code
stringlengths
81
54k
code_codestyle
int64
0
721
style_context
stringlengths
91
41.9k
style_context_codestyle
int64
0
699
label
int64
0
1
"""simple docstring""" from pathlib import PurePosixPath from typing import Optional import fsspec from fsspec import AbstractFileSystem from huggingface_hub.hf_api import DatasetInfo from ..utils.file_utils import get_authentication_headers_for_url from ..utils.hub import hf_hub_url class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): __UpperCAmelCase : Any = '' __UpperCAmelCase : List[str] = 'hf-legacy' # "hf://"" is reserved for hffs def __init__( self , _a = None , _a = None , **_a , ): super().__init__(self , **lowercase_ ) __a = repo_info __a = token __a = None def __UpperCAmelCase ( self ): if self.dir_cache is None: __a = {} for hf_file in self.repo_info.siblings: # TODO(QL): add sizes __a = { '''name''': hf_file.rfilename, '''size''': None, '''type''': '''file''', } self.dir_cache.update( { str(lowercase_ ): {'''name''': str(lowercase_ ), '''size''': None, '''type''': '''directory'''} for d in list(PurePosixPath(hf_file.rfilename ).parents )[:-1] } ) def __UpperCAmelCase ( self , _a , _a = "rb" , **_a , ): if not isinstance(self.repo_info , lowercase_ ): raise NotImplementedError(f'''Open is only implemented for dataset repositories, but got {self.repo_info}''' ) __a = hf_hub_url(self.repo_info.id , lowercase_ , revision=self.repo_info.sha ) return fsspec.open( lowercase_ , mode=lowercase_ , headers=get_authentication_headers_for_url(lowercase_ , use_auth_token=self.token ) , client_kwargs={'''trust_env''': True} , ).open() def __UpperCAmelCase ( self , _a , **_a ): self._get_dirs() __a = self._strip_protocol(lowercase_ ) if path in self.dir_cache: return self.dir_cache[path] else: raise FileNotFoundError(lowercase_ ) def __UpperCAmelCase ( self , _a , _a=False , **_a ): self._get_dirs() __a = PurePosixPath(path.strip('''/''' ) ) __a = {} for p, f in self.dir_cache.items(): __a = PurePosixPath(p.strip('''/''' ) ) __a = p.parent if root == path: __a = f __a = list(paths.values() ) if detail: return out else: return sorted(f['''name'''] for f in out )
715
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_mask2former": [ "MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "Mask2FormerConfig", ], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["Mask2FormerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST", "Mask2FormerForUniversalSegmentation", "Mask2FormerModel", "Mask2FormerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_maskaformer import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskaFormerConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_maskaformer import MaskaFormerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_maskaformer import ( MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST, MaskaFormerForUniversalSegmentation, MaskaFormerModel, MaskaFormerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
65
0
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "tiiuae/falcon-40b": "https://huggingface.co/tiiuae/falcon-40b/resolve/main/config.json", "tiiuae/falcon-7b": "https://huggingface.co/tiiuae/falcon-7b/resolve/main/config.json", } class __lowerCAmelCase ( __lowerCAmelCase ): '''simple docstring''' __UpperCAmelCase : List[str] = '''falcon''' __UpperCAmelCase : str = ['''past_key_values'''] def __init__( self , _a=65_024 , _a=4_544 , _a=32 , _a=71 , _a=1E-5 , _a=0.02 , _a=True , _a=0.0 , _a=0.0 , _a=None , _a=False , _a=False , _a=True , _a=True , _a=False , _a=11 , _a=11 , **_a , ): __a = vocab_size # Backward compatibility with n_embed kwarg __a = kwargs.pop('''n_embed''' , lowerCAmelCase_ ) __a = hidden_size if n_embed is None else n_embed __a = num_hidden_layers __a = num_attention_heads __a = layer_norm_epsilon __a = initializer_range __a = use_cache __a = hidden_dropout __a = attention_dropout __a = bos_token_id __a = eos_token_id __a = num_attention_heads if num_kv_heads is None else num_kv_heads __a = alibi __a = new_decoder_architecture __a = multi_query # Ignored when new_decoder_architecture is True __a = parallel_attn __a = bias super().__init__(bos_token_id=lowerCAmelCase_ , eos_token_id=lowerCAmelCase_ , **lowerCAmelCase_ ) @property def __UpperCAmelCase ( self ): return self.hidden_size // self.num_attention_heads @property def __UpperCAmelCase ( self ): return not self.alibi
716
"""simple docstring""" from sklearn.metrics import fa_score import datasets lowercase_ = "\nThe F1 score is the harmonic mean of the precision and recall. It can be computed with the equation:\nF1 = 2 * (precision * recall) / (precision + recall)\n" lowercase_ = "\nArgs:\n predictions (`list` of `int`): Predicted labels.\n references (`list` of `int`): Ground truth labels.\n labels (`list` of `int`): The set of labels to include when `average` is not set to `'binary'`, and the order of the labels if `average` is `None`. Labels present in the data can be excluded, for example to calculate a multiclass average ignoring a majority negative class. Labels not present in the data will result in 0 components in a macro average. For multilabel targets, labels are column indices. By default, all labels in `predictions` and `references` are used in sorted order. Defaults to None.\n pos_label (`int`): The class to be considered the positive class, in the case where `average` is set to `binary`. Defaults to 1.\n average (`string`): This parameter is required for multiclass/multilabel targets. If set to `None`, the scores for each class are returned. Otherwise, this determines the type of averaging performed on the data. Defaults to `'binary'`.\n\n - 'binary': Only report results for the class specified by `pos_label`. This is applicable only if the classes found in `predictions` and `references` are binary.\n - 'micro': Calculate metrics globally by counting the total true positives, false negatives and false positives.\n - 'macro': Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.\n - 'weighted': Calculate metrics for each label, and find their average weighted by support (the number of true instances for each label). This alters `'macro'` to account for label imbalance. This option can result in an F-score that is not between precision and recall.\n - 'samples': Calculate metrics for each instance, and find their average (only meaningful for multilabel classification).\n sample_weight (`list` of `float`): Sample weights Defaults to None.\n\nReturns:\n f1 (`float` or `array` of `float`): F1 score or list of f1 scores, depending on the value passed to `average`. Minimum possible value is 0. Maximum possible value is 1. Higher f1 scores are better.\n\nExamples:\n\n Example 1-A simple binary example\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0])\n >>> print(results)\n {'f1': 0.5}\n\n Example 2-The same simple binary example as in Example 1, but with `pos_label` set to `0`.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], pos_label=0)\n >>> print(round(results['f1'], 2))\n 0.67\n\n Example 3-The same simple binary example as in Example 1, but with `sample_weight` included.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], sample_weight=[0.9, 0.5, 3.9, 1.2, 0.3])\n >>> print(round(results['f1'], 2))\n 0.35\n\n Example 4-A multiclass example, with different values for the `average` input.\n >>> predictions = [0, 2, 1, 0, 0, 1]\n >>> references = [0, 1, 2, 0, 1, 2]\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"macro\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"micro\")\n >>> print(round(results['f1'], 2))\n 0.33\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"weighted\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=None)\n >>> print(results)\n {'f1': array([0.8, 0. , 0. ])}\n" lowercase_ = "\n@article{scikit-learn,\n title={Scikit-learn: Machine Learning in {P}ython},\n author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.\n and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.\n and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and\n Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},\n journal={Journal of Machine Learning Research},\n volume={12},\n pages={2825--2830},\n year={2011}\n}\n" @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class __lowerCAmelCase ( datasets.Metric ): '''simple docstring''' def __UpperCAmelCase ( self ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { '''predictions''': datasets.Sequence(datasets.Value('''int32''' ) ), '''references''': datasets.Sequence(datasets.Value('''int32''' ) ), } if self.config_name == '''multilabel''' else { '''predictions''': datasets.Value('''int32''' ), '''references''': datasets.Value('''int32''' ), } ) , reference_urls=['''https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html'''] , ) def __UpperCAmelCase ( self , _a , _a , _a=None , _a=1 , _a="binary" , _a=None ): __a = fa_score( _a , _a , labels=_a , pos_label=_a , average=_a , sample_weight=_a ) return {"f1": float(_a ) if score.size == 1 else score}
65
0
"""simple docstring""" import argparse import hashlib # hashlib is only used inside the Test class import struct class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = data __a = [0X6745_2301, 0XEFCD_AB89, 0X98BA_DCFE, 0X1032_5476, 0XC3D2_E1F0] @staticmethod def __UpperCAmelCase ( _a , _a ): return ((n << b) | (n >> (32 - b))) & 0XFFFF_FFFF def __UpperCAmelCase ( self ): __a = b'''\x80''' + b'''\x00''' * (63 - (len(self.data ) + 8) % 64) __a = self.data + padding + struct.pack('''>Q''' , 8 * len(self.data ) ) return padded_data def __UpperCAmelCase ( self ): return [ self.padded_data[i : i + 64] for i in range(0 , len(self.padded_data ) , 64 ) ] def __UpperCAmelCase ( self , _a ): __a = list(struct.unpack('''>16L''' , UpperCAmelCase__ ) ) + [0] * 64 for i in range(16 , 80 ): __a = self.rotate((w[i - 3] ^ w[i - 8] ^ w[i - 14] ^ w[i - 16]) , 1 ) return w def __UpperCAmelCase ( self ): __a = self.padding() __a = self.split_blocks() for block in self.blocks: __a = self.expand_block(UpperCAmelCase__ ) __a = self.h for i in range(0 , 80 ): if 0 <= i < 20: __a = (b & c) | ((~b) & d) __a = 0X5A82_7999 elif 20 <= i < 40: __a = b ^ c ^ d __a = 0X6ED9_EBA1 elif 40 <= i < 60: __a = (b & c) | (b & d) | (c & d) __a = 0X8F1B_BCDC elif 60 <= i < 80: __a = b ^ c ^ d __a = 0XCA62_C1D6 __a = ( self.rotate(UpperCAmelCase__ , 5 ) + f + e + k + expanded_block[i] & 0XFFFF_FFFF, a, self.rotate(UpperCAmelCase__ , 30 ), c, d, ) __a = ( self.h[0] + a & 0XFFFF_FFFF, self.h[1] + b & 0XFFFF_FFFF, self.h[2] + c & 0XFFFF_FFFF, self.h[3] + d & 0XFFFF_FFFF, self.h[4] + e & 0XFFFF_FFFF, ) return ("{:08x}" * 5).format(*self.h ) def lowercase ( ) -> int: __a = b'''Test String''' assert SHAaHash(lowerCAmelCase__ ).final_hash() == hashlib.shaa(lowerCAmelCase__ ).hexdigest() # noqa: S324 def lowercase ( ) -> Optional[Any]: __a = argparse.ArgumentParser(description='''Process some strings or files''' ) parser.add_argument( '''--string''' , dest='''input_string''' , default='''Hello World!! Welcome to Cryptography''' , help='''Hash the string''' , ) parser.add_argument('''--file''' , dest='''input_file''' , help='''Hash contents of a file''' ) __a = parser.parse_args() __a = args.input_string # In any case hash input should be a bytestring if args.input_file: with open(args.input_file , '''rb''' ) as f: __a = f.read() else: __a = bytes(lowerCAmelCase__ , '''utf-8''' ) print(SHAaHash(lowerCAmelCase__ ).final_hash() ) if __name__ == "__main__": main() import doctest doctest.testmod()
717
"""simple docstring""" from ..utils import DummyObject, requires_backends class __lowerCAmelCase ( metaclass=__SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = ['onnx'] def __init__( self , *_a , **_a ): requires_backends(self , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] )
65
0
"""simple docstring""" from PIL import Image def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : int ) -> Image: __a = (259 * (level + 255)) / (255 * (259 - level)) def contrast(lowerCAmelCase__ : Union[str, Any] ) -> int: return int(128 + factor * (c - 128) ) return img.point(lowerCAmelCase__ ) if __name__ == "__main__": # Load image with Image.open("image_data/lena.jpg") as img: # Change contrast to 170 lowercase_ = change_contrast(img, 1_7_0) cont_img.save("image_data/lena_high_contrast.png", format="png")
718
"""simple docstring""" 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 lowercase_ = 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 __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=16 , _a=13 , _a=7 , _a=14 , _a=10 , _a=19 , _a=5 , _a=4 , _a=True , _a=16 , _a=2 , _a=4 , _a=4 , _a="gelu" , _a=0.1 , _a=0.1 , _a=[1, 2, 3, 4, 5] , _a=25 , _a=5 , ): __a = d_model __a = parent __a = batch_size __a = prediction_length __a = context_length __a = cardinality __a = num_time_features __a = lags_sequence __a = embedding_dimension __a = is_training __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = context_length __a = prediction_length + label_length __a = label_length __a = moving_average __a = autocorrelation_factor def __UpperCAmelCase ( self ): 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 __UpperCAmelCase ( self , _a ): __a = config.context_length + max(config.lags_sequence ) __a = ids_tensor([self.batch_size, 1] , config.cardinality[0] ) __a = floats_tensor([self.batch_size, _past_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, _past_length] ) __a = floats_tensor([self.batch_size, _past_length] ) > 0.5 # decoder inputs __a = floats_tensor([self.batch_size, config.prediction_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, config.prediction_length] ) __a = { '''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 __UpperCAmelCase ( self ): __a = self.get_config() __a = self.prepare_autoformer_inputs_dict(_a ) return config, inputs_dict def __UpperCAmelCase ( self ): __a , __a = self.prepare_config_and_inputs() return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = AutoformerModel(config=_a ).to(_a ).eval() __a = model(**_a ) __a = outputs.encoder_last_hidden_state __a = outputs.last_hidden_state with tempfile.TemporaryDirectory() as tmpdirname: __a = model.get_encoder() encoder.save_pretrained(_a ) __a = AutoformerEncoder.from_pretrained(_a ).to(_a ) __a , __a , __a , __a , __a = model.create_network_inputs(**_a ) __a , __a = model.decomposition_layer(transformer_inputs[:, : config.context_length, ...] ) __a = torch.cat( (transformer_inputs[:, : config.context_length, ...], feature[:, : config.context_length, ...]) , dim=-1 , ) __a = encoder(inputs_embeds=_a )[0] self.parent.assertTrue((encoder_last_hidden_state_a - encoder_last_hidden_state).abs().max().item() < 1E-3 ) __a = ( torch.mean(transformer_inputs[:, : config.context_length, ...] , dim=1 ) .unsqueeze(1 ) .repeat(1 , config.prediction_length , 1 ) ) __a = torch.zeros( [transformer_inputs.shape[0], config.prediction_length, transformer_inputs.shape[2]] , device=enc_input.device , ) __a = torch.cat( ( torch.cat((seasonal_input[:, -config.label_length :, ...], zeros) , dim=1 ), feature[:, config.context_length - config.label_length :, ...], ) , dim=-1 , ) __a = 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: __a = model.get_decoder() decoder.save_pretrained(_a ) __a = AutoformerDecoder.from_pretrained(_a ).to(_a ) __a = decoder( trend=_a , inputs_embeds=_a , encoder_hidden_states=_a , )[0] self.parent.assertTrue((last_hidden_state_a - last_hidden_state).abs().max().item() < 1E-3 ) @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Any = (AutoformerModel, AutoformerForPrediction) if is_torch_available() else () __UpperCAmelCase : int = (AutoformerForPrediction,) if is_torch_available() else () __UpperCAmelCase : Any = {'feature-extraction': AutoformerModel} if is_torch_available() else {} __UpperCAmelCase : int = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False __UpperCAmelCase : Any = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False def __UpperCAmelCase ( self ): __a = AutoformerModelTester(self ) __a = ConfigTester(self , config_class=_a , has_text_modality=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs() for model_class in self.all_model_classes: __a = model_class(_a ) with tempfile.TemporaryDirectory() as tmpdirname: model.save_pretrained(_a ) __a , __a = model_class.from_pretrained(_a , output_loading_info=_a ) self.assertEqual(info['''missing_keys'''] , [] ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.check_encoder_decoder_model_standalone(*_a ) @unittest.skip(reason='''Model has no tokens embeddings''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = inspect.signature(getattr(_a , '''forward''' ) ) # The main input is the name of the argument after `self` __a = list(model_signature.parameters.keys() )[1] self.assertEqual(AutoformerModel.main_input_name , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ) __a = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic __a = [*signature.parameters.keys()] __a = [ '''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(_a )] , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() __a = True __a = getattr(self.model_tester , '''seq_length''' , _a ) __a = getattr(self.model_tester , '''decoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''encoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''d_model''' , _a ) __a = getattr(self.model_tester , '''num_attention_heads''' , _a ) __a = d_model // num_attention_heads for model_class in self.all_model_classes: __a = True __a = False __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) # check that output_attentions also work using config del inputs_dict["output_attentions"] __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) self.assertListEqual( list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, encoder_seq_length, dim] , ) __a = len(_a ) __a = 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(_a , _a ) # decoder attentions __a = outputs.decoder_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = outputs.cross_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = True __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) self.assertEqual(out_len + 2 , len(_a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , 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 __UpperCAmelCase ( self ): super().test_retain_grad_hidden_states_attentions() def lowercase ( lowerCAmelCase__ : Optional[int]="train-batch.pt" ) -> List[str]: __a = hf_hub_download(repo_id='''hf-internal-testing/tourism-monthly-batch''' , filename=lowerCAmelCase__ , repo_type='''dataset''' ) __a = torch.load(lowerCAmelCase__ , map_location=lowerCAmelCase__ ) return batch @require_torch @slow class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = AutoformerModel.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch() with torch.no_grad(): __a = 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] __a = torch.Size( (64, model.config.prediction_length + model.config.label_length, model.config.feature_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[0.3593, -1.3398, 0.6330], [0.2279, 1.5396, -0.1792], [0.0450, 1.3225, -0.2335]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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 __a = torch.Size((64, model.config.context_length, model.config.d_model) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[-0.0734, -0.9036, 0.8358], [4.7186, 2.4113, 1.9581], [1.7953, 2.3558, 1.2970]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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'''] , ) __a = torch.Size((64, model.config.num_parallel_samples, model.config.prediction_length) ) self.assertEqual(outputs.sequences.shape , _a ) __a = torch.tensor([3130.6763, 4056.5293, 7053.0786] , device=_a ) __a = outputs.sequences.mean(dim=1 ) self.assertTrue(torch.allclose(mean_prediction[0, -3:] , _a , rtol=1E-1 ) )
