.convtr''': '''decoder.layers.9.conv''',
'''decoder.model.10.block.1.conv.conv''': '''decoder.layers.10.block.1.conv''',
'''decoder.model.10.block.3.conv.conv''': '''decoder.layers.10.block.3.conv''',
'''decoder.model.10.shortcut.conv.conv''': '''decoder.layers.10.shortcut.conv''',
'''decoder.model.12.convtr.convtr''': '''decoder.layers.12.conv''',
'''decoder.model.13.block.1.conv.conv''': '''decoder.layers.13.block.1.conv''',
'''decoder.model.13.block.3.conv.conv''': '''decoder.layers.13.block.3.conv''',
'''decoder.model.13.shortcut.conv.conv''': '''decoder.layers.13.shortcut.conv''',
'''decoder.model.15.conv.conv''': '''decoder.layers.15.conv''',
}
UpperCamelCase_ = {
'''decoder.model.0.conv.norm''': '''decoder.layers.0.norm''',
'''decoder.model.3.convtr.norm''': '''decoder.layers.3.norm''',
'''decoder.model.4.block.1.conv.norm''': '''decoder.layers.4.block.1.norm''',
'''decoder.model.4.block.3.conv.norm''': '''decoder.layers.4.block.3.norm''',
'''decoder.model.4.shortcut.conv.norm''': '''decoder.layers.4.shortcut.norm''',
'''decoder.model.6.convtr.norm''': '''decoder.layers.6.norm''',
'''decoder.model.7.block.1.conv.norm''': '''decoder.layers.7.block.1.norm''',
'''decoder.model.7.block.3.conv.norm''': '''decoder.layers.7.block.3.norm''',
'''decoder.model.7.shortcut.conv.norm''': '''decoder.layers.7.shortcut.norm''',
'''decoder.model.9.convtr.norm''': '''decoder.layers.9.norm''',
'''decoder.model.10.block.1.conv.norm''': '''decoder.layers.10.block.1.norm''',
'''decoder.model.10.block.3.conv.norm''': '''decoder.layers.10.block.3.norm''',
'''decoder.model.10.shortcut.conv.norm''': '''decoder.layers.10.shortcut.norm''',
'''decoder.model.12.convtr.norm''': '''decoder.layers.12.norm''',
'''decoder.model.13.block.1.conv.norm''': '''decoder.layers.13.block.1.norm''',
'''decoder.model.13.block.3.conv.norm''': '''decoder.layers.13.block.3.norm''',
'''decoder.model.13.shortcut.conv.norm''': '''decoder.layers.13.shortcut.norm''',
'''decoder.model.15.conv.norm''': '''decoder.layers.15.norm''',
}
UpperCamelCase_ = {
**MAPPING_QUANTIZER,
**MAPPING_ENCODER,
**MAPPING_DECODER,
}
UpperCamelCase_ = {
**MAPPING_QUANTIZER,
**MAPPING_ENCODER,
**MAPPING_ENCODER_48K,
**MAPPING_DECODER,
**MAPPING_DECODER_48K,
}
UpperCamelCase_ = []
UpperCamelCase_ = []
def lowerCamelCase_ ( _a : Any , _a : Any , _a : Tuple , _a : List[Any] , _a : str ):
'''simple docstring'''
for attribute in key.split(""".""" ):
UpperCAmelCase_ : Any = getattr(_a , _a )
if weight_type is not None:
UpperCAmelCase_ : str = getattr(_a , _a ).shape
else:
UpperCAmelCase_ : Optional[int] = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F'''Shape of hf {key + '.' + weight_type if weight_type is not None else ''} is {hf_shape}, but should be'''
F''' {value.shape} for {full_name}''' )
if weight_type == "weight":
UpperCAmelCase_ : List[Any] = value
elif weight_type == "weight_g":
UpperCAmelCase_ : Tuple = value
elif weight_type == "weight_v":
UpperCAmelCase_ : str = value
elif weight_type == "bias":
UpperCAmelCase_ : str = value
elif weight_type == "running_mean":
UpperCAmelCase_ : Optional[Any] = value
elif weight_type == "running_var":
UpperCAmelCase_ : str = value
elif weight_type == "num_batches_tracked":
UpperCAmelCase_ : Optional[Any] = value
elif weight_type == "weight_ih_l0":
UpperCAmelCase_ : Dict = value
elif weight_type == "weight_hh_l0":
UpperCAmelCase_ : Union[str, Any] = value
elif weight_type == "bias_ih_l0":
UpperCAmelCase_ : Any = value
elif weight_type == "bias_hh_l0":
UpperCAmelCase_ : Optional[int] = value
elif weight_type == "weight_ih_l1":
UpperCAmelCase_ : str = value
elif weight_type == "weight_hh_l1":
UpperCAmelCase_ : int = value
elif weight_type == "bias_ih_l1":
UpperCAmelCase_ : int = value
elif weight_type == "bias_hh_l1":
UpperCAmelCase_ : Optional[Any] = value
else:
UpperCAmelCase_ : Dict = value
logger.info(F'''{key + ('.' + weight_type if weight_type is not None else '')} was initialized from {full_name}.''' )
def lowerCamelCase_ ( _a : Optional[int] , _a : int ):
'''simple docstring'''
for key in ignore_keys:
if key.endswith(""".*""" ):
if name.startswith(key[:-1] ):
return True
elif ".*." in key:
UpperCAmelCase_ , UpperCAmelCase_ : str = key.split(""".*.""" )
if prefix in name and suffix in name:
return True
elif key in name:
return True
return False
def lowerCamelCase_ ( _a : List[Any] , _a : List[Any] , _a : Union[str, Any] ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = []
if model_name == "encodec_24khz" or "encodec_32khz":
UpperCAmelCase_ : Tuple = MAPPING_24K
elif model_name == "encodec_48khz":
UpperCAmelCase_ : List[str] = MAPPING_48K
else:
raise ValueError(F'''Unsupported model: {model_name}''' )
for name, value in orig_dict.items():
if should_ignore(_a , _a ):
logger.info(F'''{name} was ignored''' )
continue
UpperCAmelCase_ : List[str] = False
for key, mapped_key in MAPPING.items():
if "*" in key:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = key.split(""".*.""" )
if prefix in name and suffix in name:
UpperCAmelCase_ : Optional[Any] = suffix
if key in name:
# HACK otherwise .embed gets initialized with .embed_avg too
if key.endswith("""embed""" ) and name.endswith("""embed_avg""" ):
continue
UpperCAmelCase_ : int = True
if "*" in mapped_key:
UpperCAmelCase_ : List[str] = name.split(_a )[0].split(""".""" )[-2]
UpperCAmelCase_ : int = mapped_key.replace("""*""" , _a )
if "weight_g" in name:
UpperCAmelCase_ : List[str] = """weight_g"""
elif "weight_v" in name:
UpperCAmelCase_ : Tuple = """weight_v"""
elif "weight_ih_l0" in name:
UpperCAmelCase_ : Optional[Any] = """weight_ih_l0"""
elif "weight_hh_l0" in name:
UpperCAmelCase_ : List[str] = """weight_hh_l0"""
elif "bias_ih_l0" in name:
UpperCAmelCase_ : Dict = """bias_ih_l0"""
elif "bias_hh_l0" in name:
UpperCAmelCase_ : Union[str, Any] = """bias_hh_l0"""
elif "weight_ih_l1" in name:
UpperCAmelCase_ : Union[str, Any] = """weight_ih_l1"""
elif "weight_hh_l1" in name:
UpperCAmelCase_ : str = """weight_hh_l1"""
elif "bias_ih_l1" in name:
UpperCAmelCase_ : Optional[int] = """bias_ih_l1"""
elif "bias_hh_l1" in name:
UpperCAmelCase_ : Any = """bias_hh_l1"""
elif "bias" in name:
UpperCAmelCase_ : Union[str, Any] = """bias"""
elif "weight" in name:
UpperCAmelCase_ : List[Any] = """weight"""
elif "running_mean" in name:
UpperCAmelCase_ : Optional[int] = """running_mean"""
elif "running_var" in name:
UpperCAmelCase_ : Any = """running_var"""
elif "num_batches_tracked" in name:
UpperCAmelCase_ : Tuple = """num_batches_tracked"""
else:
UpperCAmelCase_ : Tuple = None
set_recursively(_a , _a , _a , _a , _a )
continue
if not is_used:
unused_weights.append(_a )
logger.warning(F'''Unused weights: {unused_weights}''' )
@torch.no_grad()
def lowerCamelCase_ ( _a : Union[str, Any] , _a : Any , _a : Optional[Any] , _a : int=None , _a : Optional[int]=None , ):
'''simple docstring'''
if config_path is not None:
UpperCAmelCase_ : List[Any] = EncodecConfig.from_pretrained(_a )
else:
UpperCAmelCase_ : str = EncodecConfig()
if model_name == "encodec_24khz":
pass # config is already correct
elif model_name == "encodec_32khz":
UpperCAmelCase_ : Dict = [8, 5, 4, 4]
UpperCAmelCase_ : Dict = [2.2]
UpperCAmelCase_ : Optional[int] = 64
UpperCAmelCase_ : Tuple = 3_2000
UpperCAmelCase_ : List[str] = 2048
UpperCAmelCase_ : Optional[int] = False
UpperCAmelCase_ : Dict = False
UpperCAmelCase_ : Tuple = False
elif model_name == "encodec_48khz":
UpperCAmelCase_ : Optional[Any] = [8, 5, 4, 2]
UpperCAmelCase_ : str = [3.0, 6.0, 1_2.0, 2_4.0]
UpperCAmelCase_ : int = 4_8000
UpperCAmelCase_ : int = 2
UpperCAmelCase_ : Optional[Any] = False
UpperCAmelCase_ : Optional[int] = """time_group_norm"""
UpperCAmelCase_ : Dict = True
UpperCAmelCase_ : int = 1.0
UpperCAmelCase_ : Any = 0.0_1
else:
raise ValueError(F'''Unknown model name: {model_name}''' )
UpperCAmelCase_ : int = EncodecModel(_a )
UpperCAmelCase_ : Tuple = EncodecFeatureExtractor(
feature_size=config.audio_channels , sampling_rate=config.sampling_rate , chunk_length_s=config.chunk_length_s , overlap=config.overlap , )
feature_extractor.save_pretrained(_a )
UpperCAmelCase_ : Tuple = torch.load(_a )
if "best_state" in original_checkpoint:
# we might have a training state saved, in which case discard the yaml results and just retain the weights
UpperCAmelCase_ : Optional[int] = original_checkpoint["""best_state"""]
recursively_load_weights(_a , _a , _a )
model.save_pretrained(_a )
if repo_id:
print("""Pushing to the hub...""" )
feature_extractor.push_to_hub(_a )
model.push_to_hub(_a )
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
parser.add_argument(
'''--model''',
default='''encodec_24khz''',
type=str,
help='''The model to convert. Should be one of \'encodec_24khz\', \'encodec_32khz\', \'encodec_48khz\'.''',
)
parser.add_argument('''--checkpoint_path''', required=True, default=None, type=str, help='''Path to original checkpoint''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--pytorch_dump_folder_path''', required=True, default=None, type=str, help='''Path to the output PyTorch model.'''
)
parser.add_argument(
'''--push_to_hub''', default=None, type=str, help='''Where to upload the converted model on the 🤗 hub.'''
)
UpperCamelCase_ = parser.parse_args()
convert_checkpoint(
args.model,
args.checkpoint_path,
args.pytorch_dump_folder_path,
args.config_path,
args.push_to_hub,
)
| 345 |
import collections
import inspect
import unittest
from transformers import SwinvaConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import SwinvaForImageClassification, SwinvaForMaskedImageModeling, SwinvaModel
from transformers.models.swinva.modeling_swinva import SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: int=13 ,lowerCamelCase_: int=32 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: Any=3 ,lowerCamelCase_: str=16 ,lowerCamelCase_: Optional[Any]=[1, 2, 1] ,lowerCamelCase_: Tuple=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[Any]=2.0 ,lowerCamelCase_: str=True ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[Any]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Union[str, Any]=False ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=10 ,lowerCamelCase_: Tuple=8 ,) -> List[Any]:
UpperCAmelCase_ : List[str] = parent
UpperCAmelCase_ : int = batch_size
UpperCAmelCase_ : int = image_size
UpperCAmelCase_ : Union[str, Any] = patch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : int = embed_dim
UpperCAmelCase_ : Union[str, Any] = depths
UpperCAmelCase_ : List[str] = num_heads
UpperCAmelCase_ : int = window_size
UpperCAmelCase_ : List[str] = mlp_ratio
UpperCAmelCase_ : Tuple = qkv_bias
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : str = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = drop_path_rate
UpperCAmelCase_ : List[str] = hidden_act
UpperCAmelCase_ : int = use_absolute_embeddings
UpperCAmelCase_ : Any = patch_norm
UpperCAmelCase_ : Optional[int] = layer_norm_eps
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Dict = scope
UpperCAmelCase_ : int = use_labels
UpperCAmelCase_ : Optional[Any] = type_sequence_label_size
UpperCAmelCase_ : List[str] = encoder_stride
def A__ ( self: Any ) -> int:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : List[Any] = None
if self.use_labels:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : str = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Union[str, Any]:
return SwinvaConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,)
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ) -> str:
UpperCAmelCase_ : str = SwinvaModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : List[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Any = SwinvaForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.logits.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : str = 1
UpperCAmelCase_ : Optional[Any] = SwinvaForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : int = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ) -> int:
UpperCAmelCase_ : Union[str, Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = SwinvaForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: str ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = config_and_inputs
UpperCAmelCase_ : Optional[int] = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Tuple = (
(SwinvaModel, SwinvaForImageClassification, SwinvaForMaskedImageModeling) if is_torch_available() else ()
)
A__ : Optional[Any] = (
{"feature-extraction": SwinvaModel, "image-classification": SwinvaForImageClassification}
if is_torch_available()
else {}
)
A__ : List[Any] = False
A__ : Tuple = False
A__ : int = False
A__ : Union[str, Any] = False
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = SwinvaModelTester(self )
UpperCAmelCase_ : str = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 )
def A__ ( self: Optional[int] ) -> List[Any]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: Any ) -> Dict:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
@unittest.skip(reason="""Got `CUDA error: misaligned address` with PyTorch 2.0.0.""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""Swinv2 does not use inputs_embeds""" )
def A__ ( self: Tuple ) -> List[str]:
pass
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : Tuple = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Any = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : int = [*signature.parameters.keys()]
UpperCAmelCase_ : Tuple = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Any = True
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
UpperCAmelCase_ : Union[str, Any] = False
UpperCAmelCase_ : str = True
UpperCAmelCase_ : List[Any] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[Any] = outputs.attentions
UpperCAmelCase_ : List[str] = len(self.model_tester.depths )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
UpperCAmelCase_ : str = True
UpperCAmelCase_ : Optional[Any] = config.window_size**2
UpperCAmelCase_ : Optional[int] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[Any] = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
UpperCAmelCase_ : Optional[Any] = len(lowerCamelCase_ )
# Check attention is always last and order is fine
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = True
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
if hasattr(self.model_tester ,"""num_hidden_states_types""" ):
UpperCAmelCase_ : List[Any] = self.model_tester.num_hidden_states_types
else:
# also another +1 for reshaped_hidden_states
UpperCAmelCase_ : List[str] = 2
self.assertEqual(out_len + added_hidden_states ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
def A__ ( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : int = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[str] = outputs.hidden_states
UpperCAmelCase_ : Optional[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# Swinv2 has a different seq_length
UpperCAmelCase_ : Optional[Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : int = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Optional[int] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = reshaped_hidden_states[0].shape
UpperCAmelCase_ : Optional[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Dict = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes:
UpperCAmelCase_ : Any = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Union[str, Any] = 3
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : List[str] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : List[Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Optional[Any] = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : List[str] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@slow
def A__ ( self: str ) -> Tuple:
for model_name in SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Dict = SwinvaModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : List[str] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and "logit_scale" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Dict ) -> Optional[Any]:
return (
AutoImageProcessor.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" )
if is_vision_available()
else None
)
@slow
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ : Tuple = SwinvaForImageClassification.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" ).to(
lowerCamelCase_ )
UpperCAmelCase_ : Any = self.default_image_processor
UpperCAmelCase_ : List[str] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Optional[int] = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : Dict = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.tensor([-0.3_9_4_7, -0.4_3_0_6, 0.0_0_2_6] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
| 345 | 1 |
import inspect
import tempfile
import unittest
from huggingface_hub import hf_hub_download
from transformers import is_torch_available
from transformers.testing_utils import is_flaky, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
UpperCamelCase_ = 1E-4
if is_torch_available():
import torch
from transformers import AutoformerConfig, AutoformerForPrediction, AutoformerModel
from transformers.models.autoformer.modeling_autoformer import AutoformerDecoder, AutoformerEncoder
@require_torch
class _snake_case :
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict=16 ,lowerCamelCase_: Optional[Any]=13 ,lowerCamelCase_: Dict=7 ,lowerCamelCase_: str=14 ,lowerCamelCase_: Optional[Any]=10 ,lowerCamelCase_: Tuple=19 ,lowerCamelCase_: List[Any]=5 ,lowerCamelCase_: Tuple=4 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Tuple=16 ,lowerCamelCase_: Tuple=2 ,lowerCamelCase_: Optional[int]=4 ,lowerCamelCase_: Union[str, Any]=4 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Dict=[1, 2, 3, 4, 5] ,lowerCamelCase_: Union[str, Any]=25 ,lowerCamelCase_: Any=5 ,) -> str:
UpperCAmelCase_ : Union[str, Any] = d_model
UpperCAmelCase_ : Any = parent
UpperCAmelCase_ : List[Any] = batch_size
UpperCAmelCase_ : Union[str, Any] = prediction_length
UpperCAmelCase_ : str = context_length
UpperCAmelCase_ : List[Any] = cardinality
UpperCAmelCase_ : Dict = num_time_features
UpperCAmelCase_ : Any = lags_sequence
UpperCAmelCase_ : Any = embedding_dimension
UpperCAmelCase_ : Union[str, Any] = is_training
UpperCAmelCase_ : List[str] = hidden_size
UpperCAmelCase_ : int = num_hidden_layers
UpperCAmelCase_ : List[str] = num_attention_heads
UpperCAmelCase_ : Any = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : List[Any] = hidden_dropout_prob
UpperCAmelCase_ : Optional[int] = attention_probs_dropout_prob
UpperCAmelCase_ : Any = context_length
UpperCAmelCase_ : str = prediction_length + label_length
UpperCAmelCase_ : Any = label_length
UpperCAmelCase_ : Union[str, Any] = moving_average
UpperCAmelCase_ : Dict = autocorrelation_factor
def A__ ( self: List[Any] ) -> Union[str, Any]:
return AutoformerConfig(
d_model=self.d_model ,encoder_layers=self.num_hidden_layers ,decoder_layers=self.num_hidden_layers ,encoder_attention_heads=self.num_attention_heads ,decoder_attention_heads=self.num_attention_heads ,encoder_ffn_dim=self.intermediate_size ,decoder_ffn_dim=self.intermediate_size ,dropout=self.hidden_dropout_prob ,attention_dropout=self.attention_probs_dropout_prob ,prediction_length=self.prediction_length ,context_length=self.context_length ,label_length=self.label_length ,lags_sequence=self.lags_sequence ,num_time_features=self.num_time_features ,num_static_categorical_features=1 ,cardinality=[self.cardinality] ,embedding_dimension=[self.embedding_dimension] ,moving_average=self.moving_average ,)
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[str] ) -> Any:
UpperCAmelCase_ : Any = config.context_length + max(config.lags_sequence )
UpperCAmelCase_ : Any = ids_tensor([self.batch_size, 1] ,config.cardinality[0] )
UpperCAmelCase_ : int = floats_tensor([self.batch_size, _past_length, config.num_time_features] )
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, _past_length] )
UpperCAmelCase_ : Dict = floats_tensor([self.batch_size, _past_length] ) > 0.5
# decoder inputs
UpperCAmelCase_ : Optional[Any] = floats_tensor([self.batch_size, config.prediction_length, config.num_time_features] )
UpperCAmelCase_ : List[str] = floats_tensor([self.batch_size, config.prediction_length] )
UpperCAmelCase_ : Dict = {
"""past_values""": past_values,
"""static_categorical_features""": static_categorical_features,
"""past_time_features""": past_time_features,
"""past_observed_mask""": past_observed_mask,
"""future_time_features""": future_time_features,
"""future_values""": future_values,
}
return inputs_dict
def A__ ( self: List[Any] ) -> int:
UpperCAmelCase_ : Union[str, Any] = self.get_config()
UpperCAmelCase_ : Any = self.prepare_autoformer_inputs_dict(lowerCamelCase_ )
return config, inputs_dict
def A__ ( self: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
return config, inputs_dict
def A__ ( self: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : str = AutoformerModel(config=lowerCamelCase_ ).to(lowerCamelCase_ ).eval()
UpperCAmelCase_ : List[str] = model(**lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = outputs.encoder_last_hidden_state
UpperCAmelCase_ : Dict = outputs.last_hidden_state
with tempfile.TemporaryDirectory() as tmpdirname:
UpperCAmelCase_ : Optional[Any] = model.get_encoder()
encoder.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = AutoformerEncoder.from_pretrained(lowerCamelCase_ ).to(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = model.create_network_inputs(**lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = model.decomposition_layer(transformer_inputs[:, : config.context_length, ...] )
UpperCAmelCase_ : Dict = torch.cat(
(transformer_inputs[:, : config.context_length, ...], feature[:, : config.context_length, ...]) ,dim=-1 ,)
UpperCAmelCase_ : Tuple = encoder(inputs_embeds=lowerCamelCase_ )[0]
self.parent.assertTrue((encoder_last_hidden_state_a - encoder_last_hidden_state).abs().max().item() < 1e-3 )
UpperCAmelCase_ : str = (
torch.mean(transformer_inputs[:, : config.context_length, ...] ,dim=1 )
.unsqueeze(1 )
.repeat(1 ,config.prediction_length ,1 )
)
UpperCAmelCase_ : int = torch.zeros(
[transformer_inputs.shape[0], config.prediction_length, transformer_inputs.shape[2]] ,device=enc_input.device ,)
UpperCAmelCase_ : List[str] = torch.cat(
(
torch.cat((seasonal_input[:, -config.label_length :, ...], zeros) ,dim=1 ),
feature[:, config.context_length - config.label_length :, ...],
) ,dim=-1 ,)
UpperCAmelCase_ : List[str] = torch.cat(
(
torch.cat((trend_input[:, -config.label_length :, ...], mean) ,dim=1 ),
feature[:, config.context_length - config.label_length :, ...],
) ,dim=-1 ,)
with tempfile.TemporaryDirectory() as tmpdirname:
UpperCAmelCase_ : Optional[Any] = model.get_decoder()
decoder.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = AutoformerDecoder.from_pretrained(lowerCamelCase_ ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = decoder(
trend=lowerCamelCase_ ,inputs_embeds=lowerCamelCase_ ,encoder_hidden_states=lowerCamelCase_ ,)[0]
self.parent.assertTrue((last_hidden_state_a - last_hidden_state).abs().max().item() < 1e-3 )
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Union[str, Any] = (AutoformerModel, AutoformerForPrediction) if is_torch_available() else ()
A__ : List[Any] = (AutoformerForPrediction,) if is_torch_available() else ()
A__ : List[Any] = {"feature-extraction": AutoformerModel} if is_torch_available() else {}
A__ : Union[str, Any] = False
A__ : int = False
A__ : Optional[Any] = False
A__ : Tuple = False
A__ : Optional[int] = False
A__ : Any = False
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : List[str] = AutoformerModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Optional[Any]:
self.config_tester.run_common_tests()
def A__ ( self: Optional[Any] ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
for model_class in self.all_model_classes:
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = model_class.from_pretrained(lowerCamelCase_ ,output_loading_info=lowerCamelCase_ )
self.assertEqual(info["""missing_keys"""] ,[] )
def A__ ( self: Tuple ) -> Union[str, Any]:
UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
self.model_tester.check_encoder_decoder_model_standalone(*lowerCamelCase_ )
@unittest.skip(reason="""Model has no tokens embeddings""" )
def A__ ( self: Tuple ) -> Any:
pass
def A__ ( self: Dict ) -> Any:
UpperCAmelCase_ : Optional[int] = inspect.signature(getattr(lowerCamelCase_ ,"""forward""" ) )
# The main input is the name of the argument after `self`
UpperCAmelCase_ : Dict = list(model_signature.parameters.keys() )[1]
self.assertEqual(AutoformerModel.main_input_name ,lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[str] = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Optional[int] = [*signature.parameters.keys()]
UpperCAmelCase_ : Tuple = [
"""past_values""",
"""past_time_features""",
"""past_observed_mask""",
"""static_categorical_features""",
"""static_real_features""",
"""future_values""",
"""future_time_features""",
]
if model.__class__.__name__ in ["AutoformerForPrediction"]:
expected_arg_names.append("""future_observed_mask""" )
expected_arg_names.extend(
[
"""decoder_attention_mask""",
"""head_mask""",
"""decoder_head_mask""",
"""cross_attn_head_mask""",
"""encoder_outputs""",
"""past_key_values""",
"""output_hidden_states""",
"""output_attentions""",
"""use_cache""",
"""return_dict""",
] )
self.assertListEqual(arg_names[: len(lowerCamelCase_ )] ,lowerCamelCase_ )
def A__ ( self: List[Any] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = True
UpperCAmelCase_ : Union[str, Any] = getattr(self.model_tester ,"""seq_length""" ,lowerCamelCase_ )
UpperCAmelCase_ : Dict = getattr(self.model_tester ,"""decoder_seq_length""" ,lowerCamelCase_ )
UpperCAmelCase_ : Dict = getattr(self.model_tester ,"""encoder_seq_length""" ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = getattr(self.model_tester ,"""d_model""" ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = getattr(self.model_tester ,"""num_attention_heads""" ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = d_model // num_attention_heads
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[int] = True
UpperCAmelCase_ : str = False
UpperCAmelCase_ : str = True
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Dict = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Union[str, Any] = outputs.encoder_attentions
self.assertEqual(len(lowerCamelCase_ ) ,self.model_tester.num_hidden_layers )
self.assertListEqual(
list(attentions[0].shape[-3:] ) ,[self.model_tester.num_attention_heads, encoder_seq_length, dim] ,)
UpperCAmelCase_ : Any = len(lowerCamelCase_ )
UpperCAmelCase_ : Dict = 7
if "last_hidden_state" in outputs:
correct_outlen += 1
if "trend" in outputs:
correct_outlen += 1
if "past_key_values" in outputs:
correct_outlen += 1 # past_key_values have been returned
if "loss" in outputs:
correct_outlen += 1
if "params" in outputs:
correct_outlen += 1
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
# decoder attentions
UpperCAmelCase_ : List[Any] = outputs.decoder_attentions
self.assertIsInstance(lowerCamelCase_ ,(list, tuple) )
self.assertEqual(len(lowerCamelCase_ ) ,self.model_tester.num_hidden_layers )
self.assertListEqual(
list(decoder_attentions[0].shape[-3:] ) ,[self.model_tester.num_attention_heads, decoder_seq_length, dim] ,)
# cross attentions
UpperCAmelCase_ : List[Any] = outputs.cross_attentions
self.assertIsInstance(lowerCamelCase_ ,(list, tuple) )
self.assertEqual(len(lowerCamelCase_ ) ,self.model_tester.num_hidden_layers )
self.assertListEqual(
list(cross_attentions[0].shape[-3:] ) ,[self.model_tester.num_attention_heads, decoder_seq_length, dim] ,)
# Check attention is always last and order is fine
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
self.assertEqual(out_len + 2 ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,self.model_tester.num_hidden_layers )
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) ,[self.model_tester.num_attention_heads, encoder_seq_length, dim] ,)
@is_flaky()
def A__ ( self: str ) -> Optional[int]:
super().test_retain_grad_hidden_states_attentions()
def lowerCamelCase_ ( _a : Optional[int]="train-batch.pt" ):
'''simple docstring'''
UpperCAmelCase_ : Optional[Any] = hf_hub_download(repo_id="""hf-internal-testing/tourism-monthly-batch""" , filename=_a , repo_type="""dataset""" )
UpperCAmelCase_ : Any = torch.load(_a , map_location=_a )
return batch
@require_torch
@slow
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[str] = AutoformerModel.from_pretrained("""huggingface/autoformer-tourism-monthly""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = prepare_batch()
with torch.no_grad():
UpperCAmelCase_ : str = model(
past_values=batch["""past_values"""] ,past_time_features=batch["""past_time_features"""] ,past_observed_mask=batch["""past_observed_mask"""] ,static_categorical_features=batch["""static_categorical_features"""] ,future_values=batch["""future_values"""] ,future_time_features=batch["""future_time_features"""] ,)[0]
UpperCAmelCase_ : Optional[Any] = torch.Size(
(64, model.config.prediction_length + model.config.label_length, model.config.feature_size) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Dict = torch.tensor(
[[0.3_5_9_3, -1.3_3_9_8, 0.6_3_3_0], [0.2_2_7_9, 1.5_3_9_6, -0.1_7_9_2], [0.0_4_5_0, 1.3_2_2_5, -0.2_3_3_5]] ,device=lowerCamelCase_ )
self.assertTrue(torch.allclose(output[0, :3, :3] ,lowerCamelCase_ ,atol=lowerCamelCase_ ) )
def A__ ( self: Tuple ) -> List[str]:
UpperCAmelCase_ : List[str] = AutoformerForPrediction.from_pretrained("""huggingface/autoformer-tourism-monthly""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = prepare_batch("""val-batch.pt""" )
with torch.no_grad():
UpperCAmelCase_ : List[Any] = model(
past_values=batch["""past_values"""] ,past_time_features=batch["""past_time_features"""] ,past_observed_mask=batch["""past_observed_mask"""] ,static_categorical_features=batch["""static_categorical_features"""] ,).encoder_last_hidden_state
UpperCAmelCase_ : Optional[int] = torch.Size((64, model.config.context_length, model.config.d_model) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = torch.tensor(
[[-0.0_7_3_4, -0.9_0_3_6, 0.8_3_5_8], [4.7_1_8_6, 2.4_1_1_3, 1.9_5_8_1], [1.7_9_5_3, 2.3_5_5_8, 1.2_9_7_0]] ,device=lowerCamelCase_ )
self.assertTrue(torch.allclose(output[0, :3, :3] ,lowerCamelCase_ ,atol=lowerCamelCase_ ) )
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Any = AutoformerForPrediction.from_pretrained("""huggingface/autoformer-tourism-monthly""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = prepare_batch("""val-batch.pt""" )
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.generate(
static_categorical_features=batch["""static_categorical_features"""] ,past_time_features=batch["""past_time_features"""] ,past_values=batch["""past_values"""] ,future_time_features=batch["""future_time_features"""] ,past_observed_mask=batch["""past_observed_mask"""] ,)
UpperCAmelCase_ : Optional[Any] = torch.Size((64, model.config.num_parallel_samples, model.config.prediction_length) )
self.assertEqual(outputs.sequences.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([3_1_3_0.6_7_6_3, 4_0_5_6.5_2_9_3, 7_0_5_3.0_7_8_6] ,device=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = outputs.sequences.mean(dim=1 )
self.assertTrue(torch.allclose(mean_prediction[0, -3:] ,lowerCamelCase_ ,rtol=1e-1 ) )
| 345 |
import json
import os
from functools import lru_cache
from typing import Dict, List, Optional, Tuple, Union
import regex as re
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...tokenization_utils_base import BatchEncoding, EncodedInput
from ...utils import PaddingStrategy, logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {'''vocab_file''': '''vocab.json''', '''merges_file''': '''merges.txt'''}
# See all LED models at https://huggingface.co/models?filter=LED
UpperCamelCase_ = {
'''vocab_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/vocab.json''',
},
'''merges_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/merges.txt''',
},
'''tokenizer_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/tokenizer.json''',
},
}
UpperCamelCase_ = {
'''allenai/led-base-16384''': 16384,
}
@lru_cache()
# Copied from transformers.models.bart.tokenization_bart.bytes_to_unicode
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : int = (
list(range(ord("""!""" ) , ord("""~""" ) + 1 ) ) + list(range(ord("""¡""" ) , ord("""¬""" ) + 1 ) ) + list(range(ord("""®""" ) , ord("""ÿ""" ) + 1 ) )
)
UpperCAmelCase_ : Dict = bs[:]
UpperCAmelCase_ : Any = 0
for b in range(2**8 ):
if b not in bs:
bs.append(_a )
cs.append(2**8 + n )
n += 1
UpperCAmelCase_ : Any = [chr(_a ) for n in cs]
return dict(zip(_a , _a ) )
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = set()
UpperCAmelCase_ : List[Any] = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
UpperCAmelCase_ : Optional[int] = char
return pairs
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = VOCAB_FILES_NAMES
A__ : List[str] = PRETRAINED_VOCAB_FILES_MAP
A__ : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Optional[int] = ["input_ids", "attention_mask"]
def __init__( self: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any]="replace" ,lowerCamelCase_: Optional[Any]="<s>" ,lowerCamelCase_: List[Any]="</s>" ,lowerCamelCase_: List[str]="</s>" ,lowerCamelCase_: int="<s>" ,lowerCamelCase_: int="<unk>" ,lowerCamelCase_: str="<pad>" ,lowerCamelCase_: Optional[Any]="<mask>" ,lowerCamelCase_: List[str]=False ,**lowerCamelCase_: Tuple ,) -> Any:
UpperCAmelCase_ : Union[str, Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else bos_token
UpperCAmelCase_ : int = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else eos_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else sep_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else cls_token
UpperCAmelCase_ : Optional[Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else unk_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else pad_token
# Mask token behave like a normal word, i.e. include the space before it
UpperCAmelCase_ : str = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else mask_token
super().__init__(
errors=lowerCamelCase_ ,bos_token=lowerCamelCase_ ,eos_token=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,add_prefix_space=lowerCamelCase_ ,**lowerCamelCase_ ,)
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as vocab_handle:
UpperCAmelCase_ : Union[str, Any] = json.load(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {v: k for k, v in self.encoder.items()}
UpperCAmelCase_ : Any = errors # how to handle errors in decoding
UpperCAmelCase_ : int = bytes_to_unicode()
UpperCAmelCase_ : Dict = {v: k for k, v in self.byte_encoder.items()}
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as merges_handle:
UpperCAmelCase_ : Any = merges_handle.read().split("""\n""" )[1:-1]
UpperCAmelCase_ : int = [tuple(merge.split() ) for merge in bpe_merges]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : Tuple = {}
UpperCAmelCase_ : Optional[int] = add_prefix_space
# Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions
UpperCAmelCase_ : int = re.compile(R"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""" )
@property
# Copied from transformers.models.bart.tokenization_bart.BartTokenizer.vocab_size
def A__ ( self: List[str] ) -> List[str]:
return len(self.encoder )
def A__ ( self: Any ) -> Union[str, Any]:
return dict(self.encoder ,**self.added_tokens_encoder )
def A__ ( self: Tuple ,lowerCamelCase_: Dict ) -> Optional[Any]:
if token in self.cache:
return self.cache[token]
UpperCAmelCase_ : Union[str, Any] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = get_pairs(lowerCamelCase_ )
if not pairs:
return token
while True:
UpperCAmelCase_ : Union[str, Any] = min(lowerCamelCase_ ,key=lambda lowerCamelCase_ : self.bpe_ranks.get(lowerCamelCase_ ,float("""inf""" ) ) )
if bigram not in self.bpe_ranks:
break
UpperCAmelCase_ , UpperCAmelCase_ : Any = bigram
UpperCAmelCase_ : Optional[Any] = []
UpperCAmelCase_ : List[str] = 0
while i < len(lowerCamelCase_ ):
try:
UpperCAmelCase_ : str = word.index(lowerCamelCase_ ,lowerCamelCase_ )
except ValueError:
new_word.extend(word[i:] )
break
else:
new_word.extend(word[i:j] )
UpperCAmelCase_ : Union[str, Any] = j
if word[i] == first and i < len(lowerCamelCase_ ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
UpperCAmelCase_ : List[str] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = new_word
if len(lowerCamelCase_ ) == 1:
break
else:
UpperCAmelCase_ : List[str] = get_pairs(lowerCamelCase_ )
UpperCAmelCase_ : int = """ """.join(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = word
return word
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> List[str]:
UpperCAmelCase_ : str = []
for token in re.findall(self.pat ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = """""".join(
self.byte_encoder[b] for b in token.encode("""utf-8""" ) ) # Maps all our bytes to unicode strings, avoiding control tokens of the BPE (spaces in our case)
bpe_tokens.extend(bpe_token for bpe_token in self.bpe(lowerCamelCase_ ).split(""" """ ) )
return bpe_tokens
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ) -> Optional[int]:
return self.encoder.get(lowerCamelCase_ ,self.encoder.get(self.unk_token ) )
def A__ ( self: List[str] ,lowerCamelCase_: str ) -> Optional[Any]:
return self.decoder.get(lowerCamelCase_ )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ) -> List[Any]:
UpperCAmelCase_ : str = """""".join(lowerCamelCase_ )
UpperCAmelCase_ : int = bytearray([self.byte_decoder[c] for c in text] ).decode("""utf-8""" ,errors=self.errors )
return text
def A__ ( self: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCamelCase_ ):
logger.error(F'''Vocabulary path ({save_directory}) should be a directory''' )
return
UpperCAmelCase_ : List[Any] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : List[str] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""merges_file"""] )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as f:
f.write(json.dumps(self.encoder ,indent=2 ,sort_keys=lowerCamelCase_ ,ensure_ascii=lowerCamelCase_ ) + """\n""" )
UpperCAmelCase_ : str = 0
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as writer:
writer.write("""#version: 0.2\n""" )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() ,key=lambda lowerCamelCase_ : kv[1] ):
if index != token_index:
logger.warning(
F'''Saving vocabulary to {merge_file}: BPE merge indices are not consecutive.'''
""" Please check that the tokenizer is not corrupted!""" )
UpperCAmelCase_ : Tuple = token_index
writer.write(""" """.join(lowerCamelCase_ ) + """\n""" )
index += 1
return vocab_file, merge_file
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
UpperCAmelCase_ : int = [self.cls_token_id]
UpperCAmelCase_ : Optional[int] = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ,lowerCamelCase_: bool = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=lowerCamelCase_ ,token_ids_a=lowerCamelCase_ ,already_has_special_tokens=lowerCamelCase_ )
if token_ids_a is None:
return [1] + ([0] * len(lowerCamelCase_ )) + [1]
return [1] + ([0] * len(lowerCamelCase_ )) + [1, 1] + ([0] * len(lowerCamelCase_ )) + [1]
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Optional[Any] = [self.sep_token_id]
UpperCAmelCase_ : Optional[Any] = [self.cls_token_id]
if token_ids_a is None:
return len(cls + token_ids_a + sep ) * [0]
return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0]
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str=False ,**lowerCamelCase_: List[str] ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = kwargs.pop("""add_prefix_space""" ,self.add_prefix_space )
if (is_split_into_words or add_prefix_space) and (len(lowerCamelCase_ ) > 0 and not text[0].isspace()):
UpperCAmelCase_ : Dict = """ """ + text
return (text, kwargs)
def A__ ( self: List[str] ,lowerCamelCase_: Union[Dict[str, EncodedInput], BatchEncoding] ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: PaddingStrategy = PaddingStrategy.DO_NOT_PAD ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: Optional[bool] = None ,) -> dict:
UpperCAmelCase_ : Optional[int] = super()._pad(
encoded_inputs=lowerCamelCase_ ,max_length=lowerCamelCase_ ,padding_strategy=lowerCamelCase_ ,pad_to_multiple_of=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,)
# Load from model defaults
if return_attention_mask is None:
UpperCAmelCase_ : str = """attention_mask""" in self.model_input_names
if return_attention_mask and "global_attention_mask" in encoded_inputs:
UpperCAmelCase_ : str = encoded_inputs[self.model_input_names[0]]
# `global_attention_mask` need to have the same length as other (sequential) inputs.
UpperCAmelCase_ : List[Any] = len(encoded_inputs["""global_attention_mask"""] ) != len(lowerCamelCase_ )
if needs_to_be_padded:
UpperCAmelCase_ : Dict = len(lowerCamelCase_ ) - len(encoded_inputs["""global_attention_mask"""] )
if self.padding_side == "right":
# Use `-1` since `0` in `global_attention_mask` means `local attention` instead of `not to attend`
UpperCAmelCase_ : str = (
encoded_inputs["""global_attention_mask"""] + [-1] * difference
)
elif self.padding_side == "left":
UpperCAmelCase_ : List[str] = [-1] * difference + encoded_inputs[
"""global_attention_mask"""
]
else:
raise ValueError("""Invalid padding strategy:""" + str(self.padding_side ) )
return encoded_inputs
| 345 | 1 |
from typing import Dict, List, Optional, Tuple, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_torch_available, is_torch_tensor, logging
if is_torch_available():
import torch
UpperCamelCase_ = logging.get_logger(__name__)
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : List[Any] = ["pixel_values"]
def __init__( self: List[str] ,lowerCamelCase_: bool = True ,lowerCamelCase_: Optional[Dict[str, int]] = None ,lowerCamelCase_: PILImageResampling = PILImageResampling.BILINEAR ,lowerCamelCase_: bool = True ,lowerCamelCase_: Dict[str, int] = None ,lowerCamelCase_: bool = True ,lowerCamelCase_: Union[int, float] = 1 / 255 ,lowerCamelCase_: bool = True ,lowerCamelCase_: Optional[Union[float, List[float]]] = None ,lowerCamelCase_: Optional[Union[float, List[float]]] = None ,**lowerCamelCase_: List[str] ,) -> None:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : List[str] = size if size is not None else {"""shortest_edge""": 256}
UpperCAmelCase_ : List[Any] = get_size_dict(lowerCamelCase_ ,default_to_square=lowerCamelCase_ )
UpperCAmelCase_ : str = crop_size if crop_size is not None else {"""height""": 224, """width""": 224}
UpperCAmelCase_ : Dict = get_size_dict(lowerCamelCase_ ,param_name="""crop_size""" )
UpperCAmelCase_ : List[Any] = do_resize
UpperCAmelCase_ : Dict = size
UpperCAmelCase_ : Optional[int] = resample
UpperCAmelCase_ : str = do_center_crop
UpperCAmelCase_ : Union[str, Any] = crop_size
UpperCAmelCase_ : Optional[Any] = do_rescale
UpperCAmelCase_ : int = rescale_factor
UpperCAmelCase_ : Union[str, Any] = do_normalize
UpperCAmelCase_ : Tuple = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
UpperCAmelCase_ : List[Any] = image_std if image_std is not None else IMAGENET_STANDARD_STD
def A__ ( self: Tuple ,lowerCamelCase_: np.ndarray ,lowerCamelCase_: Dict[str, int] ,lowerCamelCase_: PILImageResampling = PILImageResampling.BICUBIC ,lowerCamelCase_: Optional[Union[str, ChannelDimension]] = None ,**lowerCamelCase_: List[str] ,) -> np.ndarray:
UpperCAmelCase_ : Optional[int] = get_size_dict(lowerCamelCase_ ,default_to_square=lowerCamelCase_ )
if "shortest_edge" not in size:
raise ValueError(F'''The `size` parameter must contain the key `shortest_edge`. Got {size.keys()}''' )
UpperCAmelCase_ : List[Any] = get_resize_output_image_size(lowerCamelCase_ ,size=size["""shortest_edge"""] ,default_to_square=lowerCamelCase_ )
return resize(lowerCamelCase_ ,size=lowerCamelCase_ ,resample=lowerCamelCase_ ,data_format=lowerCamelCase_ ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: np.ndarray ,lowerCamelCase_: Dict[str, int] ,lowerCamelCase_: Optional[Union[str, ChannelDimension]] = None ,**lowerCamelCase_: List[Any] ,) -> np.ndarray:
UpperCAmelCase_ : Optional[int] = get_size_dict(lowerCamelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(F'''The `size` parameter must contain the keys `height` and `width`. Got {size.keys()}''' )
return center_crop(lowerCamelCase_ ,size=(size["""height"""], size["""width"""]) ,data_format=lowerCamelCase_ ,**lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: np.ndarray ,lowerCamelCase_: float ,lowerCamelCase_: Optional[Union[str, ChannelDimension]] = None ,**lowerCamelCase_: int ) -> np.ndarray:
return rescale(lowerCamelCase_ ,scale=lowerCamelCase_ ,data_format=lowerCamelCase_ ,**lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: np.ndarray ,lowerCamelCase_: Union[float, List[float]] ,lowerCamelCase_: Union[float, List[float]] ,lowerCamelCase_: Optional[Union[str, ChannelDimension]] = None ,**lowerCamelCase_: int ,) -> np.ndarray:
return normalize(lowerCamelCase_ ,mean=lowerCamelCase_ ,std=lowerCamelCase_ ,data_format=lowerCamelCase_ ,**lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: ImageInput ,lowerCamelCase_: Optional[bool] = None ,lowerCamelCase_: Dict[str, int] = None ,lowerCamelCase_: PILImageResampling = None ,lowerCamelCase_: bool = None ,lowerCamelCase_: Dict[str, int] = None ,lowerCamelCase_: Optional[bool] = None ,lowerCamelCase_: Optional[float] = None ,lowerCamelCase_: Optional[bool] = None ,lowerCamelCase_: Optional[Union[float, List[float]]] = None ,lowerCamelCase_: Optional[Union[float, List[float]]] = None ,lowerCamelCase_: Optional[Union[str, TensorType]] = None ,lowerCamelCase_: Union[str, ChannelDimension] = ChannelDimension.FIRST ,**lowerCamelCase_: Dict ,) -> int:
UpperCAmelCase_ : str = do_resize if do_resize is not None else self.do_resize
UpperCAmelCase_ : List[Any] = size if size is not None else self.size
UpperCAmelCase_ : Dict = get_size_dict(lowerCamelCase_ ,default_to_square=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resample if resample is not None else self.resample
UpperCAmelCase_ : Union[str, Any] = do_center_crop if do_center_crop is not None else self.do_center_crop
UpperCAmelCase_ : str = crop_size if crop_size is not None else self.crop_size
UpperCAmelCase_ : List[str] = get_size_dict(lowerCamelCase_ ,param_name="""crop_size""" )
UpperCAmelCase_ : Optional[Any] = do_rescale if do_rescale is not None else self.do_rescale
UpperCAmelCase_ : Any = rescale_factor if rescale_factor is not None else self.rescale_factor
UpperCAmelCase_ : str = do_normalize if do_normalize is not None else self.do_normalize
UpperCAmelCase_ : List[str] = image_mean if image_mean is not None else self.image_mean
UpperCAmelCase_ : List[Any] = image_std if image_std is not None else self.image_std
UpperCAmelCase_ : List[Any] = make_list_of_images(lowerCamelCase_ )
if not valid_images(lowerCamelCase_ ):
raise ValueError(
"""Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, """
"""torch.Tensor, tf.Tensor or jax.ndarray.""" )
if do_resize and size is None:
raise ValueError("""Size must be specified if do_resize is True.""" )
if do_center_crop and crop_size is None:
raise ValueError("""Crop size must be specified if do_center_crop is True.""" )
if do_rescale and rescale_factor is None:
raise ValueError("""Rescale factor must be specified if do_rescale is True.""" )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError("""Image mean and std must be specified if do_normalize is True.""" )
# All transformations expect numpy arrays.
UpperCAmelCase_ : Dict = [to_numpy_array(lowerCamelCase_ ) for image in images]
if do_resize:
UpperCAmelCase_ : List[Any] = [self.resize(image=lowerCamelCase_ ,size=lowerCamelCase_ ,resample=lowerCamelCase_ ) for image in images]
if do_center_crop:
UpperCAmelCase_ : Optional[Any] = [self.center_crop(image=lowerCamelCase_ ,size=lowerCamelCase_ ) for image in images]
if do_rescale:
UpperCAmelCase_ : str = [self.rescale(image=lowerCamelCase_ ,scale=lowerCamelCase_ ) for image in images]
if do_normalize:
UpperCAmelCase_ : Optional[Any] = [self.normalize(image=lowerCamelCase_ ,mean=lowerCamelCase_ ,std=lowerCamelCase_ ) for image in images]
UpperCAmelCase_ : int = [to_channel_dimension_format(lowerCamelCase_ ,lowerCamelCase_ ) for image in images]
UpperCAmelCase_ : Optional[int] = {"""pixel_values""": images}
return BatchFeature(data=lowerCamelCase_ ,tensor_type=lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Tuple] = None ) -> Optional[int]:
UpperCAmelCase_ : List[str] = outputs.logits
# Resize logits and compute semantic segmentation maps
if target_sizes is not None:
if len(lowerCamelCase_ ) != len(lowerCamelCase_ ):
raise ValueError(
"""Make sure that you pass in as many target sizes as the batch dimension of the logits""" )
if is_torch_tensor(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = target_sizes.numpy()
UpperCAmelCase_ : List[str] = []
for idx in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : str = torch.nn.functional.interpolate(
logits[idx].unsqueeze(dim=0 ) ,size=target_sizes[idx] ,mode="""bilinear""" ,align_corners=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = resized_logits[0].argmax(dim=0 )
semantic_segmentation.append(lowerCamelCase_ )
else:
UpperCAmelCase_ : Union[str, Any] = logits.argmax(dim=1 )
UpperCAmelCase_ : str = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )]
return semantic_segmentation
| 345 |
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Optional[Any] = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[Any] ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : int = UNetaDConditionModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=4 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,)
return model
@property
def A__ ( self: str ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = AutoencoderKL(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Tuple:
def extract(*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: str ):
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = torch.ones([0] )
def A__ ( self: List[Any] ,lowerCamelCase_: str ) -> int:
self.pixel_values.to(lowerCamelCase_ )
return self
return Out()
return extract
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : int = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : int = self.dummy_cond_unet
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(
beta_start=0.0_0_0_8_5 ,beta_end=0.0_1_2 ,beta_schedule="""scaled_linear""" ,clip_sample=lowerCamelCase_ ,set_alpha_to_one=lowerCamelCase_ ,)
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : int = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : List[Any] = output.images
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Dict = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : int = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Dict = self.dummy_cond_unet
UpperCAmelCase_ : List[Any] = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : Union[str, Any] = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : str = output.images
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(
"""hf-internal-testing/tiny-stable-diffusion-lms-pipe""" ,safety_checker=lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(pipe.scheduler ,lowerCamelCase_ )
assert pipe.safety_checker is None
UpperCAmelCase_ : List[Any] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(lowerCamelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
UpperCAmelCase_ : Optional[int] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Tuple = self.dummy_cond_unet
UpperCAmelCase_ : Dict = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# put models in fp16
UpperCAmelCase_ : Optional[Any] = unet.half()
UpperCAmelCase_ : Optional[int] = vae.half()
UpperCAmelCase_ : int = bert.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = sd_pipe([prompt] ,num_inference_steps=2 ,output_type="""np""" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Tuple = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : str = (
"""portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"""
""" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"""
""" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"""
""" children from bahnhof zoo, detailed """
)
UpperCAmelCase_ : Optional[int] = 4003660346
UpperCAmelCase_ : int = 7
# without safety guidance (sld_guidance_scale = 0)
UpperCAmelCase_ : Dict = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Optional[int] = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Dict = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
# without safety guidance (strong configuration)
UpperCAmelCase_ : Union[str, Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Any = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Tuple = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : str = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Any = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = """padme amidala taking a bath artwork, safe for work, no nudity"""
UpperCAmelCase_ : List[Any] = 2734971755
UpperCAmelCase_ : Optional[Any] = 7
UpperCAmelCase_ : int = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : Tuple = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
UpperCAmelCase_ : Any = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : List[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" )
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = (
"""the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."""
""" leyendecker"""
)
UpperCAmelCase_ : Optional[Any] = 1044355234
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : List[Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Any = output.images
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-7
UpperCAmelCase_ : Optional[int] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Any = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Any = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 345 | 1 |
import os
import tempfile
import unittest
from transformers import NezhaConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_torch, require_torch_gpu, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
NezhaForMaskedLM,
NezhaForMultipleChoice,
NezhaForNextSentencePrediction,
NezhaForPreTraining,
NezhaForQuestionAnswering,
NezhaForSequenceClassification,
NezhaForTokenClassification,
NezhaModel,
)
from transformers.models.nezha.modeling_nezha import NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST
class _snake_case :
'''simple docstring'''
def __init__( self: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any]=13 ,lowerCamelCase_: Union[str, Any]=7 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Optional[Any]=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=99 ,lowerCamelCase_: str=32 ,lowerCamelCase_: List[str]=5 ,lowerCamelCase_: int=4 ,lowerCamelCase_: Dict=37 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Optional[Any]=0.1 ,lowerCamelCase_: List[Any]=0.1 ,lowerCamelCase_: Optional[Any]=128 ,lowerCamelCase_: List[str]=32 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Any=2 ,lowerCamelCase_: int=0.0_2 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Dict=4 ,lowerCamelCase_: List[Any]=None ,) -> Any:
UpperCAmelCase_ : List[str] = parent
UpperCAmelCase_ : str = batch_size
UpperCAmelCase_ : str = seq_length
UpperCAmelCase_ : List[str] = is_training
UpperCAmelCase_ : Optional[int] = use_input_mask
UpperCAmelCase_ : Union[str, Any] = use_token_type_ids
UpperCAmelCase_ : Any = use_labels
UpperCAmelCase_ : int = vocab_size
UpperCAmelCase_ : Optional[Any] = hidden_size
UpperCAmelCase_ : Dict = num_hidden_layers
UpperCAmelCase_ : Tuple = num_attention_heads
UpperCAmelCase_ : Any = intermediate_size
UpperCAmelCase_ : Optional[int] = hidden_act
UpperCAmelCase_ : Union[str, Any] = hidden_dropout_prob
UpperCAmelCase_ : Any = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[Any] = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = initializer_range
UpperCAmelCase_ : Optional[int] = num_labels
UpperCAmelCase_ : Dict = num_choices
UpperCAmelCase_ : int = scope
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Dict = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : Any = None
if self.use_input_mask:
UpperCAmelCase_ : Any = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : Dict = None
if self.use_token_type_ids:
UpperCAmelCase_ : Dict = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : List[str] = None
UpperCAmelCase_ : Tuple = None
UpperCAmelCase_ : List[str] = None
if self.use_labels:
UpperCAmelCase_ : List[str] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : List[Any] = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : int = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: Optional[int] ) -> str:
return NezhaConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: Any ) -> Tuple:
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Tuple = self.prepare_config_and_inputs()
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
UpperCAmelCase_ : List[Any] = ids_tensor([self.batch_size, self.seq_length] ,vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def A__ ( self: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: str ,lowerCamelCase_: str ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = NezhaModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Any = model(lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: int ,lowerCamelCase_: Any ,lowerCamelCase_: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[int] ,) -> str:
UpperCAmelCase_ : Union[str, Any] = True
UpperCAmelCase_ : str = NezhaModel(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,encoder_hidden_states=lowerCamelCase_ ,encoder_attention_mask=lowerCamelCase_ ,)
UpperCAmelCase_ : Optional[Any] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,encoder_hidden_states=lowerCamelCase_ ,)
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: str ,lowerCamelCase_: List[str] ,lowerCamelCase_: int ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: int ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = NezhaForMaskedLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = NezhaForNextSentencePrediction(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 2) )
def A__ ( self: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : Tuple = NezhaForPreTraining(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,next_sentence_label=lowerCamelCase_ ,)
self.parent.assertEqual(result.prediction_logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape ,(self.batch_size, 2) )
def A__ ( self: Optional[int] ,lowerCamelCase_: int ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ) -> Tuple:
UpperCAmelCase_ : List[Any] = NezhaForQuestionAnswering(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,start_positions=lowerCamelCase_ ,end_positions=lowerCamelCase_ ,)
self.parent.assertEqual(result.start_logits.shape ,(self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape ,(self.batch_size, self.seq_length) )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: int ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ) -> Tuple:
UpperCAmelCase_ : str = self.num_labels
UpperCAmelCase_ : List[Any] = NezhaForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : str = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_labels) )
def A__ ( self: Any ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Optional[int] ) -> Dict:
UpperCAmelCase_ : Optional[int] = self.num_labels
UpperCAmelCase_ : Optional[Any] = NezhaForTokenClassification(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Any ,lowerCamelCase_: str ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.num_choices
UpperCAmelCase_ : Optional[int] = NezhaForMultipleChoice(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = input_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Optional[Any] = token_type_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : str = input_mask.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_choices) )
def A__ ( self: Any ) -> Optional[Any]:
UpperCAmelCase_ : Optional[int] = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : List[str] = config_and_inputs
UpperCAmelCase_ : Any = {"""input_ids""": input_ids, """token_type_ids""": token_type_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = (
(
NezhaModel,
NezhaForMaskedLM,
NezhaForMultipleChoice,
NezhaForNextSentencePrediction,
NezhaForPreTraining,
NezhaForQuestionAnswering,
NezhaForSequenceClassification,
NezhaForTokenClassification,
)
if is_torch_available()
else ()
)
A__ : List[str] = (
{
"feature-extraction": NezhaModel,
"fill-mask": NezhaForMaskedLM,
"question-answering": NezhaForQuestionAnswering,
"text-classification": NezhaForSequenceClassification,
"token-classification": NezhaForTokenClassification,
"zero-shot": NezhaForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : List[str] = True
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: List[Any]=False ) -> Any:
UpperCAmelCase_ : Tuple = super()._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ,return_labels=lowerCamelCase_ )
if return_labels:
if model_class in get_values(lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : str = torch.zeros(
self.model_tester.batch_size ,dtype=torch.long ,device=lowerCamelCase_ )
return inputs_dict
def A__ ( self: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : List[Any] = NezhaModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Optional[int] ) -> Optional[Any]:
self.config_tester.run_common_tests()
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_decoder()
self.model_tester.create_and_check_model_as_decoder(*lowerCamelCase_ )
def A__ ( self: Dict ) -> int:
# This regression test was failing with PyTorch < 1.3
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Optional[int] = self.model_tester.prepare_config_and_inputs_for_decoder()
UpperCAmelCase_ : str = None
self.model_tester.create_and_check_model_as_decoder(
lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,)
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> Union[str, Any]:
UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> Dict:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_next_sequence_prediction(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_pretraining(*lowerCamelCase_ )
def A__ ( self: str ) -> Any:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCamelCase_ )
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCamelCase_ )
def A__ ( self: int ) -> List[str]:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCamelCase_ )
@slow
def A__ ( self: Tuple ) -> Union[str, Any]:
for model_name in NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Union[str, Any] = NezhaModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
@slow
@require_torch_gpu
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
# NezhaForMultipleChoice behaves incorrectly in JIT environments.
if model_class == NezhaForMultipleChoice:
return
UpperCAmelCase_ : Optional[Any] = True
UpperCAmelCase_ : Tuple = model_class(config=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = torch.jit.trace(
lowerCamelCase_ ,(inputs_dict["""input_ids"""].to("""cpu""" ), inputs_dict["""attention_mask"""].to("""cpu""" )) )
with tempfile.TemporaryDirectory() as tmp:
torch.jit.save(lowerCamelCase_ ,os.path.join(lowerCamelCase_ ,"""bert.pt""" ) )
UpperCAmelCase_ : List[str] = torch.jit.load(os.path.join(lowerCamelCase_ ,"""bert.pt""" ) ,map_location=lowerCamelCase_ )
loaded(inputs_dict["""input_ids"""].to(lowerCamelCase_ ) ,inputs_dict["""attention_mask"""].to(lowerCamelCase_ ) )
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = NezhaModel.from_pretrained("""sijunhe/nezha-cn-base""" )
UpperCAmelCase_ : int = torch.tensor([[0, 1, 2, 3, 4, 5]] )
UpperCAmelCase_ : str = torch.tensor([[0, 1, 1, 1, 1, 1]] )
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ )[0]
UpperCAmelCase_ : Optional[Any] = torch.Size((1, 6, 768) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.tensor([[[0.0_6_8_5, 0.2_4_4_1, 0.1_1_0_2], [0.0_6_0_0, 0.1_9_0_6, 0.1_3_4_9], [0.0_2_2_1, 0.0_8_1_9, 0.0_5_8_6]]] )
self.assertTrue(torch.allclose(output[:, 1:4, 1:4] ,lowerCamelCase_ ,atol=1e-4 ) )
@slow
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = NezhaForMaskedLM.from_pretrained("""sijunhe/nezha-cn-base""" )
UpperCAmelCase_ : Dict = torch.tensor([[0, 1, 2, 3, 4, 5]] )
UpperCAmelCase_ : Optional[int] = torch.tensor([[1, 1, 1, 1, 1, 1]] )
with torch.no_grad():
UpperCAmelCase_ : int = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ )[0]
UpperCAmelCase_ : List[Any] = torch.Size((1, 6, 21128) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : str = torch.tensor(
[[-2.7_9_3_9, -1.7_9_0_2, -2.2_1_8_9], [-2.8_5_8_5, -1.8_9_0_8, -2.3_7_2_3], [-2.6_4_9_9, -1.7_7_5_0, -2.2_5_5_8]] )
self.assertTrue(torch.allclose(output[:, 1:4, 1:4] ,lowerCamelCase_ ,atol=1e-4 ) )
| 345 |
import unittest
from transformers import MobileBertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
MobileBertModel,
)
class _snake_case :
'''simple docstring'''
def __init__( self: Optional[int] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Tuple=13 ,lowerCamelCase_: int=7 ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Dict=True ,lowerCamelCase_: str=True ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: int=99 ,lowerCamelCase_: List[str]=64 ,lowerCamelCase_: Tuple=32 ,lowerCamelCase_: List[str]=5 ,lowerCamelCase_: str=4 ,lowerCamelCase_: str=37 ,lowerCamelCase_: Union[str, Any]="gelu" ,lowerCamelCase_: Union[str, Any]=0.1 ,lowerCamelCase_: str=0.1 ,lowerCamelCase_: List[str]=512 ,lowerCamelCase_: Dict=16 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: List[str]=0.0_2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: Union[str, Any]=4 ,lowerCamelCase_: str=None ,) -> List[str]:
UpperCAmelCase_ : Any = parent
UpperCAmelCase_ : List[Any] = batch_size
UpperCAmelCase_ : Union[str, Any] = seq_length
UpperCAmelCase_ : Optional[int] = is_training
UpperCAmelCase_ : Dict = use_input_mask
UpperCAmelCase_ : Any = use_token_type_ids
UpperCAmelCase_ : Tuple = use_labels
UpperCAmelCase_ : List[Any] = vocab_size
UpperCAmelCase_ : str = hidden_size
UpperCAmelCase_ : List[str] = embedding_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : List[Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : str = hidden_dropout_prob
UpperCAmelCase_ : List[str] = attention_probs_dropout_prob
UpperCAmelCase_ : Any = max_position_embeddings
UpperCAmelCase_ : List[str] = type_vocab_size
UpperCAmelCase_ : Any = type_sequence_label_size
UpperCAmelCase_ : Optional[Any] = initializer_range
UpperCAmelCase_ : Optional[int] = num_labels
UpperCAmelCase_ : Optional[int] = num_choices
UpperCAmelCase_ : List[str] = scope
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size, self.seq_length] ,self.vocab_size )
UpperCAmelCase_ : List[str] = None
if self.use_input_mask:
UpperCAmelCase_ : Tuple = random_attention_mask([self.batch_size, self.seq_length] )
UpperCAmelCase_ : Dict = None
if self.use_token_type_ids:
UpperCAmelCase_ : str = ids_tensor([self.batch_size, self.seq_length] ,self.type_vocab_size )
UpperCAmelCase_ : int = None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : Union[str, Any] = None
if self.use_labels:
UpperCAmelCase_ : List[str] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] ,self.num_labels )
UpperCAmelCase_ : int = ids_tensor([self.batch_size] ,self.num_choices )
UpperCAmelCase_ : Tuple = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def A__ ( self: Any ) -> Dict:
return MobileBertConfig(
vocab_size=self.vocab_size ,hidden_size=self.hidden_size ,num_hidden_layers=self.num_hidden_layers ,num_attention_heads=self.num_attention_heads ,intermediate_size=self.intermediate_size ,embedding_size=self.embedding_size ,hidden_act=self.hidden_act ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,max_position_embeddings=self.max_position_embeddings ,type_vocab_size=self.type_vocab_size ,is_decoder=lowerCamelCase_ ,initializer_range=self.initializer_range ,)
def A__ ( self: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> int:
UpperCAmelCase_ : Any = MobileBertModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,token_type_ids=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape ,(self.batch_size, self.hidden_size) )
def A__ ( self: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Dict ) -> int:
UpperCAmelCase_ : Union[str, Any] = MobileBertForMaskedLM(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
def A__ ( self: str ,lowerCamelCase_: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : List[Any] = MobileBertForNextSentencePrediction(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 2) )
def A__ ( self: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Tuple = MobileBertForPreTraining(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,next_sentence_label=lowerCamelCase_ ,)
self.parent.assertEqual(result.prediction_logits.shape ,(self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape ,(self.batch_size, 2) )
def A__ ( self: Any ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = MobileBertForQuestionAnswering(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,start_positions=lowerCamelCase_ ,end_positions=lowerCamelCase_ ,)
self.parent.assertEqual(result.start_logits.shape ,(self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape ,(self.batch_size, self.seq_length) )
def A__ ( self: List[str] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> str:
UpperCAmelCase_ : Optional[Any] = self.num_labels
UpperCAmelCase_ : Union[str, Any] = MobileBertForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_labels) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Any:
UpperCAmelCase_ : str = self.num_labels
UpperCAmelCase_ : Optional[int] = MobileBertForTokenClassification(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.seq_length, self.num_labels) )
def A__ ( self: Tuple ,lowerCamelCase_: str ,lowerCamelCase_: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Union[str, Any] = self.num_choices
UpperCAmelCase_ : Tuple = MobileBertForMultipleChoice(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = input_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Union[str, Any] = token_type_ids.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : str = input_mask.unsqueeze(1 ).expand(-1 ,self.num_choices ,-1 ).contiguous()
UpperCAmelCase_ : Optional[int] = model(
lowerCamelCase_ ,attention_mask=lowerCamelCase_ ,token_type_ids=lowerCamelCase_ ,labels=lowerCamelCase_ ,)
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.num_choices) )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ : str = self.prepare_config_and_inputs()
(
(
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) , (
UpperCAmelCase_
) ,
) : Union[str, Any] = config_and_inputs
UpperCAmelCase_ : Dict = {"""input_ids""": input_ids, """token_type_ids""": token_type_ids, """attention_mask""": input_mask}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = (
(
MobileBertModel,
MobileBertForMaskedLM,
MobileBertForMultipleChoice,
MobileBertForNextSentencePrediction,
MobileBertForPreTraining,
MobileBertForQuestionAnswering,
MobileBertForSequenceClassification,
MobileBertForTokenClassification,
)
if is_torch_available()
else ()
)
A__ : List[str] = (
{
"feature-extraction": MobileBertModel,
"fill-mask": MobileBertForMaskedLM,
"question-answering": MobileBertForQuestionAnswering,
"text-classification": MobileBertForSequenceClassification,
"token-classification": MobileBertForTokenClassification,
"zero-shot": MobileBertForSequenceClassification,
}
if is_torch_available()
else {}
)
A__ : List[str] = True
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Tuple ,lowerCamelCase_: int=False ) -> Union[str, Any]:
UpperCAmelCase_ : List[Any] = super()._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ,return_labels=lowerCamelCase_ )
if return_labels:
if model_class in get_values(lowerCamelCase_ ):
UpperCAmelCase_ : Any = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) ,dtype=torch.long ,device=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = torch.zeros(
self.model_tester.batch_size ,dtype=torch.long ,device=lowerCamelCase_ )
return inputs_dict
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[str] = MobileBertModelTester(self )
UpperCAmelCase_ : Union[str, Any] = ConfigTester(self ,config_class=lowerCamelCase_ ,hidden_size=37 )
def A__ ( self: Optional[Any] ) -> List[Any]:
self.config_tester.run_common_tests()
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_model(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_masked_lm(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Tuple:
UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_multiple_choice(*lowerCamelCase_ )
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_next_sequence_prediction(*lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_pretraining(*lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_question_answering(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_sequence_classification(*lowerCamelCase_ )
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_mobilebert_for_token_classification(*lowerCamelCase_ )
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return torch.tensor(
_a , dtype=torch.long , device=_a , )
UpperCamelCase_ = 1E-3
@require_torch
@require_sentencepiece
@require_tokenizers
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: List[Any] ) -> str:
UpperCAmelCase_ : Any = MobileBertModel.from_pretrained("""google/mobilebert-uncased""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = _long_tensor([[101, 7110, 1005, 1056, 2023, 11333, 17413, 1029, 102]] )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )[0]
UpperCAmelCase_ : Union[str, Any] = torch.Size((1, 9, 512) )
self.assertEqual(output.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = torch.tensor(
[
[
[-2.473_6526e07, 8.269_1656e04, 1.652_1838e05],
[-5.754_1704e-01, 3.905_6022e00, 4.401_1507e00],
[2.604_7359e00, 1.567_7652e00, -1.732_4188e-01],
]
] ,device=lowerCamelCase_ ,)