65
0
"""simple docstring""" import os import time import pytest from datasets.utils.filelock import FileLock, Timeout def lowercase ( lowerCAmelCase__ : Optional[int] ) -> Dict: __a = FileLock(str(tmpdir / '''foo.lock''' ) ) __a = FileLock(str(tmpdir / '''foo.lock''' ) ) __a = 0.01 with locka.acquire(): with pytest.raises(__a ): __a = time.time() locka.acquire(__a ) assert time.time() - _start > timeout def lowercase ( lowerCAmelCase__ : Dict ) -> str: __a = '''a''' * 1000 + '''.lock''' __a = FileLock(str(tmpdir / filename ) ) assert locka._lock_file.endswith('''.lock''' ) assert not locka._lock_file.endswith(__a ) assert len(os.path.basename(locka._lock_file ) ) <= 255 __a = FileLock(tmpdir / filename ) with locka.acquire(): with pytest.raises(__a ): locka.acquire(0 )
719
"""simple docstring""" import inspect from typing import Optional, Union import numpy as np import PIL import torch from torch.nn import functional as F from torchvision import transforms from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DiffusionPipeline, DPMSolverMultistepScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, randn_tensor, ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> Any: if isinstance(lowerCAmelCase__ , torch.Tensor ): return image elif isinstance(lowerCAmelCase__ , PIL.Image.Image ): __a = [image] if isinstance(image[0] , PIL.Image.Image ): __a = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['''lanczos'''] ) )[None, :] for i in image] __a = np.concatenate(lowerCAmelCase__ , axis=0 ) __a = np.array(lowerCAmelCase__ ).astype(np.floataa ) / 2_55.0 __a = image.transpose(0 , 3 , 1 , 2 ) __a = 2.0 * image - 1.0 __a = torch.from_numpy(lowerCAmelCase__ ) elif isinstance(image[0] , torch.Tensor ): __a = torch.cat(lowerCAmelCase__ , dim=0 ) return image def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=0.99_95 ) -> int: if not isinstance(lowerCAmelCase__ , np.ndarray ): __a = True __a = va.device __a = va.cpu().numpy() __a = va.cpu().numpy() __a = np.sum(va * va / (np.linalg.norm(lowerCAmelCase__ ) * np.linalg.norm(lowerCAmelCase__ )) ) if np.abs(lowerCAmelCase__ ) > DOT_THRESHOLD: __a = (1 - t) * va + t * va else: __a = np.arccos(lowerCAmelCase__ ) __a = np.sin(lowerCAmelCase__ ) __a = theta_a * t __a = np.sin(lowerCAmelCase__ ) __a = np.sin(theta_a - theta_t ) / sin_theta_a __a = sin_theta_t / sin_theta_a __a = sa * va + sa * va if inputs_are_torch: __a = torch.from_numpy(lowerCAmelCase__ ).to(lowerCAmelCase__ ) return va def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> int: __a = F.normalize(lowerCAmelCase__ , dim=-1 ) __a = F.normalize(lowerCAmelCase__ , dim=-1 ) return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> List[str]: for param in model.parameters(): __a = value class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a , _a , _a , _a=None , _a=None , _a=None , ): super().__init__() self.register_modules( vae=_a , text_encoder=_a , clip_model=_a , tokenizer=_a , unet=_a , scheduler=_a , feature_extractor=_a , coca_model=_a , coca_tokenizer=_a , coca_transform=_a , ) __a = ( feature_extractor.size if isinstance(feature_extractor.size , _a ) else feature_extractor.size['''shortest_edge'''] ) __a = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std ) set_requires_grad(self.text_encoder , _a ) set_requires_grad(self.clip_model , _a ) def __UpperCAmelCase ( self , _a = "auto" ): if slice_size == "auto": # half the attention head size is usually a good trade-off between # speed and memory __a = self.unet.config.attention_head_dim // 2 self.unet.set_attention_slice(_a ) def __UpperCAmelCase ( self ): self.enable_attention_slicing(_a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self , _a , _a , _a ): # get the original timestep using init_timestep __a = min(int(num_inference_steps * strength ) , _a ) __a = max(num_inference_steps - init_timestep , 0 ) __a = self.scheduler.timesteps[t_start:] return timesteps, num_inference_steps - t_start def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a=None ): if not isinstance(_a , torch.Tensor ): raise ValueError(f'''`image` has to be of type `torch.Tensor` but is {type(_a )}''' ) __a = image.to(device=_a , dtype=_a ) if isinstance(_a , _a ): __a = [ self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_a ) ] __a = torch.cat(_a , dim=0 ) else: __a = self.vae.encode(_a ).latent_dist.sample(_a ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 0.1_8215 * init_latents __a = init_latents.repeat_interleave(_a , dim=0 ) __a = randn_tensor(init_latents.shape , generator=_a , device=_a , dtype=_a ) # get latents __a = self.scheduler.add_noise(_a , _a , _a ) __a = init_latents return latents def __UpperCAmelCase ( self , _a ): __a = self.coca_transform(_a ).unsqueeze(0 ) with torch.no_grad(), torch.cuda.amp.autocast(): __a = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) ) __a = self.coca_tokenizer.decode(generated[0].cpu().numpy() ) return generated.split('''<end_of_text>''' )[0].replace('''<start_of_text>''' , '''''' ).rstrip(''' .,''' ) def __UpperCAmelCase ( self , _a , _a ): __a = self.feature_extractor.preprocess(_a ) __a = torch.from_numpy(clip_image_input['''pixel_values'''][0] ).unsqueeze(0 ).to(self.device ).half() __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = image_embeddings_clip.repeat_interleave(_a , dim=0 ) return image_embeddings_clip @torch.enable_grad() def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , ): __a = latents.detach().requires_grad_() __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ): __a = self.scheduler.alphas_cumprod[timestep] __a = 1 - alpha_prod_t # compute predicted original sample from predicted noise also called # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf __a = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5 __a = torch.sqrt(_a ) __a = pred_original_sample * (fac) + latents * (1 - fac) elif isinstance(self.scheduler , _a ): __a = self.scheduler.sigmas[index] __a = latents - sigma * noise_pred else: raise ValueError(f'''scheduler type {type(self.scheduler )} not supported''' ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * sample __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = transforms.Resize(self.feature_extractor_size )(_a ) __a = self.normalize(_a ).to(latents.dtype ) __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = spherical_dist_loss(_a , _a ).mean() * clip_guidance_scale __a = -torch.autograd.grad(_a , _a )[0] if isinstance(self.scheduler , _a ): __a = latents.detach() + grads * (sigma**2) __a = noise_pred_original else: __a = noise_pred_original - torch.sqrt(_a ) * grads return noise_pred, latents @torch.no_grad() def __call__( self , _a , _a , _a = None , _a = None , _a = 512 , _a = 512 , _a = 0.6 , _a = 50 , _a = 7.5 , _a = 1 , _a = 0.0 , _a = 100 , _a = None , _a = "pil" , _a = True , _a = 0.8 , _a = 0.1 , _a = 0.1 , ): if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError(f'''You have passed {batch_size} batch_size, but only {len(_a )} generators.''' ) if height % 8 != 0 or width % 8 != 0: raise ValueError(f'''`height` and `width` have to be divisible by 8 but are {height} and {width}.''' ) if isinstance(_a , torch.Generator ) and batch_size > 1: __a = [generator] + [None] * (batch_size - 1) __a = [ ('''model''', self.coca_model is None), ('''tokenizer''', self.coca_tokenizer is None), ('''transform''', self.coca_transform is None), ] __a = [x[0] for x in coca_is_none if x[1]] __a = ''', '''.join(_a ) # generate prompts with coca model if prompt is None if content_prompt is None: if len(_a ): raise ValueError( f'''Content prompt is None and CoCa [{coca_is_none_str}] is None.''' f'''Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) if style_prompt is None: if len(_a ): raise ValueError( f'''Style prompt is None and CoCa [{coca_is_none_str}] is None.''' f''' Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) # get prompt text embeddings for content and style __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0] __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0] __a = slerp(_a , _a , _a ) # duplicate text embeddings for each generation per prompt __a = text_embeddings.repeat_interleave(_a , dim=0 ) # set timesteps __a = '''offset''' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() ) __a = {} if accepts_offset: __a = 1 self.scheduler.set_timesteps(_a , **_a ) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand self.scheduler.timesteps.to(self.device ) __a , __a = self.get_timesteps(_a , _a , self.device ) __a = timesteps[:1].repeat(_a ) # Preprocess image __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = slerp(_a , _a , _a ) if clip_guidance_scale > 0: __a = self.get_clip_image_embeddings(_a , _a ) __a = self.get_clip_image_embeddings(_a , _a ) __a = slerp( _a , _a , _a ) # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` # corresponds to doing no classifier free guidance. __a = guidance_scale > 1.0 # get unconditional embeddings for classifier free guidance if do_classifier_free_guidance: __a = content_text_input.input_ids.shape[-1] __a = self.tokenizer([''''''] , padding='''max_length''' , max_length=_a , return_tensors='''pt''' ) __a = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0] # duplicate unconditional embeddings for each generation per prompt __a = uncond_embeddings.repeat_interleave(_a , dim=0 ) # For classifier free guidance, we need to do two forward passes. # Here we concatenate the unconditional and text embeddings into a single batch # to avoid doing two forward passes __a = torch.cat([uncond_embeddings, text_embeddings] ) # get the initial random noise unless the user supplied it # Unlike in other pipelines, latents need to be generated in the target device # for 1-to-1 results reproducibility with the CompVis implementation. # However this currently doesn't work in `mps`. __a = (batch_size, self.unet.config.in_channels, height // 8, width // 8) __a = text_embeddings.dtype if latents is None: if self.device.type == "mps": # randn does not work reproducibly on mps __a = torch.randn(_a , generator=_a , device='''cpu''' , dtype=_a ).to( self.device ) else: __a = torch.randn(_a , generator=_a , device=self.device , dtype=_a ) else: if latents.shape != latents_shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {latents_shape}''' ) __a = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler __a = latents * self.scheduler.init_noise_sigma # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 # and should be between [0, 1] __a = '''eta''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) __a = {} if accepts_eta: __a = eta # check if the scheduler accepts generator __a = '''generator''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) if accepts_generator: __a = generator with self.progress_bar(total=_a ): for i, t in enumerate(_a ): # expand the latents if we are doing classifier free guidance __a = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample # perform classifier free guidance if do_classifier_free_guidance: __a , __a = noise_pred.chunk(2 ) __a = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) # perform clip guidance if clip_guidance_scale > 0: __a = ( text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings ) __a , __a = self.cond_fn( _a , _a , _a , _a , _a , _a , _a , ) # compute the previous noisy sample x_t -> x_t-1 __a = self.scheduler.step(_a , _a , _a , **_a ).prev_sample # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * latents __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image, None) return StableDiffusionPipelineOutput(images=_a , nsfw_content_detected=_a )
65
0
"""simple docstring""" import argparse import torch from transformers import ( WavaVecaConfig, WavaVecaFeatureExtractor, WavaVecaForAudioFrameClassification, WavaVecaForSequenceClassification, WavaVecaForXVector, logging, ) logging.set_verbosity_info() lowercase_ = logging.get_logger(__name__) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : List[str] ) -> Optional[int]: __a = WavaVecaForSequenceClassification.from_pretrained(lowerCAmelCase__ , config=lowerCAmelCase__ ) __a = downstream_dict["projector.weight"] __a = downstream_dict["projector.bias"] __a = downstream_dict["model.post_net.linear.weight"] __a = downstream_dict["model.post_net.linear.bias"] return model def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple ) -> Optional[Any]: __a = WavaVecaForAudioFrameClassification.from_pretrained(lowerCAmelCase__ , config=lowerCAmelCase__ ) __a = downstream_dict["model.linear.weight"] __a = downstream_dict["model.linear.bias"] return model def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = WavaVecaForXVector.from_pretrained(lowerCAmelCase__ , config=lowerCAmelCase__ ) __a = downstream_dict["connector.weight"] __a = downstream_dict["connector.bias"] for i, kernel_size in enumerate(hf_config.tdnn_kernel ): __a = downstream_dict[ f'''model.framelevel_feature_extractor.module.{i}.kernel.weight''' ] __a = downstream_dict[f'''model.framelevel_feature_extractor.module.{i}.kernel.bias'''] __a = downstream_dict["model.utterancelevel_feature_extractor.linear1.weight"] __a = downstream_dict["model.utterancelevel_feature_extractor.linear1.bias"] __a = downstream_dict["model.utterancelevel_feature_extractor.linear2.weight"] __a = downstream_dict["model.utterancelevel_feature_extractor.linear2.bias"] __a = downstream_dict["objective.W"] return model @torch.no_grad() def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : int ) -> Optional[Any]: __a = torch.load(lowerCAmelCase__ , map_location='''cpu''' ) __a = checkpoint["Downstream"] __a = WavaVecaConfig.from_pretrained(lowerCAmelCase__ ) __a = WavaVecaFeatureExtractor.from_pretrained( lowerCAmelCase__ , return_attention_mask=lowerCAmelCase__ , do_normalize=lowerCAmelCase__ ) __a = hf_config.architectures[0] if arch.endswith('''ForSequenceClassification''' ): __a = convert_classification(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) elif arch.endswith('''ForAudioFrameClassification''' ): __a = convert_diarization(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) elif arch.endswith('''ForXVector''' ): __a = convert_xvector(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) else: raise NotImplementedError(f'''S3PRL weights conversion is not supported for {arch}''' ) if hf_config.use_weighted_layer_sum: __a = checkpoint["Featurizer"]["weights"] hf_feature_extractor.save_pretrained(lowerCAmelCase__ ) hf_model.save_pretrained(lowerCAmelCase__ ) if __name__ == "__main__": lowercase_ = argparse.ArgumentParser() parser.add_argument( "--base_model_name", default=None, type=str, help="Name of the huggingface pretrained base model." ) parser.add_argument("--config_path", default=None, type=str, help="Path to the huggingface classifier config.") parser.add_argument("--checkpoint_path", default=None, type=str, help="Path to the s3prl checkpoint.") parser.add_argument("--model_dump_path", default=None, type=str, help="Path to the final converted model.") lowercase_ = parser.parse_args() convert_saprl_checkpoint(args.base_model_name, args.config_path, args.checkpoint_path, args.model_dump_path)