# MobileBERT results range from 10e0 to 10e8. Even a 0.0000001% difference with a value of 10e8 results in a
# ~1 difference, it's therefore not a good idea to measure using addition.
# Here, we instead divide the expected result with the result in order to obtain ~1. We then check that the
# result is held between bounds: 1 - TOLERANCE < expected_result / result < 1 + TOLERANCE
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) >= 1 - TOLERANCE )
UpperCAmelCase_ : Dict = torch.all((expected_slice / output[..., :3, :3]) <= 1 + TOLERANCE )
self.assertTrue(lower_bound and upper_bound )
| 345 | 1 |
# Logistic Regression from scratch
# In[62]:
# In[63]:
# importing all the required libraries
import numpy as np
from matplotlib import pyplot as plt
from sklearn import datasets
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
return 1 / (1 + np.exp(-z ))
def lowerCamelCase_ ( _a : Any , _a : List[str] ):
'''simple docstring'''
return (-y * np.log(_a ) - (1 - y) * np.log(1 - h )).mean()
def lowerCamelCase_ ( _a : List[Any] , _a : Dict , _a : Optional[int] ):
'''simple docstring'''
UpperCAmelCase_ : str = np.dot(_a , _a )
return np.sum(y * scores - np.log(1 + np.exp(_a ) ) )
def lowerCamelCase_ ( _a : Any , _a : Optional[Any] , _a : int , _a : Dict=7_0000 ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = np.zeros(x.shape[1] )
for iterations in range(_a ):
UpperCAmelCase_ : int = np.dot(_a , _a )
UpperCAmelCase_ : List[Any] = sigmoid_function(_a )
UpperCAmelCase_ : Union[str, Any] = np.dot(x.T , h - y ) / y.size
UpperCAmelCase_ : List[Any] = theta - alpha * gradient # updating the weights
UpperCAmelCase_ : List[str] = np.dot(_a , _a )
UpperCAmelCase_ : List[Any] = sigmoid_function(_a )
UpperCAmelCase_ : Tuple = cost_function(_a , _a )
if iterations % 100 == 0:
print(F'''loss: {j} \t''' ) # printing the loss after every 100 iterations
return theta
# In[68]:
if __name__ == "__main__":
UpperCamelCase_ = datasets.load_iris()
UpperCamelCase_ = iris.data[:, :2]
UpperCamelCase_ = (iris.target != 0) * 1
UpperCamelCase_ = 0.1
UpperCamelCase_ = logistic_reg(alpha, x, y, max_iterations=70000)
print('''theta: ''', theta) # printing the theta i.e our weights vector
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
return sigmoid_function(
np.dot(_a , _a ) ) # predicting the value of probability from the logistic regression algorithm
plt.figure(figsize=(10, 6))
plt.scatter(x[y == 0][:, 0], x[y == 0][:, 1], color='''b''', label='''0''')
plt.scatter(x[y == 1][:, 0], x[y == 1][:, 1], color='''r''', label='''1''')
((UpperCamelCase_) ,(UpperCamelCase_)) = (x[:, 0].min(), x[:, 0].max())
((UpperCamelCase_) ,(UpperCamelCase_)) = (x[:, 1].min(), x[:, 1].max())
((UpperCamelCase_) ,(UpperCamelCase_)) = np.meshgrid(np.linspace(xa_min, xa_max), np.linspace(xa_min, xa_max))
UpperCamelCase_ = np.c_[xxa.ravel(), xxa.ravel()]
UpperCamelCase_ = predict_prob(grid).reshape(xxa.shape)
plt.contour(xxa, xxa, probs, [0.5], linewidths=1, colors='''black''')
plt.legend()
plt.show()
| 345 |
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import AutoTokenizer, BarkProcessor
from transformers.testing_utils import require_torch, slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: str ) -> int:
UpperCAmelCase_ : List[Any] = """ylacombe/bark-small"""
UpperCAmelCase_ : Tuple = tempfile.mkdtemp()
UpperCAmelCase_ : Union[str, Any] = """en_speaker_1"""
UpperCAmelCase_ : Optional[Any] = """This is a test string"""
UpperCAmelCase_ : int = """speaker_embeddings_path.json"""
UpperCAmelCase_ : Any = """speaker_embeddings"""
def A__ ( self: Tuple ,**lowerCamelCase_: List[str] ) -> List[Any]:
return AutoTokenizer.from_pretrained(self.checkpoint ,**lowerCamelCase_ )
def A__ ( self: str ) -> Union[str, Any]:
shutil.rmtree(self.tmpdirname )
def A__ ( self: List[Any] ) -> int:
UpperCAmelCase_ : int = self.get_tokenizer()
UpperCAmelCase_ : Tuple = BarkProcessor(tokenizer=lowerCamelCase_ )
processor.save_pretrained(self.tmpdirname )
UpperCAmelCase_ : Optional[int] = BarkProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer.get_vocab() )
@slow
def A__ ( self: List[Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
processor.save_pretrained(
self.tmpdirname ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,speaker_embeddings_directory=self.speaker_embeddings_directory ,)
UpperCAmelCase_ : Optional[Any] = self.get_tokenizer(bos_token="""(BOS)""" ,eos_token="""(EOS)""" )
UpperCAmelCase_ : List[Any] = BarkProcessor.from_pretrained(
self.tmpdirname ,self.speaker_embeddings_dict_path ,bos_token="""(BOS)""" ,eos_token="""(EOS)""" ,)
self.assertEqual(processor.tokenizer.get_vocab() ,tokenizer_add_kwargs.get_vocab() )
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = BarkProcessor.from_pretrained(
pretrained_processor_name_or_path=self.checkpoint ,speaker_embeddings_dict_path=self.speaker_embeddings_dict_path ,)
UpperCAmelCase_ : Optional[int] = 35
UpperCAmelCase_ : Optional[int] = 2
UpperCAmelCase_ : Dict = 8
UpperCAmelCase_ : Optional[int] = {
"""semantic_prompt""": np.ones(lowerCamelCase_ ),
"""coarse_prompt""": np.ones((nb_codebooks_coarse, seq_len) ),
"""fine_prompt""": np.ones((nb_codebooks_total, seq_len) ),
}
# test providing already loaded voice_preset
UpperCAmelCase_ : str = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from npz file
UpperCAmelCase_ : List[Any] = os.path.join(self.tmpdirname ,"""file.npz""" )
np.savez(lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string ,voice_preset=lowerCamelCase_ )
UpperCAmelCase_ : int = inputs["""history_prompt"""]
for key in voice_preset:
self.assertListEqual(voice_preset[key].tolist() ,processed_voice_preset.get(lowerCamelCase_ ,np.array([] ) ).tolist() )
# test loading voice preset from the hub
UpperCAmelCase_ : Union[str, Any] = processor(text=self.input_string ,voice_preset=self.voice_preset )
def A__ ( self: Dict ) -> Tuple:
UpperCAmelCase_ : Any = self.get_tokenizer()
UpperCAmelCase_ : Dict = BarkProcessor(tokenizer=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = processor(text=self.input_string )
UpperCAmelCase_ : str = tokenizer(
self.input_string ,padding="""max_length""" ,max_length=256 ,add_special_tokens=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,return_token_type_ids=lowerCamelCase_ ,)
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] ,encoded_processor[key].squeeze().tolist() )
| 345 | 1 |
import logging
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
from seqaseq_trainer import SeqaSeqTrainer
from seqaseq_training_args import SeqaSeqTrainingArguments
import transformers
from transformers import (
AutoConfig,
AutoModelForSeqaSeqLM,
AutoTokenizer,
HfArgumentParser,
MBartTokenizer,
MBartTokenizerFast,
set_seed,
)
from transformers.trainer_utils import EvaluationStrategy, is_main_process
from transformers.training_args import ParallelMode
from utils import (
SeqaSeqDataCollator,
SeqaSeqDataset,
assert_all_frozen,
build_compute_metrics_fn,
check_output_dir,
freeze_embeds,
freeze_params,
lmap,
save_json,
use_task_specific_params,
write_txt_file,
)
UpperCamelCase_ = logging.getLogger(__name__)
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Pretrained config name or path if not the same as model_name"} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"} )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"} , )
A__ : bool = field(default=__snake_case , metadata={"help": "Whether tp freeze the encoder."} )
A__ : bool = field(default=__snake_case , metadata={"help": "Whether to freeze the embeddings."} )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
metadata={"help": "The input data dir. Should contain the .tsv files (or other data files) for the task."} )
A__ : Optional[str] = field(
default="summarization" , metadata={"help": "Task name, summarization (or summarization_{dataset} for pegasus) or translation"} , )
A__ : Optional[int] = field(
default=1_024 , metadata={
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
} , )
A__ : Optional[int] = field(
default=128 , metadata={
"help": (
"The maximum total sequence length for target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
} , )
A__ : Optional[int] = field(
default=142 , metadata={
"help": (
"The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. "
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
)
} , )
A__ : Optional[int] = field(
default=142 , metadata={
"help": (
"The maximum total sequence length for test target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
} , )
A__ : Optional[int] = field(default=-1 , metadata={"help": "# training examples. -1 means use all."} )
A__ : Optional[int] = field(default=-1 , metadata={"help": "# validation examples. -1 means use all."} )
A__ : Optional[int] = field(default=-1 , metadata={"help": "# test examples. -1 means use all."} )
A__ : Optional[str] = field(default=__snake_case , metadata={"help": "Source language id for translation."} )
A__ : Optional[str] = field(default=__snake_case , metadata={"help": "Target language id for translation."} )
A__ : Optional[int] = field(default=__snake_case , metadata={"help": "# num_beams to use for evaluation."} )
A__ : bool = field(
default=__snake_case , metadata={"help": "If only pad tokens should be ignored. This assumes that `config.pad_token_id` is defined."} , )
def lowerCamelCase_ ( _a : str , _a : int , _a : Optional[int] ):
'''simple docstring'''
logger.info(F'''***** {split} metrics *****''' )
for key in sorted(metrics.keys() ):
logger.info(F''' {key} = {metrics[key]}''' )
save_json(_a , os.path.join(_a , F'''{split}_results.json''' ) )
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : Any = HfArgumentParser((ModelArguments, DataTrainingArguments, SeqaSeqTrainingArguments) )
if len(sys.argv ) == 2 and sys.argv[1].endswith(""".json""" ):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : str = parser.parse_args_into_dataclasses()
check_output_dir(_a )
# Setup logging
logging.basicConfig(
format="""%(asctime)s - %(levelname)s - %(name)s - %(message)s""" , datefmt="""%m/%d/%Y %H:%M:%S""" , level=logging.INFO if training_args.local_rank in [-1, 0] else logging.WARN , )
logger.warning(
"""Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s""" , training_args.local_rank , training_args.device , training_args.n_gpu , bool(training_args.parallel_mode == ParallelMode.DISTRIBUTED ) , training_args.fpaa , )
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
logger.info("""Training/evaluation parameters %s""" , _a )
# Set seed
set_seed(training_args.seed )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
UpperCAmelCase_ : Union[str, Any] = AutoConfig.from_pretrained(
model_args.config_name if model_args.config_name else model_args.model_name_or_path , cache_dir=model_args.cache_dir , )
UpperCAmelCase_ : str = ("""encoder_layerdrop""", """decoder_layerdrop""", """dropout""", """attention_dropout""")
for p in extra_model_params:
if getattr(_a , _a , _a ):
assert hasattr(_a , _a ), F'''({config.__class__.__name__}) doesn\'t have a `{p}` attribute'''
setattr(_a , _a , getattr(_a , _a ) )
UpperCAmelCase_ : int = AutoTokenizer.from_pretrained(
model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path , cache_dir=model_args.cache_dir , )
UpperCAmelCase_ : Tuple = AutoModelForSeqaSeqLM.from_pretrained(
model_args.model_name_or_path , from_tf=""".ckpt""" in model_args.model_name_or_path , config=_a , cache_dir=model_args.cache_dir , )
# use task specific params
use_task_specific_params(_a , data_args.task )
# set num_beams for evaluation
if data_args.eval_beams is None:
UpperCAmelCase_ : List[Any] = model.config.num_beams
# set decoder_start_token_id for MBart
if model.config.decoder_start_token_id is None and isinstance(_a , (MBartTokenizer, MBartTokenizerFast) ):
assert (
data_args.tgt_lang is not None and data_args.src_lang is not None
), "mBart requires --tgt_lang and --src_lang"
if isinstance(_a , _a ):
UpperCAmelCase_ : Dict = tokenizer.lang_code_to_id[data_args.tgt_lang]
else:
UpperCAmelCase_ : Optional[Any] = tokenizer.convert_tokens_to_ids(data_args.tgt_lang )
if model_args.freeze_embeds:
freeze_embeds(_a )
if model_args.freeze_encoder:
freeze_params(model.get_encoder() )
assert_all_frozen(model.get_encoder() )
UpperCAmelCase_ : List[Any] = SeqaSeqDataset
# Get datasets
UpperCAmelCase_ : Optional[Any] = (
dataset_class(
_a , type_path="""train""" , data_dir=data_args.data_dir , n_obs=data_args.n_train , max_target_length=data_args.max_target_length , max_source_length=data_args.max_source_length , prefix=model.config.prefix or """""" , )
if training_args.do_train
else None
)
UpperCAmelCase_ : Union[str, Any] = (
dataset_class(
_a , type_path="""val""" , data_dir=data_args.data_dir , n_obs=data_args.n_val , max_target_length=data_args.val_max_target_length , max_source_length=data_args.max_source_length , prefix=model.config.prefix or """""" , )
if training_args.do_eval or training_args.evaluation_strategy != EvaluationStrategy.NO
else None
)
UpperCAmelCase_ : Any = (
dataset_class(
_a , type_path="""test""" , data_dir=data_args.data_dir , n_obs=data_args.n_test , max_target_length=data_args.test_max_target_length , max_source_length=data_args.max_source_length , prefix=model.config.prefix or """""" , )
if training_args.do_predict
else None
)
# Initialize our Trainer
UpperCAmelCase_ : Tuple = (
build_compute_metrics_fn(data_args.task , _a ) if training_args.predict_with_generate else None
)
UpperCAmelCase_ : Dict = SeqaSeqTrainer(
model=_a , args=_a , data_args=_a , train_dataset=_a , eval_dataset=_a , data_collator=SeqaSeqDataCollator(
_a , _a , model.config.decoder_start_token_id , training_args.tpu_num_cores ) , compute_metrics=_a , tokenizer=_a , )
UpperCAmelCase_ : Optional[Any] = {}
# Training
if training_args.do_train:
logger.info("""*** Train ***""" )
UpperCAmelCase_ : Dict = trainer.train(
model_path=model_args.model_name_or_path if os.path.isdir(model_args.model_name_or_path ) else None )
UpperCAmelCase_ : str = train_result.metrics
UpperCAmelCase_ : Union[str, Any] = data_args.n_train
trainer.save_model() # this also saves the tokenizer
if trainer.is_world_process_zero():
handle_metrics("""train""" , _a , training_args.output_dir )
all_metrics.update(_a )
# Need to save the state, since Trainer.save_model saves only the tokenizer with the model
trainer.state.save_to_json(os.path.join(training_args.output_dir , """trainer_state.json""" ) )
# For convenience, we also re-save the tokenizer to the same directory,
# so that you can share your model easily on huggingface.co/models =)
tokenizer.save_pretrained(training_args.output_dir )
# Evaluation
if training_args.do_eval:
logger.info("""*** Evaluate ***""" )
UpperCAmelCase_ : Tuple = trainer.evaluate(metric_key_prefix="""val""" )
UpperCAmelCase_ : List[str] = data_args.n_val
UpperCAmelCase_ : Optional[Any] = round(metrics["""val_loss"""] , 4 )
if trainer.is_world_process_zero():
handle_metrics("""val""" , _a , training_args.output_dir )
all_metrics.update(_a )
if training_args.do_predict:
logger.info("""*** Predict ***""" )
UpperCAmelCase_ : Tuple = trainer.predict(test_dataset=_a , metric_key_prefix="""test""" )
UpperCAmelCase_ : Tuple = test_output.metrics
UpperCAmelCase_ : Any = data_args.n_test
if trainer.is_world_process_zero():
UpperCAmelCase_ : Optional[int] = round(metrics["""test_loss"""] , 4 )
handle_metrics("""test""" , _a , training_args.output_dir )
all_metrics.update(_a )
if training_args.predict_with_generate:
UpperCAmelCase_ : List[str] = tokenizer.batch_decode(
test_output.predictions , skip_special_tokens=_a , clean_up_tokenization_spaces=_a )
UpperCAmelCase_ : Union[str, Any] = lmap(str.strip , _a )
write_txt_file(_a , os.path.join(training_args.output_dir , """test_generations.txt""" ) )
if trainer.is_world_process_zero():
save_json(_a , os.path.join(training_args.output_dir , """all_results.json""" ) )
return all_metrics
def lowerCamelCase_ ( _a : Union[str, Any] ):
'''simple docstring'''
main()
if __name__ == "__main__":
main()
| 345 |
import unittest
from queue import Empty
from threading import Thread
from transformers import AutoTokenizer, TextIteratorStreamer, TextStreamer, is_torch_available
from transformers.testing_utils import CaptureStdout, require_torch, torch_device
from ..test_modeling_common import ids_tensor
if is_torch_available():
import torch
from transformers import AutoModelForCausalLM
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Union[str, Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : str = -1
UpperCAmelCase_ : Dict = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Any = tokenizer.decode(greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : Optional[int] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Dict ) -> Optional[Any]:
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : List[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : Dict = tokenizer.decode(greedy_ids[0] )
UpperCAmelCase_ : str = TextIteratorStreamer(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : str = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
UpperCAmelCase_ : int = """"""
for new_text in streamer:
streamer_text += new_text
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[Any] ) -> Dict:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = -1
UpperCAmelCase_ : Tuple = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ )
UpperCAmelCase_ : str = greedy_ids[:, input_ids.shape[1] :]
UpperCAmelCase_ : Dict = tokenizer.decode(new_greedy_ids[0] )
with CaptureStdout() as cs:
UpperCAmelCase_ : List[Any] = TextStreamer(lowerCamelCase_ ,skip_prompt=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=10 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
UpperCAmelCase_ : List[str] = cs.out[:-1]
self.assertEqual(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: str ) -> str:
# Tests that we can pass `decode_kwargs` to the streamer to control how the tokens are decoded. Must be tested
# with actual models -- the dummy models' tokenizers are not aligned with their models, and
# `skip_special_tokens=True` has no effect on them
UpperCAmelCase_ : Union[str, Any] = AutoTokenizer.from_pretrained("""distilgpt2""" )
UpperCAmelCase_ : Optional[Any] = AutoModelForCausalLM.from_pretrained("""distilgpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Any = -1
UpperCAmelCase_ : Union[str, Any] = torch.ones((1, 5) ,device=lowerCamelCase_ ).long() * model.config.bos_token_id
with CaptureStdout() as cs:
UpperCAmelCase_ : Union[str, Any] = TextStreamer(lowerCamelCase_ ,skip_special_tokens=lowerCamelCase_ )
model.generate(lowerCamelCase_ ,max_new_tokens=1 ,do_sample=lowerCamelCase_ ,streamer=lowerCamelCase_ )
# The prompt contains a special token, so the streamer should not print it. As such, the output text, when
# re-tokenized, must only contain one token
UpperCAmelCase_ : List[str] = cs.out[:-1] # Remove the final "\n"
UpperCAmelCase_ : Dict = tokenizer(lowerCamelCase_ ,return_tensors="""pt""" )
self.assertEqual(streamer_text_tokenized.input_ids.shape ,(1, 1) )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
UpperCAmelCase_ : Any = AutoModelForCausalLM.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = -1
UpperCAmelCase_ : Optional[Any] = ids_tensor((1, 5) ,vocab_size=model.config.vocab_size ).to(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = TextIteratorStreamer(lowerCamelCase_ ,timeout=0.0_0_1 )
UpperCAmelCase_ : Any = {"""input_ids""": input_ids, """max_new_tokens""": 10, """do_sample""": False, """streamer""": streamer}
UpperCAmelCase_ : Dict = Thread(target=model.generate ,kwargs=lowerCamelCase_ )
thread.start()
# The streamer will timeout after 0.001 seconds, so an exception will be raised
with self.assertRaises(lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = """"""
for new_text in streamer:
streamer_text += new_text
| 345 | 1 |
import os
from pathlib import Path
import numpy as np
import pytest
from pack_dataset import pack_data_dir
from parameterized import parameterized
from save_len_file import save_len_file
from torch.utils.data import DataLoader
from transformers import AutoTokenizer
from transformers.models.mbart.modeling_mbart import shift_tokens_right
from transformers.testing_utils import TestCasePlus, slow
from utils import FAIRSEQ_AVAILABLE, DistributedSortishSampler, LegacySeqaSeqDataset, SeqaSeqDataset
UpperCamelCase_ = '''bert-base-cased'''
UpperCamelCase_ = '''google/pegasus-xsum'''
UpperCamelCase_ = [''' Sam ate lunch today.''', '''Sams lunch ingredients.''']
UpperCamelCase_ = ['''A very interesting story about what I ate for lunch.''', '''Avocado, celery, turkey, coffee''']
UpperCamelCase_ = '''patrickvonplaten/t5-tiny-random'''
UpperCamelCase_ = '''sshleifer/bart-tiny-random'''
UpperCamelCase_ = '''sshleifer/tiny-mbart'''
UpperCamelCase_ = '''sshleifer/tiny-marian-en-de'''
def lowerCamelCase_ ( _a : Path , _a : list ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = """\n""".join(_a )
Path(_a ).open("""w""" ).writelines(_a )
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
for split in ["train", "val", "test"]:
_dump_articles(os.path.join(_a , F'''{split}.source''' ) , _a )
_dump_articles(os.path.join(_a , F'''{split}.target''' ) , _a )
return tmp_dir
class _snake_case ( __snake_case ):
'''simple docstring'''
@parameterized.expand(
[
MBART_TINY,
MARIAN_TINY,
T5_TINY,
BART_TINY,
PEGASUS_XSUM,
] ,)
@slow
def A__ ( self: List[Any] ,lowerCamelCase_: Tuple ) -> int:
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = make_test_data_dir(tmp_dir=self.get_auto_remove_tmp_dir() )
UpperCAmelCase_ : Optional[int] = max(len(tokenizer.encode(lowerCamelCase_ ) ) for a in ARTICLES )
UpperCAmelCase_ : List[str] = max(len(tokenizer.encode(lowerCamelCase_ ) ) for a in SUMMARIES )
UpperCAmelCase_ : str = 4
UpperCAmelCase_ : List[str] = 8
assert max_len_target > max_src_len # Will be truncated
assert max_len_source > max_src_len # Will be truncated
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = """ro_RO""", """de_DE""" # ignored for all but mbart, but never causes error.
UpperCAmelCase_ : List[str] = SeqaSeqDataset(
lowerCamelCase_ ,data_dir=lowerCamelCase_ ,type_path="""train""" ,max_source_length=lowerCamelCase_ ,max_target_length=lowerCamelCase_ ,src_lang=lowerCamelCase_ ,tgt_lang=lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = DataLoader(lowerCamelCase_ ,batch_size=2 ,collate_fn=train_dataset.collate_fn )
for batch in dataloader:
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert batch["attention_mask"].shape == batch["input_ids"].shape
# show that articles were trimmed.
assert batch["input_ids"].shape[1] == max_src_len
# show that targets are the same len
assert batch["labels"].shape[1] == max_tgt_len
if tok_name != MBART_TINY:
continue
# check language codes in correct place
UpperCAmelCase_ : List[str] = shift_tokens_right(batch["""labels"""] ,tokenizer.pad_token_id )
assert batch["decoder_input_ids"][0, 0].item() == tokenizer.lang_code_to_id[tgt_lang]
assert batch["decoder_input_ids"][0, -1].item() == tokenizer.eos_token_id
assert batch["input_ids"][0, -2].item() == tokenizer.eos_token_id
assert batch["input_ids"][0, -1].item() == tokenizer.lang_code_to_id[src_lang]
break # No need to test every batch
@parameterized.expand([BART_TINY, BERT_BASE_CASED] )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Dict:
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : str = make_test_data_dir(tmp_dir=self.get_auto_remove_tmp_dir() )
UpperCAmelCase_ : Dict = max(len(tokenizer.encode(lowerCamelCase_ ) ) for a in ARTICLES )
UpperCAmelCase_ : Optional[Any] = max(len(tokenizer.encode(lowerCamelCase_ ) ) for a in SUMMARIES )
UpperCAmelCase_ : int = 4
UpperCAmelCase_ : Optional[int] = LegacySeqaSeqDataset(
lowerCamelCase_ ,data_dir=lowerCamelCase_ ,type_path="""train""" ,max_source_length=20 ,max_target_length=lowerCamelCase_ ,)
UpperCAmelCase_ : Optional[Any] = DataLoader(lowerCamelCase_ ,batch_size=2 ,collate_fn=train_dataset.collate_fn )
for batch in dataloader:
assert batch["attention_mask"].shape == batch["input_ids"].shape
# show that articles were trimmed.
assert batch["input_ids"].shape[1] == max_len_source
assert 20 >= batch["input_ids"].shape[1] # trimmed significantly
# show that targets were truncated
assert batch["labels"].shape[1] == trunc_target # Truncated
assert max_len_target > trunc_target # Truncated
break # No need to test every batch
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : List[Any] = AutoTokenizer.from_pretrained("""facebook/mbart-large-cc25""" )
UpperCAmelCase_ : Dict = Path(make_test_data_dir(tmp_dir=self.get_auto_remove_tmp_dir() ) )
UpperCAmelCase_ : List[str] = tmp_dir.joinpath("""train.source""" ).open().readlines()
UpperCAmelCase_ : int = Path(make_test_data_dir(tmp_dir=self.get_auto_remove_tmp_dir() ) )
pack_data_dir(lowerCamelCase_ ,lowerCamelCase_ ,128 ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = {x.name for x in tmp_dir.iterdir()}
UpperCAmelCase_ : Optional[Any] = {x.name for x in save_dir.iterdir()}
UpperCAmelCase_ : Optional[Any] = save_dir.joinpath("""train.source""" ).open().readlines()
# orig: [' Sam ate lunch today.\n', 'Sams lunch ingredients.']
# desired_packed: [' Sam ate lunch today.\n Sams lunch ingredients.']
assert len(lowerCamelCase_ ) < len(lowerCamelCase_ )
assert len(lowerCamelCase_ ) == 1
assert len(packed_examples[0] ) == sum(len(lowerCamelCase_ ) for x in orig_examples )
assert orig_paths == new_paths
@pytest.mark.skipif(not FAIRSEQ_AVAILABLE ,reason="""This test requires fairseq""" )
def A__ ( self: Tuple ) -> Any:
if not FAIRSEQ_AVAILABLE:
return
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : int = self._get_dataset(max_len=64 )
UpperCAmelCase_ : Optional[Any] = 64
UpperCAmelCase_ : Tuple = ds.make_dynamic_sampler(lowerCamelCase_ ,required_batch_size_multiple=lowerCamelCase_ )
UpperCAmelCase_ : str = [len(lowerCamelCase_ ) for x in batch_sampler]
assert len(set(lowerCamelCase_ ) ) > 1 # it's not dynamic batch size if every batch is the same length
assert sum(lowerCamelCase_ ) == len(lowerCamelCase_ ) # no dropped or added examples
UpperCAmelCase_ : str = DataLoader(lowerCamelCase_ ,batch_sampler=lowerCamelCase_ ,collate_fn=ds.collate_fn ,num_workers=2 )
UpperCAmelCase_ : Union[str, Any] = []
UpperCAmelCase_ : Dict = []
for batch in data_loader:
UpperCAmelCase_ : Optional[Any] = batch["""input_ids"""].shape
UpperCAmelCase_ : List[Any] = src_shape[0]
assert bs % required_batch_size_multiple == 0 or bs < required_batch_size_multiple
UpperCAmelCase_ : str = np.product(batch["""input_ids"""].shape )
num_src_per_batch.append(lowerCamelCase_ )
if num_src_tokens > (max_tokens * 1.1):
failures.append(lowerCamelCase_ )
assert num_src_per_batch[0] == max(lowerCamelCase_ )
if failures:
raise AssertionError(F'''too many tokens in {len(lowerCamelCase_ )} batches''' )
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self._get_dataset(max_len=512 )
UpperCAmelCase_ : List[Any] = 2
UpperCAmelCase_ : int = ds.make_sortish_sampler(lowerCamelCase_ ,shuffle=lowerCamelCase_ )
UpperCAmelCase_ : int = DataLoader(lowerCamelCase_ ,batch_size=lowerCamelCase_ ,collate_fn=ds.collate_fn ,num_workers=2 )
UpperCAmelCase_ : Optional[int] = DataLoader(lowerCamelCase_ ,batch_size=lowerCamelCase_ ,collate_fn=ds.collate_fn ,num_workers=2 ,sampler=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = tokenizer.pad_token_id
def count_pad_tokens(lowerCamelCase_: List[Any] ,lowerCamelCase_: str="input_ids" ):
return [batch[k].eq(lowerCamelCase_ ).sum().item() for batch in data_loader]
assert sum(count_pad_tokens(lowerCamelCase_ ,k="""labels""" ) ) < sum(count_pad_tokens(lowerCamelCase_ ,k="""labels""" ) )
assert sum(count_pad_tokens(lowerCamelCase_ ) ) < sum(count_pad_tokens(lowerCamelCase_ ) )
assert len(lowerCamelCase_ ) == len(lowerCamelCase_ )
def A__ ( self: Tuple ,lowerCamelCase_: int=1000 ,lowerCamelCase_: Optional[int]=128 ) -> Optional[Any]:
if os.getenv("""USE_REAL_DATA""" ,lowerCamelCase_ ):
UpperCAmelCase_ : str = """examples/seq2seq/wmt_en_ro"""
UpperCAmelCase_ : Dict = max_len * 2 * 64
if not Path(lowerCamelCase_ ).joinpath("""train.len""" ).exists():
save_len_file(lowerCamelCase_ ,lowerCamelCase_ )
else:
UpperCAmelCase_ : int = """examples/seq2seq/test_data/wmt_en_ro"""
UpperCAmelCase_ : Dict = max_len * 4
save_len_file(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : str = SeqaSeqDataset(
lowerCamelCase_ ,data_dir=lowerCamelCase_ ,type_path="""train""" ,max_source_length=lowerCamelCase_ ,max_target_length=lowerCamelCase_ ,n_obs=lowerCamelCase_ ,)
return ds, max_tokens, tokenizer
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Dict = self._get_dataset()
UpperCAmelCase_ : Union[str, Any] = set(DistributedSortishSampler(lowerCamelCase_ ,256 ,num_replicas=2 ,rank=0 ,add_extra_examples=lowerCamelCase_ ) )
UpperCAmelCase_ : Dict = set(DistributedSortishSampler(lowerCamelCase_ ,256 ,num_replicas=2 ,rank=1 ,add_extra_examples=lowerCamelCase_ ) )
assert idsa.intersection(lowerCamelCase_ ) == set()
@parameterized.expand(
[
MBART_TINY,
MARIAN_TINY,
T5_TINY,
BART_TINY,
PEGASUS_XSUM,
] ,)
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = AutoTokenizer.from_pretrained(lowerCamelCase_ ,use_fast=lowerCamelCase_ )
if tok_name == MBART_TINY:
UpperCAmelCase_ : Any = SeqaSeqDataset(
lowerCamelCase_ ,data_dir=make_test_data_dir(tmp_dir=self.get_auto_remove_tmp_dir() ) ,type_path="""train""" ,max_source_length=4 ,max_target_length=8 ,src_lang="""EN""" ,tgt_lang="""FR""" ,)
UpperCAmelCase_ : Tuple = train_dataset.dataset_kwargs
assert "src_lang" in kwargs and "tgt_lang" in kwargs
else:
UpperCAmelCase_ : Optional[int] = SeqaSeqDataset(
lowerCamelCase_ ,data_dir=make_test_data_dir(tmp_dir=self.get_auto_remove_tmp_dir() ) ,type_path="""train""" ,max_source_length=4 ,max_target_length=8 ,)
UpperCAmelCase_ : Optional[int] = train_dataset.dataset_kwargs
assert "add_prefix_space" not in kwargs if tok_name != BART_TINY else "add_prefix_space" in kwargs
assert len(lowerCamelCase_ ) == 1 if tok_name == BART_TINY else len(lowerCamelCase_ ) == 0
| 345 |
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel
from diffusers import DDIMScheduler, LDMPipeline, UNetaDModel, VQModel
from diffusers.utils.testing_utils import enable_full_determinism, require_torch, slow, torch_device
enable_full_determinism()
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = UNetaDModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=3 ,out_channels=3 ,down_block_types=("""DownBlock2D""", """AttnDownBlock2D""") ,up_block_types=("""AttnUpBlock2D""", """UpBlock2D""") ,)
return model
@property
def A__ ( self: Tuple ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : List[str] = VQModel(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=3 ,)
return model
@property
def A__ ( self: Tuple ) -> Any:
torch.manual_seed(0 )
UpperCAmelCase_ : int = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
def A__ ( self: str ) -> Optional[Any]:
UpperCAmelCase_ : str = self.dummy_uncond_unet
UpperCAmelCase_ : List[Any] = DDIMScheduler()
UpperCAmelCase_ : List[Any] = self.dummy_vq_model
UpperCAmelCase_ : Optional[int] = LDMPipeline(unet=lowerCamelCase_ ,vqvae=lowerCamelCase_ ,scheduler=lowerCamelCase_ )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.manual_seed(0 )
UpperCAmelCase_ : int = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ).images
UpperCAmelCase_ : List[str] = torch.manual_seed(0 )
UpperCAmelCase_ : Union[str, Any] = ldm(generator=lowerCamelCase_ ,num_inference_steps=2 ,output_type="""numpy""" ,return_dict=lowerCamelCase_ )[0]
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : str = np.array([0.8_5_1_2, 0.8_1_8, 0.6_4_1_1, 0.6_8_0_8, 0.4_4_6_5, 0.5_6_1_8, 0.4_6, 0.6_2_3_1, 0.5_1_7_2] )
UpperCAmelCase_ : Tuple = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < tolerance
@slow
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = LDMPipeline.from_pretrained("""CompVis/ldm-celebahq-256""" )
ldm.to(lowerCamelCase_ )
ldm.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = ldm(generator=lowerCamelCase_ ,num_inference_steps=5 ,output_type="""numpy""" ).images
UpperCAmelCase_ : Optional[Any] = image[0, -3:, -3:, -1]
assert image.shape == (1, 256, 256, 3)
UpperCAmelCase_ : int = np.array([0.4_3_9_9, 0.4_4_9_7_5, 0.4_6_8_2_5, 0.4_7_4, 0.4_3_5_9, 0.4_5_8_1, 0.4_5_0_9_5, 0.4_3_4_1, 0.4_4_4_7] )
UpperCAmelCase_ : Union[str, Any] = 1e-2 if torch_device != """mps""" else 3e-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
| 345 | 1 |
from typing import Optional
from torch import nn
from .transformer_ad import TransformeraDModel, TransformeraDModelOutput
class _snake_case ( nn.Module ):
'''simple docstring'''
def __init__( self: Dict ,lowerCamelCase_: int = 16 ,lowerCamelCase_: int = 88 ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: int = 1 ,lowerCamelCase_: float = 0.0 ,lowerCamelCase_: int = 32 ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: bool = False ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: str = "geglu" ,lowerCamelCase_: Optional[int] = None ,) -> Any:
super().__init__()
UpperCAmelCase_ : str = nn.ModuleList(
[
TransformeraDModel(
num_attention_heads=lowerCamelCase_ ,attention_head_dim=lowerCamelCase_ ,in_channels=lowerCamelCase_ ,num_layers=lowerCamelCase_ ,dropout=lowerCamelCase_ ,norm_num_groups=lowerCamelCase_ ,cross_attention_dim=lowerCamelCase_ ,attention_bias=lowerCamelCase_ ,sample_size=lowerCamelCase_ ,num_vector_embeds=lowerCamelCase_ ,activation_fn=lowerCamelCase_ ,num_embeds_ada_norm=lowerCamelCase_ ,)
for _ in range(2 )
] )
# Variables that can be set by a pipeline:
# The ratio of transformer1 to transformer2's output states to be combined during inference
UpperCAmelCase_ : Any = 0.5
# The shape of `encoder_hidden_states` is expected to be
# `(batch_size, condition_lengths[0]+condition_lengths[1], num_features)`
UpperCAmelCase_ : Optional[Any] = [77, 257]
# Which transformer to use to encode which condition.
# E.g. `(1, 0)` means that we'll use `transformers[1](conditions[0])` and `transformers[0](conditions[1])`
UpperCAmelCase_ : int = [1, 0]
def A__ ( self: int ,lowerCamelCase_: Any ,lowerCamelCase_: int ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: List[str]=None ,lowerCamelCase_: Dict=None ,lowerCamelCase_: bool = True ,) -> Union[str, Any]:
UpperCAmelCase_ : Any = hidden_states
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : List[str] = 0
# attention_mask is not used yet
for i in range(2 ):
# for each of the two transformers, pass the corresponding condition tokens
UpperCAmelCase_ : Tuple = encoder_hidden_states[:, tokens_start : tokens_start + self.condition_lengths[i]]
UpperCAmelCase_ : int = self.transformer_index_for_condition[i]
UpperCAmelCase_ : List[Any] = self.transformers[transformer_index](
lowerCamelCase_ ,encoder_hidden_states=lowerCamelCase_ ,timestep=lowerCamelCase_ ,cross_attention_kwargs=lowerCamelCase_ ,return_dict=lowerCamelCase_ ,)[0]
encoded_states.append(encoded_state - input_states )
tokens_start += self.condition_lengths[i]
UpperCAmelCase_ : Dict = encoded_states[0] * self.mix_ratio + encoded_states[1] * (1 - self.mix_ratio)
UpperCAmelCase_ : Tuple = output_states + input_states
if not return_dict:
return (output_states,)
return TransformeraDModelOutput(sample=lowerCamelCase_ )
| 345 |
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = [0] * len(_a )
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Dict = 0
for values in graph.values():
for i in values:
indegree[i] += 1
for i in range(len(_a ) ):
if indegree[i] == 0:
queue.append(_a )
while queue:
UpperCAmelCase_ : List[str] = queue.pop(0 )
cnt += 1
topo.append(_a )
for x in graph[vertex]:
indegree[x] -= 1
if indegree[x] == 0:
queue.append(_a )
if cnt != len(_a ):
print("""Cycle exists""" )
else:
print(_a )
# Adjacency List of Graph
UpperCamelCase_ = {0: [1, 2], 1: [3], 2: [3], 3: [4, 5], 4: [], 5: []}
topological_sort(graph)
| 345 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_torch_available,
)
UpperCamelCase_ = {
'''configuration_resnet''': ['''RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''ResNetConfig''', '''ResNetOnnxConfig''']
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''RESNET_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''ResNetForImageClassification''',
'''ResNetModel''',
'''ResNetPreTrainedModel''',
'''ResNetBackbone''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''TF_RESNET_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TFResNetForImageClassification''',
'''TFResNetModel''',
'''TFResNetPreTrainedModel''',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''FlaxResNetForImageClassification''',
'''FlaxResNetModel''',
'''FlaxResNetPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_resnet import RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ResNetConfig, ResNetOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_resnet import (
RESNET_PRETRAINED_MODEL_ARCHIVE_LIST,
ResNetBackbone,
ResNetForImageClassification,
ResNetModel,
ResNetPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_resnet import (
TF_RESNET_PRETRAINED_MODEL_ARCHIVE_LIST,
TFResNetForImageClassification,
TFResNetModel,
TFResNetPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_resnet import FlaxResNetForImageClassification, FlaxResNetModel, FlaxResNetPreTrainedModel
else:
import sys
UpperCamelCase_ = _LazyModule(__name__, globals()['''__file__'''], _import_structure)
| 345 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/swinv2-tiny-patch4-window8-256''': (
'''https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256/resolve/main/config.json'''
),
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "swinv2"
A__ : int = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self: List[str] ,lowerCamelCase_: List[str]=224 ,lowerCamelCase_: List[str]=4 ,lowerCamelCase_: List[Any]=3 ,lowerCamelCase_: Optional[Any]=96 ,lowerCamelCase_: Any=[2, 2, 6, 2] ,lowerCamelCase_: Dict=[3, 6, 12, 24] ,lowerCamelCase_: str=7 ,lowerCamelCase_: Optional[Any]=4.0 ,lowerCamelCase_: Tuple=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: str=False ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: Union[str, Any]=1e-5 ,lowerCamelCase_: str=32 ,**lowerCamelCase_: List[str] ,) -> Tuple:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = image_size
UpperCAmelCase_ : Tuple = patch_size
UpperCAmelCase_ : Dict = num_channels
UpperCAmelCase_ : List[Any] = embed_dim
UpperCAmelCase_ : Dict = depths
UpperCAmelCase_ : Dict = len(lowerCamelCase_ )
UpperCAmelCase_ : str = num_heads
UpperCAmelCase_ : Tuple = window_size
UpperCAmelCase_ : int = mlp_ratio
UpperCAmelCase_ : str = qkv_bias
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : Tuple = attention_probs_dropout_prob
UpperCAmelCase_ : int = drop_path_rate
UpperCAmelCase_ : Optional[Any] = hidden_act
UpperCAmelCase_ : List[str] = use_absolute_embeddings
UpperCAmelCase_ : Dict = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Union[str, Any] = encoder_stride
# we set the hidden_size attribute in order to make Swinv2 work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
UpperCAmelCase_ : List[str] = int(embed_dim * 2 ** (len(lowerCamelCase_ ) - 1) )
UpperCAmelCase_ : Any = (0, 0, 0, 0)
| 345 | 1 |
import json
import os
import tempfile
import unittest
import numpy as np
from datasets import load_dataset
from transformers.testing_utils import require_torch, require_vision, slow
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import ImageGPTImageProcessor
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def __init__( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[Any]=7 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Optional[int]=18 ,lowerCamelCase_: Tuple=30 ,lowerCamelCase_: Any=400 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: List[Any]=None ,lowerCamelCase_: int=True ,) -> Union[str, Any]:
UpperCAmelCase_ : Any = size if size is not None else {"""height""": 18, """width""": 18}
UpperCAmelCase_ : Tuple = parent
UpperCAmelCase_ : List[str] = batch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : Union[str, Any] = image_size
UpperCAmelCase_ : Optional[Any] = min_resolution
UpperCAmelCase_ : str = max_resolution
UpperCAmelCase_ : List[str] = do_resize
UpperCAmelCase_ : Any = size
UpperCAmelCase_ : Dict = do_normalize
def A__ ( self: str ) -> Any:
return {
# here we create 2 clusters for the sake of simplicity
"clusters": np.asarray(
[
[0.8_8_6_6_4_4_3_6_3_4_0_3_3_2_0_3, 0.6_6_1_8_8_2_9_3_6_9_5_4_4_9_8_3, 0.3_8_9_1_7_4_6_4_0_1_7_8_6_8_0_4],
[-0.6_0_4_2_5_5_9_1_4_6_8_8_1_1_0_4, -0.0_2_2_9_5_0_0_8_8_6_0_5_2_8_4_6_9, 0.5_4_2_3_7_9_7_3_6_9_0_0_3_2_9_6],
] ),
"do_resize": self.do_resize,
"size": self.size,
"do_normalize": self.do_normalize,
}
@require_torch
@require_vision
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = ImageGPTImageProcessor if is_vision_available() else None
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ : Optional[int] = ImageGPTImageProcessingTester(self )
@property
def A__ ( self: str ) -> List[Any]:
return self.image_processor_tester.prepare_image_processor_dict()
def A__ ( self: Optional[Any] ) -> Dict:
UpperCAmelCase_ : str = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(lowerCamelCase_ ,"""clusters""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""do_resize""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""size""" ) )
self.assertTrue(hasattr(lowerCamelCase_ ,"""do_normalize""" ) )
def A__ ( self: Dict ) -> Any:
UpperCAmelCase_ : Optional[Any] = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size ,{"""height""": 18, """width""": 18} )
UpperCAmelCase_ : Union[str, Any] = self.image_processing_class.from_dict(self.image_processor_dict ,size=42 )
self.assertEqual(image_processor.size ,{"""height""": 42, """width""": 42} )
def A__ ( self: Optional[Any] ) -> List[Any]:
UpperCAmelCase_ : str = self.image_processing_class(**self.image_processor_dict )
UpperCAmelCase_ : Any = json.loads(image_processor.to_json_string() )
for key, value in self.image_processor_dict.items():
if key == "clusters":
self.assertTrue(np.array_equal(lowerCamelCase_ ,obj[key] ) )
else:
self.assertEqual(obj[key] ,lowerCamelCase_ )
def A__ ( self: Dict ) -> Tuple:
UpperCAmelCase_ : Optional[int] = self.image_processing_class(**self.image_processor_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
UpperCAmelCase_ : Optional[Any] = os.path.join(lowerCamelCase_ ,"""image_processor.json""" )
image_processor_first.to_json_file(lowerCamelCase_ )
UpperCAmelCase_ : Any = self.image_processing_class.from_json_file(lowerCamelCase_ ).to_dict()
UpperCAmelCase_ : Union[str, Any] = image_processor_first.to_dict()
for key, value in image_processor_first.items():
if key == "clusters":
self.assertTrue(np.array_equal(lowerCamelCase_ ,image_processor_second[key] ) )
else:
self.assertEqual(image_processor_first[key] ,lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : Union[str, Any] = self.image_processing_class(**self.image_processor_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
image_processor_first.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : int = self.image_processing_class.from_pretrained(lowerCamelCase_ ).to_dict()
UpperCAmelCase_ : int = image_processor_first.to_dict()
for key, value in image_processor_first.items():
if key == "clusters":
self.assertTrue(np.array_equal(lowerCamelCase_ ,image_processor_second[key] ) )
else:
self.assertEqual(image_processor_first[key] ,lowerCamelCase_ )
@unittest.skip("""ImageGPT requires clusters at initialization""" )
def A__ ( self: Dict ) -> Optional[int]:
pass
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : Tuple = load_dataset("""hf-internal-testing/fixtures_image_utils""" , split="""test""" )
UpperCAmelCase_ : Union[str, Any] = Image.open(dataset[4]["""file"""] )
UpperCAmelCase_ : Dict = Image.open(dataset[5]["""file"""] )
UpperCAmelCase_ : Optional[int] = [imagea, imagea]
return images
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : int = ImageGPTImageProcessor.from_pretrained("""openai/imagegpt-small""" )
UpperCAmelCase_ : Optional[int] = prepare_images()
# test non-batched
UpperCAmelCase_ : Union[str, Any] = image_processing(images[0] ,return_tensors="""pt""" )
self.assertIsInstance(encoding.input_ids ,torch.LongTensor )
self.assertEqual(encoding.input_ids.shape ,(1, 1024) )
UpperCAmelCase_ : Optional[Any] = [306, 191, 191]
self.assertEqual(encoding.input_ids[0, :3].tolist() ,lowerCamelCase_ )
# test batched
UpperCAmelCase_ : Any = image_processing(lowerCamelCase_ ,return_tensors="""pt""" )
self.assertIsInstance(encoding.input_ids ,torch.LongTensor )
self.assertEqual(encoding.input_ids.shape ,(2, 1024) )
UpperCAmelCase_ : List[str] = [303, 13, 13]
self.assertEqual(encoding.input_ids[1, -3:].tolist() ,lowerCamelCase_ )
| 345 |
import os
import sys
import tempfile
import unittest
import unittest.mock as mock
from pathlib import Path
from huggingface_hub import HfFolder, delete_repo
from huggingface_hub.file_download import http_get
from requests.exceptions import HTTPError
from transformers import (
AlbertTokenizer,
AutoTokenizer,
BertTokenizer,
BertTokenizerFast,
GPTaTokenizerFast,
is_tokenizers_available,
)
from transformers.testing_utils import TOKEN, USER, is_staging_test, require_tokenizers
from transformers.tokenization_utils import Trie
sys.path.append(str(Path(__file__).parent.parent / '''utils'''))
from test_module.custom_tokenization import CustomTokenizer # noqa E402
if is_tokenizers_available():
from test_module.custom_tokenization_fast import CustomTokenizerFast
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: int ) -> str:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : List[str] = mock.Mock()
UpperCAmelCase_ : List[Any] = 500
UpperCAmelCase_ : Union[str, Any] = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : Any = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Union[str, Any] = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = BertTokenizer.from_pretrained("""hf-internal-testing/tiny-random-bert""" )
# This check we did call the fake head request
mock_head.assert_called()
@require_tokenizers
def A__ ( self: str ) -> int:
# A mock response for an HTTP head request to emulate server down
UpperCAmelCase_ : str = mock.Mock()
UpperCAmelCase_ : Optional[int] = 500
UpperCAmelCase_ : int = {}
UpperCAmelCase_ : Union[str, Any] = HTTPError
UpperCAmelCase_ : List[Any] = {}
# Download this model to make sure it's in the cache.
UpperCAmelCase_ : Optional[int] = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("""requests.Session.request""" ,return_value=lowerCamelCase_ ) as mock_head:
UpperCAmelCase_ : Any = GPTaTokenizerFast.from_pretrained("""gpt2""" )
# This check we did call the fake head request
mock_head.assert_called()
def A__ ( self: str ) -> Dict:
# This test is for deprecated behavior and can be removed in v5
try:
UpperCAmelCase_ : Any = tempfile.mktemp()
with open(lowerCamelCase_ ,"""wb""" ) as f:
http_get("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = AlbertTokenizer.from_pretrained(lowerCamelCase_ )
finally:
os.remove(lowerCamelCase_ )
# Supporting this legacy load introduced a weird bug where the tokenizer would load local files if they are in
# the current folder and have the right name.
if os.path.isfile("""tokenizer.json""" ):
# We skip the test if the user has a `tokenizer.json` in this folder to avoid deleting it.
return
try:
with open("""tokenizer.json""" ,"""wb""" ) as f:
http_get("""https://huggingface.co/hf-internal-testing/tiny-random-bert/blob/main/tokenizer.json""" ,lowerCamelCase_ )
UpperCAmelCase_ : str = AutoTokenizer.from_pretrained("""hf-internal-testing/tiny-random-gpt2""" )
# The tiny random BERT has a vocab size of 1024, tiny gpt2 as a vocab size of 1000
self.assertEqual(tokenizer.vocab_size ,1000 )
# Tokenizer should depend on the remote checkpoint, not the local tokenizer.json file.
finally:
os.remove("""tokenizer.json""" )
def A__ ( self: List[str] ) -> Tuple:
# This test is for deprecated behavior and can be removed in v5
UpperCAmelCase_ : str = AlbertTokenizer.from_pretrained("""https://huggingface.co/albert-base-v1/resolve/main/spiece.model""" )
@is_staging_test
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
A__ : str = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "bla", "blou"]
@classmethod
def A__ ( cls: Dict ) -> Optional[int]:
UpperCAmelCase_ : List[str] = TOKEN
HfFolder.save_token(lowerCamelCase_ )
@classmethod
def A__ ( cls: Optional[Any] ) -> List[str]:
try:
delete_repo(token=cls._token ,repo_id="""test-tokenizer""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""valid_org/test-tokenizer-org""" )
except HTTPError:
pass
try:
delete_repo(token=cls._token ,repo_id="""test-dynamic-tokenizer""" )
except HTTPError:
pass
def A__ ( self: Any ) -> Optional[int]:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Tuple = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : List[Any] = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""test-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""test-tokenizer""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(lowerCamelCase_ ,repo_id="""test-tokenizer""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained(F'''{USER}/test-tokenizer''' )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
def A__ ( self: Optional[int] ) -> Any:
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[Any] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Dict = BertTokenizer(lowerCamelCase_ )
tokenizer.push_to_hub("""valid_org/test-tokenizer-org""" ,use_auth_token=self._token )
UpperCAmelCase_ : Dict = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
# Reset repo
delete_repo(token=self._token ,repo_id="""valid_org/test-tokenizer-org""" )
# Push to hub via save_pretrained
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer.save_pretrained(
lowerCamelCase_ ,repo_id="""valid_org/test-tokenizer-org""" ,push_to_hub=lowerCamelCase_ ,use_auth_token=self._token )
UpperCAmelCase_ : List[Any] = BertTokenizer.from_pretrained("""valid_org/test-tokenizer-org""" )
self.assertDictEqual(new_tokenizer.vocab ,tokenizer.vocab )
@require_tokenizers
def A__ ( self: Optional[int] ) -> Optional[Any]:
CustomTokenizer.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : Any = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : Optional[Any] = CustomTokenizer(lowerCamelCase_ )
# No fast custom tokenizer
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : Optional[Any] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the CustomTokenizer class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
# Fast and slow custom tokenizer
CustomTokenizerFast.register_for_auto_class()
with tempfile.TemporaryDirectory() as tmp_dir:
UpperCAmelCase_ : List[str] = os.path.join(lowerCamelCase_ ,"""vocab.txt""" )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in self.vocab_tokens] ) )
UpperCAmelCase_ : str = BertTokenizerFast.from_pretrained(lowerCamelCase_ )
bert_tokenizer.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = CustomTokenizerFast.from_pretrained(lowerCamelCase_ )
tokenizer.push_to_hub("""test-dynamic-tokenizer""" ,use_auth_token=self._token )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(F'''{USER}/test-dynamic-tokenizer''' ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizerFast""" )
UpperCAmelCase_ : List[str] = AutoTokenizer.from_pretrained(
F'''{USER}/test-dynamic-tokenizer''' ,use_fast=lowerCamelCase_ ,trust_remote_code=lowerCamelCase_ )
# Can't make an isinstance check because the new_model.config is from the FakeConfig class of a dynamic module
self.assertEqual(tokenizer.__class__.__name__ ,"""CustomTokenizer""" )
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Any = Trie()
trie.add("""Hello 友達""" )
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {""" """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
trie.add("""Hello""" )
trie.data
self.assertEqual(trie.data ,{"""H""": {"""e""": {"""l""": {"""l""": {"""o""": {"""""": 1, """ """: {"""友""": {"""達""": {"""""": 1}}}}}}}}} )
def A__ ( self: Tuple ) -> Optional[int]:
UpperCAmelCase_ : str = Trie()
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS] This is a extra_id_100"""] )
trie.add("""[CLS]""" )
trie.add("""extra_id_1""" )
trie.add("""extra_id_100""" )
self.assertEqual(trie.split("""[CLS] This is a extra_id_100""" ) ,["""[CLS]""", """ This is a """, """extra_id_100"""] )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ : Dict = Trie()
trie.add("""A""" )
self.assertEqual(trie.split("""ABC""" ) ,["""A""", """BC"""] )
self.assertEqual(trie.split("""BCA""" ) ,["""BC""", """A"""] )
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""TOKEN]""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> Union[str, Any]:
UpperCAmelCase_ : List[str] = Trie()
trie.add("""A""" )
trie.add("""P""" )
trie.add("""[SPECIAL_TOKEN]""" )
self.assertEqual(trie.split("""This is something [SPECIAL_TOKEN]""" ) ,["""This is something """, """[SPECIAL_TOKEN]"""] )
def A__ ( self: int ) -> List[str]:
UpperCAmelCase_ : int = Trie()
trie.add("""AB""" )
trie.add("""B""" )
trie.add("""C""" )
self.assertEqual(trie.split("""ABC""" ) ,["""AB""", """C"""] )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = Trie()
trie.add("""ABC""" )
trie.add("""B""" )
trie.add("""CD""" )
self.assertEqual(trie.split("""ABCD""" ) ,["""ABC""", """D"""] )
def A__ ( self: List[Any] ) -> Any:
# Even if the offsets are wrong, we necessarily output correct string
# parts.
UpperCAmelCase_ : Tuple = Trie()
UpperCAmelCase_ : Optional[Any] = trie.cut_text("""ABC""" ,[0, 0, 2, 1, 2, 3] )
self.assertEqual(lowerCamelCase_ ,["""AB""", """C"""] )
| 345 | 1 |
def lowerCamelCase_ ( _a : int ):
'''simple docstring'''
if divisor % 5 == 0 or divisor % 2 == 0:
return 0
UpperCAmelCase_ : Optional[Any] = 1
UpperCAmelCase_ : int = 1
while repunit:
UpperCAmelCase_ : Union[str, Any] = (10 * repunit + 1) % divisor
repunit_index += 1
return repunit_index
def lowerCamelCase_ ( _a : int = 100_0000 ):
'''simple docstring'''
UpperCAmelCase_ : int = limit - 1
if divisor % 2 == 0:
divisor += 1
while least_divisible_repunit(_a ) <= limit:
divisor += 2
return divisor
if __name__ == "__main__":
print(F"{solution() = }")
| 345 |
from ..utils import DummyObject, requires_backends
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: int ,**lowerCamelCase_: List[str] ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: List[str] ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: int ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Any = ["flax"]
def __init__( self: int ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[str] ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Dict = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[Any] ) -> Any:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: List[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[str] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: List[str] ,**lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : int = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Union[str, Any] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[Any] ) -> str:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Optional[int] ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Optional[Any] ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : List[Any] = ["flax"]
def __init__( self: Union[str, Any] ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> List[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Dict:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: str ,*lowerCamelCase_: Any ,**lowerCamelCase_: int ) -> Tuple:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: Union[str, Any] ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: Dict ) -> Optional[int]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : str = ["flax"]
def __init__( self: Optional[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: List[str] ) -> Optional[Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: int ) -> List[str]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: str ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: int ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: Optional[int] ) -> List[str]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[int] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: str ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Tuple = ["flax"]
def __init__( self: Any ,*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: Dict ) -> str:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Tuple ,*lowerCamelCase_: Union[str, Any] ,**lowerCamelCase_: List[str] ) -> int:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: List[Any] ,*lowerCamelCase_: str ,**lowerCamelCase_: str ) -> Any:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[Any] = ["flax"]
def __init__( self: Dict ,*lowerCamelCase_: int ,**lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: int ,**lowerCamelCase_: Tuple ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: Optional[Any] ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Optional[int] ) -> int:
requires_backends(cls ,["""flax"""] )
class _snake_case ( metaclass=__snake_case ):
'''simple docstring'''
A__ : Optional[int] = ["flax"]
def __init__( self: List[str] ,*lowerCamelCase_: Dict ,**lowerCamelCase_: Dict ) -> int:
requires_backends(self ,["""flax"""] )
@classmethod
def A__ ( cls: Dict ,*lowerCamelCase_: List[Any] ,**lowerCamelCase_: Dict ) -> Union[str, Any]:
requires_backends(cls ,["""flax"""] )
@classmethod
def A__ ( cls: int ,*lowerCamelCase_: Any ,**lowerCamelCase_: Any ) -> Optional[Any]:
requires_backends(cls ,["""flax"""] )
| 345 | 1 |
from math import pi, sqrt
def lowerCamelCase_ ( _a : float ):
'''simple docstring'''
if num <= 0:
raise ValueError("""math domain error""" )
if num > 1_7_1.5:
raise OverflowError("""math range error""" )
elif num - int(_a ) not in (0, 0.5):
raise NotImplementedError("""num must be an integer or a half-integer""" )
elif num == 0.5:
return sqrt(_a )
else:
return 1.0 if num == 1 else (num - 1) * gamma(num - 1 )
def lowerCamelCase_ ( ):
'''simple docstring'''
assert gamma(0.5 ) == sqrt(_a )
assert gamma(1 ) == 1.0
assert gamma(2 ) == 1.0
if __name__ == "__main__":
from doctest import testmod
testmod()
UpperCamelCase_ = 1.0
while num:
UpperCamelCase_ = float(input('''Gamma of: '''))
print(F"gamma({num}) = {gamma(num)}")
print('''\nEnter 0 to exit...''')