720
"""simple docstring""" # tests directory-specific settings - this file is run automatically # by pytest before any tests are run import sys import warnings from os.path import abspath, dirname, join # allow having multiple repository checkouts and not needing to remember to rerun # 'pip install -e .[dev]' when switching between checkouts and running tests. lowercase_ = abspath(join(dirname(dirname(dirname(__file__))), "src")) sys.path.insert(1, git_repo_path) # silence FutureWarning warnings in tests since often we can't act on them until # they become normal warnings - i.e. the tests still need to test the current functionality warnings.simplefilter(action="ignore", category=FutureWarning) def lowercase ( lowerCAmelCase__ : List[Any] ) -> str: from transformers.testing_utils import pytest_addoption_shared pytest_addoption_shared(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Union[str, Any]: from transformers.testing_utils import pytest_terminal_summary_main __a = terminalreporter.config.getoption('''--make-reports''' ) if make_reports: pytest_terminal_summary_main(lowerCAmelCase__ , id=lowerCAmelCase__ )
65
0
"""simple docstring""" import os import sys lowercase_ = os.path.join(os.path.dirname(__file__), "src") sys.path.append(SRC_DIR) from transformers import ( AutoConfig, AutoModel, AutoModelForCausalLM, AutoModelForMaskedLM, AutoModelForQuestionAnswering, AutoModelForSequenceClassification, AutoTokenizer, add_start_docstrings, ) lowercase_ = [ "torch", "numpy", "tokenizers", "filelock", "requests", "tqdm", "regex", "sentencepiece", "sacremoses", "importlib_metadata", "huggingface_hub", ] @add_start_docstrings(AutoConfig.__doc__ ) def lowercase ( *lowerCAmelCase__ : Optional[Any] , **lowerCAmelCase__ : Optional[Any] ) -> Any: return AutoConfig.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ ) @add_start_docstrings(AutoTokenizer.__doc__ ) def lowercase ( *lowerCAmelCase__ : int , **lowerCAmelCase__ : List[Any] ) -> List[str]: return AutoTokenizer.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ ) @add_start_docstrings(AutoModel.__doc__ ) def lowercase ( *lowerCAmelCase__ : List[Any] , **lowerCAmelCase__ : Optional[int] ) -> List[str]: return AutoModel.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ ) @add_start_docstrings(AutoModelForCausalLM.__doc__ ) def lowercase ( *lowerCAmelCase__ : int , **lowerCAmelCase__ : str ) -> Dict: return AutoModelForCausalLM.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ ) @add_start_docstrings(AutoModelForMaskedLM.__doc__ ) def lowercase ( *lowerCAmelCase__ : Tuple , **lowerCAmelCase__ : Dict ) -> str: return AutoModelForMaskedLM.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ ) @add_start_docstrings(AutoModelForSequenceClassification.__doc__ ) def lowercase ( *lowerCAmelCase__ : str , **lowerCAmelCase__ : List[Any] ) -> int: return AutoModelForSequenceClassification.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ ) @add_start_docstrings(AutoModelForQuestionAnswering.__doc__ ) def lowercase ( *lowerCAmelCase__ : List[Any] , **lowerCAmelCase__ : List[Any] ) -> Optional[int]: return AutoModelForQuestionAnswering.from_pretrained(*lowerCAmelCase__ , **lowerCAmelCase__ )
721
"""simple docstring""" 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. lowercase_ = {"LayoutLMv2Config", "LayoutLMv3Config"} @is_pipeline_test class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING __UpperCAmelCase : Any = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING if model_mapping is not None: __UpperCAmelCase : Any = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP} if tf_model_mapping is not None: __UpperCAmelCase : Dict = { config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP } def __UpperCAmelCase ( self , _a , _a , _a ): __a = ZeroShotClassificationPipeline( model=_a , tokenizer=_a , candidate_labels=['''polics''', '''health'''] ) return classifier, ["Who are you voting for in 2020?", "My stomach hurts."] def __UpperCAmelCase ( self , _a , _a ): __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # No kwarg __a = classifier('''Who are you voting for in 2020?''' , ['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics, public health''' ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health'''] ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''This text is about {}''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # https://github.com/huggingface/transformers/issues/13846 __a = classifier(['''I am happy'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(1 ) ] , ) __a = classifier(['''I am happy''', '''I am sad'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(2 ) ] , ) with self.assertRaises(_a ): classifier('''''' , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier(_a , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels='''''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels=_a ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''Not formatting template''' , ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template=_a , ) self.run_entailment_id(_a ) def __UpperCAmelCase ( self , _a ): __a = zero_shot_classifier.model.config __a = config.labelaid __a = zero_shot_classifier.entailment_id __a = {'''LABEL_0''': 0, '''LABEL_1''': 1, '''LABEL_2''': 2} self.assertEqual(zero_shot_classifier.entailment_id , -1 ) __a = {'''entailment''': 0, '''neutral''': 1, '''contradiction''': 2} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 0, '''NON-ENTAIL''': 1} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 2, '''NEUTRAL''': 1, '''CONTR''': 0} self.assertEqual(zero_shot_classifier.entailment_id , 2 ) __a = original_labelaid self.assertEqual(_a , zero_shot_classifier.entailment_id ) @require_torch def __UpperCAmelCase ( self ): __a = 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 __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''pt''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @require_tf def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''tf''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @slow @require_torch def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''pt''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , ) @slow @require_tf def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''tf''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , )
65
0
"""simple docstring""" import json from typing import TYPE_CHECKING, List, Optional, Tuple from tokenizers import pre_tokenizers from ...tokenization_utils_base import BatchEncoding from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging if TYPE_CHECKING: from transformers.pipelines.conversational import Conversation lowercase_ = logging.get_logger(__name__) lowercase_ = {"tokenizer_file": "tokenizer.json"} lowercase_ = { "tokenizer_file": { "bigscience/tokenizer": "https://huggingface.co/bigscience/tokenizer/blob/main/tokenizer.json", "bigscience/bloom-560m": "https://huggingface.co/bigscience/bloom-560m/blob/main/tokenizer.json", "bigscience/bloom-1b1": "https://huggingface.co/bigscience/bloom-1b1/blob/main/tokenizer.json", "bigscience/bloom-1b7": "https://huggingface.co/bigscience/bloom-1b7/blob/main/tokenizer.json", "bigscience/bloom-3b": "https://huggingface.co/bigscience/bloom-3b/blob/main/tokenizer.json", "bigscience/bloom-7b1": "https://huggingface.co/bigscience/bloom-7b1/blob/main/tokenizer.json", "bigscience/bloom": "https://huggingface.co/bigscience/bloom/blob/main/tokenizer.json", }, } class __lowerCAmelCase ( __UpperCAmelCase ): '''simple docstring''' __UpperCAmelCase : Any = VOCAB_FILES_NAMES __UpperCAmelCase : Optional[int] = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : int = ["input_ids", "attention_mask"] __UpperCAmelCase : List[Any] = None def __init__( self , _a=None , _a=None , _a=None , _a="<unk>" , _a="<s>" , _a="</s>" , _a="<pad>" , _a=False , _a=False , **_a , ): super().__init__( _lowerCamelCase , _lowerCamelCase , tokenizer_file=_lowerCamelCase , unk_token=_lowerCamelCase , bos_token=_lowerCamelCase , eos_token=_lowerCamelCase , pad_token=_lowerCamelCase , add_prefix_space=_lowerCamelCase , clean_up_tokenization_spaces=_lowerCamelCase , **_lowerCamelCase , ) __a = json.loads(self.backend_tokenizer.pre_tokenizer.__getstate__() ) if pre_tok_state.get('''add_prefix_space''' , _lowerCamelCase ) != add_prefix_space: __a = getattr(_lowerCamelCase , pre_tok_state.pop('''type''' ) ) __a = add_prefix_space __a = pre_tok_class(**_lowerCamelCase ) __a = add_prefix_space def __UpperCAmelCase ( self , *_a , **_a ): __a = kwargs.get('''is_split_into_words''' , _lowerCamelCase ) if not (self.add_prefix_space or not is_split_into_words): raise Exception( f'''You need to instantiate {self.__class__.__name__} with add_prefix_space=True to use it with''' ''' pretokenized inputs.''' ) return super()._batch_encode_plus(*_lowerCamelCase , **_lowerCamelCase ) def __UpperCAmelCase ( self , *_a , **_a ): __a = kwargs.get('''is_split_into_words''' , _lowerCamelCase ) if not (self.add_prefix_space or not is_split_into_words): raise Exception( f'''You need to instantiate {self.__class__.__name__} with add_prefix_space=True to use it with''' ''' pretokenized inputs.''' ) return super()._encode_plus(*_lowerCamelCase , **_lowerCamelCase ) def __UpperCAmelCase ( self , _a , _a = None ): __a = self._tokenizer.model.save(_lowerCamelCase , name=_lowerCamelCase ) return tuple(_lowerCamelCase ) def __UpperCAmelCase ( self , _a ): __a = [] for is_user, text in conversation.iter_texts(): input_ids.extend(self.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase ) + [self.eos_token_id] ) if len(_lowerCamelCase ) > self.model_max_length: __a = input_ids[-self.model_max_length :] return input_ids
700
"""simple docstring""" import pytest from datasets import inspect_metric, list_metrics, load_metric @pytest.fixture def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' , set() ) @pytest.fixture def lowercase ( lowerCAmelCase__ : Any ) -> Any: class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = metric_id class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Any = [MetricMock(__SCREAMING_SNAKE_CASE ) for metric_id in ['accuracy', 'mse', 'precision', 'codeparrot/apps_metric']] def __UpperCAmelCase ( self ): return self._metrics monkeypatch.setattr('''datasets.inspect.huggingface_hub''' , HfhMock() ) @pytest.mark.parametrize( '''func, args''' , [(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Tuple ) -> Optional[int]: if "tmp_path" in args: __a = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args ) with pytest.warns(lowerCAmelCase__ , match='''https://huggingface.co/docs/evaluate''' ): func(*lowerCAmelCase__ )
65
0
"""simple docstring""" import itertools import random import unittest import numpy as np from transformers import BatchFeature, SpeechTaFeatureExtractor from transformers.testing_utils import require_torch from transformers.utils.import_utils import is_torch_available from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin if is_torch_available(): import torch lowercase_ = random.Random() def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any]=1.0 , lowerCAmelCase__ : List[Any]=None , lowerCAmelCase__ : Union[str, Any]=None ) -> List[Any]: if rng is None: __a = global_rng __a = [] for batch_idx in range(shape[0] ): values.append([] ) for _ in range(shape[1] ): values[-1].append(rng.random() * scale ) return values @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __init__( self , _a , _a=7 , _a=400 , _a=2_000 , _a=1 , _a=0.0 , _a=16_000 , _a=True , _a=80 , _a=16 , _a=64 , _a="hann_window" , _a=80 , _a=7_600 , _a=1E-10 , _a=True , ): __a = parent __a = batch_size __a = min_seq_length __a = max_seq_length __a = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1) __a = feature_size __a = padding_value __a = sampling_rate __a = do_normalize __a = num_mel_bins __a = hop_length __a = win_length __a = win_function __a = fmin __a = fmax __a = mel_floor __a = return_attention_mask def __UpperCAmelCase ( self ): return { "feature_size": self.feature_size, "padding_value": self.padding_value, "sampling_rate": self.sampling_rate, "do_normalize": self.do_normalize, "num_mel_bins": self.num_mel_bins, "hop_length": self.hop_length, "win_length": self.win_length, "win_function": self.win_function, "fmin": self.fmin, "fmax": self.fmax, "mel_floor": self.mel_floor, "return_attention_mask": self.return_attention_mask, } def __UpperCAmelCase ( self , _a=False , _a=False ): def _flatten(_a ): return list(itertools.chain(*__lowerCamelCase ) ) if equal_length: __a = floats_list((self.batch_size, self.max_seq_length) ) else: # make sure that inputs increase in size __a = [ _flatten(floats_list((x, self.feature_size) ) ) for x in range(self.min_seq_length , self.max_seq_length , self.seq_length_diff ) ] if numpify: __a = [np.asarray(__lowerCamelCase ) for x in speech_inputs] return speech_inputs def __UpperCAmelCase ( self , _a=False , _a=False ): if equal_length: __a = [floats_list((self.max_seq_length, self.num_mel_bins) ) for _ in range(self.batch_size )] else: # make sure that inputs increase in size __a = [ floats_list((x, self.num_mel_bins) ) for x in range(self.min_seq_length , self.max_seq_length , self.seq_length_diff ) ] if numpify: __a = [np.asarray(__lowerCamelCase ) for x in speech_inputs] return speech_inputs @require_torch class __lowerCAmelCase ( lowerCamelCase__ , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Tuple = SpeechTaFeatureExtractor def __UpperCAmelCase ( self ): __a = SpeechTaFeatureExtractionTester(self ) def __UpperCAmelCase ( self , _a ): self.assertTrue(np.all(np.mean(__lowerCamelCase , axis=0 ) < 1E-3 ) ) self.assertTrue(np.all(np.abs(np.var(__lowerCamelCase , axis=0 ) - 1 ) < 1E-3 ) ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) # create three inputs of length 800, 1000, and 1200 __a = [floats_list((1, x) )[0] for x in range(800 , 1_400 , 200 )] __a = [np.asarray(__lowerCamelCase ) for speech_input in speech_inputs] # Test not batched input __a = feat_extract(speech_inputs[0] , return_tensors='''np''' ).input_values __a = feat_extract(np_speech_inputs[0] , return_tensors='''np''' ).input_values self.assertTrue(np.allclose(__lowerCamelCase , __lowerCamelCase , atol=1E-3 ) ) # Test batched __a = feat_extract(__lowerCamelCase , return_tensors='''np''' ).input_values __a = feat_extract(__lowerCamelCase , return_tensors='''np''' ).input_values for enc_seq_a, enc_seq_a in zip(__lowerCamelCase , __lowerCamelCase ): self.assertTrue(np.allclose(__lowerCamelCase , __lowerCamelCase , atol=1E-3 ) ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) __a = [floats_list((1, x) )[0] for x in range(800 , 1_400 , 200 )] __a = ['''longest''', '''max_length''', '''do_not_pad'''] __a = [None, 1_600, None] for max_length, padding in zip(__lowerCamelCase , __lowerCamelCase ): __a = feat_extract(__lowerCamelCase , padding=__lowerCamelCase , max_length=__lowerCamelCase , return_tensors='''np''' ) __a = processed.input_values self._check_zero_mean_unit_variance(input_values[0][:800] ) self.assertTrue(input_values[0][800:].sum() < 1E-6 ) self._check_zero_mean_unit_variance(input_values[1][:1_000] ) self.assertTrue(input_values[0][1_000:].sum() < 1E-6 ) self._check_zero_mean_unit_variance(input_values[2][:1_200] ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) __a = range(800 , 1_400 , 200 ) __a = [floats_list((1, x) )[0] for x in lengths] __a = ['''longest''', '''max_length''', '''do_not_pad'''] __a = [None, 1_600, None] for max_length, padding in zip(__lowerCamelCase , __lowerCamelCase ): __a = feat_extract(__lowerCamelCase , max_length=__lowerCamelCase , padding=__lowerCamelCase ) __a = processed.input_values self._check_zero_mean_unit_variance(input_values[0][:800] ) self._check_zero_mean_unit_variance(input_values[1][:1_000] ) self._check_zero_mean_unit_variance(input_values[2][:1_200] ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) __a = [floats_list((1, x) )[0] for x in range(800 , 1_400 , 200 )] __a = feat_extract( __lowerCamelCase , truncation=__lowerCamelCase , max_length=1_000 , padding='''max_length''' , return_tensors='''np''' ) __a = processed.input_values self._check_zero_mean_unit_variance(input_values[0, :800] ) self._check_zero_mean_unit_variance(input_values[1] ) self._check_zero_mean_unit_variance(input_values[2] ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) __a = [floats_list((1, x) )[0] for x in range(800 , 1_400 , 200 )] __a = feat_extract( __lowerCamelCase , truncation=__lowerCamelCase , max_length=1_000 , padding='''longest''' , return_tensors='''np''' ) __a = processed.input_values self._check_zero_mean_unit_variance(input_values[0, :800] ) self._check_zero_mean_unit_variance(input_values[1, :1_000] ) self._check_zero_mean_unit_variance(input_values[2] ) # make sure that if max_length < longest -> then pad to max_length self.assertTrue(input_values.shape == (3, 1_000) ) __a = [floats_list((1, x) )[0] for x in range(800 , 1_400 , 200 )] __a = feat_extract( __lowerCamelCase , truncation=__lowerCamelCase , max_length=2_000 , padding='''longest''' , return_tensors='''np''' ) __a = processed.input_values self._check_zero_mean_unit_variance(input_values[0, :800] ) self._check_zero_mean_unit_variance(input_values[1, :1_000] ) self._check_zero_mean_unit_variance(input_values[2] ) # make sure that if max_length > longest -> then pad to longest self.assertTrue(input_values.shape == (3, 1_200) ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) __a = np.random.rand(100 ).astype(np.floataa ) __a = np_speech_inputs.tolist() for inputs in [py_speech_inputs, np_speech_inputs]: __a = feature_extractor.pad([{'''input_values''': inputs}] , return_tensors='''np''' ) self.assertTrue(np_processed.input_values.dtype == np.floataa ) __a = feature_extractor.pad([{'''input_values''': inputs}] , return_tensors='''pt''' ) self.assertTrue(pt_processed.input_values.dtype == torch.floataa ) def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() ) # create three inputs of length 800, 1000, and 1200 __a = [floats_list((1, x) )[0] for x in range(800 , 1_400 , 200 )] __a = [np.asarray(__lowerCamelCase ) for speech_input in speech_inputs] # Test feature size __a = feature_extractor(audio_target=__lowerCamelCase , padding=__lowerCamelCase , return_tensors='''np''' ).input_values self.assertTrue(input_values.ndim == 3 ) self.assertTrue(input_values.shape[-1] == feature_extractor.num_mel_bins ) # Test not batched input __a = feature_extractor(speech_inputs[0] , return_tensors='''np''' ).input_values __a = feature_extractor(np_speech_inputs[0] , return_tensors='''np''' ).input_values self.assertTrue(np.allclose(__lowerCamelCase , __lowerCamelCase , atol=1E-3 ) ) # Test batched __a = feature_extractor(__lowerCamelCase , return_tensors='''np''' ).input_values __a = feature_extractor(__lowerCamelCase , return_tensors='''np''' ).input_values for enc_seq_a, enc_seq_a in zip(__lowerCamelCase , __lowerCamelCase ): self.assertTrue(np.allclose(__lowerCamelCase , __lowerCamelCase , atol=1E-3 ) ) # Test 2-D numpy arrays are batched. __a = [floats_list((1, x) )[0] for x in (800, 800, 800)] __a = np.asarray(__lowerCamelCase ) __a = feature_extractor(__lowerCamelCase , return_tensors='''np''' ).input_values __a = feature_extractor(__lowerCamelCase , return_tensors='''np''' ).input_values for enc_seq_a, enc_seq_a in zip(__lowerCamelCase , __lowerCamelCase ): self.assertTrue(np.allclose(__lowerCamelCase , __lowerCamelCase , atol=1E-3 ) ) def __UpperCAmelCase ( self ): __a = self.feat_extract_tester.prepare_inputs_for_target() __a = self.feature_extraction_class(**self.feat_extract_dict ) __a = feat_extract.model_input_names[0] __a = BatchFeature({input_name: speech_inputs} ) self.assertTrue(all(len(__lowerCamelCase ) == len(__lowerCamelCase ) for x, y in zip(__lowerCamelCase , processed_features[input_name] ) ) ) __a = self.feat_extract_tester.prepare_inputs_for_target(equal_length=__lowerCamelCase ) __a = BatchFeature({input_name: speech_inputs} , tensor_type='''np''' ) __a = processed_features[input_name] if len(batch_features_input.shape ) < 3: __a = batch_features_input[:, :, None] self.assertTrue( batch_features_input.shape == (self.feat_extract_tester.batch_size, len(speech_inputs[0] ), self.feat_extract_tester.num_mel_bins) ) @require_torch def __UpperCAmelCase ( self ): __a = self.feat_extract_tester.prepare_inputs_for_target(equal_length=__lowerCamelCase ) __a = self.feature_extraction_class(**self.feat_extract_dict ) __a = feat_extract.model_input_names[0] __a = BatchFeature({input_name: speech_inputs} , tensor_type='''pt''' ) __a = processed_features[input_name] if len(batch_features_input.shape ) < 3: __a = batch_features_input[:, :, None] self.assertTrue( batch_features_input.shape == (self.feat_extract_tester.batch_size, len(speech_inputs[0] ), self.feat_extract_tester.num_mel_bins) ) @require_torch def __UpperCAmelCase ( self ): __a = self.feature_extraction_class(**self.feat_extract_dict ) __a = self.feat_extract_tester.prepare_inputs_for_target() __a = feat_extract.model_input_names[0] __a = BatchFeature({input_name: speech_inputs} ) __a = feat_extract.num_mel_bins # hack! __a = feat_extract.pad(__lowerCamelCase , padding='''longest''' , return_tensors='''np''' )[input_name] __a = feat_extract.pad(__lowerCamelCase , padding='''longest''' , return_tensors='''pt''' )[input_name] self.assertTrue(abs(input_np.astype(np.floataa ).sum() - input_pt.numpy().astype(np.floataa ).sum() ) < 1E-2 ) def __UpperCAmelCase ( self ): __a = self.feat_extract_dict __a = True __a = self.feature_extraction_class(**__lowerCamelCase ) __a = self.feat_extract_tester.prepare_inputs_for_target() __a = [len(__lowerCamelCase ) for x in speech_inputs] __a = feat_extract.model_input_names[0] __a = BatchFeature({input_name: speech_inputs} ) __a = feat_extract.num_mel_bins # hack! __a = feat_extract.pad(__lowerCamelCase , padding='''longest''' , return_tensors='''np''' ) self.assertIn('''attention_mask''' , __lowerCamelCase ) self.assertListEqual(list(processed.attention_mask.shape ) , list(processed[input_name].shape[:2] ) ) self.assertListEqual(processed.attention_mask.sum(-1 ).tolist() , __lowerCamelCase ) def __UpperCAmelCase ( self ): __a = self.feat_extract_dict __a = True __a = self.feature_extraction_class(**__lowerCamelCase ) __a = self.feat_extract_tester.prepare_inputs_for_target() __a = [len(__lowerCamelCase ) for x in speech_inputs] __a = feat_extract.model_input_names[0] __a = BatchFeature({input_name: speech_inputs} ) __a = min(__lowerCamelCase ) __a = feat_extract.num_mel_bins # hack! __a = feat_extract.pad( __lowerCamelCase , padding='''max_length''' , max_length=__lowerCamelCase , truncation=__lowerCamelCase , return_tensors='''np''' ) self.assertIn('''attention_mask''' , __lowerCamelCase ) self.assertListEqual( list(processed_pad.attention_mask.shape ) , [processed_pad[input_name].shape[0], max_length] ) self.assertListEqual( processed_pad.attention_mask[:, :max_length].sum(-1 ).tolist() , [max_length for x in speech_inputs] ) def __UpperCAmelCase ( self , _a ): from datasets import load_dataset __a = load_dataset('''hf-internal-testing/librispeech_asr_dummy''' , '''clean''' , split='''validation''' ) # automatic decoding with librispeech __a = ds.sort('''id''' ).select(range(__lowerCamelCase ) )[:num_samples]['''audio'''] return [x["array"] for x in speech_samples] def __UpperCAmelCase ( self ): __a = torch.tensor( [2.3_804E-03, 2.0_752E-03, 1.9_836E-03, 2.1_057E-03, 1.6_174E-03, 3.0_518E-04, 9.1_553E-05, 3.3_569E-04, 9.7_656E-04, 1.8_311E-03, 2.0_142E-03, 2.1_057E-03, 1.7_395E-03, 4.5_776E-04, -3.9_673E-04, 4.5_776E-04, 1.0_071E-03, 9.1_553E-05, 4.8_828E-04, 1.1_597E-03, 7.3_242E-04, 9.4_604E-04, 1.8_005E-03, 1.8_311E-03, 8.8_501E-04, 4.2_725E-04, 4.8_828E-04, 7.3_242E-04, 1.0_986E-03, 2.1_057E-03] ) # fmt: on __a = self._load_datasamples(1 ) __a = SpeechTaFeatureExtractor() __a = feature_extractor(__lowerCamelCase , return_tensors='''pt''' ).input_values self.assertEquals(input_values.shape , (1, 93_680) ) self.assertTrue(torch.allclose(input_values[0, :30] , __lowerCamelCase , atol=1E-6 ) ) def __UpperCAmelCase ( self ): __a = torch.tensor( [-2.6870, -3.0104, -3.1356, -3.5352, -3.0044, -3.0353, -3.4719, -3.6777, -3.1520, -2.9435, -2.6553, -2.8795, -2.9944, -2.5921, -3.0279, -3.0386, -3.0864, -3.1291, -3.2353, -2.7444, -2.6831, -2.7287, -3.1761, -3.1571, -3.2726, -3.0582, -3.1007, -3.4533, -3.4695, -3.0998] ) # fmt: on __a = self._load_datasamples(1 ) __a = SpeechTaFeatureExtractor() __a = feature_extractor(audio_target=__lowerCamelCase , return_tensors='''pt''' ).input_values self.assertEquals(input_values.shape , (1, 366, 80) ) self.assertTrue(torch.allclose(input_values[0, 0, :30] , __lowerCamelCase , atol=1E-4 ) )
701
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> int: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or number < 0: raise ValueError('''Input must be a non-negative integer''' ) __a = 0 while number: # This way we arrive at next set bit (next 1) instead of looping # through each bit and checking for 1s hence the # loop won't run 32 times it will only run the number of `1` times number &= number - 1 count += 1 return count if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import copy import re class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : List[Any] = 'hp' __UpperCAmelCase : Dict = {} __UpperCAmelCase : List[str] = None @classmethod def __UpperCAmelCase ( cls , _a , _a ): __a = prefix __a = defaults cls.build_naming_info() @staticmethod def __UpperCAmelCase ( _a , _a ): if len(_lowercase ) == 0: return "" __a = None if any(char.isdigit() for char in word ): raise Exception(f'''Parameters should not contain numbers: \'{word}\' contains a number''' ) if word in info["short_word"]: return info["short_word"][word] for prefix_len in range(1 , len(_lowercase ) + 1 ): __a = word[:prefix_len] if prefix in info["reverse_short_word"]: continue else: __a = prefix break if short_word is None: # Paranoid fallback def int_to_alphabetic(_a ): __a = '''''' while integer != 0: __a = chr(ord('''A''' ) + integer % 10 ) + s integer //= 10 return s __a = 0 while True: __a = word + '''#''' + int_to_alphabetic(_lowercase ) if sword in info["reverse_short_word"]: continue else: __a = sword break __a = short_word __a = word return short_word @staticmethod def __UpperCAmelCase ( _a , _a ): __a = param_name.split('''_''' ) __a = [TrialShortNamer.shortname_for_word(_lowercase , _lowercase ) for word in words] # We try to create a separatorless short name, but if there is a collision we have to fallback # to a separated short name __a = ['''''', '''_'''] for separator in separators: __a = separator.join(_lowercase ) if shortname not in info["reverse_short_param"]: __a = shortname __a = param_name return shortname return param_name @staticmethod def __UpperCAmelCase ( _a , _a ): __a = TrialShortNamer.shortname_for_key(_lowercase , _lowercase ) __a = short_name __a = param_name @classmethod def __UpperCAmelCase ( cls ): if cls.NAMING_INFO is not None: return __a = { '''short_word''': {}, '''reverse_short_word''': {}, '''short_param''': {}, '''reverse_short_param''': {}, } __a = list(cls.DEFAULTS.keys() ) for k in field_keys: cls.add_new_param_name(_lowercase , _lowercase ) __a = info @classmethod def __UpperCAmelCase ( cls , _a ): cls.build_naming_info() assert cls.PREFIX is not None __a = [copy.copy(cls.PREFIX )] for k, v in params.items(): if k not in cls.DEFAULTS: raise Exception(f'''You should provide a default value for the param name {k} with value {v}''' ) if v == cls.DEFAULTS[k]: # The default value is not added to the name continue __a = cls.NAMING_INFO['''short_param'''][k] if isinstance(_lowercase , _lowercase ): __a = 1 if v else 0 __a = '''''' if isinstance(_lowercase , (int, float) ) else '''-''' __a = f'''{key}{sep}{v}''' name.append(_lowercase ) return "_".join(_lowercase ) @classmethod def __UpperCAmelCase ( cls , _a ): __a = repr[len(cls.PREFIX ) + 1 :] if repr == "": __a = [] else: __a = repr.split('''_''' ) __a = {} for value in values: if "-" in value: __a , __a = value.split('''-''' ) else: __a = re.sub('''[0-9.]''' , '''''' , _lowercase ) __a = float(re.sub('''[^0-9.]''' , '''''' , _lowercase ) ) __a = cls.NAMING_INFO['''reverse_short_param'''][p_k] __a = p_v for k in cls.DEFAULTS: if k not in parameters: __a = cls.DEFAULTS[k] return parameters
702
"""simple docstring""" import pytest from datasets import Dataset, DatasetDict, Features, NamedSplit, Value from datasets.io.text import TextDatasetReader from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.split == split if split else "train" @pytest.mark.parametrize('''path_type''' , [str, list] ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict: if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = text_path elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [text_path] __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for split in splits: __a = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str: __a = tmp_path / '''cache''' # CSV file loses col_1 string dtype information: default now is "int64" instead of "string" __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict: if split: __a = {split: text_path} else: __a = '''train''' __a = {'''train''': text_path, '''test''': text_path} __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() )
65
0
import unittest from transformers import AlbertConfig, is_torch_available from transformers.models.auto import get_values from transformers.testing_utils import 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, AlbertForMaskedLM, AlbertForMultipleChoice, AlbertForPreTraining, AlbertForQuestionAnswering, AlbertForSequenceClassification, AlbertForTokenClassification, AlbertModel, ) from transformers.models.albert.modeling_albert import ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=13 , _a=7 , _a=True , _a=True , _a=True , _a=True , _a=99 , _a=16 , _a=36 , _a=6 , _a=6 , _a=6 , _a=37 , _a="gelu" , _a=0.1 , _a=0.1 , _a=512 , _a=16 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ): __a = parent __a = batch_size __a = seq_length __a = is_training __a = use_input_mask __a = use_token_type_ids __a = use_labels __a = vocab_size __a = embedding_size __a = hidden_size __a = num_hidden_layers __a = num_hidden_groups __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = type_sequence_label_size __a = initializer_range __a = num_labels __a = num_choices __a = scope def __UpperCAmelCase ( self ): __a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) __a = None if self.use_input_mask: __a = random_attention_mask([self.batch_size, self.seq_length] ) __a = None if self.use_token_type_ids: __a = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size ) __a = None __a = None __a = None if self.use_labels: __a = ids_tensor([self.batch_size] , self.type_sequence_label_size ) __a = ids_tensor([self.batch_size, self.seq_length] , self.num_labels ) __a = ids_tensor([self.batch_size] , self.num_choices ) __a = self.get_config() return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels def __UpperCAmelCase ( self ): return AlbertConfig( 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 , initializer_range=self.initializer_range , num_hidden_groups=self.num_hidden_groups , ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = AlbertModel(config=lowercase_ ) model.to(lowercase_ ) model.eval() __a = model(lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ ) __a = model(lowercase_ , token_type_ids=lowercase_ ) __a = model(lowercase_ ) 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = AlbertForPreTraining(config=lowercase_ ) model.to(lowercase_ ) model.eval() __a = model( lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ , labels=lowercase_ , sentence_order_label=lowercase_ , ) self.parent.assertEqual(result.prediction_logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) self.parent.assertEqual(result.sop_logits.shape , (self.batch_size, config.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = AlbertForMaskedLM(config=lowercase_ ) model.to(lowercase_ ) model.eval() __a = model(lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ , labels=lowercase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = AlbertForQuestionAnswering(config=lowercase_ ) model.to(lowercase_ ) model.eval() __a = model( lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ , start_positions=lowercase_ , end_positions=lowercase_ , ) 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = AlbertForSequenceClassification(lowercase_ ) model.to(lowercase_ ) model.eval() __a = model(lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ , labels=lowercase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = AlbertForTokenClassification(config=lowercase_ ) model.to(lowercase_ ) model.eval() __a = model(lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ , labels=lowercase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_choices __a = AlbertForMultipleChoice(config=lowercase_ ) model.to(lowercase_ ) model.eval() __a = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = model( lowercase_ , attention_mask=lowercase_ , token_type_ids=lowercase_ , labels=lowercase_ , ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) ) def __UpperCAmelCase ( self ): __a = self.prepare_config_and_inputs() ( __a ) = config_and_inputs __a = {"input_ids": input_ids, "token_type_ids": token_type_ids, "attention_mask": input_mask} return config, inputs_dict @require_torch class __lowerCAmelCase ( snake_case__ , snake_case__ , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Any = ( ( AlbertModel, AlbertForPreTraining, AlbertForMaskedLM, AlbertForMultipleChoice, AlbertForSequenceClassification, AlbertForTokenClassification, AlbertForQuestionAnswering, ) if is_torch_available() else () ) __UpperCAmelCase : Optional[int] = ( { 'feature-extraction': AlbertModel, 'fill-mask': AlbertForMaskedLM, 'question-answering': AlbertForQuestionAnswering, 'text-classification': AlbertForSequenceClassification, 'token-classification': AlbertForTokenClassification, 'zero-shot': AlbertForSequenceClassification, } if is_torch_available() else {} ) __UpperCAmelCase : Dict = True def __UpperCAmelCase ( self , _a , _a , _a=False ): __a = super()._prepare_for_class(lowercase_ , lowercase_ , return_labels=lowercase_ ) if return_labels: if model_class in get_values(lowercase_ ): __a = torch.zeros( (self.model_tester.batch_size, self.model_tester.seq_length) , dtype=torch.long , device=lowercase_ ) __a = torch.zeros( self.model_tester.batch_size , dtype=torch.long , device=lowercase_ ) return inputs_dict def __UpperCAmelCase ( self ): __a = AlbertModelTester(self ) __a = ConfigTester(self , config_class=lowercase_ , hidden_size=37 ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*lowercase_ ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_pretraining(*lowercase_ ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_masked_lm(*lowercase_ ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_multiple_choice(*lowercase_ ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_question_answering(*lowercase_ ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_sequence_classification(*lowercase_ ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() for type in ["absolute", "relative_key", "relative_key_query"]: __a = type self.model_tester.create_and_check_model(*lowercase_ ) @slow def __UpperCAmelCase ( self ): for model_name in ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: __a = AlbertModel.from_pretrained(lowercase_ ) self.assertIsNotNone(lowercase_ ) @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @slow def __UpperCAmelCase ( self ): __a = AlbertModel.from_pretrained('''albert-base-v2''' ) __a = torch.tensor([[0, 345, 232, 328, 740, 140, 1_695, 69, 6_078, 1_588, 2]] ) __a = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]] ) with torch.no_grad(): __a = model(lowercase_ , attention_mask=lowercase_ )[0] __a = torch.Size((1, 11, 768) ) self.assertEqual(output.shape , lowercase_ ) __a = torch.tensor( [[[-0.6513, 1.5035, -0.2766], [-0.6515, 1.5046, -0.2780], [-0.6512, 1.5049, -0.2784]]] ) self.assertTrue(torch.allclose(output[:, 1:4, 1:4] , lowercase_ , atol=1E-4 ) )