| 345 |
import random
from typing import Any
def lowerCamelCase_ ( _a : list ):
'''simple docstring'''
for _ in range(len(_a ) ):
UpperCAmelCase_ : Tuple = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ : List[Any] = random.randint(0 , len(_a ) - 1 )
UpperCAmelCase_ , UpperCAmelCase_ : int = data[b], data[a]
return data
if __name__ == "__main__":
UpperCamelCase_ = [0, 1, 2, 3, 4, 5, 6, 7]
UpperCamelCase_ = ['''python''', '''says''', '''hello''', '''!''']
print('''Fisher-Yates Shuffle:''')
print('''List''', integers, strings)
print('''FY Shuffle''', fisher_yates_shuffle(integers), fisher_yates_shuffle(strings))
| 345 | 1 |
import json
from typing import List, Optional, Tuple
from tokenizers import normalizers
from ...tokenization_utils_base import BatchEncoding
from ...tokenization_utils_fast import PreTrainedTokenizerFast
from ...utils import PaddingStrategy, logging
from .tokenization_realm import RealmTokenizer
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {'''vocab_file''': '''vocab.txt''', '''tokenizer_file''': '''tokenizer.json'''}
UpperCamelCase_ = {
'''vocab_file''': {
'''google/realm-cc-news-pretrained-embedder''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-embedder/resolve/main/vocab.txt'''
),
'''google/realm-cc-news-pretrained-encoder''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-encoder/resolve/main/vocab.txt'''
),
'''google/realm-cc-news-pretrained-scorer''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-scorer/resolve/main/vocab.txt'''
),
'''google/realm-cc-news-pretrained-openqa''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-openqa/aresolve/main/vocab.txt'''
),
'''google/realm-orqa-nq-openqa''': '''https://huggingface.co/google/realm-orqa-nq-openqa/resolve/main/vocab.txt''',
'''google/realm-orqa-nq-reader''': '''https://huggingface.co/google/realm-orqa-nq-reader/resolve/main/vocab.txt''',
'''google/realm-orqa-wq-openqa''': '''https://huggingface.co/google/realm-orqa-wq-openqa/resolve/main/vocab.txt''',
'''google/realm-orqa-wq-reader''': '''https://huggingface.co/google/realm-orqa-wq-reader/resolve/main/vocab.txt''',
},
'''tokenizer_file''': {
'''google/realm-cc-news-pretrained-embedder''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-embedder/resolve/main/tokenizer.jsont'''
),
'''google/realm-cc-news-pretrained-encoder''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-encoder/resolve/main/tokenizer.json'''
),
'''google/realm-cc-news-pretrained-scorer''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-scorer/resolve/main/tokenizer.json'''
),
'''google/realm-cc-news-pretrained-openqa''': (
'''https://huggingface.co/google/realm-cc-news-pretrained-openqa/aresolve/main/tokenizer.json'''
),
'''google/realm-orqa-nq-openqa''': (
'''https://huggingface.co/google/realm-orqa-nq-openqa/resolve/main/tokenizer.json'''
),
'''google/realm-orqa-nq-reader''': (
'''https://huggingface.co/google/realm-orqa-nq-reader/resolve/main/tokenizer.json'''
),
'''google/realm-orqa-wq-openqa''': (
'''https://huggingface.co/google/realm-orqa-wq-openqa/resolve/main/tokenizer.json'''
),
'''google/realm-orqa-wq-reader''': (
'''https://huggingface.co/google/realm-orqa-wq-reader/resolve/main/tokenizer.json'''
),
},
}
UpperCamelCase_ = {
'''google/realm-cc-news-pretrained-embedder''': 512,
'''google/realm-cc-news-pretrained-encoder''': 512,
'''google/realm-cc-news-pretrained-scorer''': 512,
'''google/realm-cc-news-pretrained-openqa''': 512,
'''google/realm-orqa-nq-openqa''': 512,
'''google/realm-orqa-nq-reader''': 512,
'''google/realm-orqa-wq-openqa''': 512,
'''google/realm-orqa-wq-reader''': 512,
}
UpperCamelCase_ = {
'''google/realm-cc-news-pretrained-embedder''': {'''do_lower_case''': True},
'''google/realm-cc-news-pretrained-encoder''': {'''do_lower_case''': True},
'''google/realm-cc-news-pretrained-scorer''': {'''do_lower_case''': True},
'''google/realm-cc-news-pretrained-openqa''': {'''do_lower_case''': True},
'''google/realm-orqa-nq-openqa''': {'''do_lower_case''': True},
'''google/realm-orqa-nq-reader''': {'''do_lower_case''': True},
'''google/realm-orqa-wq-openqa''': {'''do_lower_case''': True},
'''google/realm-orqa-wq-reader''': {'''do_lower_case''': True},
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = VOCAB_FILES_NAMES
A__ : Optional[Any] = PRETRAINED_VOCAB_FILES_MAP
A__ : Optional[int] = PRETRAINED_INIT_CONFIGURATION
A__ : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Tuple = RealmTokenizer
def __init__( self: Tuple ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: str=None ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[Any]="[UNK]" ,lowerCamelCase_: str="[SEP]" ,lowerCamelCase_: int="[PAD]" ,lowerCamelCase_: Union[str, Any]="[CLS]" ,lowerCamelCase_: List[Any]="[MASK]" ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: Tuple=None ,**lowerCamelCase_: Any ,) -> str:
super().__init__(
lowerCamelCase_ ,tokenizer_file=lowerCamelCase_ ,do_lower_case=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,tokenize_chinese_chars=lowerCamelCase_ ,strip_accents=lowerCamelCase_ ,**lowerCamelCase_ ,)
UpperCAmelCase_ : int = json.loads(self.backend_tokenizer.normalizer.__getstate__() )
if (
normalizer_state.get("""lowercase""" ,lowerCamelCase_ ) != do_lower_case
or normalizer_state.get("""strip_accents""" ,lowerCamelCase_ ) != strip_accents
or normalizer_state.get("""handle_chinese_chars""" ,lowerCamelCase_ ) != tokenize_chinese_chars
):
UpperCAmelCase_ : Optional[Any] = getattr(lowerCamelCase_ ,normalizer_state.pop("""type""" ) )
UpperCAmelCase_ : Tuple = do_lower_case
UpperCAmelCase_ : Union[str, Any] = strip_accents
UpperCAmelCase_ : List[Any] = tokenize_chinese_chars
UpperCAmelCase_ : Dict = normalizer_class(**lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = do_lower_case
def A__ ( self: Optional[Any] ,lowerCamelCase_: int ,**lowerCamelCase_: str ) -> List[Any]:
UpperCAmelCase_ : Dict = PaddingStrategy.MAX_LENGTH
UpperCAmelCase_ : int = text
UpperCAmelCase_ : Optional[Any] = kwargs.pop("""text_pair""" ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = kwargs.pop("""return_tensors""" ,lowerCamelCase_ )
UpperCAmelCase_ : Tuple = {
"""input_ids""": [],
"""attention_mask""": [],
"""token_type_ids""": [],
}
for idx, candidate_text in enumerate(lowerCamelCase_ ):
if batch_text_pair is not None:
UpperCAmelCase_ : int = batch_text_pair[idx]
else:
UpperCAmelCase_ : Dict = None
UpperCAmelCase_ : List[str] = super().__call__(lowerCamelCase_ ,lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = encoded_candidates.get("""input_ids""" )
UpperCAmelCase_ : List[Any] = encoded_candidates.get("""attention_mask""" )
UpperCAmelCase_ : Optional[int] = encoded_candidates.get("""token_type_ids""" )
if encoded_input_ids is not None:
output_data["input_ids"].append(lowerCamelCase_ )
if encoded_attention_mask is not None:
output_data["attention_mask"].append(lowerCamelCase_ )
if encoded_token_type_ids is not None:
output_data["token_type_ids"].append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = {key: item for key, item in output_data.items() if len(lowerCamelCase_ ) != 0}
return BatchEncoding(lowerCamelCase_ ,tensor_type=lowerCamelCase_ )
def A__ ( self: str ,lowerCamelCase_: Dict ,lowerCamelCase_: Dict=None ) -> List[Any]:
UpperCAmelCase_ : Union[str, Any] = [self.cls_token_id] + token_ids_a + [self.sep_token_id]
if token_ids_a:
output += token_ids_a + [self.sep_token_id]
return output
def A__ ( self: int ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Tuple = [self.sep_token_id]
UpperCAmelCase_ : List[Any] = [self.cls_token_id]
if token_ids_a is None:
return len(cls + token_ids_a + sep ) * [0]
return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1]
def A__ ( self: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
UpperCAmelCase_ : Optional[Any] = self._tokenizer.model.save(lowerCamelCase_ ,name=lowerCamelCase_ )
return tuple(lowerCamelCase_ )
| 345 |
import flax.linen as nn
import jax.numpy as jnp
from .attention_flax import FlaxTransformeraDModel
from .resnet_flax import FlaxDownsampleaD, FlaxResnetBlockaD, FlaxUpsampleaD
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = []
UpperCAmelCase_ : Optional[int] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : List[Any] = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : List[Any] = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : int = resnets
UpperCAmelCase_ : Tuple = attentions
if self.add_downsample:
UpperCAmelCase_ : List[Any] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: int=True ) -> int:
UpperCAmelCase_ : List[Any] = ()
for resnet, attn in zip(self.resnets ,self.attentions ):
UpperCAmelCase_ : str = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[Any] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> int:
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if i == 0 else self.out_channels
UpperCAmelCase_ : Dict = FlaxResnetBlockaD(
in_channels=lowerCamelCase_ ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnets
if self.add_downsample:
UpperCAmelCase_ : List[str] = FlaxDownsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[Any]=True ) -> Any:
UpperCAmelCase_ : Union[str, Any] = ()
for resnet in self.resnets:
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
output_states += (hidden_states,)
if self.add_downsample:
UpperCAmelCase_ : List[str] = self.downsamplers_a(lowerCamelCase_ )
output_states += (hidden_states,)
return hidden_states, output_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = True
A__ : bool = False
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: str ) -> Any:
UpperCAmelCase_ : Dict = []
UpperCAmelCase_ : List[str] = []
for i in range(self.num_layers ):
UpperCAmelCase_ : int = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : int = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Optional[Any] = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : int = FlaxTransformeraDModel(
in_channels=self.out_channels ,n_heads=self.num_attention_heads ,d_head=self.out_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,only_cross_attention=self.only_cross_attention ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : List[str] = resnets
UpperCAmelCase_ : Dict = attentions
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Optional[int] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: Any ,lowerCamelCase_: str ,lowerCamelCase_: List[str]=True ) -> List[str]:
for resnet, attn in zip(self.resnets ,self.attentions ):
# pop res hidden states
UpperCAmelCase_ : List[str] = res_hidden_states_tuple[-1]
UpperCAmelCase_ : Union[str, Any] = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : Optional[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : Tuple = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Dict = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : int
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : bool = True
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> Dict:
UpperCAmelCase_ : Any = []
for i in range(self.num_layers ):
UpperCAmelCase_ : str = self.in_channels if (i == self.num_layers - 1) else self.out_channels
UpperCAmelCase_ : Optional[int] = self.prev_output_channel if i == 0 else self.out_channels
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=resnet_in_channels + res_skip_channels ,out_channels=self.out_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : str = resnets
if self.add_upsample:
UpperCAmelCase_ : Union[str, Any] = FlaxUpsampleaD(self.out_channels ,dtype=self.dtype )
def __call__( self: Dict ,lowerCamelCase_: Dict ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any=True ) -> List[str]:
for resnet in self.resnets:
# pop res hidden states
UpperCAmelCase_ : Dict = res_hidden_states_tuple[-1]
UpperCAmelCase_ : str = res_hidden_states_tuple[:-1]
UpperCAmelCase_ : List[Any] = jnp.concatenate((hidden_states, res_hidden_states) ,axis=-1 )
UpperCAmelCase_ : List[str] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
if self.add_upsample:
UpperCAmelCase_ : Optional[Any] = self.upsamplers_a(lowerCamelCase_ )
return hidden_states
class _snake_case ( nn.Module ):
'''simple docstring'''
A__ : int
A__ : float = 0.0
A__ : int = 1
A__ : int = 1
A__ : bool = False
A__ : bool = False
A__ : jnp.dtype = jnp.floataa
def A__ ( self: Dict ) -> List[str]:
# there is always at least one resnet
UpperCAmelCase_ : List[Any] = [
FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
]
UpperCAmelCase_ : Any = []
for _ in range(self.num_layers ):
UpperCAmelCase_ : Optional[Any] = FlaxTransformeraDModel(
in_channels=self.in_channels ,n_heads=self.num_attention_heads ,d_head=self.in_channels // self.num_attention_heads ,depth=1 ,use_linear_projection=self.use_linear_projection ,use_memory_efficient_attention=self.use_memory_efficient_attention ,dtype=self.dtype ,)
attentions.append(lowerCamelCase_ )
UpperCAmelCase_ : Any = FlaxResnetBlockaD(
in_channels=self.in_channels ,out_channels=self.in_channels ,dropout_prob=self.dropout ,dtype=self.dtype ,)
resnets.append(lowerCamelCase_ )
UpperCAmelCase_ : Dict = resnets
UpperCAmelCase_ : Any = attentions
def __call__( self: str ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any]=True ) -> List[Any]:
UpperCAmelCase_ : List[Any] = self.resnets[0](lowerCamelCase_ ,lowerCamelCase_ )
for attn, resnet in zip(self.attentions ,self.resnets[1:] ):
UpperCAmelCase_ : Optional[Any] = attn(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = resnet(lowerCamelCase_ ,lowerCamelCase_ ,deterministic=lowerCamelCase_ )
return hidden_states
| 345 | 1 |
import collections
import inspect
import unittest
from transformers import FocalNetConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
FocalNetBackbone,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetModel,
)
from transformers.models.focalnet.modeling_focalnet import FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=13 ,lowerCamelCase_: Optional[int]=32 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Optional[Any]=[32, 64, 128] ,lowerCamelCase_: Optional[int]=[1, 2, 1] ,lowerCamelCase_: Union[str, Any]=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[str]=2.0 ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: int=True ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: str=True ,lowerCamelCase_: Dict=10 ,lowerCamelCase_: str=8 ,lowerCamelCase_: Union[str, Any]=["stage1", "stage2"] ,lowerCamelCase_: Optional[Any]=[1, 2] ,) -> str:
UpperCAmelCase_ : List[Any] = parent
UpperCAmelCase_ : Tuple = batch_size
UpperCAmelCase_ : Any = image_size
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : List[str] = num_channels
UpperCAmelCase_ : Dict = embed_dim
UpperCAmelCase_ : Dict = hidden_sizes
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = num_heads
UpperCAmelCase_ : List[Any] = window_size
UpperCAmelCase_ : Union[str, Any] = mlp_ratio
UpperCAmelCase_ : int = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Union[str, Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[int] = drop_path_rate
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : List[Any] = use_absolute_embeddings
UpperCAmelCase_ : List[Any] = patch_norm
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Optional[Any] = scope
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = encoder_stride
UpperCAmelCase_ : Optional[int] = out_features
UpperCAmelCase_ : Optional[int] = out_indices
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : int = None
if self.use_labels:
UpperCAmelCase_ : str = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Any = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Tuple:
return FocalNetConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,hidden_sizes=self.hidden_sizes ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,out_features=self.out_features ,out_indices=self.out_indices ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : Dict = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : Optional[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[str]:
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size, 8, 8] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,len(config.out_features ) )
self.parent.assertListEqual(model.channels ,config.hidden_sizes[:-1] )
# verify backbone works with out_features=None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size * 2, 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,1 )
self.parent.assertListEqual(model.channels ,[config.hidden_sizes[-1]] )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Any = FocalNetForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.reconstruction.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : int = 1
UpperCAmelCase_ : List[str] = FocalNetForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.reconstruction.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ) -> int:
UpperCAmelCase_ : List[Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
# test greyscale images
UpperCAmelCase_ : List[Any] = 1
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = config_and_inputs
UpperCAmelCase_ : int = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (
(
FocalNetModel,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetBackbone,
)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (
{"feature-extraction": FocalNetModel, "image-classification": FocalNetForImageClassification}
if is_torch_available()
else {}
)
A__ : Optional[Any] = False
A__ : Any = False
A__ : List[str] = False
A__ : Any = False
A__ : Any = False
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : Dict = FocalNetModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: List[str] ) -> Union[str, Any]:
return
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCamelCase_ )
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@unittest.skip(reason="""FocalNet does not use inputs_embeds""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""FocalNet does not use feedforward chunking""" )
def A__ ( self: Optional[Any] ) -> Optional[Any]:
pass
def A__ ( self: Optional[Any] ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : List[Any] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Dict = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Any = [*signature.parameters.keys()]
UpperCAmelCase_ : List[str] = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.hidden_states
UpperCAmelCase_ : List[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# FocalNet has a different seq_length
UpperCAmelCase_ : int = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Optional[int] = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Union[str, Any] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = reshaped_hidden_states[0].shape
UpperCAmelCase_ : List[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Union[str, Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Tuple = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : Union[str, Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Union[str, Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Any = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Optional[int] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
@slow
def A__ ( self: Optional[int] ) -> Optional[Any]:
for model_name in FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Tuple = FocalNetModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[Any] = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Optional[int] ) -> str:
# TODO update organization
return AutoImageProcessor.from_pretrained("""microsoft/focalnet-tiny""" ) if is_vision_available() else None
@slow
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification.from_pretrained("""microsoft/focalnet-tiny""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.default_image_processor
UpperCAmelCase_ : Union[str, Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Dict = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : str = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_1_6_6, -0.4_3_6_8, 0.2_1_9_1] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
self.assertTrue(outputs.logits.argmax(dim=-1 ).item() ,281 )
@require_torch
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (FocalNetBackbone,) if is_torch_available() else ()
A__ : int = FocalNetConfig
A__ : List[str] = False
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : str = FocalNetModelTester(self )
| 345 |
import pickle
import numpy as np
from matplotlib import pyplot as plt
class _snake_case :
'''simple docstring'''
def __init__( self: Any ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Tuple=0.2 ,lowerCamelCase_: Union[str, Any]=0.2 ) -> List[str]:
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : str = bp_numa
UpperCAmelCase_ : List[Any] = bp_numa
UpperCAmelCase_ : Optional[int] = conva_get[:2]
UpperCAmelCase_ : List[Any] = conva_get[2]
UpperCAmelCase_ : str = size_pa
UpperCAmelCase_ : Optional[int] = rate_w
UpperCAmelCase_ : Dict = rate_t
UpperCAmelCase_ : List[Any] = [
np.mat(-1 * np.random.rand(self.conva[0] ,self.conva[0] ) + 0.5 )
for i in range(self.conva[1] )
]
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : int = np.mat(-1 * np.random.rand(self.num_bpa ,self.num_bpa ) + 0.5 )
UpperCAmelCase_ : Dict = -2 * np.random.rand(self.conva[1] ) + 1
UpperCAmelCase_ : str = -2 * np.random.rand(self.num_bpa ) + 1
UpperCAmelCase_ : Union[str, Any] = -2 * np.random.rand(self.num_bpa ) + 1
def A__ ( self: str ,lowerCamelCase_: Optional[Any] ) -> Tuple:
# save model dict with pickle
UpperCAmelCase_ : Dict = {
"""num_bp1""": self.num_bpa,
"""num_bp2""": self.num_bpa,
"""num_bp3""": self.num_bpa,
"""conv1""": self.conva,
"""step_conv1""": self.step_conva,
"""size_pooling1""": self.size_poolinga,
"""rate_weight""": self.rate_weight,
"""rate_thre""": self.rate_thre,
"""w_conv1""": self.w_conva,
"""wkj""": self.wkj,
"""vji""": self.vji,
"""thre_conv1""": self.thre_conva,
"""thre_bp2""": self.thre_bpa,
"""thre_bp3""": self.thre_bpa,
}
with open(lowerCamelCase_ ,"""wb""" ) as f:
pickle.dump(lowerCamelCase_ ,lowerCamelCase_ )
print(F'''Model saved： {save_path}''' )
@classmethod
def A__ ( cls: List[str] ,lowerCamelCase_: str ) -> List[str]:
# read saved model
with open(lowerCamelCase_ ,"""rb""" ) as f:
UpperCAmelCase_ : Any = pickle.load(lowerCamelCase_ ) # noqa: S301
UpperCAmelCase_ : Union[str, Any] = model_dic.get("""conv1""" )
conv_get.append(model_dic.get("""step_conv1""" ) )
UpperCAmelCase_ : List[str] = model_dic.get("""size_pooling1""" )
UpperCAmelCase_ : Tuple = model_dic.get("""num_bp1""" )
UpperCAmelCase_ : Optional[Any] = model_dic.get("""num_bp2""" )
UpperCAmelCase_ : List[str] = model_dic.get("""num_bp3""" )
UpperCAmelCase_ : List[Any] = model_dic.get("""rate_weight""" )
UpperCAmelCase_ : Dict = model_dic.get("""rate_thre""" )
# create model instance
UpperCAmelCase_ : List[Any] = CNN(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# modify model parameter
UpperCAmelCase_ : Any = model_dic.get("""w_conv1""" )
UpperCAmelCase_ : int = model_dic.get("""wkj""" )
UpperCAmelCase_ : int = model_dic.get("""vji""" )
UpperCAmelCase_ : Optional[int] = model_dic.get("""thre_conv1""" )
UpperCAmelCase_ : List[str] = model_dic.get("""thre_bp2""" )
UpperCAmelCase_ : Dict = model_dic.get("""thre_bp3""" )
return conv_ins
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> Tuple:
return 1 / (1 + np.exp(-1 * x ))
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
return round(lowerCamelCase_ ,3 )
def A__ ( self: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: str ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Any:
# convolution process
UpperCAmelCase_ : Optional[Any] = convs[0]
UpperCAmelCase_ : int = convs[1]
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )[0]
# get the data slice of original image data, data_focus
UpperCAmelCase_ : Dict = []
for i_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
for j_focus in range(0 ,size_data - size_conv + 1 ,lowerCamelCase_ ):
UpperCAmelCase_ : Union[str, Any] = data[
i_focus : i_focus + size_conv, j_focus : j_focus + size_conv
]
data_focus.append(lowerCamelCase_ )
# calculate the feature map of every single kernel, and saved as list of matrix
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = int((size_data - size_conv) / conv_step + 1 )
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[int] = []
for i_focus in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = (
np.sum(np.multiply(data_focus[i_focus] ,w_convs[i_map] ) )
- thre_convs[i_map]
)
featuremap.append(self.sig(lowerCamelCase_ ) )
UpperCAmelCase_ : Union[str, Any] = np.asmatrix(lowerCamelCase_ ).reshape(
lowerCamelCase_ ,lowerCamelCase_ )
data_featuremap.append(lowerCamelCase_ )
# expanding the data slice to One dimenssion
UpperCAmelCase_ : Optional[Any] = []
for each_focus in data_focus:
focusa_list.extend(self.Expand_Mat(lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[int] = np.asarray(lowerCamelCase_ )
return focus_list, data_featuremap
def A__ ( self: Tuple ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any]="average_pool" ) -> List[Any]:
# pooling process
UpperCAmelCase_ : Optional[Any] = len(featuremaps[0] )
UpperCAmelCase_ : Any = int(size_map / size_pooling )
UpperCAmelCase_ : Optional[int] = []
for i_map in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Any = featuremaps[i_map]
UpperCAmelCase_ : Tuple = []
for i_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j_focus in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : str = feature_map[
i_focus : i_focus + size_pooling,
j_focus : j_focus + size_pooling,
]
if pooling_type == "average_pool":
# average pooling
map_pooled.append(np.average(lowerCamelCase_ ) )
elif pooling_type == "max_pooling":
# max pooling
map_pooled.append(np.max(lowerCamelCase_ ) )
UpperCAmelCase_ : int = np.asmatrix(lowerCamelCase_ ).reshape(lowerCamelCase_ ,lowerCamelCase_ )
featuremap_pooled.append(lowerCamelCase_ )
return featuremap_pooled
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> Optional[int]:
# expanding three dimension data to one dimension list
UpperCAmelCase_ : List[Any] = []
for i in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : Tuple = np.shape(data[i] )
UpperCAmelCase_ : Optional[int] = data[i].reshape(1 ,shapes[0] * shapes[1] )
UpperCAmelCase_ : Optional[int] = data_listed.getA().tolist()[0]
data_expanded.extend(lowerCamelCase_ )
UpperCAmelCase_ : int = np.asarray(lowerCamelCase_ )
return data_expanded
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> Union[str, Any]:
# expanding matrix to one dimension list
UpperCAmelCase_ : List[Any] = np.asarray(lowerCamelCase_ )
UpperCAmelCase_ : str = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = data_mat.reshape(1 ,shapes[0] * shapes[1] )
return data_expanded
def A__ ( self: str ,lowerCamelCase_: Dict ,lowerCamelCase_: int ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : Any = []
UpperCAmelCase_ : Tuple = 0
for i_map in range(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = np.ones((size_map, size_map) )
for i in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
for j in range(0 ,lowerCamelCase_ ,lowerCamelCase_ ):
UpperCAmelCase_ : Any = pd_pool[
i_pool
]
UpperCAmelCase_ : List[str] = i_pool + 1
UpperCAmelCase_ : Optional[Any] = np.multiply(
lowerCamelCase_ ,np.multiply(out_map[i_map] ,(1 - out_map[i_map]) ) )
pd_all.append(lowerCamelCase_ )
return pd_all
def A__ ( self: str ,lowerCamelCase_: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any=bool ) -> Optional[int]:
# model traning
print("""----------------------Start Training-------------------------""" )
print((""" - - Shape: Train_Data """, np.shape(lowerCamelCase_ )) )
print((""" - - Shape: Teach_Data """, np.shape(lowerCamelCase_ )) )
UpperCAmelCase_ : str = 0
UpperCAmelCase_ : Tuple = []
UpperCAmelCase_ : Any = 10000
while rp < n_repeat and mse >= error_accuracy:
UpperCAmelCase_ : List[str] = 0
print(F'''-------------Learning Time {rp}--------------''' )
for p in range(len(lowerCamelCase_ ) ):
# print('------------Learning Image: %d--------------'%p)
UpperCAmelCase_ : str = np.asmatrix(datas_train[p] )
UpperCAmelCase_ : Optional[Any] = np.asarray(datas_teach[p] )
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : List[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : int = np.shape(lowerCamelCase_ )
UpperCAmelCase_ : Dict = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = data_bp_input
UpperCAmelCase_ : Optional[Any] = np.dot(lowerCamelCase_ ,self.vji.T ) - self.thre_bpa
UpperCAmelCase_ : int = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = np.dot(lowerCamelCase_ ,self.wkj.T ) - self.thre_bpa
UpperCAmelCase_ : Optional[Any] = self.sig(lowerCamelCase_ )
# --------------Model Leaning ------------------------
# calculate error and gradient---------------
UpperCAmelCase_ : List[str] = np.multiply(
(data_teach - bp_outa) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : List[Any] = np.multiply(
np.dot(lowerCamelCase_ ,self.wkj ) ,np.multiply(lowerCamelCase_ ,(1 - bp_outa) ) )
UpperCAmelCase_ : Any = np.dot(lowerCamelCase_ ,self.vji )
UpperCAmelCase_ : Tuple = pd_i_all / (self.size_poolinga * self.size_poolinga)
UpperCAmelCase_ : List[str] = pd_conva_pooled.T.getA().tolist()
UpperCAmelCase_ : str = self._calculate_gradient_from_pool(
lowerCamelCase_ ,lowerCamelCase_ ,shape_featuremapa[0] ,shape_featuremapa[1] ,self.size_poolinga ,)
# weight and threshold learning process---------
# convolution layer
for k_conv in range(self.conva[1] ):
UpperCAmelCase_ : List[str] = self._expand_mat(pd_conva_all[k_conv] )
UpperCAmelCase_ : Optional[Any] = self.rate_weight * np.dot(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : int = self.w_conva[k_conv] + delta_w.reshape(
(self.conva[0], self.conva[0]) )
UpperCAmelCase_ : str = (
self.thre_conva[k_conv]
- np.sum(pd_conva_all[k_conv] ) * self.rate_thre
)
# all connected layer
UpperCAmelCase_ : int = self.wkj + pd_k_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : Tuple = self.vji + pd_j_all.T * bp_outa * self.rate_weight
UpperCAmelCase_ : int = self.thre_bpa - pd_k_all * self.rate_thre
UpperCAmelCase_ : str = self.thre_bpa - pd_j_all * self.rate_thre
# calculate the sum error of all single image
UpperCAmelCase_ : int = np.sum(abs(data_teach - bp_outa ) )
error_count += errors
# print(' ----Teach ',data_teach)
# print(' ----BP_output ',bp_out3)
UpperCAmelCase_ : int = rp + 1
UpperCAmelCase_ : Any = error_count / patterns
all_mse.append(lowerCamelCase_ )
def draw_error():
UpperCAmelCase_ : Any = [error_accuracy for i in range(int(n_repeat * 1.2 ) )]
plt.plot(lowerCamelCase_ ,"""+-""" )