703
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> bool: return numa ^ numa < 0 if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import math from dataclasses import dataclass from typing import Optional, Tuple, Union import torch from ..configuration_utils import ConfigMixin, register_to_config from ..utils import BaseOutput, randn_tensor from .scheduling_utils import SchedulerMixin, SchedulerOutput @dataclass class __lowerCAmelCase ( _UpperCAmelCase ): '''simple docstring''' snake_case : torch.FloatTensor snake_case : torch.FloatTensor class __lowerCAmelCase ( _UpperCAmelCase , _UpperCAmelCase ): '''simple docstring''' snake_case : Any = 1 @register_to_config def __init__( self , _a = 2_000 , _a = 0.15 , _a = 0.01 , _a = 1_348.0 , _a = 1E-5 , _a = 1 , ): # standard deviation of the initial noise distribution __a = sigma_max # setable values __a = None self.set_sigmas(A_ , A_ , A_ , A_ ) def __UpperCAmelCase ( self , _a , _a = None ): return sample def __UpperCAmelCase ( self , _a , _a = None , _a = None ): __a = sampling_eps if sampling_eps is not None else self.config.sampling_eps __a = torch.linspace(1 , A_ , A_ , device=A_ ) def __UpperCAmelCase ( self , _a , _a = None , _a = None , _a = None ): __a = sigma_min if sigma_min is not None else self.config.sigma_min __a = sigma_max if sigma_max is not None else self.config.sigma_max __a = sampling_eps if sampling_eps is not None else self.config.sampling_eps if self.timesteps is None: self.set_timesteps(A_ , A_ ) __a = sigma_min * (sigma_max / sigma_min) ** (self.timesteps / sampling_eps) __a = torch.exp(torch.linspace(math.log(A_ ) , math.log(A_ ) , A_ ) ) __a = torch.tensor([sigma_min * (sigma_max / sigma_min) ** t for t in self.timesteps] ) def __UpperCAmelCase ( self , _a , _a ): return torch.where( timesteps == 0 , torch.zeros_like(t.to(timesteps.device ) ) , self.discrete_sigmas[timesteps - 1].to(timesteps.device ) , ) def __UpperCAmelCase ( self , _a , _a , _a , _a = None , _a = True , ): if self.timesteps is None: raise ValueError( '''`self.timesteps` is not set, you need to run \'set_timesteps\' after creating the scheduler''' ) __a = timestep * torch.ones( sample.shape[0] , device=sample.device ) # torch.repeat_interleave(timestep, sample.shape[0]) __a = (timestep * (len(self.timesteps ) - 1)).long() # mps requires indices to be in the same device, so we use cpu as is the default with cuda __a = timesteps.to(self.discrete_sigmas.device ) __a = self.discrete_sigmas[timesteps].to(sample.device ) __a = self.get_adjacent_sigma(A_ , A_ ).to(sample.device ) __a = torch.zeros_like(A_ ) __a = (sigma**2 - adjacent_sigma**2) ** 0.5 # equation 6 in the paper: the model_output modeled by the network is grad_x log pt(x) # also equation 47 shows the analog from SDE models to ancestral sampling methods __a = diffusion.flatten() while len(diffusion.shape ) < len(sample.shape ): __a = diffusion.unsqueeze(-1 ) __a = drift - diffusion**2 * model_output # equation 6: sample noise for the diffusion term of __a = randn_tensor( sample.shape , layout=sample.layout , generator=A_ , device=sample.device , dtype=sample.dtype ) __a = sample - drift # subtract because `dt` is a small negative timestep # TODO is the variable diffusion the correct scaling term for the noise? __a = prev_sample_mean + diffusion * noise # add impact of diffusion field g if not return_dict: return (prev_sample, prev_sample_mean) return SdeVeOutput(prev_sample=A_ , prev_sample_mean=A_ ) def __UpperCAmelCase ( self , _a , _a , _a = None , _a = True , ): if self.timesteps is None: raise ValueError( '''`self.timesteps` is not set, you need to run \'set_timesteps\' after creating the scheduler''' ) # For small batch sizes, the paper "suggest replacing norm(z) with sqrt(d), where d is the dim. of z" # sample noise for correction __a = randn_tensor(sample.shape , layout=sample.layout , generator=A_ ).to(sample.device ) # compute step size from the model_output, the noise, and the snr __a = torch.norm(model_output.reshape(model_output.shape[0] , -1 ) , dim=-1 ).mean() __a = torch.norm(noise.reshape(noise.shape[0] , -1 ) , dim=-1 ).mean() __a = (self.config.snr * noise_norm / grad_norm) ** 2 * 2 __a = step_size * torch.ones(sample.shape[0] ).to(sample.device ) # self.repeat_scalar(step_size, sample.shape[0]) # compute corrected sample: model_output term and noise term __a = step_size.flatten() while len(step_size.shape ) < len(sample.shape ): __a = step_size.unsqueeze(-1 ) __a = sample + step_size * model_output __a = prev_sample_mean + ((step_size * 2) ** 0.5) * noise if not return_dict: return (prev_sample,) return SchedulerOutput(prev_sample=A_ ) def __UpperCAmelCase ( self , _a , _a , _a , ): # Make sure sigmas and timesteps have the same device and dtype as original_samples __a = timesteps.to(original_samples.device ) __a = self.discrete_sigmas.to(original_samples.device )[timesteps] __a = ( noise * sigmas[:, None, None, None] if noise is not None else torch.randn_like(A_ ) * sigmas[:, None, None, None] ) __a = noise + original_samples return noisy_samples def __len__( self ): return self.config.num_train_timesteps
704
"""simple docstring""" import unittest from transformers import SPIECE_UNDERLINE from transformers.models.speechta import SpeechTaTokenizer from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow from transformers.tokenization_utils import AddedToken from ...test_tokenization_common import TokenizerTesterMixin lowercase_ = get_tests_dir("fixtures/test_sentencepiece_bpe_char.model") @require_sentencepiece @require_tokenizers class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : int = SpeechTaTokenizer __UpperCAmelCase : Tuple = False __UpperCAmelCase : Dict = True def __UpperCAmelCase ( self ): super().setUp() # We have a SentencePiece fixture for testing __a = SpeechTaTokenizer(_a ) __a = AddedToken('''<mask>''' , lstrip=_a , rstrip=_a ) __a = mask_token tokenizer.add_special_tokens({'''mask_token''': mask_token} ) tokenizer.add_tokens(['''<ctc_blank>'''] ) tokenizer.save_pretrained(self.tmpdirname ) def __UpperCAmelCase ( self , _a ): __a = '''this is a test''' __a = '''this is a test''' return input_text, output_text def __UpperCAmelCase ( self , _a , _a=False , _a=20 , _a=5 ): __a , __a = self.get_input_output_texts(_a ) __a = tokenizer.encode(_a , add_special_tokens=_a ) __a = tokenizer.decode(_a , clean_up_tokenization_spaces=_a ) return text, ids def __UpperCAmelCase ( self ): __a = '''<pad>''' __a = 1 self.assertEqual(self.get_tokenizer()._convert_token_to_id(_a ) , _a ) self.assertEqual(self.get_tokenizer()._convert_id_to_token(_a ) , _a ) def __UpperCAmelCase ( self ): __a = list(self.get_tokenizer().get_vocab().keys() ) self.assertEqual(vocab_keys[0] , '''<s>''' ) self.assertEqual(vocab_keys[1] , '''<pad>''' ) self.assertEqual(vocab_keys[-4] , '''œ''' ) self.assertEqual(vocab_keys[-2] , '''<mask>''' ) self.assertEqual(vocab_keys[-1] , '''<ctc_blank>''' ) self.assertEqual(len(_a ) , 81 ) def __UpperCAmelCase ( self ): self.assertEqual(self.get_tokenizer().vocab_size , 79 ) def __UpperCAmelCase ( self ): __a = self.get_tokenizers(do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) # We usually have added tokens from the start in tests because our vocab fixtures are # smaller than the original vocabs - let's not assert this # self.assertEqual(vocab_size, all_size) __a = ['''aaaaa bbbbbb''', '''cccccccccdddddddd'''] __a = tokenizer.add_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size + len(_a ) ) __a = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 4 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) __a = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''} __a = tokenizer.add_special_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size_a + len(_a ) ) __a = tokenizer.encode( '''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 6 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[0] , tokens[1] ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokens[-4] ) self.assertEqual(tokens[0] , tokenizer.eos_token_id ) self.assertEqual(tokens[-3] , tokenizer.pad_token_id ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tokenizer.tokenize('''This is a test''' ) # fmt: off self.assertListEqual(_a , [SPIECE_UNDERLINE, '''T''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''a''', SPIECE_UNDERLINE, '''t''', '''e''', '''s''', '''t'''] ) # fmt: on self.assertListEqual( tokenizer.convert_tokens_to_ids(_a ) , [4, 32, 11, 10, 12, 4, 10, 12, 4, 7, 4, 6, 5, 12, 6] , ) __a = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''92000''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) __a = tokenizer.convert_tokens_to_ids(_a ) # fmt: off self.assertListEqual(_a , [4, 30, 4, 20, 7, 12, 4, 25, 8, 13, 9, 4, 10, 9, 4, 3, 23, 4, 7, 9, 14, 4, 6, 11, 10, 12, 4, 10, 12, 4, 19, 7, 15, 12, 73, 26] ) # fmt: on __a = tokenizer.convert_ids_to_tokens(_a ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''<unk>''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) @slow def __UpperCAmelCase ( self ): # Use custom sequence because this tokenizer does not handle numbers. __a = [ '''Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides ''' '''general-purpose architectures (BERT, GPT, RoBERTa, XLM, DistilBert, XLNet...) for Natural ''' '''Language Understanding (NLU) and Natural Language Generation (NLG) with over thirty-two pretrained ''' '''models in one hundred plus languages and deep interoperability between Jax, PyTorch and TensorFlow.''', '''BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly ''' '''conditioning on both left and right context in all layers.''', '''The quick brown fox jumps over the lazy dog.''', ] # fmt: off __a = { '''input_ids''': [ [4, 32, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 64, 19, 8, 13, 18, 5, 13, 15, 22, 4, 28, 9, 8, 20, 9, 4, 7, 12, 4, 24, 22, 6, 8, 13, 17, 11, 39, 6, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 7, 9, 14, 4, 24, 22, 6, 8, 13, 17, 11, 39, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 39, 25, 5, 13, 6, 63, 4, 24, 13, 8, 27, 10, 14, 5, 12, 4, 21, 5, 9, 5, 13, 7, 15, 39, 24, 16, 13, 24, 8, 12, 5, 4, 7, 13, 17, 11, 10, 6, 5, 17, 6, 16, 13, 5, 12, 4, 64, 40, 47, 54, 32, 23, 4, 53, 49, 32, 23, 4, 54, 8, 40, 47, 54, 32, 7, 23, 4, 69, 52, 43, 23, 4, 51, 10, 12, 6, 10, 15, 40, 5, 13, 6, 23, 4, 69, 52, 48, 5, 6, 26, 26, 26, 63, 4, 19, 8, 13, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 61, 9, 14, 5, 13, 12, 6, 7, 9, 14, 10, 9, 21, 4, 64, 48, 52, 61, 63, 4, 7, 9, 14, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 53, 5, 9, 5, 13, 7, 6, 10, 8, 9, 4, 64, 48, 52, 53, 63, 4, 20, 10, 6, 11, 4, 8, 27, 5, 13, 4, 6, 11, 10, 13, 6, 22, 39, 6, 20, 8, 4, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 4, 18, 8, 14, 5, 15, 12, 4, 10, 9, 4, 8, 9, 5, 4, 11, 16, 9, 14, 13, 5, 14, 4, 24, 15, 16, 12, 4, 15, 7, 9, 21, 16, 7, 21, 5, 12, 4, 7, 9, 14, 4, 14, 5, 5, 24, 4, 10, 9, 6, 5, 13, 8, 24, 5, 13, 7, 25, 10, 15, 10, 6, 22, 4, 25, 5, 6, 20, 5, 5, 9, 4, 58, 7, 37, 23, 4, 49, 22, 32, 8, 13, 17, 11, 4, 7, 9, 14, 4, 32, 5, 9, 12, 8, 13, 55, 15, 8, 20, 26, 2], [4, 40, 47, 54, 32, 4, 10, 12, 4, 14, 5, 12, 10, 21, 9, 5, 14, 4, 6, 8, 4, 24, 13, 5, 39, 6, 13, 7, 10, 9, 4, 14, 5, 5, 24, 4, 25, 10, 14, 10, 13, 5, 17, 6, 10, 8, 9, 7, 15, 4, 13, 5, 24, 13, 5, 12, 5, 9, 6, 7, 6, 10, 8, 9, 12, 4, 19, 13, 8, 18, 4, 16, 9, 15, 7, 25, 5, 15, 5, 14, 4, 6, 5, 37, 6, 4, 25, 22, 4, 46, 8, 10, 9, 6, 15, 22, 4, 17, 8, 9, 14, 10, 6, 10, 8, 9, 10, 9, 21, 4, 8, 9, 4, 25, 8, 6, 11, 4, 15, 5, 19, 6, 4, 7, 9, 14, 4, 13, 10, 21, 11, 6, 4, 17, 8, 9, 6, 5, 37, 6, 4, 10, 9, 4, 7, 15, 15, 4, 15, 7, 22, 5, 13, 12, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [4, 32, 11, 5, 4, 45, 16, 10, 17, 28, 4, 25, 13, 8, 20, 9, 4, 19, 8, 37, 4, 46, 16, 18, 24, 12, 4, 8, 27, 5, 13, 4, 6, 11, 5, 4, 15, 7, 57, 22, 4, 14, 8, 21, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], ], '''attention_mask''': [ [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], ] } # fmt: on self.tokenizer_integration_test_util( expected_encoding=_a , model_name='''microsoft/speecht5_asr''' , revision='''c5ef64c71905caeccde0e4462ef3f9077224c524''' , sequences=_a , )
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> bool: return number & 1 == 0 if __name__ == "__main__": import doctest doctest.testmod()
705
"""simple docstring""" from typing import List, Optional, Union import numpy as np from ....audio_utils import mel_filter_bank, optimal_fft_length, spectrogram, window_function from ....feature_extraction_sequence_utils import SequenceFeatureExtractor from ....feature_extraction_utils import BatchFeature from ....file_utils import PaddingStrategy, TensorType from ....utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Any = ['input_features', 'attention_mask'] def __init__( self , _a=80 , _a=16_000 , _a=0.0 , _a=10 , _a=25 , _a="hamming_window" , _a=3_2768.0 , _a=0.97 , _a=1.0 , _a=True , _a=True , _a=False , **_a , ): super().__init__(feature_size=_a , sampling_rate=_a , padding_value=_a , **_a ) __a = feature_size __a = sampling_rate __a = padding_value __a = hop_length __a = win_length __a = frame_signal_scale __a = preemphasis_coeff __a = mel_floor __a = normalize_means __a = normalize_vars __a = win_function __a = return_attention_mask __a = win_length * sampling_rate // 1_000 __a = hop_length * sampling_rate // 1_000 __a = optimal_fft_length(self.sample_size ) __a = (self.n_fft // 2) + 1 def __UpperCAmelCase ( self , _a ): if self.win_function == "hamming_window": __a = window_function(window_length=self.sample_size , name=self.win_function , periodic=_a ) else: __a = window_function(window_length=self.sample_size , name=self.win_function ) __a = mel_filter_bank( num_frequency_bins=self.n_freqs , num_mel_filters=self.feature_size , min_frequency=0.0 , max_frequency=self.sampling_rate / 2.0 , sampling_rate=self.sampling_rate , ) __a = spectrogram( one_waveform * self.frame_signal_scale , window=_a , frame_length=self.sample_size , hop_length=self.sample_stride , fft_length=self.n_fft , center=_a , preemphasis=self.preemphasis_coeff , mel_filters=_a , mel_floor=self.mel_floor , log_mel='''log''' , ) return msfc_features.T def __UpperCAmelCase ( self , _a , _a , _a ): # make sure we normalize float32 arrays if self.normalize_means: __a = x[:input_length].mean(axis=0 ) __a = np.subtract(_a , _a ) if self.normalize_vars: __a = x[:input_length].std(axis=0 ) __a = np.divide(_a , _a ) if input_length < x.shape[0]: __a = padding_value # make sure array is in float32 __a = x.astype(np.floataa ) return x def __UpperCAmelCase ( self , _a , _a = None ): __a = attention_mask.sum(-1 ) if attention_mask is not None else [x.shape[0] for x in input_features] return [self._normalize_one(_a , _a , self.padding_value ) for x, n in zip(_a , _a )] def __call__( self , _a , _a = False , _a = None , _a = False , _a = None , _a = None , _a = None , _a = None , **_a , ): if sampling_rate is not None: if sampling_rate != self.sampling_rate: raise ValueError( f'''The model corresponding to this feature extractor: {self} was trained using a sampling rate of''' f''' {self.sampling_rate}. Please make sure that the provided `raw_speech` input was sampled with''' f''' {self.sampling_rate} and not {sampling_rate}.''' ) else: logger.warning( '''It is strongly recommended to pass the ``sampling_rate`` argument to this function. ''' '''Failing to do so can result in silent errors that might be hard to debug.''' ) __a = isinstance(_a , np.ndarray ) and len(raw_speech.shape ) > 1 if is_batched_numpy and len(raw_speech.shape ) > 2: raise ValueError(f'''Only mono-channel audio is supported for input to {self}''' ) __a = is_batched_numpy or ( isinstance(_a , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) )) ) if is_batched: __a = [np.asarray(_a , dtype=np.floataa ) for speech in raw_speech] elif not is_batched and not isinstance(_a , np.ndarray ): __a = np.asarray(_a , dtype=np.floataa ) elif isinstance(_a , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ): __a = raw_speech.astype(np.floataa ) # always return batch if not is_batched: __a = [raw_speech] # extract fbank features __a = [self._extract_mfsc_features(_a ) for one_waveform in raw_speech] # convert into correct format for padding __a = BatchFeature({'''input_features''': features} ) __a = self.pad( _a , padding=_a , max_length=_a , truncation=_a , pad_to_multiple_of=_a , return_attention_mask=_a , **_a , ) # make sure list is in array format __a = padded_inputs.get('''input_features''' ) if isinstance(input_features[0] , _a ): __a = [np.asarray(_a , dtype=np.floataa ) for feature in input_features] __a = padded_inputs.get('''attention_mask''' ) if attention_mask is not None: __a = [np.asarray(_a , dtype=np.intaa ) for array in attention_mask] if self.normalize_means or self.normalize_vars: __a = ( np.array(_a , dtype=np.intaa ) if self._get_padding_strategies(_a , max_length=_a ) is not PaddingStrategy.DO_NOT_PAD and padding else None ) __a = self.normalize( padded_inputs['''input_features'''] , attention_mask=_a ) if return_tensors is not None: __a = padded_inputs.convert_to_tensors(_a ) return padded_inputs
65
0
import unittest from transformers import MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING, is_vision_available, pipeline from transformers.testing_utils import ( is_pipeline_test, nested_simplify, require_tf, require_torch, require_vision, slow, ) from .test_pipelines_common import ANY if is_vision_available(): from PIL import Image else: class __lowerCAmelCase : '''simple docstring''' @staticmethod def __UpperCAmelCase ( *_a , **_a ): pass @is_pipeline_test @require_vision @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : List[Any] = MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING def __UpperCAmelCase ( self , _a , _a , _a ): __a = pipeline( '''zero-shot-object-detection''' , model='''hf-internal-testing/tiny-random-owlvit-object-detection''' ) __a = [ { '''image''': '''./tests/fixtures/tests_samples/COCO/000000039769.png''', '''candidate_labels''': ['''cat''', '''remote''', '''couch'''], } ] return object_detector, examples def __UpperCAmelCase ( self , _a , _a ): __a = object_detector(examples[0] , threshold=0.0 ) __a = len(__UpperCamelCase ) self.assertGreater(__UpperCamelCase , 0 ) self.assertEqual( __UpperCamelCase , [ { '''score''': ANY(__UpperCamelCase ), '''label''': ANY(__UpperCamelCase ), '''box''': {'''xmin''': ANY(__UpperCamelCase ), '''ymin''': ANY(__UpperCamelCase ), '''xmax''': ANY(__UpperCamelCase ), '''ymax''': ANY(__UpperCamelCase )}, } for i in range(__UpperCamelCase ) ] , ) @require_tf @unittest.skip('''Zero Shot Object Detection not implemented in TF''' ) def __UpperCAmelCase ( self ): pass @require_torch def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-object-detection''' , model='''hf-internal-testing/tiny-random-owlvit-object-detection''' ) __a = object_detector( '''./tests/fixtures/tests_samples/COCO/000000039769.png''' , candidate_labels=['''cat''', '''remote''', '''couch'''] , threshold=0.64 , ) self.assertEqual( nested_simplify(__UpperCamelCase , decimals=4 ) , [ {'''score''': 0.7235, '''label''': '''cat''', '''box''': {'''xmin''': 204, '''ymin''': 167, '''xmax''': 232, '''ymax''': 190}}, {'''score''': 0.7218, '''label''': '''remote''', '''box''': {'''xmin''': 204, '''ymin''': 167, '''xmax''': 232, '''ymax''': 190}}, {'''score''': 0.7184, '''label''': '''couch''', '''box''': {'''xmin''': 204, '''ymin''': 167, '''xmax''': 232, '''ymax''': 190}}, {'''score''': 0.6748, '''label''': '''remote''', '''box''': {'''xmin''': 571, '''ymin''': 83, '''xmax''': 598, '''ymax''': 103}}, {'''score''': 0.6656, '''label''': '''cat''', '''box''': {'''xmin''': 571, '''ymin''': 83, '''xmax''': 598, '''ymax''': 103}}, {'''score''': 0.6614, '''label''': '''couch''', '''box''': {'''xmin''': 571, '''ymin''': 83, '''xmax''': 598, '''ymax''': 103}}, {'''score''': 0.6456, '''label''': '''remote''', '''box''': {'''xmin''': 494, '''ymin''': 105, '''xmax''': 521, '''ymax''': 127}}, {'''score''': 0.642, '''label''': '''remote''', '''box''': {'''xmin''': 67, '''ymin''': 274, '''xmax''': 93, '''ymax''': 297}}, {'''score''': 0.6419, '''label''': '''cat''', '''box''': {'''xmin''': 494, '''ymin''': 105, '''xmax''': 521, '''ymax''': 127}}, ] , ) __a = object_detector( [ { '''image''': '''./tests/fixtures/tests_samples/COCO/000000039769.png''', '''candidate_labels''': ['''cat''', '''remote''', '''couch'''], } ] , threshold=0.64 , ) self.assertEqual( nested_simplify(__UpperCamelCase , decimals=4 ) , [ [ {'''score''': 0.7235, '''label''': '''cat''', '''box''': {'''xmin''': 204, '''ymin''': 167, '''xmax''': 232, '''ymax''': 190}}, {'''score''': 0.7218, '''label''': '''remote''', '''box''': {'''xmin''': 204, '''ymin''': 167, '''xmax''': 232, '''ymax''': 190}}, {'''score''': 0.7184, '''label''': '''couch''', '''box''': {'''xmin''': 204, '''ymin''': 167, '''xmax''': 232, '''ymax''': 190}}, {'''score''': 0.6748, '''label''': '''remote''', '''box''': {'''xmin''': 571, '''ymin''': 83, '''xmax''': 598, '''ymax''': 103}}, {'''score''': 0.6656, '''label''': '''cat''', '''box''': {'''xmin''': 571, '''ymin''': 83, '''xmax''': 598, '''ymax''': 103}}, {'''score''': 0.6614, '''label''': '''couch''', '''box''': {'''xmin''': 571, '''ymin''': 83, '''xmax''': 598, '''ymax''': 103}}, {'''score''': 0.6456, '''label''': '''remote''', '''box''': {'''xmin''': 494, '''ymin''': 105, '''xmax''': 521, '''ymax''': 127}}, {'''score''': 0.642, '''label''': '''remote''', '''box''': {'''xmin''': 67, '''ymin''': 274, '''xmax''': 93, '''ymax''': 297}}, {'''score''': 0.6419, '''label''': '''cat''', '''box''': {'''xmin''': 494, '''ymin''': 105, '''xmax''': 521, '''ymax''': 127}}, ] ] , ) @require_torch @slow def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-object-detection''' ) __a = object_detector( '''http://images.cocodataset.org/val2017/000000039769.jpg''' , candidate_labels=['''cat''', '''remote''', '''couch'''] , ) self.assertEqual( nested_simplify(__UpperCamelCase , decimals=4 ) , [ {'''score''': 0.2868, '''label''': '''cat''', '''box''': {'''xmin''': 324, '''ymin''': 20, '''xmax''': 640, '''ymax''': 373}}, {'''score''': 0.277, '''label''': '''remote''', '''box''': {'''xmin''': 40, '''ymin''': 72, '''xmax''': 177, '''ymax''': 115}}, {'''score''': 0.2537, '''label''': '''cat''', '''box''': {'''xmin''': 1, '''ymin''': 55, '''xmax''': 315, '''ymax''': 472}}, {'''score''': 0.1474, '''label''': '''remote''', '''box''': {'''xmin''': 335, '''ymin''': 74, '''xmax''': 371, '''ymax''': 187}}, {'''score''': 0.1208, '''label''': '''couch''', '''box''': {'''xmin''': 4, '''ymin''': 0, '''xmax''': 642, '''ymax''': 476}}, ] , ) __a = object_detector( [ { '''image''': '''http://images.cocodataset.org/val2017/000000039769.jpg''', '''candidate_labels''': ['''cat''', '''remote''', '''couch'''], }, { '''image''': '''http://images.cocodataset.org/val2017/000000039769.jpg''', '''candidate_labels''': ['''cat''', '''remote''', '''couch'''], }, ] , ) self.assertEqual( nested_simplify(__UpperCamelCase , decimals=4 ) , [ [ {'''score''': 0.2868, '''label''': '''cat''', '''box''': {'''xmin''': 324, '''ymin''': 20, '''xmax''': 640, '''ymax''': 373}}, {'''score''': 0.277, '''label''': '''remote''', '''box''': {'''xmin''': 40, '''ymin''': 72, '''xmax''': 177, '''ymax''': 115}}, {'''score''': 0.2537, '''label''': '''cat''', '''box''': {'''xmin''': 1, '''ymin''': 55, '''xmax''': 315, '''ymax''': 472}}, {'''score''': 0.1474, '''label''': '''remote''', '''box''': {'''xmin''': 335, '''ymin''': 74, '''xmax''': 371, '''ymax''': 187}}, {'''score''': 0.1208, '''label''': '''couch''', '''box''': {'''xmin''': 4, '''ymin''': 0, '''xmax''': 642, '''ymax''': 476}}, ], [ {'''score''': 0.2868, '''label''': '''cat''', '''box''': {'''xmin''': 324, '''ymin''': 20, '''xmax''': 640, '''ymax''': 373}}, {'''score''': 0.277, '''label''': '''remote''', '''box''': {'''xmin''': 40, '''ymin''': 72, '''xmax''': 177, '''ymax''': 115}}, {'''score''': 0.2537, '''label''': '''cat''', '''box''': {'''xmin''': 1, '''ymin''': 55, '''xmax''': 315, '''ymax''': 472}}, {'''score''': 0.1474, '''label''': '''remote''', '''box''': {'''xmin''': 335, '''ymin''': 74, '''xmax''': 371, '''ymax''': 187}}, {'''score''': 0.1208, '''label''': '''couch''', '''box''': {'''xmin''': 4, '''ymin''': 0, '''xmax''': 642, '''ymax''': 476}}, ], ] , ) @require_tf @unittest.skip('''Zero Shot Object Detection not implemented in TF''' ) def __UpperCAmelCase ( self ): pass @require_torch @slow def __UpperCAmelCase ( self ): __a = 0.2 __a = pipeline('''zero-shot-object-detection''' ) __a = object_detector( '''http://images.cocodataset.org/val2017/000000039769.jpg''' , candidate_labels=['''cat''', '''remote''', '''couch'''] , threshold=__UpperCamelCase , ) self.assertEqual( nested_simplify(__UpperCamelCase , decimals=4 ) , [ {'''score''': 0.2868, '''label''': '''cat''', '''box''': {'''xmin''': 324, '''ymin''': 20, '''xmax''': 640, '''ymax''': 373}}, {'''score''': 0.277, '''label''': '''remote''', '''box''': {'''xmin''': 40, '''ymin''': 72, '''xmax''': 177, '''ymax''': 115}}, {'''score''': 0.2537, '''label''': '''cat''', '''box''': {'''xmin''': 1, '''ymin''': 55, '''xmax''': 315, '''ymax''': 472}}, ] , ) @require_torch @slow def __UpperCAmelCase ( self ): __a = 2 __a = pipeline('''zero-shot-object-detection''' ) __a = object_detector( '''http://images.cocodataset.org/val2017/000000039769.jpg''' , candidate_labels=['''cat''', '''remote''', '''couch'''] , top_k=__UpperCamelCase , ) self.assertEqual( nested_simplify(__UpperCamelCase , decimals=4 ) , [ {'''score''': 0.2868, '''label''': '''cat''', '''box''': {'''xmin''': 324, '''ymin''': 20, '''xmax''': 640, '''ymax''': 373}}, {'''score''': 0.277, '''label''': '''remote''', '''box''': {'''xmin''': 40, '''ymin''': 72, '''xmax''': 177, '''ymax''': 115}}, ] , )
706
"""simple docstring""" import json import os import subprocess import unittest from ast import literal_eval import pytest from parameterized import parameterized, 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_model_parallelism.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, { 'framework': 'pytorch', 'script': 'run_glue.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, ] ) class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): 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=_a , ) assert hasattr(self , '''env''' ) def __UpperCAmelCase ( self , _a ): # configuration for running training on smdistributed Model Parallel __a = { '''enabled''': True, '''processes_per_host''': 8, } __a = { '''enabled''': True, '''parameters''': { '''microbatches''': 4, '''placement_strategy''': '''spread''', '''pipeline''': '''interleaved''', '''optimize''': '''speed''', '''partitions''': 4, '''ddp''': True, }, } __a = {'''smdistributed''': {'''modelparallel''': smp_options}, '''mpi''': mpi_options} __a = '''trainer''' if self.script == '''run_glue.py''' else '''smtrainer''' # 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}-{instance_count}-smp-{name_extension}''' , instance_count=_a , instance_type=self.instance_type , debugger_hook_config=_a , hyperparameters={ **self.env.hyperparameters, '''model_name_or_path''': self.model_name_or_path, '''max_steps''': 500, } , metric_definitions=self.env.metric_definitions , distribution=_a , py_version='''py36''' , ) def __UpperCAmelCase ( self , _a ): TrainingJobAnalytics(_a ).export_csv(f'''{self.env.test_path}/{job_name}_metrics.csv''' ) @parameterized.expand([(1,)] ) def __UpperCAmelCase ( self , _a ): # create estimator __a = self.create_estimator(_a ) # run training estimator.fit() # result dataframe __a = TrainingJobAnalytics(estimator.latest_training_job.name ).dataframe() # extract kpis __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_accuracy''']['''value'''] ) __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_loss''']['''value'''] ) # get train time from SageMaker job, this includes starting, preprocessing, stopping __a = ( Session().describe_training_job(estimator.latest_training_job.name ).get('''TrainingTimeInSeconds''' , 999_999 ) ) # 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} , _a )
65
0
"""simple docstring""" import argparse import re from pathlib import Path import requests import torch from PIL import Image from torchvision.transforms import CenterCrop, Compose, Normalize, Resize, ToTensor from transformers import ( EfficientFormerConfig, EfficientFormerForImageClassificationWithTeacher, EfficientFormerImageProcessor, ) from transformers.image_utils import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD, PILImageResampling def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any] ) -> Dict: __a = old_name if "patch_embed" in old_name: __a = old_name.split('''.''' ) if layer == "0": __a = old_name.replace('''0''' , '''convolution1''' ) elif layer == "1": __a = old_name.replace('''1''' , '''batchnorm_before''' ) elif layer == "3": __a = old_name.replace('''3''' , '''convolution2''' ) else: __a = old_name.replace('''4''' , '''batchnorm_after''' ) if "network" in old_name and re.search(r'''\d\.\d''' , lowerCAmelCase__ ): __a = r"""\b\d{2}\b""" if bool(re.search(lowerCAmelCase__ , lowerCAmelCase__ ) ): __a = re.search(r'''\d\.\d\d.''' , lowerCAmelCase__ ).group() else: __a = re.search(r'''\d\.\d.''' , lowerCAmelCase__ ).group() if int(match[0] ) < 6: __a = old_name.replace(lowerCAmelCase__ , '''''' ) __a = trimmed_name.replace('''network''' , match[0] + '''.meta4D_layers.blocks.''' + match[2:-1] ) __a = """intermediate_stages.""" + trimmed_name else: __a = old_name.replace(lowerCAmelCase__ , '''''' ) if int(match[2] ) < num_meta4D_last_stage: __a = trimmed_name.replace('''network''' , '''meta4D_layers.blocks.''' + match[2] ) else: __a = str(int(match[2] ) - num_meta4D_last_stage ) __a = trimmed_name.replace('''network''' , '''meta3D_layers.blocks.''' + layer_index ) if "norm1" in old_name: __a = trimmed_name.replace('''norm1''' , '''layernorm1''' ) elif "norm2" in old_name: __a = trimmed_name.replace('''norm2''' , '''layernorm2''' ) elif "fc1" in old_name: __a = trimmed_name.replace('''fc1''' , '''linear_in''' ) elif "fc2" in old_name: __a = trimmed_name.replace('''fc2''' , '''linear_out''' ) __a = """last_stage.""" + trimmed_name elif "network" in old_name and re.search(r'''.