plt.plot(lowerCamelCase_ ,"""r--""" )
plt.xlabel("""Learning Times""" )
plt.ylabel("""All_mse""" )
plt.grid(lowerCamelCase_ ,alpha=0.5 )
plt.show()
print("""------------------Training Complished---------------------""" )
print((""" - - Training epoch: """, rp, F''' - - Mse: {mse:.6f}''') )
if draw_e:
draw_error()
return mse
def A__ ( self: Optional[int] ,lowerCamelCase_: Any ) -> Tuple:
# model predict
UpperCAmelCase_ : Union[str, Any] = []
print("""-------------------Start Testing-------------------------""" )
print((""" - - Shape: Test_Data """, np.shape(lowerCamelCase_ )) )
for p in range(len(lowerCamelCase_ ) ):
UpperCAmelCase_ : int = np.asmatrix(datas_test[p] )
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Optional[Any] = self.pooling(lowerCamelCase_ ,self.size_poolinga )
UpperCAmelCase_ : str = self._expand(lowerCamelCase_ )
UpperCAmelCase_ : str = data_bp_input
UpperCAmelCase_ : Union[str, Any] = bp_outa * self.vji.T - self.thre_bpa
UpperCAmelCase_ : Optional[int] = self.sig(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = bp_outa * self.wkj.T - self.thre_bpa
UpperCAmelCase_ : List[Any] = self.sig(lowerCamelCase_ )
produce_out.extend(bp_outa.getA().tolist() )
UpperCAmelCase_ : int = [list(map(self.do_round ,lowerCamelCase_ ) ) for each in produce_out]
return np.asarray(lowerCamelCase_ )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Tuple:
# return the data of image after convoluting process so we can check it out
UpperCAmelCase_ : Optional[int] = np.asmatrix(lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.convolute(
lowerCamelCase_ ,self.conva ,self.w_conva ,self.thre_conva ,conv_step=self.step_conva ,)
UpperCAmelCase_ : Dict = self.pooling(lowerCamelCase_ ,self.size_poolinga )
return data_conveda, data_pooleda
if __name__ == "__main__":
pass
| 345 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
UpperCamelCase_ = {
'''configuration_roformer''': ['''ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''RoFormerConfig''', '''RoFormerOnnxConfig'''],
'''tokenization_roformer''': ['''RoFormerTokenizer'''],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = ['''RoFormerTokenizerFast''']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''RoFormerForCausalLM''',
'''RoFormerForMaskedLM''',
'''RoFormerForMultipleChoice''',
'''RoFormerForQuestionAnswering''',
'''RoFormerForSequenceClassification''',
'''RoFormerForTokenClassification''',
'''RoFormerLayer''',
'''RoFormerModel''',
'''RoFormerPreTrainedModel''',
'''load_tf_weights_in_roformer''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''TF_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TFRoFormerForCausalLM''',
'''TFRoFormerForMaskedLM''',
'''TFRoFormerForMultipleChoice''',
'''TFRoFormerForQuestionAnswering''',
'''TFRoFormerForSequenceClassification''',
'''TFRoFormerForTokenClassification''',
'''TFRoFormerLayer''',
'''TFRoFormerModel''',
'''TFRoFormerPreTrainedModel''',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase_ = [
'''FLAX_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''FlaxRoFormerForMaskedLM''',
'''FlaxRoFormerForMultipleChoice''',
'''FlaxRoFormerForQuestionAnswering''',
'''FlaxRoFormerForSequenceClassification''',
'''FlaxRoFormerForTokenClassification''',
'''FlaxRoFormerModel''',
'''FlaxRoFormerPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_roformer import ROFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, RoFormerConfig, RoFormerOnnxConfig
from .tokenization_roformer import RoFormerTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_roformer_fast import RoFormerTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_roformer import (
ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
RoFormerForCausalLM,
RoFormerForMaskedLM,
RoFormerForMultipleChoice,
RoFormerForQuestionAnswering,
RoFormerForSequenceClassification,
RoFormerForTokenClassification,
RoFormerLayer,
RoFormerModel,
RoFormerPreTrainedModel,
load_tf_weights_in_roformer,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_roformer import (
TF_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
TFRoFormerForCausalLM,
TFRoFormerForMaskedLM,
TFRoFormerForMultipleChoice,
TFRoFormerForQuestionAnswering,
TFRoFormerForSequenceClassification,
TFRoFormerForTokenClassification,
TFRoFormerLayer,
TFRoFormerModel,
TFRoFormerPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_roformer import (
FLAX_ROFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
FlaxRoFormerForMaskedLM,
FlaxRoFormerForMultipleChoice,
FlaxRoFormerForQuestionAnswering,
FlaxRoFormerForSequenceClassification,
FlaxRoFormerForTokenClassification,
FlaxRoFormerModel,
FlaxRoFormerPreTrainedModel,
)
else:
import sys
UpperCamelCase_ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 345 |
import json
import os
import unittest
from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Optional[Any] = CTRLTokenizer
A__ : Optional[Any] = False
A__ : str = False
def A__ ( self: Optional[int] ) -> List[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
UpperCAmelCase_ : Dict = ["""adapt""", """re@@""", """a@@""", """apt""", """c@@""", """t""", """<unk>"""]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : List[Any] = ["""#version: 0.2""", """a p""", """ap t</w>""", """r e""", """a d""", """ad apt</w>""", """"""]
UpperCAmelCase_ : Optional[Any] = {"""unk_token""": """<unk>"""}
UpperCAmelCase_ : Union[str, Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : Optional[Any] = os.path.join(self.tmpdirname ,VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase_ ) + """\n""" )
with open(self.merges_file ,"""w""" ,encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase_ ) )
def A__ ( self: Optional[int] ,**lowerCamelCase_: Any ) -> str:
kwargs.update(self.special_tokens_map )
return CTRLTokenizer.from_pretrained(self.tmpdirname ,**lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: int ) -> str:
UpperCAmelCase_ : List[str] = """adapt react readapt apt"""
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
return input_text, output_text
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : Union[str, Any] = CTRLTokenizer(self.vocab_file ,self.merges_file ,**self.special_tokens_map )
UpperCAmelCase_ : List[Any] = """adapt react readapt apt"""
UpperCAmelCase_ : Optional[int] = """adapt re@@ a@@ c@@ t re@@ adapt apt""".split()
UpperCAmelCase_ : Tuple = tokenizer.tokenize(lowerCamelCase_ )
self.assertListEqual(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = tokens + [tokenizer.unk_token]
UpperCAmelCase_ : List[str] = [0, 1, 2, 4, 5, 1, 0, 3, 6]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase_ ) ,lowerCamelCase_ )
| 345 | 1 |
import unittest
from transformers import is_torch_available
from transformers.testing_utils import require_torch
if is_torch_available():
import torch
from transformers.activations import gelu_new, gelu_python, get_activation
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : List[Any] = torch.tensor([-100, -1, -0.1, 0, 0.1, 1.0, 100] )
UpperCAmelCase_ : List[str] = get_activation("""gelu""" )
self.assertTrue(torch.allclose(gelu_python(lowerCamelCase_ ) ,torch_builtin(lowerCamelCase_ ) ) )
self.assertFalse(torch.allclose(gelu_python(lowerCamelCase_ ) ,gelu_new(lowerCamelCase_ ) ) )
def A__ ( self: int ) -> Optional[int]:
UpperCAmelCase_ : Any = torch.tensor([-100, -1, -0.1, 0, 0.1, 1.0, 100] )
UpperCAmelCase_ : Tuple = get_activation("""gelu""" )
UpperCAmelCase_ : Union[str, Any] = get_activation("""gelu_10""" )
UpperCAmelCase_ : Tuple = torch_builtin(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = geluaa(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = torch.where(y_gelu_aa < 1_0.0 ,1 ,0 )
self.assertTrue(torch.max(lowerCamelCase_ ).item() == 1_0.0 )
self.assertTrue(torch.allclose(y_gelu * clipped_mask ,y_gelu_aa * clipped_mask ) )
def A__ ( self: Tuple ) -> List[str]:
get_activation("""gelu""" )
get_activation("""gelu_10""" )
get_activation("""gelu_fast""" )
get_activation("""gelu_new""" )
get_activation("""gelu_python""" )
get_activation("""gelu_pytorch_tanh""" )
get_activation("""linear""" )
get_activation("""mish""" )
get_activation("""quick_gelu""" )
get_activation("""relu""" )
get_activation("""sigmoid""" )
get_activation("""silu""" )
get_activation("""swish""" )
get_activation("""tanh""" )
with self.assertRaises(lowerCamelCase_ ):
get_activation("""bogus""" )
with self.assertRaises(lowerCamelCase_ ):
get_activation(lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : str = get_activation("""gelu""" )
UpperCAmelCase_ : Optional[Any] = 1
UpperCAmelCase_ : Optional[int] = get_activation("""gelu""" )
self.assertEqual(acta.a ,1 )
with self.assertRaises(lowerCamelCase_ ):
UpperCAmelCase_ : Optional[Any] = acta.a
| 345 |
from __future__ import annotations
from typing import Dict
from ...configuration_utils import PretrainedConfig
UpperCamelCase_ = {
'''susnato/ernie-m-base_pytorch''': '''https://huggingface.co/susnato/ernie-m-base_pytorch/blob/main/config.json''',
'''susnato/ernie-m-large_pytorch''': '''https://huggingface.co/susnato/ernie-m-large_pytorch/blob/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Union[str, Any] = "ernie_m"
A__ : Dict[str, str] = {"dropout": "classifier_dropout", "num_classes": "num_labels"}
def __init__( self: str ,lowerCamelCase_: int = 250002 ,lowerCamelCase_: int = 768 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 12 ,lowerCamelCase_: int = 3072 ,lowerCamelCase_: str = "gelu" ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: float = 0.1 ,lowerCamelCase_: int = 514 ,lowerCamelCase_: float = 0.0_2 ,lowerCamelCase_: int = 1 ,lowerCamelCase_: float = 1e-05 ,lowerCamelCase_: Any=None ,lowerCamelCase_: List[Any]=False ,lowerCamelCase_: Tuple=0.0 ,**lowerCamelCase_: Optional[int] ,) -> Optional[Any]:
super().__init__(pad_token_id=lowerCamelCase_ ,**lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = vocab_size
UpperCAmelCase_ : Any = hidden_size
UpperCAmelCase_ : Optional[Any] = num_hidden_layers
UpperCAmelCase_ : Union[str, Any] = num_attention_heads
UpperCAmelCase_ : List[Any] = intermediate_size
UpperCAmelCase_ : List[Any] = hidden_act
UpperCAmelCase_ : Any = hidden_dropout_prob
UpperCAmelCase_ : List[Any] = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = initializer_range
UpperCAmelCase_ : Union[str, Any] = layer_norm_eps
UpperCAmelCase_ : List[Any] = classifier_dropout
UpperCAmelCase_ : str = is_decoder
UpperCAmelCase_ : List[str] = act_dropout
| 345 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
from ...utils.backbone_utils import BackboneConfigMixin, get_aligned_output_features_output_indices
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''shi-labs/nat-mini-in1k-224''': '''https://huggingface.co/shi-labs/nat-mini-in1k-224/resolve/main/config.json''',
# See all Nat models at https://huggingface.co/models?filter=nat
}
class _snake_case ( __snake_case , __snake_case ):
'''simple docstring'''
A__ : Tuple = "nat"
A__ : int = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self: List[Any] ,lowerCamelCase_: Optional[Any]=4 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: List[Any]=64 ,lowerCamelCase_: Optional[int]=[3, 4, 6, 5] ,lowerCamelCase_: List[str]=[2, 4, 8, 16] ,lowerCamelCase_: Tuple=7 ,lowerCamelCase_: str=3.0 ,lowerCamelCase_: Optional[Any]=True ,lowerCamelCase_: int=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Dict=0.1 ,lowerCamelCase_: Union[str, Any]="gelu" ,lowerCamelCase_: Any=0.0_2 ,lowerCamelCase_: List[str]=1e-5 ,lowerCamelCase_: int=0.0 ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: List[str]=None ,**lowerCamelCase_: Optional[int] ,) -> Union[str, Any]:
super().__init__(**lowerCamelCase_ )
UpperCAmelCase_ : Tuple = patch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : List[str] = embed_dim
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = len(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = num_heads
UpperCAmelCase_ : Optional[Any] = kernel_size
UpperCAmelCase_ : Dict = mlp_ratio
UpperCAmelCase_ : Dict = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Any = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = drop_path_rate
UpperCAmelCase_ : Tuple = hidden_act
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : List[str] = initializer_range
# we set the hidden_size attribute in order to make Nat work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
UpperCAmelCase_ : Tuple = int(embed_dim * 2 ** (len(lowerCamelCase_ ) - 1) )
UpperCAmelCase_ : Dict = layer_scale_init_value
UpperCAmelCase_ : Optional[int] = ["""stem"""] + [F'''stage{idx}''' for idx in range(1 ,len(lowerCamelCase_ ) + 1 )]
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = get_aligned_output_features_output_indices(
out_features=lowerCamelCase_ ,out_indices=lowerCamelCase_ ,stage_names=self.stage_names )
| 345 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
UpperCamelCase_ = logging.getLogger(__name__)
torch.set_grad_enabled(False)
UpperCamelCase_ = '''cuda''' if torch.cuda.is_available() else '''cpu'''
def lowerCamelCase_ ( _a : str , _a : Any=100 , _a : int=" " ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = text.split(_a )
return [character.join(text[i : i + n] ).strip() for i in range(0 , len(_a ) , _a )]
def lowerCamelCase_ ( _a : dict ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Dict = [], []
for title, text in zip(documents["""title"""] , documents["""text"""] ):
if text is not None:
for passage in split_text(_a ):
titles.append(title if title is not None else """""" )
texts.append(_a )
return {"title": titles, "text": texts}
def lowerCamelCase_ ( _a : dict , _a : DPRContextEncoder , _a : DPRContextEncoderTokenizerFast ):
'''simple docstring'''
UpperCAmelCase_ : List[str] = ctx_tokenizer(
documents["""title"""] , documents["""text"""] , truncation=_a , padding="""longest""" , return_tensors="""pt""" )["""input_ids"""]
UpperCAmelCase_ : Tuple = ctx_encoder(input_ids.to(device=_a ) , return_dict=_a ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def lowerCamelCase_ ( _a : "RagExampleArguments" , _a : "ProcessingArguments" , _a : "IndexHnswArguments" , ):
'''simple docstring'''
logger.info("""Step 1 - Create the dataset""" )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
UpperCAmelCase_ : Optional[int] = load_dataset(
"""csv""" , data_files=[rag_example_args.csv_path] , split="""train""" , delimiter="""\t""" , column_names=["""title""", """text"""] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
UpperCAmelCase_ : Tuple = dataset.map(_a , batched=_a , num_proc=processing_args.num_proc )
# And compute the embeddings
UpperCAmelCase_ : List[str] = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=_a )
UpperCAmelCase_ : Dict = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
UpperCAmelCase_ : Any = Features(
{"""text""": Value("""string""" ), """title""": Value("""string""" ), """embeddings""": Sequence(Value("""float32""" ) )} ) # optional, save as float32 instead of float64 to save space
UpperCAmelCase_ : List[str] = dataset.map(
partial(_a , ctx_encoder=_a , ctx_tokenizer=_a ) , batched=_a , batch_size=processing_args.batch_size , features=_a , )
# And finally save your dataset
UpperCAmelCase_ : Union[str, Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset""" )
dataset.save_to_disk(_a )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info("""Step 2 - Index the dataset""" )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
UpperCAmelCase_ : Union[str, Any] = faiss.IndexHNSWFlat(index_hnsw_args.d , index_hnsw_args.m , faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index("""embeddings""" , custom_index=_a )
# And save the index
UpperCAmelCase_ : Optional[Any] = os.path.join(rag_example_args.output_dir , """my_knowledge_dataset_hnsw_index.faiss""" )
dataset.get_index("""embeddings""" ).save(_a )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _snake_case :
'''simple docstring'''
A__ : str = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" / "my_knowledge_dataset.csv" ) , metadata={"help": "Path to a tab-separated csv file with columns 'title' and 'text'"} , )
A__ : Optional[str] = field(
default=__snake_case , metadata={"help": "Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."} , )
A__ : str = field(
default="facebook/rag-sequence-nq" , metadata={"help": "The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"} , )
A__ : str = field(
default="facebook/dpr-ctx_encoder-multiset-base" , metadata={
"help": (
"The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"
" 'facebook/dpr-ctx_encoder-multiset-base'"
)
} , )
A__ : Optional[str] = field(
default=str(Path(__snake_case ).parent / "test_run" / "dummy-kb" ) , metadata={"help": "Path to a directory where the dataset passages and the index will be saved"} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : Optional[int] = field(
default=__snake_case , metadata={
"help": "The number of processes to use to split the documents into passages. Default is single process."
} , )
A__ : int = field(
default=16 , metadata={
"help": "The batch size to use when computing the passages embeddings using the DPR context encoder."
} , )
@dataclass
class _snake_case :
'''simple docstring'''
A__ : int = field(
default=768 , metadata={"help": "The dimension of the embeddings to pass to the HNSW Faiss index."} , )
A__ : int = field(
default=128 , metadata={
"help": (
"The number of bi-directional links created for every new element during the HNSW index construction."
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
UpperCamelCase_ = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
UpperCamelCase_ ,UpperCamelCase_ ,UpperCamelCase_ = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
UpperCamelCase_ = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 345 | 1 |
# NOTE: This file is deprecated and will be removed in a future version.
# It only exists so that temporarely `from diffusers.pipelines import DiffusionPipeline` works
from ...utils import deprecate
from ..controlnet.pipeline_flax_controlnet import FlaxStableDiffusionControlNetPipeline # noqa: F401
deprecate(
'''stable diffusion controlnet''',
'''0.22.0''',
'''Importing `FlaxStableDiffusionControlNetPipeline` from diffusers.pipelines.stable_diffusion.flax_pipeline_stable_diffusion_controlnet is deprecated. Please import `from diffusers import FlaxStableDiffusionControlNetPipeline` instead.''',
standard_warn=False,
stacklevel=3,
)
| 345 |
import gc
import unittest
import torch
from parameterized import parameterized
from diffusers import AutoencoderKL
from diffusers.utils import floats_tensor, load_hf_numpy, require_torch_gpu, slow, torch_all_close, torch_device
from diffusers.utils.import_utils import is_xformers_available
from diffusers.utils.testing_utils import enable_full_determinism
from .test_modeling_common import ModelTesterMixin, UNetTesterMixin
enable_full_determinism()
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Dict = AutoencoderKL
A__ : Optional[int] = "sample"
A__ : Tuple = 1E-2
@property
def A__ ( self: List[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = 4
UpperCAmelCase_ : str = 3
UpperCAmelCase_ : Any = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ).to(lowerCamelCase_ )
return {"sample": image}
@property
def A__ ( self: List[str] ) -> Tuple:
return (3, 32, 32)
@property
def A__ ( self: Optional[Any] ) -> Any:
return (3, 32, 32)
def A__ ( self: Any ) -> Tuple:
UpperCAmelCase_ : List[Any] = {
"""block_out_channels""": [32, 64],
"""in_channels""": 3,
"""out_channels""": 3,
"""down_block_types""": ["""DownEncoderBlock2D""", """DownEncoderBlock2D"""],
"""up_block_types""": ["""UpDecoderBlock2D""", """UpDecoderBlock2D"""],
"""latent_channels""": 4,
}
UpperCAmelCase_ : int = self.dummy_input
return init_dict, inputs_dict
def A__ ( self: Optional[Any] ) -> int:
pass
def A__ ( self: str ) -> Any:
pass
@unittest.skipIf(torch_device == """mps""" ,"""Gradient checkpointing skipped on MPS""" )
def A__ ( self: Union[str, Any] ) -> Dict:
# enable deterministic behavior for gradient checkpointing
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.prepare_init_args_and_inputs_for_common()
UpperCAmelCase_ : List[Any] = self.model_class(**lowerCamelCase_ )
model.to(lowerCamelCase_ )
assert not model.is_gradient_checkpointing and model.training
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model.zero_grad()
UpperCAmelCase_ : Any = torch.randn_like(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = (out - labels).mean()
loss.backward()
# re-instantiate the model now enabling gradient checkpointing
UpperCAmelCase_ : str = self.model_class(**lowerCamelCase_ )
# clone model
model_a.load_state_dict(model.state_dict() )
model_a.to(lowerCamelCase_ )
model_a.enable_gradient_checkpointing()
assert model_a.is_gradient_checkpointing and model_a.training
UpperCAmelCase_ : Optional[int] = model_a(**lowerCamelCase_ ).sample
# run the backwards pass on the model. For backwards pass, for simplicity purpose,
# we won't calculate the loss and rather backprop on out.sum()
model_a.zero_grad()
UpperCAmelCase_ : Dict = (out_a - labels).mean()
loss_a.backward()
# compare the output and parameters gradients
self.assertTrue((loss - loss_a).abs() < 1e-5 )
UpperCAmelCase_ : Dict = dict(model.named_parameters() )
UpperCAmelCase_ : Union[str, Any] = dict(model_a.named_parameters() )
for name, param in named_params.items():
self.assertTrue(torch_all_close(param.grad.data ,named_params_a[name].grad.data ,atol=5e-5 ) )
def A__ ( self: Optional[Any] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" ,output_loading_info=lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
self.assertEqual(len(loading_info["""missing_keys"""] ) ,0 )
model.to(lowerCamelCase_ )
UpperCAmelCase_ : Dict = model(**self.dummy_input )
assert image is not None, "Make sure output is not None"
def A__ ( self: Optional[int] ) -> int:
UpperCAmelCase_ : Dict = AutoencoderKL.from_pretrained("""fusing/autoencoder-kl-dummy""" )
UpperCAmelCase_ : Tuple = model.to(lowerCamelCase_ )
model.eval()
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.manual_seed(0 )
else:
UpperCAmelCase_ : Optional[int] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : str = torch.randn(
1 ,model.config.in_channels ,model.config.sample_size ,model.config.sample_size ,generator=torch.manual_seed(0 ) ,)
UpperCAmelCase_ : int = image.to(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ,generator=lowerCamelCase_ ).sample
UpperCAmelCase_ : Optional[int] = output[0, -1, -3:, -3:].flatten().cpu()
# Since the VAE Gaussian prior's generator is seeded on the appropriate device,
# the expected output slices are not the same for CPU and GPU.
if torch_device == "mps":
UpperCAmelCase_ : Tuple = torch.tensor(
[
-4.0078e-01,
-3.8323e-04,
-1.2681e-01,
-1.1462e-01,
2.0095e-01,
1.0893e-01,
-8.8247e-02,
-3.0361e-01,
-9.8644e-03,
] )
elif torch_device == "cpu":
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.1_3_5_2, 0.0_8_7_8, 0.0_4_1_9, -0.0_8_1_8, -0.1_0_6_9, 0.0_6_8_8, -0.1_4_5_8, -0.4_4_4_6, -0.0_0_2_6] )
else:
UpperCAmelCase_ : List[str] = torch.tensor(
[-0.2_4_2_1, 0.4_6_4_2, 0.2_5_0_7, -0.0_4_3_8, 0.0_6_8_2, 0.3_1_6_0, -0.2_0_1_8, -0.0_7_2_7, 0.2_4_8_5] )
self.assertTrue(torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,rtol=1e-2 ) )
@slow
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Any ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
return F'''gaussian_noise_s={seed}_shape={'_'.join([str(lowerCamelCase_ ) for s in shape] )}.npy'''
def A__ ( self: Union[str, Any] ) -> Optional[int]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ,lowerCamelCase_: Optional[int]=0 ,lowerCamelCase_: List[Any]=(4, 3, 512, 512) ,lowerCamelCase_: Optional[Any]=False ) -> Optional[int]:
UpperCAmelCase_ : Tuple = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : Tuple = torch.from_numpy(load_hf_numpy(self.get_file_format(lowerCamelCase_ ,lowerCamelCase_ ) ) ).to(lowerCamelCase_ ).to(lowerCamelCase_ )
return image
def A__ ( self: List[Any] ,lowerCamelCase_: List[str]="CompVis/stable-diffusion-v1-4" ,lowerCamelCase_: Union[str, Any]=False ) -> Any:
UpperCAmelCase_ : Optional[Any] = """fp16""" if fpaa else None
UpperCAmelCase_ : str = torch.floataa if fpaa else torch.floataa
UpperCAmelCase_ : int = AutoencoderKL.from_pretrained(
lowerCamelCase_ ,subfolder="""vae""" ,torch_dtype=lowerCamelCase_ ,revision=lowerCamelCase_ ,)
model.to(lowerCamelCase_ ).eval()
return model
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any]=0 ) -> Optional[int]:
if torch_device == "mps":
return torch.manual_seed(lowerCamelCase_ )
return torch.Generator(device=lowerCamelCase_ ).manual_seed(lowerCamelCase_ )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_3, 0.9_8_7_8, -0.0_4_9_5, -0.0_7_9_0, -0.2_7_0_9, 0.8_3_7_5, -0.2_0_6_0, -0.0_8_2_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_6, 0.1_1_6_8, 0.1_3_3_2, -0.4_8_4_0, -0.2_5_0_8, -0.0_7_9_1, -0.0_4_9_3, -0.4_0_8_9], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Dict ) -> Tuple:
UpperCAmelCase_ : List[Any] = self.get_sd_vae_model()
UpperCAmelCase_ : int = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Optional[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Tuple = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[33, [-0.0_5_1_3, 0.0_2_8_9, 1.3_7_9_9, 0.2_1_6_6, -0.2_5_7_3, -0.0_8_7_1, 0.5_1_0_3, -0.0_9_9_9]],
[47, [-0.4_1_2_8, -0.1_3_2_0, -0.3_7_0_4, 0.1_9_6_5, -0.4_1_1_6, -0.2_3_3_2, -0.3_3_4_0, 0.2_2_4_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> Tuple:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,fpaa=lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,generator=lowerCamelCase_ ,sample_posterior=lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : Tuple = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : Optional[int] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.1_6_0_9, 0.9_8_6_6, -0.0_4_8_7, -0.0_7_7_7, -0.2_7_1_6, 0.8_3_6_8, -0.2_0_5_5, -0.0_8_1_4], [-0.2_3_9_5, 0.0_0_9_8, 0.0_1_0_2, -0.0_7_0_9, -0.2_8_4_0, -0.0_2_7_4, -0.0_7_1_8, -0.1_8_2_4]],
[47, [-0.2_3_7_7, 0.1_1_4_7, 0.1_3_3_3, -0.4_8_4_1, -0.2_5_0_6, -0.0_8_0_5, -0.0_4_9_1, -0.4_0_8_5], [0.0_3_5_0, 0.0_8_4_7, 0.0_4_6_7, 0.0_3_4_4, -0.0_8_4_2, -0.0_5_4_7, -0.0_6_3_3, -0.1_1_3_1]],
# fmt: on
] )
def A__ ( self: Tuple ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: List[str] ) -> Dict:
UpperCAmelCase_ : Optional[int] = self.get_sd_vae_model()
UpperCAmelCase_ : Dict = self.get_sd_image(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : str = model(lowerCamelCase_ ).sample
assert sample.shape == image.shape
UpperCAmelCase_ : List[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu()
UpperCAmelCase_ : Any = torch.tensor(expected_slice_mps if torch_device == """mps""" else expected_slice )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=3e-3 )
@parameterized.expand(
[
# fmt: off
[13, [-0.2_0_5_1, -0.1_8_0_3, -0.2_3_1_1, -0.2_1_1_4, -0.3_2_9_2, -0.3_5_7_4, -0.2_9_5_3, -0.3_3_2_3]],
[37, [-0.2_6_3_2, -0.2_6_2_5, -0.2_1_9_9, -0.2_7_4_1, -0.4_5_3_9, -0.4_9_9_0, -0.3_7_2_0, -0.4_9_2_5]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: str ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[int] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : str = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : Any = sample[-1, -2:, :2, -2:].flatten().cpu()
UpperCAmelCase_ : Union[str, Any] = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-3 )
@parameterized.expand(
[
# fmt: off
[27, [-0.0_3_6_9, 0.0_2_0_7, -0.0_7_7_6, -0.0_6_8_2, -0.1_7_4_7, -0.1_9_3_0, -0.1_4_6_5, -0.2_0_3_9]],
[16, [-0.1_6_2_8, -0.2_1_3_4, -0.2_7_4_7, -0.2_6_4_2, -0.3_7_7_4, -0.4_4_0_4, -0.3_6_8_7, -0.4_2_7_7]],
# fmt: on
] )
@require_torch_gpu
def A__ ( self: str ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Any ) -> Optional[Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
UpperCAmelCase_ : str = sample[-1, -2:, :2, -2:].flatten().float().cpu()
UpperCAmelCase_ : str = torch.tensor(lowerCamelCase_ )
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=5e-3 )
@parameterized.expand([(13,), (16,), (27,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ) -> int:
UpperCAmelCase_ : Optional[Any] = self.get_sd_vae_model(fpaa=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) ,fpaa=lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : List[str] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-1 )
@parameterized.expand([(13,), (16,), (37,)] )
@require_torch_gpu
@unittest.skipIf(not is_xformers_available() ,reason="""xformers is not required when using PyTorch 2.0.""" )
def A__ ( self: Optional[Any] ,lowerCamelCase_: Dict ) -> Union[str, Any]:
UpperCAmelCase_ : Tuple = self.get_sd_vae_model()
UpperCAmelCase_ : Any = self.get_sd_image(lowerCamelCase_ ,shape=(3, 4, 64, 64) )
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model.decode(lowerCamelCase_ ).sample
model.enable_xformers_memory_efficient_attention()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model.decode(lowerCamelCase_ ).sample
assert list(sample.shape ) == [3, 3, 512, 512]
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=1e-2 )
@parameterized.expand(
[
# fmt: off
[33, [-0.3_0_0_1, 0.0_9_1_8, -2.6_9_8_4, -3.9_7_2_0, -3.2_0_9_9, -5.0_3_5_3, 1.7_3_3_8, -0.2_0_6_5, 3.4_2_6_7]],
[47, [-1.5_0_3_0, -4.3_8_7_1, -6.0_3_5_5, -9.1_1_5_7, -1.6_6_6_1, -2.7_8_5_3, 2.1_6_0_7, -5.0_8_2_3, 2.5_6_3_3]],
# fmt: on
] )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Dict = self.get_sd_vae_model()
UpperCAmelCase_ : Optional[Any] = self.get_sd_image(lowerCamelCase_ )
UpperCAmelCase_ : str = self.get_generator(lowerCamelCase_ )
with torch.no_grad():
UpperCAmelCase_ : int = model.encode(lowerCamelCase_ ).latent_dist
UpperCAmelCase_ : Optional[Any] = dist.sample(generator=lowerCamelCase_ )
assert list(sample.shape ) == [image.shape[0], 4] + [i // 8 for i in image.shape[2:]]
UpperCAmelCase_ : Tuple = sample[0, -1, -3:, -3:].flatten().cpu()
UpperCAmelCase_ : Optional[Any] = torch.tensor(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = 3e-3 if torch_device != """mps""" else 1e-2
assert torch_all_close(lowerCamelCase_ ,lowerCamelCase_ ,atol=lowerCamelCase_ )
| 345 | 1 |
from ...utils import is_torch_available, is_transformers_available
if is_transformers_available() and is_torch_available():
from .pipeline_vq_diffusion import LearnedClassifierFreeSamplingEmbeddings, VQDiffusionPipeline
| 345 |
import copy
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Audio, Features, Value
from .base import TaskTemplate