\d.''' , lowerCAmelCase__ ): __a = old_name.replace('''network''' , '''intermediate_stages''' ) if "fc" in new_name: __a = new_name.replace('''fc''' , '''convolution''' ) elif ("norm1" in new_name) and ("layernorm1" not in new_name): __a = new_name.replace('''norm1''' , '''batchnorm_before''' ) elif ("norm2" in new_name) and ("layernorm2" not in new_name): __a = new_name.replace('''norm2''' , '''batchnorm_after''' ) if "proj" in new_name: __a = new_name.replace('''proj''' , '''projection''' ) if "dist_head" in new_name: __a = new_name.replace('''dist_head''' , '''distillation_classifier''' ) elif "head" in new_name: __a = new_name.replace('''head''' , '''classifier''' ) elif "patch_embed" in new_name: __a = """efficientformer.""" + new_name elif new_name == "norm.weight" or new_name == "norm.bias": __a = new_name.replace('''norm''' , '''layernorm''' ) __a = """efficientformer.""" + new_name else: __a = """efficientformer.encoder.""" + new_name return new_name def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : str ) -> Any: for key in checkpoint.copy().keys(): __a = checkpoint.pop(lowerCAmelCase__ ) __a = val return checkpoint def lowercase ( ) -> str: __a = """http://images.cocodataset.org/val2017/000000039769.jpg""" __a = Image.open(requests.get(lowerCAmelCase__ , stream=lowerCAmelCase__ ).raw ) return image def lowercase ( lowerCAmelCase__ : Path , lowerCAmelCase__ : Path , lowerCAmelCase__ : Path , lowerCAmelCase__ : bool ) -> List[str]: __a = torch.load(lowerCAmelCase__ , map_location='''cpu''' )["""model"""] __a = EfficientFormerConfig.from_json_file(lowerCAmelCase__ ) __a = EfficientFormerForImageClassificationWithTeacher(lowerCAmelCase__ ) __a = """_""".join(checkpoint_path.split('''/''' )[-1].split('''.''' )[0].split('''_''' )[:-1] ) __a = config.depths[-1] - config.num_metaad_blocks + 1 __a = convert_torch_checkpoint(lowerCAmelCase__ , lowerCAmelCase__ ) model.load_state_dict(lowerCAmelCase__ ) model.eval() __a = { """bilinear""": PILImageResampling.BILINEAR, """bicubic""": PILImageResampling.BICUBIC, """nearest""": PILImageResampling.NEAREST, } # prepare image __a = prepare_img() __a = 256 __a = 224 __a = EfficientFormerImageProcessor( size={'''shortest_edge''': image_size} , crop_size={'''height''': crop_size, '''width''': crop_size} , resample=pillow_resamplings['''bicubic'''] , ) __a = processor(images=lowerCAmelCase__ , return_tensors='''pt''' ).pixel_values # original processing pipeline __a = Compose( [ Resize(lowerCAmelCase__ , interpolation=pillow_resamplings['''bicubic'''] ), CenterCrop(lowerCAmelCase__ ), ToTensor(), Normalize(lowerCAmelCase__ , lowerCAmelCase__ ), ] ) __a = image_transforms(lowerCAmelCase__ ).unsqueeze(0 ) assert torch.allclose(lowerCAmelCase__ , lowerCAmelCase__ ) __a = model(lowerCAmelCase__ ) __a = outputs.logits __a = (1, 1000) if "l1" in model_name: __a = torch.Tensor( [-0.13_12, 0.43_53, -1.04_99, -0.51_24, 0.41_83, -0.67_93, -1.37_77, -0.08_93, -0.73_58, -2.43_28] ) assert torch.allclose(logits[0, :10] , lowerCAmelCase__ , atol=1e-3 ) assert logits.shape == expected_shape elif "l3" in model_name: __a = torch.Tensor( [-1.31_50, -1.54_56, -1.25_56, -0.84_96, -0.71_27, -0.78_97, -0.97_28, -0.30_52, 0.37_51, -0.31_27] ) assert torch.allclose(logits[0, :10] , lowerCAmelCase__ , atol=1e-3 ) assert logits.shape == expected_shape elif "l7" in model_name: __a = torch.Tensor( [-1.02_83, -1.41_31, -0.56_44, -1.31_15, -0.57_85, -1.20_49, -0.75_28, 0.19_92, -0.38_22, -0.08_78] ) assert logits.shape == expected_shape else: raise ValueError( f'''Unknown model checkpoint: {checkpoint_path}. Supported version of efficientformer are l1, l3 and l7''' ) # Save Checkpoints Path(lowerCAmelCase__ ).mkdir(exist_ok=lowerCAmelCase__ ) model.save_pretrained(lowerCAmelCase__ ) print(f'''Checkpoint successfuly converted. Model saved at {pytorch_dump_path}''' ) processor.save_pretrained(lowerCAmelCase__ ) print(f'''Processor successfuly saved at {pytorch_dump_path}''' ) if push_to_hub: print('''Pushing model to the hub...''' ) model.push_to_hub( repo_id=f'''Bearnardd/{pytorch_dump_path}''' , commit_message='''Add model''' , use_temp_dir=lowerCAmelCase__ , ) processor.push_to_hub( repo_id=f'''Bearnardd/{pytorch_dump_path}''' , commit_message='''Add image processor''' , use_temp_dir=lowerCAmelCase__ , ) if __name__ == "__main__": lowercase_ = argparse.ArgumentParser() # Required parameters parser.add_argument( "--pytorch_model_path", default=None, type=str, required=True, help="Path to EfficientFormer pytorch checkpoint.", ) parser.add_argument( "--config_file", default=None, type=str, required=True, help="The json file for EfficientFormer model config.", ) parser.add_argument( "--pytorch_dump_path", default=None, type=str, required=True, help="Path to the output PyTorch model." ) parser.add_argument("--push_to_hub", action="store_true", help="Push model and image processor to the hub") parser.add_argument( "--no-push_to_hub", dest="push_to_hub", action="store_false", help="Do not push model and image processor to the hub", ) parser.set_defaults(push_to_hub=True) lowercase_ = parser.parse_args() convert_efficientformer_checkpoint( checkpoint_path=args.pytorch_model_path, efficientformer_config_file=args.config_file, pytorch_dump_path=args.pytorch_dump_path, push_to_hub=args.push_to_hub, )
707
"""simple docstring""" import warnings from ..trainer import Trainer from ..utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a=None , **_a ): warnings.warn( '''`SageMakerTrainer` is deprecated and will be removed in v5 of Transformers. You can use `Trainer` ''' '''instead.''' , _a , ) super().__init__(args=_a , **_a )
65
0
"""simple docstring""" from abc import ABC, abstractmethod from argparse import ArgumentParser class __lowerCAmelCase ( __A ): '''simple docstring''' @staticmethod @abstractmethod def __UpperCAmelCase ( _a ): raise NotImplementedError() @abstractmethod def __UpperCAmelCase ( self ): raise NotImplementedError()
708
"""simple docstring""" from typing import TYPE_CHECKING from ...file_utils import _LazyModule, is_tokenizers_available, is_torch_available from ...utils import OptionalDependencyNotAvailable lowercase_ = {"configuration_gpt_neox": ["GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXConfig"]} try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["GPTNeoXTokenizerFast"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST", "GPTNeoXForCausalLM", "GPTNeoXForQuestionAnswering", "GPTNeoXForSequenceClassification", "GPTNeoXForTokenClassification", "GPTNeoXLayer", "GPTNeoXModel", "GPTNeoXPreTrainedModel", ] if TYPE_CHECKING: from .configuration_gpt_neox import GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXConfig try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .tokenization_gpt_neox_fast import GPTNeoXTokenizerFast try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_gpt_neox import ( GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST, GPTNeoXForCausalLM, GPTNeoXForQuestionAnswering, GPTNeoXForSequenceClassification, GPTNeoXForTokenClassification, GPTNeoXLayer, GPTNeoXModel, GPTNeoXPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Tuple ) -> List[Any]: __a = [0 for i in range(r + 1 )] # nc0 = 1 __a = 1 for i in range(1 , n + 1 ): # to compute current row from previous row. __a = min(lowerCAmelCase__ , lowerCAmelCase__ ) while j > 0: c[j] += c[j - 1] j -= 1 return c[r] print(binomial_coefficient(n=1_0, r=5))
709
"""simple docstring""" import unittest from transformers import MraConfig, is_torch_available from transformers.testing_utils import require_torch, slow, torch_device from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask if is_torch_available(): import torch from transformers import ( MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, MraModel, ) from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=2 , _a=8 , _a=True , _a=True , _a=True , _a=True , _a=99 , _a=16 , _a=5 , _a=2 , _a=36 , _a="gelu" , _a=0.0 , _a=0.0 , _a=512 , _a=16 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ): __a = parent __a = batch_size __a = seq_length __a = is_training __a = use_input_mask __a = use_token_type_ids __a = use_labels __a = vocab_size __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = type_sequence_label_size __a = initializer_range __a = num_labels __a = num_choices __a = scope def __UpperCAmelCase ( self ): __a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) __a = None if self.use_input_mask: __a = random_attention_mask([self.batch_size, self.seq_length] ) __a = None if self.use_token_type_ids: __a = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size ) __a = None __a = None __a = None if self.use_labels: __a = ids_tensor([self.batch_size] , self.type_sequence_label_size ) __a = ids_tensor([self.batch_size, self.seq_length] , self.num_labels ) __a = ids_tensor([self.batch_size] , self.num_choices ) __a = self.get_config() return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels def __UpperCAmelCase ( self ): return MraConfig( 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=_a , initializer_range=self.initializer_range , ) def __UpperCAmelCase ( self ): __a = self.get_config() __a = 300 return config def __UpperCAmelCase ( self ): ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = self.prepare_config_and_inputs() __a = True __a = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] ) __a = 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraModel(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a ) __a = model(_a , token_type_ids=_a ) __a = model(_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , _a , _a , ): __a = True __a = MraModel(_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , encoder_attention_mask=_a , ) __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , ) __a = model(_a , attention_mask=_a , token_type_ids=_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForMaskedLM(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForQuestionAnswering(config=_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , start_positions=_a , end_positions=_a , ) 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForSequenceClassification(_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForTokenClassification(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_choices __a = MraForMultipleChoice(config=_a ) model.to(_a ) model.eval() __a = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = model( _a , attention_mask=_a , token_type_ids=_a , labels=_a , ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) ) def __UpperCAmelCase ( self ): __a = self.prepare_config_and_inputs() ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = config_and_inputs __a = {'''input_ids''': input_ids, '''token_type_ids''': token_type_ids, '''attention_mask''': input_mask} return config, inputs_dict @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = ( ( MraModel, MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, ) if is_torch_available() else () ) __UpperCAmelCase : str = False __UpperCAmelCase : Any = False __UpperCAmelCase : List[Any] = False __UpperCAmelCase : Optional[Any] = False __UpperCAmelCase : Dict = () def __UpperCAmelCase ( self ): __a = MraModelTester(self ) __a = ConfigTester(self , config_class=_a , hidden_size=37 ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() for type in ["absolute", "relative_key", "relative_key_query"]: __a = type self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_masked_lm(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_multiple_choice(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_question_answering(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_sequence_classification(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_token_classification(*_a ) @slow def __UpperCAmelCase ( self ): for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: __a = MraModel.from_pretrained(_a ) self.assertIsNotNone(_a ) @unittest.skip(reason='''MRA does not output attentions''' ) def __UpperCAmelCase ( self ): return @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @slow def __UpperCAmelCase ( self ): __a = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = torch.Size((1, 256, 768) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[-0.0140, 0.0830, -0.0381], [0.1546, 0.1402, 0.0220], [0.1162, 0.0851, 0.0165]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 256, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[9.2595, -3.6038, 11.8819], [9.3869, -3.2693, 11.0956], [11.8524, -3.4938, 13.1210]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' ) __a = torch.arange(4_096 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 4_096, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[5.4789, -2.3564, 7.5064], [7.9067, -1.3369, 9.9668], [9.0712, -1.8106, 7.0380]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) )
65
0
"""simple docstring""" import time import unittest from transformers import is_torch_available from transformers.testing_utils import require_torch, torch_device from ..test_modeling_common import ids_tensor if is_torch_available(): import torch from transformers.generation import ( MaxLengthCriteria, MaxNewTokensCriteria, MaxTimeCriteria, StoppingCriteriaList, validate_stopping_criteria, ) @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self , _a ): __a = 3 __a = 250 __a = ids_tensor((batch_size, length) , UpperCamelCase__ ) __a = torch.ones((batch_size, length) , device=UpperCamelCase__ , dtype=torch.float ) / length return input_ids, scores def __UpperCAmelCase ( self ): __a , __a = self._get_tensors(5 ) __a = StoppingCriteriaList( [ MaxLengthCriteria(max_length=10 ), MaxTimeCriteria(max_time=0.1 ), ] ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a , __a = self._get_tensors(9 ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a , __a = self._get_tensors(10 ) self.assertTrue(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) def __UpperCAmelCase ( self ): __a = MaxLengthCriteria(max_length=10 ) __a , __a = self._get_tensors(5 ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a , __a = self._get_tensors(9 ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a , __a = self._get_tensors(10 ) self.assertTrue(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) def __UpperCAmelCase ( self ): __a = MaxNewTokensCriteria(start_length=5 , max_new_tokens=5 ) __a , __a = self._get_tensors(5 ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a , __a = self._get_tensors(9 ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a , __a = self._get_tensors(10 ) self.assertTrue(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a = StoppingCriteriaList([criteria] ) self.assertEqual(criteria_list.max_length , 10 ) def __UpperCAmelCase ( self ): __a , __a = self._get_tensors(5 ) __a = MaxTimeCriteria(max_time=0.1 ) self.assertFalse(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) __a = MaxTimeCriteria(max_time=0.1 , initial_timestamp=time.time() - 0.2 ) self.assertTrue(criteria(UpperCamelCase__ , UpperCamelCase__ ) ) def __UpperCAmelCase ( self ): validate_stopping_criteria(StoppingCriteriaList([MaxLengthCriteria(10 )] ) , 10 ) with self.assertWarns(UpperCamelCase__ ): validate_stopping_criteria(StoppingCriteriaList([MaxLengthCriteria(10 )] ) , 11 ) __a = validate_stopping_criteria(StoppingCriteriaList() , 11 ) self.assertEqual(len(UpperCamelCase__ ) , 1 )
710
"""simple docstring""" from __future__ import annotations def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> list[str]: if partitions <= 0: raise ValueError('''partitions must be a positive number!''' ) if partitions > number_of_bytes: raise ValueError('''partitions can not > number_of_bytes!''' ) __a = number_of_bytes // partitions __a = [] for i in range(lowerCAmelCase__ ): __a = i * bytes_per_partition + 1 __a = ( number_of_bytes if i == partitions - 1 else (i + 1) * bytes_per_partition ) allocation_list.append(f'''{start_bytes}-{end_bytes}''' ) return allocation_list if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" from typing import List, Optional, Tuple from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging from .tokenization_herbert import HerbertTokenizer lowercase_ = logging.get_logger(__name__) lowercase_ = {"vocab_file": "vocab.json", "merges_file": "merges.txt", "tokenizer_file": "tokenizer.json"} lowercase_ = { "vocab_file": { "allegro/herbert-base-cased": "https://huggingface.co/allegro/herbert-base-cased/resolve/main/vocab.json" }, "merges_file": { "allegro/herbert-base-cased": "https://huggingface.co/allegro/herbert-base-cased/resolve/main/merges.txt" }, } lowercase_ = {"allegro/herbert-base-cased": 5_1_4} lowercase_ = {} class __lowerCAmelCase ( __UpperCAmelCase ): '''simple docstring''' __UpperCAmelCase : str = VOCAB_FILES_NAMES __UpperCAmelCase : Dict = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : List[str] = PRETRAINED_INIT_CONFIGURATION __UpperCAmelCase : int = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : Union[str, Any] = HerbertTokenizer def __init__( self , _a=None , _a=None , _a=None , _a="<s>" , _a="<unk>" , _a="<pad>" , _a="<mask>" , _a="</s>" , **_a , ): super().__init__( lowerCAmelCase_ , lowerCAmelCase_ , tokenizer_file=lowerCAmelCase_ , cls_token=lowerCAmelCase_ , unk_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , mask_token=lowerCAmelCase_ , sep_token=lowerCAmelCase_ , **lowerCAmelCase_ , ) def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.cls_token_id] __a = [self.sep_token_id] if token_ids_a is None: return cls + token_ids_a + sep return cls + token_ids_a + sep + token_ids_a + sep def __UpperCAmelCase ( self , _a , _a = None , _a = False ): 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] + ([0] * len(lowerCAmelCase_ )) + [1] def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [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 __UpperCAmelCase ( self , _a , _a = None ): __a = self._tokenizer.model.save(lowerCAmelCase_ , name=lowerCAmelCase_ ) return tuple(lowerCAmelCase_ )
711
"""simple docstring""" def lowercase ( lowerCAmelCase__ : list , lowerCAmelCase__ : list , lowerCAmelCase__ : int ) -> list: __a = len(lowerCAmelCase__ ) __a = [[0] * n for i in range(lowerCAmelCase__ )] for i in range(lowerCAmelCase__ ): __a = y_points[i] for i in range(2 , lowerCAmelCase__ ): for j in range(lowerCAmelCase__ , lowerCAmelCase__ ): __a = ( (xa - x_points[j - i + 1]) * q[j][i - 1] - (xa - x_points[j]) * q[j - 1][i - 1] ) / (x_points[j] - x_points[j - i + 1]) return [q[n - 1][n - 1], q] if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: return round(float(moles / volume ) * nfactor ) def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: return round(float((moles * 0.08_21 * temperature) / (volume) ) ) def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: return round(float((moles * 0.08_21 * temperature) / (pressure) ) ) def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: return round(float((pressure * volume) / (0.08_21 * moles) ) ) if __name__ == "__main__": import doctest doctest.testmod()
712