@dataclass(frozen=__snake_case )
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = field(default="automatic-speech-recognition" , metadata={"include_in_asdict_even_if_is_default": True} )
A__ : ClassVar[Features] = Features({"audio": Audio()} )
A__ : ClassVar[Features] = Features({"transcription": Value("string" )} )
A__ : str = "audio"
A__ : str = "transcription"
def A__ ( self: int ,lowerCamelCase_: Union[str, Any] ) -> Optional[Any]:
if self.audio_column not in features:
raise ValueError(F'''Column {self.audio_column} is not present in features.''' )
if not isinstance(features[self.audio_column] ,lowerCamelCase_ ):
raise ValueError(F'''Column {self.audio_column} is not an Audio type.''' )
UpperCAmelCase_ : Any = copy.deepcopy(self )
UpperCAmelCase_ : Union[str, Any] = self.input_schema.copy()
UpperCAmelCase_ : Any = features[self.audio_column]
UpperCAmelCase_ : Union[str, Any] = input_schema
return task_template
@property
def A__ ( self: List[str] ) -> Dict[str, str]:
return {self.audio_column: "audio", self.transcription_column: "transcription"}
| 345 | 1 |
from math import acos, sin
from typing import List, Tuple, Union
import numpy as np
import torch
from PIL import Image
from ...models import AutoencoderKL, UNetaDConditionModel
from ...schedulers import DDIMScheduler, DDPMScheduler
from ...utils import randn_tensor
from ..pipeline_utils import AudioPipelineOutput, BaseOutput, DiffusionPipeline, ImagePipelineOutput
from .mel import Mel
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = ["vqvae"]
def __init__( self: int ,lowerCamelCase_: AutoencoderKL ,lowerCamelCase_: UNetaDConditionModel ,lowerCamelCase_: Mel ,lowerCamelCase_: Union[DDIMScheduler, DDPMScheduler] ,) -> Optional[int]:
super().__init__()
self.register_modules(unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,mel=lowerCamelCase_ ,vqvae=lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> int:
return 50 if isinstance(self.scheduler ,lowerCamelCase_ ) else 1000
@torch.no_grad()
def __call__( self: Optional[Any] ,lowerCamelCase_: int = 1 ,lowerCamelCase_: str = None ,lowerCamelCase_: np.ndarray = None ,lowerCamelCase_: int = 0 ,lowerCamelCase_: int = 0 ,lowerCamelCase_: int = None ,lowerCamelCase_: torch.Generator = None ,lowerCamelCase_: float = 0 ,lowerCamelCase_: float = 0 ,lowerCamelCase_: torch.Generator = None ,lowerCamelCase_: float = 0 ,lowerCamelCase_: torch.Tensor = None ,lowerCamelCase_: torch.Tensor = None ,lowerCamelCase_: Tuple=True ,) -> Union[
Union[AudioPipelineOutput, ImagePipelineOutput],
Tuple[List[Image.Image], Tuple[int, List[np.ndarray]]],
]:
UpperCAmelCase_ : Union[str, Any] = steps or self.get_default_steps()
self.scheduler.set_timesteps(lowerCamelCase_ )
UpperCAmelCase_ : Dict = step_generator or generator
# For backwards compatibility
if type(self.unet.config.sample_size ) == int:
UpperCAmelCase_ : List[Any] = (self.unet.config.sample_size, self.unet.config.sample_size)
if noise is None:
UpperCAmelCase_ : str = randn_tensor(
(
batch_size,
self.unet.config.in_channels,
self.unet.config.sample_size[0],
self.unet.config.sample_size[1],
) ,generator=lowerCamelCase_ ,device=self.device ,)
UpperCAmelCase_ : int = noise
UpperCAmelCase_ : Tuple = None
if audio_file is not None or raw_audio is not None:
self.mel.load_audio(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = self.mel.audio_slice_to_image(lowerCamelCase_ )
UpperCAmelCase_ : int = np.frombuffer(input_image.tobytes() ,dtype="""uint8""" ).reshape(
(input_image.height, input_image.width) )
UpperCAmelCase_ : Any = (input_image / 255) * 2 - 1
UpperCAmelCase_ : List[str] = torch.tensor(input_image[np.newaxis, :, :] ,dtype=torch.float ).to(self.device )
if self.vqvae is not None:
UpperCAmelCase_ : Optional[Any] = self.vqvae.encode(torch.unsqueeze(lowerCamelCase_ ,0 ) ).latent_dist.sample(
generator=lowerCamelCase_ )[0]
UpperCAmelCase_ : Any = self.vqvae.config.scaling_factor * input_images
if start_step > 0:
UpperCAmelCase_ : Union[str, Any] = self.scheduler.add_noise(lowerCamelCase_ ,lowerCamelCase_ ,self.scheduler.timesteps[start_step - 1] )
UpperCAmelCase_ : Any = (
self.unet.config.sample_size[1] * self.mel.get_sample_rate() / self.mel.x_res / self.mel.hop_length
)
UpperCAmelCase_ : Optional[int] = int(mask_start_secs * pixels_per_second )
UpperCAmelCase_ : Union[str, Any] = int(mask_end_secs * pixels_per_second )
UpperCAmelCase_ : Optional[Any] = self.scheduler.add_noise(lowerCamelCase_ ,lowerCamelCase_ ,torch.tensor(self.scheduler.timesteps[start_step:] ) )
for step, t in enumerate(self.progress_bar(self.scheduler.timesteps[start_step:] ) ):
if isinstance(self.unet ,lowerCamelCase_ ):
UpperCAmelCase_ : Dict = self.unet(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )["""sample"""]
else:
UpperCAmelCase_ : List[str] = self.unet(lowerCamelCase_ ,lowerCamelCase_ )["""sample"""]
if isinstance(self.scheduler ,lowerCamelCase_ ):
UpperCAmelCase_ : int = self.scheduler.step(
model_output=lowerCamelCase_ ,timestep=lowerCamelCase_ ,sample=lowerCamelCase_ ,eta=lowerCamelCase_ ,generator=lowerCamelCase_ ,)["""prev_sample"""]
else:
UpperCAmelCase_ : Optional[Any] = self.scheduler.step(
model_output=lowerCamelCase_ ,timestep=lowerCamelCase_ ,sample=lowerCamelCase_ ,generator=lowerCamelCase_ ,)["""prev_sample"""]
if mask is not None:
if mask_start > 0:
UpperCAmelCase_ : Union[str, Any] = mask[:, step, :, :mask_start]
if mask_end > 0:
UpperCAmelCase_ : List[Any] = mask[:, step, :, -mask_end:]
if self.vqvae is not None:
# 0.18215 was scaling factor used in training to ensure unit variance
UpperCAmelCase_ : Dict = 1 / self.vqvae.config.scaling_factor * images
UpperCAmelCase_ : Optional[int] = self.vqvae.decode(lowerCamelCase_ )["""sample"""]
UpperCAmelCase_ : Optional[int] = (images / 2 + 0.5).clamp(0 ,1 )
UpperCAmelCase_ : List[Any] = images.cpu().permute(0 ,2 ,3 ,1 ).numpy()
UpperCAmelCase_ : Tuple = (images * 255).round().astype("""uint8""" )
UpperCAmelCase_ : Any = list(
(Image.fromarray(_[:, :, 0] ) for _ in images)
if images.shape[3] == 1
else (Image.fromarray(lowerCamelCase_ ,mode="""RGB""" ).convert("""L""" ) for _ in images) )
UpperCAmelCase_ : str = [self.mel.image_to_audio(lowerCamelCase_ ) for _ in images]
if not return_dict:
return images, (self.mel.get_sample_rate(), audios)
return BaseOutput(**AudioPipelineOutput(np.array(lowerCamelCase_ )[:, np.newaxis, :] ) ,**ImagePipelineOutput(lowerCamelCase_ ) )
@torch.no_grad()
def A__ ( self: Dict ,lowerCamelCase_: List[Image.Image] ,lowerCamelCase_: int = 50 ) -> np.ndarray:
assert isinstance(self.scheduler ,lowerCamelCase_ )
self.scheduler.set_timesteps(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = np.array(
[np.frombuffer(image.tobytes() ,dtype="""uint8""" ).reshape((1, image.height, image.width) ) for image in images] )
UpperCAmelCase_ : Union[str, Any] = (sample / 255) * 2 - 1
UpperCAmelCase_ : List[Any] = torch.Tensor(lowerCamelCase_ ).to(self.device )
for t in self.progress_bar(torch.flip(self.scheduler.timesteps ,(0,) ) ):
UpperCAmelCase_ : str = t - self.scheduler.config.num_train_timesteps // self.scheduler.num_inference_steps
UpperCAmelCase_ : Tuple = self.scheduler.alphas_cumprod[t]
UpperCAmelCase_ : Optional[Any] = (
self.scheduler.alphas_cumprod[prev_timestep]
if prev_timestep >= 0
else self.scheduler.final_alpha_cumprod
)
UpperCAmelCase_ : Optional[Any] = 1 - alpha_prod_t
UpperCAmelCase_ : Any = self.unet(lowerCamelCase_ ,lowerCamelCase_ )["""sample"""]
UpperCAmelCase_ : int = (1 - alpha_prod_t_prev) ** 0.5 * model_output
UpperCAmelCase_ : int = (sample - pred_sample_direction) * alpha_prod_t_prev ** (-0.5)
UpperCAmelCase_ : Tuple = sample * alpha_prod_t ** 0.5 + beta_prod_t ** 0.5 * model_output
return sample
@staticmethod
def A__ ( lowerCamelCase_: torch.Tensor ,lowerCamelCase_: torch.Tensor ,lowerCamelCase_: float ) -> torch.Tensor:
UpperCAmelCase_ : List[Any] = acos(torch.dot(torch.flatten(lowerCamelCase_ ) ,torch.flatten(lowerCamelCase_ ) ) / torch.norm(lowerCamelCase_ ) / torch.norm(lowerCamelCase_ ) )
return sin((1 - alpha) * theta ) * xa / sin(lowerCamelCase_ ) + sin(alpha * theta ) * xa / sin(lowerCamelCase_ )
| 345 |
from collections import OrderedDict
from typing import TYPE_CHECKING, Any, Mapping, Optional
from packaging import version
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...onnx.utils import compute_effective_axis_dimension
from ...utils import logging
if TYPE_CHECKING:
from ...processing_utils import ProcessorMixin
from ...utils import TensorType
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
'''microsoft/layoutlmv3-base''': '''https://huggingface.co/microsoft/layoutlmv3-base/resolve/main/config.json''',
}
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = "layoutlmv3"
def __init__( self: str ,lowerCamelCase_: Any=50265 ,lowerCamelCase_: int=768 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: Any=12 ,lowerCamelCase_: List[Any]=3072 ,lowerCamelCase_: str="gelu" ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Any=0.1 ,lowerCamelCase_: Tuple=512 ,lowerCamelCase_: Union[str, Any]=2 ,lowerCamelCase_: Dict=0.0_2 ,lowerCamelCase_: List[str]=1e-5 ,lowerCamelCase_: int=1 ,lowerCamelCase_: int=0 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Dict=1024 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Tuple=128 ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=32 ,lowerCamelCase_: Union[str, Any]=128 ,lowerCamelCase_: Tuple=64 ,lowerCamelCase_: Tuple=256 ,lowerCamelCase_: List[str]=True ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Any=True ,lowerCamelCase_: Dict=224 ,lowerCamelCase_: Optional[int]=3 ,lowerCamelCase_: Optional[int]=16 ,lowerCamelCase_: Dict=None ,**lowerCamelCase_: str ,) -> List[Any]:
super().__init__(
vocab_size=lowerCamelCase_ ,hidden_size=lowerCamelCase_ ,num_hidden_layers=lowerCamelCase_ ,num_attention_heads=lowerCamelCase_ ,intermediate_size=lowerCamelCase_ ,hidden_act=lowerCamelCase_ ,hidden_dropout_prob=lowerCamelCase_ ,attention_probs_dropout_prob=lowerCamelCase_ ,max_position_embeddings=lowerCamelCase_ ,type_vocab_size=lowerCamelCase_ ,initializer_range=lowerCamelCase_ ,layer_norm_eps=lowerCamelCase_ ,pad_token_id=lowerCamelCase_ ,bos_token_id=lowerCamelCase_ ,eos_token_id=lowerCamelCase_ ,**lowerCamelCase_ ,)
UpperCAmelCase_ : List[Any] = max_ad_position_embeddings
UpperCAmelCase_ : Optional[int] = coordinate_size
UpperCAmelCase_ : Optional[int] = shape_size
UpperCAmelCase_ : Optional[Any] = has_relative_attention_bias
UpperCAmelCase_ : Optional[int] = rel_pos_bins
UpperCAmelCase_ : Union[str, Any] = max_rel_pos
UpperCAmelCase_ : Dict = has_spatial_attention_bias
UpperCAmelCase_ : Optional[int] = rel_ad_pos_bins
UpperCAmelCase_ : Tuple = max_rel_ad_pos
UpperCAmelCase_ : Union[str, Any] = text_embed
UpperCAmelCase_ : Optional[Any] = visual_embed
UpperCAmelCase_ : List[str] = input_size
UpperCAmelCase_ : str = num_channels
UpperCAmelCase_ : Optional[int] = patch_size
UpperCAmelCase_ : Tuple = classifier_dropout
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : Optional[Any] = version.parse("1.12" )
@property
def A__ ( self: Dict ) -> Mapping[str, Mapping[int, str]]:
# The order of inputs is different for question answering and sequence classification
if self.task in ["question-answering", "sequence-classification"]:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels""", 2: """height""", 3: """width"""}),
] )
else:
return OrderedDict(
[
("""input_ids""", {0: """batch""", 1: """sequence"""}),
("""bbox""", {0: """batch""", 1: """sequence"""}),
("""attention_mask""", {0: """batch""", 1: """sequence"""}),
("""pixel_values""", {0: """batch""", 1: """num_channels"""}),
] )
@property
def A__ ( self: Any ) -> float:
return 1e-5
@property
def A__ ( self: int ) -> int:
return 12
def A__ ( self: List[str] ,lowerCamelCase_: "ProcessorMixin" ,lowerCamelCase_: int = -1 ,lowerCamelCase_: int = -1 ,lowerCamelCase_: bool = False ,lowerCamelCase_: Optional["TensorType"] = None ,lowerCamelCase_: int = 3 ,lowerCamelCase_: int = 40 ,lowerCamelCase_: int = 40 ,) -> Mapping[str, Any]:
setattr(processor.image_processor ,"""apply_ocr""" ,lowerCamelCase_ )
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
UpperCAmelCase_ : List[str] = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_batch ,num_token_to_add=0 )
# If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX
UpperCAmelCase_ : int = processor.tokenizer.num_special_tokens_to_add(lowerCamelCase_ )
UpperCAmelCase_ : int = compute_effective_axis_dimension(
lowerCamelCase_ ,fixed_dimension=OnnxConfig.default_fixed_sequence ,num_token_to_add=lowerCamelCase_ )
# Generate dummy inputs according to compute batch and sequence
UpperCAmelCase_ : Optional[int] = [[""" """.join([processor.tokenizer.unk_token] ) * seq_length]] * batch_size
# Generate dummy bounding boxes
UpperCAmelCase_ : List[Any] = [[[48, 84, 73, 128]]] * batch_size
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
# batch_size = compute_effective_axis_dimension(batch_size, fixed_dimension=OnnxConfig.default_fixed_batch)
UpperCAmelCase_ : Any = self._generate_dummy_images(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = dict(
processor(
lowerCamelCase_ ,text=lowerCamelCase_ ,boxes=lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,) )
return inputs
| 345 | 1 |
from __future__ import annotations
def lowerCamelCase_ ( _a : list[float] ):
'''simple docstring'''
if len(_a ) < 2:
raise ValueError("""Monogons and Digons are not polygons in the Euclidean space""" )
if any(i <= 0 for i in nums ):
raise ValueError("""All values must be greater than 0""" )
UpperCAmelCase_ : str = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 345 |
import argparse
from argparse import Namespace
import torch
from torch import nn
from transformers import XGLMConfig, XGLMForCausalLM
def lowerCamelCase_ ( _a : List[Any] ):
'''simple docstring'''
UpperCAmelCase_ : Optional[int] = [
"""decoder.version""",
"""decoder.output_projection.weight""",
"""_float_tensor""",
"""decoder.embed_positions._float_tensor""",
]
for k in ignore_keys:
state_dict.pop(_a , _a )
def lowerCamelCase_ ( _a : Any ):
'''simple docstring'''
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = emb.weight.shape
UpperCAmelCase_ : Tuple = nn.Linear(_a , _a , bias=_a )
UpperCAmelCase_ : List[Any] = emb.weight.data
return lin_layer
def lowerCamelCase_ ( _a : Dict ):
'''simple docstring'''
UpperCAmelCase_ : int = torch.load(_a , map_location="""cpu""" )
UpperCAmelCase_ : Dict = Namespace(**checkpoint["""cfg"""]["""model"""] )
UpperCAmelCase_ : Optional[int] = checkpoint["""model"""]
remove_ignore_keys_(_a )
UpperCAmelCase_ : str = state_dict["""decoder.embed_tokens.weight"""].shape[0]
UpperCAmelCase_ : List[str] = {key.replace("""decoder""" , """model""" ): val for key, val in state_dict.items()}
UpperCAmelCase_ : int = XGLMConfig(
vocab_size=_a , max_position_embeddings=args.max_target_positions , num_layers=args.decoder_layers , attention_heads=args.decoder_attention_heads , ffn_dim=args.decoder_ffn_embed_dim , d_model=args.decoder_embed_dim , layerdrop=args.decoder_layerdrop , dropout=args.dropout , attention_dropout=args.attention_dropout , activation_dropout=args.activation_dropout , activation_function="""gelu""" , scale_embedding=not args.no_scale_embedding , tie_word_embeddings=args.share_decoder_input_output_embed , )
UpperCAmelCase_ : List[str] = XGLMForCausalLM(_a )
UpperCAmelCase_ : Tuple = model.load_state_dict(_a , strict=_a )
print(_a )
UpperCAmelCase_ : Optional[Any] = make_linear_from_emb(model.model.embed_tokens )
return model
if __name__ == "__main__":
UpperCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument('''fairseq_path''', type=str, help='''path to a model.pt on local filesystem.''')
parser.add_argument('''pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
UpperCamelCase_ = parser.parse_args()
UpperCamelCase_ = convert_fairseq_xglm_checkpoint_from_disk(args.fairseq_path)
model.save_pretrained(args.pytorch_dump_folder_path)
| 345 | 1 |
import unittest
from transformers import load_tool
from .test_tools_common import ToolTesterMixin
class _snake_case ( unittest.TestCase , __snake_case ):
'''simple docstring'''
def A__ ( self: Any ) -> Tuple:
UpperCAmelCase_ : int = load_tool("""text-classification""" )
self.tool.setup()
UpperCAmelCase_ : List[str] = load_tool("""text-classification""" ,remote=lowerCamelCase_ )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ : Optional[Any] = self.tool("""That's quite cool""" ,["""positive""", """negative"""] )
self.assertEqual(lowerCamelCase_ ,"""positive""" )
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[str] = self.remote_tool("""That's quite cool""" ,["""positive""", """negative"""] )
self.assertEqual(lowerCamelCase_ ,"""positive""" )
def A__ ( self: Optional[Any] ) -> int:
UpperCAmelCase_ : Dict = self.tool(text="""That's quite cool""" ,labels=["""positive""", """negative"""] )
self.assertEqual(lowerCamelCase_ ,"""positive""" )
def A__ ( self: Dict ) -> Optional[Any]:
UpperCAmelCase_ : Tuple = self.remote_tool(text="""That's quite cool""" ,labels=["""positive""", """negative"""] )
self.assertEqual(lowerCamelCase_ ,"""positive""" )
| 345 |
import collections
import inspect
import unittest
from transformers import FocalNetConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
FocalNetBackbone,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetModel,
)
from transformers.models.focalnet.modeling_focalnet import FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any]=13 ,lowerCamelCase_: Optional[int]=32 ,lowerCamelCase_: List[str]=2 ,lowerCamelCase_: Optional[Any]=3 ,lowerCamelCase_: int=16 ,lowerCamelCase_: Optional[Any]=[32, 64, 128] ,lowerCamelCase_: Optional[int]=[1, 2, 1] ,lowerCamelCase_: Union[str, Any]=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[str]=2.0 ,lowerCamelCase_: List[Any]=True ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: List[str]=0.0 ,lowerCamelCase_: Optional[int]=0.1 ,lowerCamelCase_: Optional[int]="gelu" ,lowerCamelCase_: Any=False ,lowerCamelCase_: Dict=True ,lowerCamelCase_: Union[str, Any]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: int=True ,lowerCamelCase_: Tuple=None ,lowerCamelCase_: str=True ,lowerCamelCase_: Dict=10 ,lowerCamelCase_: str=8 ,lowerCamelCase_: Union[str, Any]=["stage1", "stage2"] ,lowerCamelCase_: Optional[Any]=[1, 2] ,) -> str:
UpperCAmelCase_ : List[Any] = parent
UpperCAmelCase_ : Tuple = batch_size
UpperCAmelCase_ : Any = image_size
UpperCAmelCase_ : str = patch_size
UpperCAmelCase_ : List[str] = num_channels
UpperCAmelCase_ : Dict = embed_dim
UpperCAmelCase_ : Dict = hidden_sizes
UpperCAmelCase_ : str = depths
UpperCAmelCase_ : int = num_heads
UpperCAmelCase_ : List[Any] = window_size
UpperCAmelCase_ : Union[str, Any] = mlp_ratio
UpperCAmelCase_ : int = qkv_bias
UpperCAmelCase_ : List[str] = hidden_dropout_prob
UpperCAmelCase_ : Union[str, Any] = attention_probs_dropout_prob
UpperCAmelCase_ : Optional[int] = drop_path_rate
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : List[Any] = use_absolute_embeddings
UpperCAmelCase_ : List[Any] = patch_norm
UpperCAmelCase_ : int = layer_norm_eps
UpperCAmelCase_ : int = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Optional[Any] = scope
UpperCAmelCase_ : Union[str, Any] = use_labels
UpperCAmelCase_ : Union[str, Any] = type_sequence_label_size
UpperCAmelCase_ : Optional[int] = encoder_stride
UpperCAmelCase_ : Optional[int] = out_features
UpperCAmelCase_ : Optional[int] = out_indices
def A__ ( self: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : int = None
if self.use_labels:
UpperCAmelCase_ : str = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : Any = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Tuple:
return FocalNetConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,hidden_sizes=self.hidden_sizes ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,out_features=self.out_features ,out_indices=self.out_indices ,)
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: str ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : Dict = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : Optional[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Any ,lowerCamelCase_: Optional[int] ) -> List[str]:
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size, 8, 8] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,len(config.out_features ) )
self.parent.assertListEqual(model.channels ,config.hidden_sizes[:-1] )
# verify backbone works with out_features=None
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : List[str] = FocalNetBackbone(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Tuple = model(lowerCamelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) ,1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) ,[self.batch_size, self.image_size * 2, 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) ,1 )
self.parent.assertListEqual(model.channels ,[config.hidden_sizes[-1]] )
def A__ ( self: Optional[int] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ) -> List[Any]:
UpperCAmelCase_ : Any = FocalNetForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.reconstruction.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : int = 1
UpperCAmelCase_ : List[str] = FocalNetForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.reconstruction.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Any ) -> int:
UpperCAmelCase_ : List[Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
# test greyscale images
UpperCAmelCase_ : List[Any] = 1
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : List[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : List[str] = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: Union[str, Any] ) -> Optional[int]:
UpperCAmelCase_ : List[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = config_and_inputs
UpperCAmelCase_ : int = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (
(
FocalNetModel,
FocalNetForImageClassification,
FocalNetForMaskedImageModeling,
FocalNetBackbone,
)
if is_torch_available()
else ()
)
A__ : Union[str, Any] = (
{"feature-extraction": FocalNetModel, "image-classification": FocalNetForImageClassification}
if is_torch_available()
else {}
)
A__ : Optional[Any] = False
A__ : Any = False
A__ : List[str] = False
A__ : Any = False
A__ : Any = False
def A__ ( self: List[str] ) -> Tuple:
UpperCAmelCase_ : Dict = FocalNetModelTester(self )
UpperCAmelCase_ : int = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 ,has_text_modality=lowerCamelCase_ )
def A__ ( self: List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: List[str] ) -> Union[str, Any]:
return
def A__ ( self: str ) -> List[str]:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
def A__ ( self: Tuple ) -> int:
UpperCAmelCase_ : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCamelCase_ )
def A__ ( self: Dict ) -> List[str]:
UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: int ) -> int:
UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@unittest.skip(reason="""FocalNet does not use inputs_embeds""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""FocalNet does not use feedforward chunking""" )
def A__ ( self: Optional[Any] ) -> Optional[Any]:
pass
def A__ ( self: Optional[Any] ) -> List[str]:
UpperCAmelCase_ , UpperCAmelCase_ : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : List[Any] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: str ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Dict = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : Any = [*signature.parameters.keys()]
UpperCAmelCase_ : List[str] = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Any ) -> List[str]:
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.hidden_states
UpperCAmelCase_ : List[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# FocalNet has a different seq_length
UpperCAmelCase_ : int = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Optional[int] = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Union[str, Any] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Tuple = reshaped_hidden_states[0].shape
UpperCAmelCase_ : List[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Union[str, Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> str:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Tuple = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : Union[str, Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : Union[str, Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Any = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes[:-1]:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : Optional[int] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
@slow
def A__ ( self: Optional[int] ) -> Optional[Any]:
for model_name in FOCALNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Tuple = FocalNetModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Optional[int] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : List[Any] = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Optional[int] ) -> str:
# TODO update organization
return AutoImageProcessor.from_pretrained("""microsoft/focalnet-tiny""" ) if is_vision_available() else None
@slow
def A__ ( self: List[Any] ) -> List[str]:
UpperCAmelCase_ : Optional[int] = FocalNetForImageClassification.from_pretrained("""microsoft/focalnet-tiny""" ).to(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = self.default_image_processor
UpperCAmelCase_ : Union[str, Any] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Dict = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Dict = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : str = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = torch.tensor([0.2_1_6_6, -0.4_3_6_8, 0.2_1_9_1] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
self.assertTrue(outputs.logits.argmax(dim=-1 ).item() ,281 )
@require_torch
class _snake_case ( __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : List[Any] = (FocalNetBackbone,) if is_torch_available() else ()
A__ : int = FocalNetConfig
A__ : List[str] = False
def A__ ( self: Any ) -> Optional[int]:
UpperCAmelCase_ : str = FocalNetModelTester(self )
| 345 | 1 |
import baseaa
import io
import json
import os
from copy import deepcopy
from ..optimizer import AcceleratedOptimizer
from ..scheduler import AcceleratedScheduler
class _snake_case :
'''simple docstring'''
def __init__( self: int ,lowerCamelCase_: Tuple ) -> List[Any]:
if isinstance(lowerCamelCase_ ,lowerCamelCase_ ):
# Don't modify user's data should they want to reuse it (e.g. in tests), because once we
# modified it, it will not be accepted here again, since `auto` values would have been overridden
UpperCAmelCase_ : List[str] = deepcopy(lowerCamelCase_ )
elif os.path.exists(lowerCamelCase_ ):
with io.open(lowerCamelCase_ ,"""r""" ,encoding="""utf-8""" ) as f:
UpperCAmelCase_ : List[str] = json.load(lowerCamelCase_ )
else:
try:
UpperCAmelCase_ : int = baseaa.urlsafe_baadecode(lowerCamelCase_ ).decode("""utf-8""" )
UpperCAmelCase_ : Optional[Any] = json.loads(lowerCamelCase_ )
except (UnicodeDecodeError, AttributeError, ValueError):
raise ValueError(
F'''Expected a string path to an existing deepspeed config, or a dictionary, or a base64 encoded string. Received: {config_file_or_dict}''' )
UpperCAmelCase_ : int = config
self.set_stage_and_offload()
def A__ ( self: Union[str, Any] ) -> str:
# zero stage - this is done as early as possible, before model is created, to allow
# ``is_deepspeed_zero3_enabled`` query and getting to the early deepspeed config object
# during ``zero.Init()`` which needs to know the dtype, and some other hparams.
UpperCAmelCase_ : Optional[int] = self.get_value("""zero_optimization.stage""" ,-1 )
# offload
UpperCAmelCase_ : Optional[Any] = False
if self.is_zeroa() or self.is_zeroa():
UpperCAmelCase_ : Any = set(["""cpu""", """nvme"""] )
UpperCAmelCase_ : int = set(
[
self.get_value("""zero_optimization.offload_optimizer.device""" ),
self.get_value("""zero_optimization.offload_param.device""" ),
] )
if len(offload_devices & offload_devices_valid ) > 0:
UpperCAmelCase_ : Optional[int] = True
def A__ ( self: int ,lowerCamelCase_: Optional[Any] ) -> Dict:
UpperCAmelCase_ : Optional[Any] = self.config
# find the config node of interest if it exists
UpperCAmelCase_ : str = ds_key_long.split(""".""" )
UpperCAmelCase_ : int = nodes.pop()
for node in nodes:
UpperCAmelCase_ : List[Any] = config.get(lowerCamelCase_ )
if config is None:
return None, ds_key
return config, ds_key
def A__ ( self: List[Any] ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: Any=None ) -> Optional[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = self.find_config_node(lowerCamelCase_ )
if config is None:
return default
return config.get(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: int ,lowerCamelCase_: Tuple ,lowerCamelCase_: int=False ) -> str:
UpperCAmelCase_ : List[Any] = self.config
# find the config node of interest if it exists
UpperCAmelCase_ : Union[str, Any] = ds_key_long.split(""".""" )
for node in nodes:
UpperCAmelCase_ : Tuple = config
UpperCAmelCase_ : Dict = config.get(lowerCamelCase_ )
if config is None:
if must_exist:
raise ValueError(F'''Can\'t find {ds_key_long} entry in the config: {self.config}''' )
else:
return
# if found remove it
if parent_config is not None:
parent_config.pop(lowerCamelCase_ )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Optional[Any] ) -> Any:
UpperCAmelCase_ : Dict = self.get_value(lowerCamelCase_ )
return False if value is None else bool(lowerCamelCase_ )
def A__ ( self: Dict ,lowerCamelCase_: Optional[Any] ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = self.get_value(lowerCamelCase_ )
return False if value is None else not bool(lowerCamelCase_ )
def A__ ( self: Dict ) -> int:
return self._stage == 2
def A__ ( self: List[str] ) -> Union[str, Any]:
return self._stage == 3
def A__ ( self: List[str] ) -> str:
return self._offload
class _snake_case :
'''simple docstring'''
def __init__( self: str ,lowerCamelCase_: Any ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = engine
def A__ ( self: Optional[Any] ,lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> List[str]:
# runs backpropagation and handles mixed precision
self.engine.backward(lowerCamelCase_ ,**lowerCamelCase_ )
# Deepspeed's `engine.step` performs the following operations:
# - gradient accumulation check
# - gradient clipping
# - optimizer step
# - zero grad
# - checking overflow
# - lr_scheduler step (only if engine.lr_scheduler is not None)
self.engine.step()
# and this plugin overrides the above calls with no-ops when Accelerate runs under
# Deepspeed, but allows normal functionality for non-Deepspeed cases thus enabling a simple
# training loop that works transparently under many training regimes.
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: Dict ,lowerCamelCase_: Tuple ) -> Union[str, Any]:
super().__init__(lowerCamelCase_ ,device_placement=lowerCamelCase_ ,scaler=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = hasattr(self.optimizer ,"""overflow""" )
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[str]=None ) -> Any:
pass # `accelerator.backward(loss)` is doing that automatically. Therefore, its implementation is not needed
def A__ ( self: str ) -> List[Any]:
pass # `accelerator.backward(loss)` is doing that automatically. Therefore, its implementation is not needed
@property
def A__ ( self: Union[str, Any] ) -> Tuple:
if self.__has_overflow__:
return self.optimizer.overflow
return False
class _snake_case ( __snake_case ):
'''simple docstring'''
def __init__( self: str ,lowerCamelCase_: int ,lowerCamelCase_: Optional[int] ) -> Optional[Any]:
super().__init__(lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: Optional[int] ) -> int:
pass # `accelerator.backward(loss)` is doing that automatically. Therefore, its implementation is not needed
class _snake_case :
'''simple docstring'''
def __init__( self: Union[str, Any] ,lowerCamelCase_: Optional[int] ,lowerCamelCase_: Tuple=0.0_0_1 ,lowerCamelCase_: Any=0 ,**lowerCamelCase_: str ) -> int:
UpperCAmelCase_ : Union[str, Any] = params
UpperCAmelCase_ : Union[str, Any] = lr
UpperCAmelCase_ : str = weight_decay
UpperCAmelCase_ : Optional[int] = kwargs
class _snake_case :
'''simple docstring'''
def __init__( self: Union[str, Any] ,lowerCamelCase_: str ,lowerCamelCase_: int=None ,lowerCamelCase_: str=0 ,**lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Dict = optimizer
UpperCAmelCase_ : Optional[int] = total_num_steps
UpperCAmelCase_ : str = warmup_num_steps
UpperCAmelCase_ : Optional[Any] = kwargs
| 345 |
import collections
import inspect
import unittest
from transformers import SwinvaConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import SwinvaForImageClassification, SwinvaForMaskedImageModeling, SwinvaModel
from transformers.models.swinva.modeling_swinva import SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _snake_case :
'''simple docstring'''
def __init__( self: Tuple ,lowerCamelCase_: List[str] ,lowerCamelCase_: int=13 ,lowerCamelCase_: int=32 ,lowerCamelCase_: Optional[int]=2 ,lowerCamelCase_: Any=3 ,lowerCamelCase_: str=16 ,lowerCamelCase_: Optional[Any]=[1, 2, 1] ,lowerCamelCase_: Tuple=[2, 2, 4] ,lowerCamelCase_: int=2 ,lowerCamelCase_: List[Any]=2.0 ,lowerCamelCase_: str=True ,lowerCamelCase_: Optional[int]=0.0 ,lowerCamelCase_: List[Any]=0.0 ,lowerCamelCase_: List[str]=0.1 ,lowerCamelCase_: Tuple="gelu" ,lowerCamelCase_: Union[str, Any]=False ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=0.0_2 ,lowerCamelCase_: int=1e-5 ,lowerCamelCase_: Optional[int]=True ,lowerCamelCase_: Union[str, Any]=None ,lowerCamelCase_: Union[str, Any]=True ,lowerCamelCase_: Optional[int]=10 ,lowerCamelCase_: Tuple=8 ,) -> List[Any]:
UpperCAmelCase_ : List[str] = parent
UpperCAmelCase_ : int = batch_size
UpperCAmelCase_ : int = image_size
UpperCAmelCase_ : Union[str, Any] = patch_size
UpperCAmelCase_ : Optional[Any] = num_channels
UpperCAmelCase_ : int = embed_dim
UpperCAmelCase_ : Union[str, Any] = depths
UpperCAmelCase_ : List[str] = num_heads
UpperCAmelCase_ : int = window_size
UpperCAmelCase_ : List[str] = mlp_ratio
UpperCAmelCase_ : Tuple = qkv_bias
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : str = attention_probs_dropout_prob
UpperCAmelCase_ : Tuple = drop_path_rate
UpperCAmelCase_ : List[str] = hidden_act
UpperCAmelCase_ : int = use_absolute_embeddings
UpperCAmelCase_ : Any = patch_norm
UpperCAmelCase_ : Optional[int] = layer_norm_eps
UpperCAmelCase_ : Tuple = initializer_range
UpperCAmelCase_ : Optional[Any] = is_training
UpperCAmelCase_ : Dict = scope
UpperCAmelCase_ : int = use_labels
UpperCAmelCase_ : Optional[Any] = type_sequence_label_size
UpperCAmelCase_ : List[str] = encoder_stride
def A__ ( self: Any ) -> int:
UpperCAmelCase_ : int = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
UpperCAmelCase_ : List[Any] = None
if self.use_labels:
UpperCAmelCase_ : Optional[int] = ids_tensor([self.batch_size] ,self.type_sequence_label_size )
UpperCAmelCase_ : str = self.get_config()
return config, pixel_values, labels
def A__ ( self: List[Any] ) -> Union[str, Any]:
return SwinvaConfig(
image_size=self.image_size ,patch_size=self.patch_size ,num_channels=self.num_channels ,embed_dim=self.embed_dim ,depths=self.depths ,num_heads=self.num_heads ,window_size=self.window_size ,mlp_ratio=self.mlp_ratio ,qkv_bias=self.qkv_bias ,hidden_dropout_prob=self.hidden_dropout_prob ,attention_probs_dropout_prob=self.attention_probs_dropout_prob ,drop_path_rate=self.drop_path_rate ,hidden_act=self.hidden_act ,use_absolute_embeddings=self.use_absolute_embeddings ,path_norm=self.patch_norm ,layer_norm_eps=self.layer_norm_eps ,initializer_range=self.initializer_range ,encoder_stride=self.encoder_stride ,)
def A__ ( self: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Union[str, Any] ,lowerCamelCase_: List[str] ) -> str:
UpperCAmelCase_ : str = SwinvaModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[Any] = model(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
UpperCAmelCase_ : List[Any] = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape ,(self.batch_size, expected_seq_len, expected_dim) )
def A__ ( self: List[Any] ,lowerCamelCase_: List[Any] ,lowerCamelCase_: int ,lowerCamelCase_: int ) -> int:
UpperCAmelCase_ : Any = SwinvaForMaskedImageModeling(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Union[str, Any] = model(lowerCamelCase_ )
self.parent.assertEqual(
result.logits.shape ,(self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
UpperCAmelCase_ : str = 1
UpperCAmelCase_ : Optional[Any] = SwinvaForMaskedImageModeling(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Dict = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
UpperCAmelCase_ : int = model(lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, 1, self.image_size, self.image_size) )
def A__ ( self: int ,lowerCamelCase_: int ,lowerCamelCase_: List[Any] ,lowerCamelCase_: Optional[Any] ) -> int:
UpperCAmelCase_ : Union[str, Any] = self.type_sequence_label_size
UpperCAmelCase_ : int = SwinvaForImageClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
UpperCAmelCase_ : Optional[int] = model(lowerCamelCase_ ,labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape ,(self.batch_size, self.type_sequence_label_size) )
def A__ ( self: str ) -> Union[str, Any]:
UpperCAmelCase_ : Optional[Any] = self.prepare_config_and_inputs()
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = config_and_inputs
UpperCAmelCase_ : Optional[int] = {"""pixel_values""": pixel_values}
return config, inputs_dict
@require_torch
class _snake_case ( __snake_case , __snake_case , unittest.TestCase ):
'''simple docstring'''
A__ : Tuple = (
(SwinvaModel, SwinvaForImageClassification, SwinvaForMaskedImageModeling) if is_torch_available() else ()
)
A__ : Optional[Any] = (
{"feature-extraction": SwinvaModel, "image-classification": SwinvaForImageClassification}
if is_torch_available()
else {}
)
A__ : List[Any] = False
A__ : Tuple = False
A__ : int = False
A__ : Union[str, Any] = False
def A__ ( self: List[str] ) -> Optional[Any]:
UpperCAmelCase_ : Any = SwinvaModelTester(self )
UpperCAmelCase_ : str = ConfigTester(self ,config_class=lowerCamelCase_ ,embed_dim=37 )
def A__ ( self: Optional[int] ) -> List[Any]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def A__ ( self: Any ) -> Dict:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase_ )
@unittest.skip(reason="""Got `CUDA error: misaligned address` with PyTorch 2.0.0.""" )
def A__ ( self: int ) -> Dict:
pass
@unittest.skip(reason="""Swinv2 does not use inputs_embeds""" )
def A__ ( self: Tuple ) -> List[str]:
pass
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : List[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(lowerCamelCase_ )
self.assertIsInstance(model.get_input_embeddings() ,(nn.Module) )
UpperCAmelCase_ : Tuple = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCamelCase_ ,nn.Linear ) )
def A__ ( self: Optional[Any] ) -> Optional[int]:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
UpperCAmelCase_ : Dict = model_class(lowerCamelCase_ )
UpperCAmelCase_ : Any = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
UpperCAmelCase_ : int = [*signature.parameters.keys()]
UpperCAmelCase_ : Tuple = ["""pixel_values"""]
self.assertListEqual(arg_names[:1] ,lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Optional[Any]:
UpperCAmelCase_ , UpperCAmelCase_ : int = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Any = True
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
UpperCAmelCase_ : Union[str, Any] = False
UpperCAmelCase_ : str = True
UpperCAmelCase_ : List[Any] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[int] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : Optional[Any] = outputs.attentions
UpperCAmelCase_ : List[str] = len(self.model_tester.depths )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
UpperCAmelCase_ : str = True
UpperCAmelCase_ : Optional[Any] = config.window_size**2
UpperCAmelCase_ : Optional[int] = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[Any] = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
UpperCAmelCase_ : Optional[Any] = len(lowerCamelCase_ )
# Check attention is always last and order is fine
UpperCAmelCase_ : Tuple = True
UpperCAmelCase_ : List[Any] = True
UpperCAmelCase_ : Tuple = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : Union[str, Any] = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
if hasattr(self.model_tester ,"""num_hidden_states_types""" ):
UpperCAmelCase_ : List[Any] = self.model_tester.num_hidden_states_types
else:
# also another +1 for reshaped_hidden_states
UpperCAmelCase_ : List[str] = 2
self.assertEqual(out_len + added_hidden_states ,len(lowerCamelCase_ ) )
UpperCAmelCase_ : Any = outputs.attentions
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) ,[self.model_tester.num_heads[0], window_size_squared, window_size_squared] ,)
def A__ ( self: List[str] ,lowerCamelCase_: Dict ,lowerCamelCase_: Tuple ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: Optional[int] ) -> List[Any]:
UpperCAmelCase_ : str = model_class(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
with torch.no_grad():
UpperCAmelCase_ : int = model(**self._prepare_for_class(lowerCamelCase_ ,lowerCamelCase_ ) )
UpperCAmelCase_ : List[str] = outputs.hidden_states
UpperCAmelCase_ : Optional[Any] = getattr(
self.model_tester ,"""expected_num_hidden_layers""" ,len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
# Swinv2 has a different seq_length
UpperCAmelCase_ : Optional[Any] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : int = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
UpperCAmelCase_ : Optional[int] = outputs.reshaped_hidden_states
self.assertEqual(len(lowerCamelCase_ ) ,lowerCamelCase_ )
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ : Union[str, Any] = reshaped_hidden_states[0].shape
UpperCAmelCase_ : Optional[Any] = (
reshaped_hidden_states[0].view(lowerCamelCase_ ,lowerCamelCase_ ,height * width ).permute(0 ,2 ,1 )
)
self.assertListEqual(
list(reshaped_hidden_states.shape[-2:] ) ,[num_patches, self.model_tester.embed_dim] ,)
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Dict = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes:
UpperCAmelCase_ : Any = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : str = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ )
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ , UpperCAmelCase_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : Union[str, Any] = 3
UpperCAmelCase_ : Optional[int] = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size ,collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
UpperCAmelCase_ : List[str] = (
config.patch_size
if isinstance(config.patch_size ,collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
UpperCAmelCase_ : List[Any] = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
UpperCAmelCase_ : Optional[Any] = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes:
UpperCAmelCase_ : Optional[Any] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
UpperCAmelCase_ : List[str] = True
self.check_hidden_states_output(lowerCamelCase_ ,lowerCamelCase_ ,lowerCamelCase_ ,(padded_height, padded_width) )
def A__ ( self: Optional[int] ) -> str:
UpperCAmelCase_ : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowerCamelCase_ )
def A__ ( self: Union[str, Any] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase_ )
@slow
def A__ ( self: str ) -> Tuple:
for model_name in SWINV2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
UpperCAmelCase_ : Dict = SwinvaModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
def A__ ( self: Any ) -> int:
UpperCAmelCase_ , UpperCAmelCase_ : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
UpperCAmelCase_ : List[str] = _config_zero_init(lowerCamelCase_ )
for model_class in self.all_model_classes:
UpperCAmelCase_ : int = model_class(config=lowerCamelCase_ )
for name, param in model.named_parameters():
if "embeddings" not in name and "logit_scale" not in name and param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() ,[0.0, 1.0] ,msg=F'''Parameter {name} of model {model_class} seems not properly initialized''' ,)
@require_vision
@require_torch
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@cached_property
def A__ ( self: Dict ) -> Optional[Any]:
return (
AutoImageProcessor.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" )
if is_vision_available()
else None
)
@slow
def A__ ( self: str ) -> List[Any]:
UpperCAmelCase_ : Tuple = SwinvaForImageClassification.from_pretrained("""microsoft/swinv2-tiny-patch4-window8-256""" ).to(
lowerCamelCase_ )
UpperCAmelCase_ : Any = self.default_image_processor
UpperCAmelCase_ : List[str] = Image.open("""./tests/fixtures/tests_samples/COCO/000000039769.png""" )
UpperCAmelCase_ : Optional[int] = image_processor(images=lowerCamelCase_ ,return_tensors="""pt""" ).to(lowerCamelCase_ )
# forward pass
with torch.no_grad():
UpperCAmelCase_ : Optional[Any] = model(**lowerCamelCase_ )
# verify the logits
UpperCAmelCase_ : Dict = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape ,lowerCamelCase_ )
UpperCAmelCase_ : Any = torch.tensor([-0.3_9_4_7, -0.4_3_0_6, 0.0_0_2_6] ).to(lowerCamelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] ,lowerCamelCase_ ,atol=1e-4 ) )
| 345 | 1 |
from ...processing_utils import ProcessorMixin
from ...tokenization_utils_base import BatchEncoding
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : List[str] = ["image_processor", "tokenizer"]
A__ : Optional[Any] = "AutoImageProcessor"
A__ : Optional[Any] = "AutoTokenizer"
def __init__( self: List[str] ,lowerCamelCase_: Any ,lowerCamelCase_: List[str] ) -> int:
super().__init__(lowerCamelCase_ ,lowerCamelCase_ )
UpperCAmelCase_ : Any = self.image_processor
def __call__( self: Dict ,lowerCamelCase_: int=None ,lowerCamelCase_: Optional[Any]=None ,lowerCamelCase_: Optional[Any]=None ,**lowerCamelCase_: Tuple ) -> Optional[int]:
if text is None and images is None:
raise ValueError("""You have to specify either text or images. Both cannot be none.""" )
if text is not None:
UpperCAmelCase_ : Union[str, Any] = self.tokenizer(lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,**lowerCamelCase_ )
if images is not None:
UpperCAmelCase_ : List[str] = self.image_processor(lowerCamelCase_ ,return_tensors=lowerCamelCase_ ,**lowerCamelCase_ )
if text is not None and images is not None:
UpperCAmelCase_ : List[Any] = image_features.pixel_values
return encoding
elif text is not None:
return encoding
else:
return BatchEncoding(data=dict(**lowerCamelCase_ ) ,tensor_type=lowerCamelCase_ )
def A__ ( self: Optional[Any] ,*lowerCamelCase_: Tuple ,**lowerCamelCase_: int ) -> Dict:
return self.tokenizer.batch_decode(*lowerCamelCase_ ,**lowerCamelCase_ )
def A__ ( self: List[Any] ,*lowerCamelCase_: Optional[int] ,**lowerCamelCase_: List[str] ) -> str:
return self.tokenizer.decode(*lowerCamelCase_ ,**lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Optional[int]:
return ["input_ids", "attention_mask", "pixel_values"]
| 345 |
import json
import os
from functools import lru_cache
from typing import Dict, List, Optional, Tuple, Union
import regex as re
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...tokenization_utils_base import BatchEncoding, EncodedInput
from ...utils import PaddingStrategy, logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {'''vocab_file''': '''vocab.json''', '''merges_file''': '''merges.txt'''}
# See all LED models at https://huggingface.co/models?filter=LED
UpperCamelCase_ = {
'''vocab_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/vocab.json''',
},
'''merges_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/merges.txt''',
},
'''tokenizer_file''': {
'''allenai/led-base-16384''': '''https://huggingface.co/allenai/led-base-16384/resolve/main/tokenizer.json''',
},
}
UpperCamelCase_ = {
'''allenai/led-base-16384''': 16384,
}
@lru_cache()
# Copied from transformers.models.bart.tokenization_bart.bytes_to_unicode
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCAmelCase_ : int = (
list(range(ord("""!""" ) , ord("""~""" ) + 1 ) ) + list(range(ord("""¡""" ) , ord("""¬""" ) + 1 ) ) + list(range(ord("""®""" ) , ord("""ÿ""" ) + 1 ) )
)
UpperCAmelCase_ : Dict = bs[:]
UpperCAmelCase_ : Any = 0
for b in range(2**8 ):
if b not in bs:
bs.append(_a )
cs.append(2**8 + n )
n += 1
UpperCAmelCase_ : Any = [chr(_a ) for n in cs]
return dict(zip(_a , _a ) )
def lowerCamelCase_ ( _a : List[str] ):
'''simple docstring'''
UpperCAmelCase_ : Union[str, Any] = set()
UpperCAmelCase_ : List[Any] = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
UpperCAmelCase_ : Optional[int] = char
return pairs
class _snake_case ( __snake_case ):
'''simple docstring'''
A__ : str = VOCAB_FILES_NAMES
A__ : List[str] = PRETRAINED_VOCAB_FILES_MAP
A__ : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A__ : Optional[int] = ["input_ids", "attention_mask"]
def __init__( self: Union[str, Any] ,lowerCamelCase_: Tuple ,lowerCamelCase_: Any ,lowerCamelCase_: Union[str, Any]="replace" ,lowerCamelCase_: Optional[Any]="<s>" ,lowerCamelCase_: List[Any]="</s>" ,lowerCamelCase_: List[str]="</s>" ,lowerCamelCase_: int="<s>" ,lowerCamelCase_: int="<unk>" ,lowerCamelCase_: str="<pad>" ,lowerCamelCase_: Optional[Any]="<mask>" ,lowerCamelCase_: List[str]=False ,**lowerCamelCase_: Tuple ,) -> Any:
UpperCAmelCase_ : Union[str, Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else bos_token
UpperCAmelCase_ : int = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else eos_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else sep_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else cls_token
UpperCAmelCase_ : Optional[Any] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else unk_token
UpperCAmelCase_ : List[str] = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else pad_token
# Mask token behave like a normal word, i.e. include the space before it
UpperCAmelCase_ : str = AddedToken(lowerCamelCase_ ,lstrip=lowerCamelCase_ ,rstrip=lowerCamelCase_ ) if isinstance(lowerCamelCase_ ,lowerCamelCase_ ) else mask_token
super().__init__(
errors=lowerCamelCase_ ,bos_token=lowerCamelCase_ ,eos_token=lowerCamelCase_ ,unk_token=lowerCamelCase_ ,sep_token=lowerCamelCase_ ,cls_token=lowerCamelCase_ ,pad_token=lowerCamelCase_ ,mask_token=lowerCamelCase_ ,add_prefix_space=lowerCamelCase_ ,**lowerCamelCase_ ,)
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as vocab_handle:
UpperCAmelCase_ : Union[str, Any] = json.load(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = {v: k for k, v in self.encoder.items()}
UpperCAmelCase_ : Any = errors # how to handle errors in decoding
UpperCAmelCase_ : int = bytes_to_unicode()
UpperCAmelCase_ : Dict = {v: k for k, v in self.byte_encoder.items()}
with open(lowerCamelCase_ ,encoding="""utf-8""" ) as merges_handle:
UpperCAmelCase_ : Any = merges_handle.read().split("""\n""" )[1:-1]
UpperCAmelCase_ : int = [tuple(merge.split() ) for merge in bpe_merges]
UpperCAmelCase_ : Union[str, Any] = dict(zip(lowerCamelCase_ ,range(len(lowerCamelCase_ ) ) ) )
UpperCAmelCase_ : Tuple = {}
UpperCAmelCase_ : Optional[int] = add_prefix_space
# Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions
UpperCAmelCase_ : int = re.compile(R"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""" )
@property
# Copied from transformers.models.bart.tokenization_bart.BartTokenizer.vocab_size
def A__ ( self: List[str] ) -> List[str]:
return len(self.encoder )
def A__ ( self: Any ) -> Union[str, Any]:
return dict(self.encoder ,**self.added_tokens_encoder )
def A__ ( self: Tuple ,lowerCamelCase_: Dict ) -> Optional[Any]:
if token in self.cache:
return self.cache[token]
UpperCAmelCase_ : Union[str, Any] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : Union[str, Any] = get_pairs(lowerCamelCase_ )
if not pairs:
return token
while True:
UpperCAmelCase_ : Union[str, Any] = min(lowerCamelCase_ ,key=lambda lowerCamelCase_ : self.bpe_ranks.get(lowerCamelCase_ ,float("""inf""" ) ) )
if bigram not in self.bpe_ranks:
break
UpperCAmelCase_ , UpperCAmelCase_ : Any = bigram
UpperCAmelCase_ : Optional[Any] = []
UpperCAmelCase_ : List[str] = 0
while i < len(lowerCamelCase_ ):
try:
UpperCAmelCase_ : str = word.index(lowerCamelCase_ ,lowerCamelCase_ )
except ValueError:
new_word.extend(word[i:] )
break
else:
new_word.extend(word[i:j] )
UpperCAmelCase_ : Union[str, Any] = j
if word[i] == first and i < len(lowerCamelCase_ ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
UpperCAmelCase_ : List[str] = tuple(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = new_word
if len(lowerCamelCase_ ) == 1:
break
else:
UpperCAmelCase_ : List[str] = get_pairs(lowerCamelCase_ )
UpperCAmelCase_ : int = """ """.join(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = word
return word
def A__ ( self: Union[str, Any] ,lowerCamelCase_: Tuple ) -> List[str]:
UpperCAmelCase_ : str = []
for token in re.findall(self.pat ,lowerCamelCase_ ):
UpperCAmelCase_ : List[Any] = """""".join(
self.byte_encoder[b] for b in token.encode("""utf-8""" ) ) # Maps all our bytes to unicode strings, avoiding control tokens of the BPE (spaces in our case)
bpe_tokens.extend(bpe_token for bpe_token in self.bpe(lowerCamelCase_ ).split(""" """ ) )
return bpe_tokens
def A__ ( self: List[Any] ,lowerCamelCase_: Optional[Any] ) -> Optional[int]:
return self.encoder.get(lowerCamelCase_ ,self.encoder.get(self.unk_token ) )
def A__ ( self: List[str] ,lowerCamelCase_: str ) -> Optional[Any]:
return self.decoder.get(lowerCamelCase_ )
def A__ ( self: List[str] ,lowerCamelCase_: List[str] ) -> List[Any]:
UpperCAmelCase_ : str = """""".join(lowerCamelCase_ )
UpperCAmelCase_ : int = bytearray([self.byte_decoder[c] for c in text] ).decode("""utf-8""" ,errors=self.errors )
return text
def A__ ( self: Optional[Any] ,lowerCamelCase_: str ,lowerCamelCase_: Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCamelCase_ ):
logger.error(F'''Vocabulary path ({save_directory}) should be a directory''' )
return
UpperCAmelCase_ : List[Any] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""vocab_file"""] )
UpperCAmelCase_ : List[str] = os.path.join(
lowerCamelCase_ ,(filename_prefix + """-""" if filename_prefix else """""") + VOCAB_FILES_NAMES["""merges_file"""] )
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as f:
f.write(json.dumps(self.encoder ,indent=2 ,sort_keys=lowerCamelCase_ ,ensure_ascii=lowerCamelCase_ ) + """\n""" )
UpperCAmelCase_ : str = 0
with open(lowerCamelCase_ ,"""w""" ,encoding="""utf-8""" ) as writer:
writer.write("""#version: 0.2\n""" )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() ,key=lambda lowerCamelCase_ : kv[1] ):
if index != token_index:
logger.warning(
F'''Saving vocabulary to {merge_file}: BPE merge indices are not consecutive.'''
""" Please check that the tokenizer is not corrupted!""" )
UpperCAmelCase_ : Tuple = token_index
writer.write(""" """.join(lowerCamelCase_ ) + """\n""" )
index += 1
return vocab_file, merge_file
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
UpperCAmelCase_ : int = [self.cls_token_id]
UpperCAmelCase_ : Optional[int] = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def A__ ( self: Union[str, Any] ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ,lowerCamelCase_: bool = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=lowerCamelCase_ ,token_ids_a=lowerCamelCase_ ,already_has_special_tokens=lowerCamelCase_ )
if token_ids_a is None:
return [1] + ([0] * len(lowerCamelCase_ )) + [1]
return [1] + ([0] * len(lowerCamelCase_ )) + [1, 1] + ([0] * len(lowerCamelCase_ )) + [1]
def A__ ( self: str ,lowerCamelCase_: List[int] ,lowerCamelCase_: Optional[List[int]] = None ) -> List[int]:
UpperCAmelCase_ : Optional[Any] = [self.sep_token_id]
UpperCAmelCase_ : Optional[Any] = [self.cls_token_id]
if token_ids_a is None:
return len(cls + token_ids_a + sep ) * [0]
return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0]
def A__ ( self: Optional[Any] ,lowerCamelCase_: Optional[Any] ,lowerCamelCase_: str=False ,**lowerCamelCase_: List[str] ) -> Optional[int]:
UpperCAmelCase_ : Optional[int] = kwargs.pop("""add_prefix_space""" ,self.add_prefix_space )
if (is_split_into_words or add_prefix_space) and (len(lowerCamelCase_ ) > 0 and not text[0].isspace()):
UpperCAmelCase_ : Dict = """ """ + text
return (text, kwargs)
def A__ ( self: List[str] ,lowerCamelCase_: Union[Dict[str, EncodedInput], BatchEncoding] ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: PaddingStrategy = PaddingStrategy.DO_NOT_PAD ,lowerCamelCase_: Optional[int] = None ,lowerCamelCase_: Optional[bool] = None ,) -> dict:
UpperCAmelCase_ : Optional[int] = super()._pad(
encoded_inputs=lowerCamelCase_ ,max_length=lowerCamelCase_ ,padding_strategy=lowerCamelCase_ ,pad_to_multiple_of=lowerCamelCase_ ,return_attention_mask=lowerCamelCase_ ,)
# Load from model defaults
if return_attention_mask is None:
UpperCAmelCase_ : str = """attention_mask""" in self.model_input_names
if return_attention_mask and "global_attention_mask" in encoded_inputs:
UpperCAmelCase_ : str = encoded_inputs[self.model_input_names[0]]
# `global_attention_mask` need to have the same length as other (sequential) inputs.
UpperCAmelCase_ : List[Any] = len(encoded_inputs["""global_attention_mask"""] ) != len(lowerCamelCase_ )
if needs_to_be_padded:
UpperCAmelCase_ : Dict = len(lowerCamelCase_ ) - len(encoded_inputs["""global_attention_mask"""] )
if self.padding_side == "right":
# Use `-1` since `0` in `global_attention_mask` means `local attention` instead of `not to attend`
UpperCAmelCase_ : str = (
encoded_inputs["""global_attention_mask"""] + [-1] * difference
)
elif self.padding_side == "left":
UpperCAmelCase_ : List[str] = [-1] * difference + encoded_inputs[
"""global_attention_mask"""
]
else:
raise ValueError("""Invalid padding strategy:""" + str(self.padding_side ) )
return encoded_inputs
| 345 | 1 |
import unittest
from transformers import is_flax_available
from transformers.testing_utils import require_flax, require_sentencepiece, require_tokenizers, require_torch, slow
if is_flax_available():
import optax
from flax.training.common_utils import onehot
from transformers import AutoTokenizer, FlaxMTaForConditionalGeneration
from transformers.models.ta.modeling_flax_ta import shift_tokens_right
@require_torch
@require_sentencepiece
@require_tokenizers
@require_flax
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
@slow
def A__ ( self: Optional[Any] ) -> Tuple:
UpperCAmelCase_ : Tuple = FlaxMTaForConditionalGeneration.from_pretrained("""google/mt5-small""" )
UpperCAmelCase_ : Union[str, Any] = AutoTokenizer.from_pretrained("""google/mt5-small""" )
UpperCAmelCase_ : Union[str, Any] = tokenizer("""Hello there""" ,return_tensors="""np""" ).input_ids
UpperCAmelCase_ : Union[str, Any] = tokenizer("""Hi I am""" ,return_tensors="""np""" ).input_ids
UpperCAmelCase_ : str = shift_tokens_right(lowerCamelCase_ ,model.config.pad_token_id ,model.config.decoder_start_token_id )
UpperCAmelCase_ : Dict = model(lowerCamelCase_ ,decoder_input_ids=lowerCamelCase_ ).logits
UpperCAmelCase_ : int = optax.softmax_cross_entropy(lowerCamelCase_ ,onehot(lowerCamelCase_ ,logits.shape[-1] ) ).mean()
UpperCAmelCase_ : List[str] = -(labels.shape[-1] * loss.item())
UpperCAmelCase_ : Optional[Any] = -8_4.9_1_2_7
self.assertTrue(abs(mtf_score - EXPECTED_SCORE ) < 1e-4 )
| 345 |
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Union[str, Any] ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def A__ ( self: List[str] ) -> Dict:
UpperCAmelCase_ : Union[str, Any] = 1
UpperCAmelCase_ : Tuple = 3
UpperCAmelCase_ : Optional[Any] = (32, 32)
UpperCAmelCase_ : Optional[int] = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(lowerCamelCase_ )
return image
@property
def A__ ( self: List[Any] ) -> Optional[Any]:
torch.manual_seed(0 )
UpperCAmelCase_ : int = UNetaDConditionModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=4 ,out_channels=4 ,down_block_types=("""DownBlock2D""", """CrossAttnDownBlock2D""") ,up_block_types=("""CrossAttnUpBlock2D""", """UpBlock2D""") ,cross_attention_dim=32 ,)
return model
@property
def A__ ( self: str ) -> List[str]:
torch.manual_seed(0 )
UpperCAmelCase_ : Optional[int] = AutoencoderKL(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=["""DownEncoderBlock2D""", """DownEncoderBlock2D"""] ,up_block_types=["""UpDecoderBlock2D""", """UpDecoderBlock2D"""] ,latent_channels=4 ,)
return model
@property
def A__ ( self: Optional[int] ) -> int:
torch.manual_seed(0 )
UpperCAmelCase_ : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1e-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(lowerCamelCase_ )
@property
def A__ ( self: Tuple ) -> Tuple:
def extract(*lowerCamelCase_: Optional[Any] ,**lowerCamelCase_: str ):
class _snake_case :
'''simple docstring'''
def __init__( self: List[Any] ) -> Optional[Any]:
UpperCAmelCase_ : List[str] = torch.ones([0] )
def A__ ( self: List[Any] ,lowerCamelCase_: str ) -> int:
self.pixel_values.to(lowerCamelCase_ )
return self
return Out()
return extract
def A__ ( self: Union[str, Any] ) -> Tuple:
UpperCAmelCase_ : int = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : int = self.dummy_cond_unet
UpperCAmelCase_ : Optional[Any] = DDIMScheduler(
beta_start=0.0_0_0_8_5 ,beta_end=0.0_1_2 ,beta_schedule="""scaled_linear""" ,clip_sample=lowerCamelCase_ ,set_alpha_to_one=lowerCamelCase_ ,)
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : int = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : str = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : List[str] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : List[Any] = output.images
UpperCAmelCase_ : str = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Dict = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : int = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[Any] ) -> Any:
UpperCAmelCase_ : Tuple = """cpu""" # ensure determinism for the device-dependent torch.Generator
UpperCAmelCase_ : Dict = self.dummy_cond_unet
UpperCAmelCase_ : List[Any] = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : str = self.dummy_vae
UpperCAmelCase_ : Union[str, Any] = self.dummy_text_encoder
UpperCAmelCase_ : str = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : int = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : Optional[Any] = sd_pipe([prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" )
UpperCAmelCase_ : str = output.images
UpperCAmelCase_ : Union[str, Any] = torch.Generator(device=lowerCamelCase_ ).manual_seed(0 )
UpperCAmelCase_ : int = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=6.0 ,num_inference_steps=2 ,output_type="""np""" ,return_dict=lowerCamelCase_ ,)[0]
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
UpperCAmelCase_ : Tuple = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: str ) -> Dict:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(
"""hf-internal-testing/tiny-stable-diffusion-lms-pipe""" ,safety_checker=lowerCamelCase_ )
assert isinstance(lowerCamelCase_ ,lowerCamelCase_ )
assert isinstance(pipe.scheduler ,lowerCamelCase_ )
assert pipe.safety_checker is None
UpperCAmelCase_ : List[Any] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCamelCase_ )
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained(lowerCamelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
UpperCAmelCase_ : Optional[int] = pipe("""example prompt""" ,num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != """cuda""" ,"""This test requires a GPU""" )
def A__ ( self: List[str] ) -> Any:
UpperCAmelCase_ : Tuple = self.dummy_cond_unet
UpperCAmelCase_ : Dict = PNDMScheduler(skip_prk_steps=lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = self.dummy_vae
UpperCAmelCase_ : List[str] = self.dummy_text_encoder
UpperCAmelCase_ : Union[str, Any] = CLIPTokenizer.from_pretrained("""hf-internal-testing/tiny-random-clip""" )
# put models in fp16
UpperCAmelCase_ : Optional[Any] = unet.half()
UpperCAmelCase_ : Optional[int] = vae.half()
UpperCAmelCase_ : int = bert.half()
# make sure here that pndm scheduler skips prk
UpperCAmelCase_ : Any = StableDiffusionPipeline(
unet=lowerCamelCase_ ,scheduler=lowerCamelCase_ ,vae=lowerCamelCase_ ,text_encoder=lowerCamelCase_ ,tokenizer=lowerCamelCase_ ,safety_checker=lowerCamelCase_ ,feature_extractor=self.dummy_extractor ,)
UpperCAmelCase_ : List[Any] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Tuple = """A painting of a squirrel eating a burger"""
UpperCAmelCase_ : Optional[int] = sd_pipe([prompt] ,num_inference_steps=2 ,output_type="""np""" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class _snake_case ( unittest.TestCase ):
'''simple docstring'''
def A__ ( self: Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def A__ ( self: List[str] ) -> List[Any]:
UpperCAmelCase_ : Tuple = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : str = (
"""portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"""
""" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"""
""" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"""
""" children from bahnhof zoo, detailed """
)
UpperCAmelCase_ : Optional[int] = 4003660346
UpperCAmelCase_ : int = 7
# without safety guidance (sld_guidance_scale = 0)
UpperCAmelCase_ : Dict = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Optional[int] = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Dict = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
# without safety guidance (strong configuration)
UpperCAmelCase_ : Union[str, Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Any = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Tuple = output.images
UpperCAmelCase_ : Union[str, Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : str = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Optional[int] ) -> Any:
UpperCAmelCase_ : Any = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" ,safety_checker=lowerCamelCase_ )
UpperCAmelCase_ : Any = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
UpperCAmelCase_ : str = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = """padme amidala taking a bath artwork, safe for work, no nudity"""
UpperCAmelCase_ : List[Any] = 2734971755
UpperCAmelCase_ : Optional[Any] = 7
UpperCAmelCase_ : int = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[int] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : Tuple = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
UpperCAmelCase_ : Any = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Tuple = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : Dict = output.images
UpperCAmelCase_ : List[Any] = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Tuple = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def A__ ( self: Union[str, Any] ) -> int:
UpperCAmelCase_ : List[Any] = StableDiffusionPipeline.from_pretrained("""runwayml/stable-diffusion-v1-5""" )
UpperCAmelCase_ : List[str] = sd_pipe.to(lowerCamelCase_ )
sd_pipe.set_progress_bar_config(disable=lowerCamelCase_ )
UpperCAmelCase_ : Any = (
"""the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."""
""" leyendecker"""
)
UpperCAmelCase_ : Optional[Any] = 1044355234
UpperCAmelCase_ : List[str] = 12
UpperCAmelCase_ : List[Any] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : List[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=0 ,)
UpperCAmelCase_ : Any = output.images
UpperCAmelCase_ : Dict = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Optional[Any] = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-7
UpperCAmelCase_ : Optional[int] = torch.manual_seed(lowerCamelCase_ )
UpperCAmelCase_ : Optional[Any] = sd_pipe(
[prompt] ,generator=lowerCamelCase_ ,guidance_scale=lowerCamelCase_ ,num_inference_steps=50 ,output_type="""np""" ,width=512 ,height=512 ,sld_guidance_scale=2000 ,sld_warmup_steps=7 ,sld_threshold=0.0_2_5 ,sld_momentum_scale=0.5 ,sld_mom_beta=0.7 ,)
UpperCAmelCase_ : List[str] = output.images
UpperCAmelCase_ : Any = image[0, -3:, -3:, -1]
UpperCAmelCase_ : Any = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 345 | 1 |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.