"""simple docstring""" import json import os from functools import lru_cache from typing import List, Optional, Tuple import regex as re from ...tokenization_utils import AddedToken, PreTrainedTokenizer from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = {"vocab_file": "vocab.json", "merges_file": "merges.txt"} lowercase_ = { "vocab_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/vocab.json", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/vocab.json" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/vocab.json" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/vocab.json" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/vocab.json" ), }, "merges_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/merges.txt", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/merges.txt" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/merges.txt" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/merges.txt" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/merges.txt" ), }, } lowercase_ = { "allenai/longformer-base-4096": 4_0_9_6, "allenai/longformer-large-4096": 4_0_9_6, "allenai/longformer-large-4096-finetuned-triviaqa": 4_0_9_6, "allenai/longformer-base-4096-extra.pos.embd.only": 4_0_9_6, "allenai/longformer-large-4096-extra.pos.embd.only": 4_0_9_6, } @lru_cache() # Copied from transformers.models.roberta.tokenization_roberta.bytes_to_unicode def lowercase ( ) -> Union[str, Any]: __a = ( list(range(ord('''!''' ) , ord('''~''' ) + 1 ) ) + list(range(ord('''¡''' ) , ord('''¬''' ) + 1 ) ) + list(range(ord('''®''' ) , ord('''ÿ''' ) + 1 ) ) ) __a = bs[:] __a = 0 for b in range(2**8 ): if b not in bs: bs.append(lowerCAmelCase__ ) cs.append(2**8 + n ) n += 1 __a = [chr(lowerCAmelCase__ ) for n in cs] return dict(zip(lowerCAmelCase__ , lowerCAmelCase__ ) ) def lowercase ( lowerCAmelCase__ : Tuple ) -> Tuple: __a = set() __a = word[0] for char in word[1:]: pairs.add((prev_char, char) ) __a = char return pairs class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Optional[int] = VOCAB_FILES_NAMES __UpperCAmelCase : int = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : Dict = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : Union[str, Any] = ['input_ids', 'attention_mask'] def __init__( self , _a , _a , _a="replace" , _a="<s>" , _a="</s>" , _a="</s>" , _a="<s>" , _a="<unk>" , _a="<pad>" , _a="<mask>" , _a=False , **_a , ): __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else bos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else eos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else sep_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else cls_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else unk_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else pad_token # Mask token behave like a normal word, i.e. include the space before it __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else mask_token super().__init__( errors=_a , bos_token=_a , eos_token=_a , unk_token=_a , sep_token=_a , cls_token=_a , pad_token=_a , mask_token=_a , add_prefix_space=_a , **_a , ) with open(_a , encoding='''utf-8''' ) as vocab_handle: __a = json.load(_a ) __a = {v: k for k, v in self.encoder.items()} __a = errors # how to handle errors in decoding __a = bytes_to_unicode() __a = {v: k for k, v in self.byte_encoder.items()} with open(_a , encoding='''utf-8''' ) as merges_handle: __a = merges_handle.read().split('''\n''' )[1:-1] __a = [tuple(merge.split() ) for merge in bpe_merges] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = {} __a = add_prefix_space # Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions __a = re.compile(R'''\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+''' ) @property def __UpperCAmelCase ( self ): return len(self.encoder ) def __UpperCAmelCase ( self ): return dict(self.encoder , **self.added_tokens_encoder ) def __UpperCAmelCase ( self , _a ): if token in self.cache: return self.cache[token] __a = tuple(_a ) __a = get_pairs(_a ) if not pairs: return token while True: __a = min(_a , key=lambda _a : self.bpe_ranks.get(_a , float('''inf''' ) ) ) if bigram not in self.bpe_ranks: break __a , __a = bigram __a = [] __a = 0 while i < len(_a ): try: __a = word.index(_a , _a ) except ValueError: new_word.extend(word[i:] ) break else: new_word.extend(word[i:j] ) __a = j if word[i] == first and i < len(_a ) - 1 and word[i + 1] == second: new_word.append(first + second ) i += 2 else: new_word.append(word[i] ) i += 1 __a = tuple(_a ) __a = new_word if len(_a ) == 1: break else: __a = get_pairs(_a ) __a = ''' '''.join(_a ) __a = word return word def __UpperCAmelCase ( self , _a ): __a = [] for token in re.findall(self.pat , _a ): __a = ''''''.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(_a ).split(''' ''' ) ) return bpe_tokens def __UpperCAmelCase ( self , _a ): return self.encoder.get(_a , self.encoder.get(self.unk_token ) ) def __UpperCAmelCase ( self , _a ): return self.decoder.get(_a ) def __UpperCAmelCase ( self , _a ): __a = ''''''.join(_a ) __a = bytearray([self.byte_decoder[c] for c in text] ).decode('''utf-8''' , errors=self.errors ) return text def __UpperCAmelCase ( self , _a , _a = None ): if not os.path.isdir(_a ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] ) __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] ) with open(_a , '''w''' , encoding='''utf-8''' ) as f: f.write(json.dumps(self.encoder , indent=2 , sort_keys=_a , ensure_ascii=_a ) + '''\n''' ) __a = 0 with open(_a , '''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 _a : 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!''' ) __a = token_index writer.write(''' '''.join(_a ) + '''\n''' ) index += 1 return vocab_file, merge_file def __UpperCAmelCase ( self , _a , _a = None ): if token_ids_a is None: return [self.cls_token_id] + token_ids_a + [self.sep_token_id] __a = [self.cls_token_id] __a = [self.sep_token_id] return cls + token_ids_a + sep + sep + token_ids_a + sep def __UpperCAmelCase ( self , _a , _a = None , _a = False ): if already_has_special_tokens: return super().get_special_tokens_mask( token_ids_a=_a , token_ids_a=_a , already_has_special_tokens=_a ) if token_ids_a is None: return [1] + ([0] * len(_a )) + [1] return [1] + ([0] * len(_a )) + [1, 1] + ([0] * len(_a )) + [1] def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [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 __UpperCAmelCase ( self , _a , _a=False , **_a ): __a = kwargs.pop('''add_prefix_space''' , self.add_prefix_space ) if (is_split_into_words or add_prefix_space) and (len(_a ) > 0 and not text[0].isspace()): __a = ''' ''' + text return (text, kwargs)
65
0
"""simple docstring""" from ... import PretrainedConfig lowercase_ = { "sijunhe/nezha-cn-base": "https://huggingface.co/sijunhe/nezha-cn-base/resolve/main/config.json", } class __lowerCAmelCase ( _a ): '''simple docstring''' __UpperCAmelCase : Tuple = NEZHA_PRETRAINED_CONFIG_ARCHIVE_MAP __UpperCAmelCase : int = 'nezha' def __init__( self , _a=21_128 , _a=768 , _a=12 , _a=12 , _a=3_072 , _a="gelu" , _a=0.1 , _a=0.1 , _a=512 , _a=64 , _a=2 , _a=0.02 , _a=1E-12 , _a=0.1 , _a=0 , _a=2 , _a=3 , _a=True , **_a , ): super().__init__(pad_token_id=_A , bos_token_id=_A , eos_token_id=_A , **_A ) __a = vocab_size __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = hidden_act __a = intermediate_size __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = max_relative_position __a = type_vocab_size __a = initializer_range __a = layer_norm_eps __a = classifier_dropout __a = use_cache
713
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "unc-nlp/lxmert-base-uncased": "https://huggingface.co/unc-nlp/lxmert-base-uncased/resolve/main/config.json", } class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = 'lxmert' __UpperCAmelCase : str = {} def __init__( self , _a=30_522 , _a=768 , _a=12 , _a=9_500 , _a=1_600 , _a=400 , _a=3_072 , _a="gelu" , _a=0.1 , _a=0.1 , _a=512 , _a=2 , _a=0.02 , _a=1E-12 , _a=9 , _a=5 , _a=5 , _a=2_048 , _a=4 , _a=6.67 , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , **_a , ): __a = vocab_size __a = hidden_size __a = num_attention_heads __a = hidden_act __a = intermediate_size __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = initializer_range __a = layer_norm_eps __a = num_qa_labels __a = num_object_labels __a = num_attr_labels __a = l_layers __a = x_layers __a = r_layers __a = visual_feat_dim __a = visual_pos_dim __a = visual_loss_normalizer __a = task_matched __a = task_mask_lm __a = task_obj_predict __a = task_qa __a = visual_obj_loss __a = visual_attr_loss __a = visual_feat_loss __a = {'''vision''': r_layers, '''cross_encoder''': x_layers, '''language''': l_layers} super().__init__(**_a )
65
0
"""simple docstring""" import os from glob import glob import imageio import torch import torchvision import wandb from img_processing import custom_to_pil, loop_post_process, preprocess, preprocess_vqgan from loaders import load_vqgan from PIL import Image from torch import nn from transformers import CLIPModel, CLIPTokenizerFast from utils import get_device, get_timestamp, show_pil class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a = "cpu" , _a = "openai/clip-vit-large-patch14" ): __a = device __a = CLIPTokenizerFast.from_pretrained(__UpperCamelCase ) __a = [0.4814_5466, 0.457_8275, 0.4082_1073] __a = [0.2686_2954, 0.2613_0258, 0.2757_7711] __a = torchvision.transforms.Normalize(self.image_mean , self.image_std ) __a = torchvision.transforms.Resize(224 ) __a = torchvision.transforms.CenterCrop(224 ) def __UpperCAmelCase ( self , _a ): __a = self.resize(__UpperCamelCase ) __a = self.center_crop(__UpperCamelCase ) __a = self.normalize(__UpperCamelCase ) return images def __call__( self , _a=None , _a=None , **_a ): __a = self.tokenizer(text=__UpperCamelCase , **__UpperCamelCase ) __a = self.preprocess_img(__UpperCamelCase ) __a = {key: value.to(self.device ) for (key, value) in encoding.items()} return encoding class __lowerCAmelCase ( nn.Module ): '''simple docstring''' def __init__( self , _a=10 , _a=0.01 , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=False , _a=True , _a="image" , _a=True , _a=False , _a=False , _a=False , ): super().__init__() __a = None __a = device if device else get_device() if vqgan: __a = vqgan else: __a = load_vqgan(self.device , conf_path=__UpperCamelCase , ckpt_path=__UpperCamelCase ) self.vqgan.eval() if clip: __a = clip else: __a = CLIPModel.from_pretrained('''openai/clip-vit-base-patch32''' ) self.clip.to(self.device ) __a = ProcessorGradientFlow(device=self.device ) __a = iterations __a = lr __a = log __a = make_grid __a = return_val __a = quantize __a = self.vqgan.decoder.z_shape def __UpperCAmelCase ( self , _a=None , _a=None , _a=5 , _a=True ): __a = [] if output_path is None: __a = '''./animation.gif''' if input_path is None: __a = self.save_path __a = sorted(glob(input_path + '''/*''' ) ) if not len(__UpperCamelCase ): raise ValueError( '''No images found in save path, aborting (did you pass save_intermediate=True to the generate''' ''' function?)''' ) if len(__UpperCamelCase ) == 1: print('''Only one image found in save path, (did you pass save_intermediate=True to the generate function?)''' ) __a = total_duration / len(__UpperCamelCase ) __a = [frame_duration] * len(__UpperCamelCase ) if extend_frames: __a = 1.5 __a = 3 for file_name in paths: if file_name.endswith('''.png''' ): images.append(imageio.imread(__UpperCamelCase ) ) imageio.mimsave(__UpperCamelCase , __UpperCamelCase , duration=__UpperCamelCase ) print(f'''gif saved to {output_path}''' ) def __UpperCAmelCase ( self , _a=None , _a=None ): if not (path or img): raise ValueError('''Input either path or tensor''' ) if img is not None: raise NotImplementedError __a = preprocess(Image.open(__UpperCamelCase ) , target_image_size=256 ).to(self.device ) __a = preprocess_vqgan(__UpperCamelCase ) __a , *__a = self.vqgan.encode(__UpperCamelCase ) return z def __UpperCAmelCase ( self , _a ): __a = self.latent.detach().requires_grad_() __a = base_latent + transform_vector if self.quantize: __a , *__a = self.vqgan.quantize(__UpperCamelCase ) else: __a = trans_latent return self.vqgan.decode(__UpperCamelCase ) def __UpperCAmelCase ( self , _a , _a , _a=None ): __a = self.clip_preprocessor(text=__UpperCamelCase , images=__UpperCamelCase , return_tensors='''pt''' , padding=__UpperCamelCase ) __a = self.clip(**__UpperCamelCase ) __a = clip_outputs.logits_per_image if weights is not None: __a = similarity_logits * weights return similarity_logits.sum() def __UpperCAmelCase ( self , _a , _a , _a ): __a = self._get_clip_similarity(pos_prompts['''prompts'''] , __UpperCamelCase , weights=(1 / pos_prompts['''weights''']) ) if neg_prompts: __a = self._get_clip_similarity(neg_prompts['''prompts'''] , __UpperCamelCase , weights=neg_prompts['''weights'''] ) else: __a = torch.tensor([1] , device=self.device ) __a = -torch.log(__UpperCamelCase ) + torch.log(__UpperCamelCase ) return loss def __UpperCAmelCase ( self , _a , _a , _a ): __a = torch.randn_like(self.latent , requires_grad=__UpperCamelCase , device=self.device ) __a = torch.optim.Adam([vector] , lr=self.lr ) for i in range(self.iterations ): optim.zero_grad() __a = self._add_vector(__UpperCamelCase ) __a = loop_post_process(__UpperCamelCase ) __a = self._get_CLIP_loss(__UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) print('''CLIP loss''' , __UpperCamelCase ) if self.log: wandb.log({'''CLIP Loss''': clip_loss} ) clip_loss.backward(retain_graph=__UpperCamelCase ) optim.step() if self.return_val == "image": yield custom_to_pil(transformed_img[0] ) else: yield vector def __UpperCAmelCase ( self , _a , _a , _a ): wandb.init(reinit=__UpperCamelCase , project='''face-editor''' ) wandb.config.update({'''Positive Prompts''': positive_prompts} ) wandb.config.update({'''Negative Prompts''': negative_prompts} ) wandb.config.update({'''lr''': self.lr, '''iterations''': self.iterations} ) if image_path: __a = Image.open(__UpperCamelCase ) __a = image.resize((256, 256) ) wandb.log('''Original Image''' , wandb.Image(__UpperCamelCase ) ) def __UpperCAmelCase ( self , _a ): if not prompts: return [] __a = [] __a = [] if isinstance(__UpperCamelCase , __UpperCamelCase ): __a = [prompt.strip() for prompt in prompts.split('''|''' )] for prompt in prompts: if isinstance(__UpperCamelCase , (tuple, list) ): __a = prompt[0] __a = float(prompt[1] ) elif ":" in prompt: __a , __a = prompt.split(''':''' ) __a = float(__UpperCamelCase ) else: __a = prompt __a = 1.0 processed_prompts.append(__UpperCamelCase ) weights.append(__UpperCamelCase ) return { "prompts": processed_prompts, "weights": torch.tensor(__UpperCamelCase , device=self.device ), } def __UpperCAmelCase ( self , _a , _a=None , _a=None , _a=True , _a=False , _a=True , _a=True , _a=None , ): if image_path: __a = self._get_latent(__UpperCamelCase ) else: __a = torch.randn(self.latent_dim , device=self.device ) if self.log: self._init_logging(__UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) assert pos_prompts, "You must provide at least one positive prompt." __a = self.process_prompts(__UpperCamelCase ) __a = self.process_prompts(__UpperCamelCase ) if save_final and save_path is None: __a = os.path.join('''./outputs/''' , '''_'''.join(pos_prompts['''prompts'''] ) ) if not os.path.exists(__UpperCamelCase ): os.makedirs(__UpperCamelCase ) else: __a = save_path + '''_''' + get_timestamp() os.makedirs(__UpperCamelCase ) __a = save_path __a = self.vqgan.decode(self.latent )[0] if show_intermediate: print('''Original Image''' ) show_pil(custom_to_pil(__UpperCamelCase ) ) __a = loop_post_process(__UpperCamelCase ) for iter, transformed_img in enumerate(self._optimize_CLIP(__UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) ): if show_intermediate: show_pil(__UpperCamelCase ) if save_intermediate: transformed_img.save(os.path.join(self.save_path , f'''iter_{iter:03d}.png''' ) ) if self.log: wandb.log({'''Image''': wandb.Image(__UpperCamelCase )} ) if show_final: show_pil(__UpperCamelCase ) if save_final: transformed_img.save(os.path.join(self.save_path , f'''iter_{iter:03d}_final.png''' ) )
714
"""simple docstring""" import itertools import math def lowercase ( lowerCAmelCase__ : int ) -> bool: 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(lowerCAmelCase__ ) + 1 ) , 6 ): if number % i == 0 or number % (i + 2) == 0: return False return True def lowercase ( ) -> int: __a = 2 while True: if is_prime(lowerCAmelCase__ ): yield num num += 1 def lowercase ( lowerCAmelCase__ : int = 10001 ) -> int: return next(itertools.islice(prime_generator() , nth - 1 , lowerCAmelCase__ ) ) if __name__ == "__main__": print(F'''{solution() = }''')
65
0
"""simple docstring""" import argparse import json from dataclasses import dataclass, field from functools import partial from pathlib import Path from typing import Callable, Dict, List, Tuple import timm import torch import torch.nn as nn from classy_vision.models.regnet import RegNet, RegNetParams, RegNetYaagf, RegNetYaagf, RegNetYaaagf from huggingface_hub import cached_download, hf_hub_url from torch import Tensor from vissl.models.model_helpers import get_trunk_forward_outputs from transformers import AutoImageProcessor, RegNetConfig, RegNetForImageClassification, RegNetModel from transformers.utils import logging logging.set_verbosity_info() lowercase_ = logging.get_logger() @dataclass class __lowerCAmelCase : __UpperCAmelCase : nn.Module __UpperCAmelCase : List[nn.Module] = field(default_factory=__SCREAMING_SNAKE_CASE ) __UpperCAmelCase : list = field(default_factory=__SCREAMING_SNAKE_CASE ) def __UpperCAmelCase ( self , _a , _a , _a ): __a = len(list(m.modules() ) ) == 1 or isinstance(_a , nn.Convad ) or isinstance(_a , nn.BatchNormad ) if has_not_submodules: self.traced.append(_a ) def __call__( self , _a ): for m in self.module.modules(): self.handles.append(m.register_forward_hook(self._forward_hook ) ) self.module(_a ) [x.remove() for x in self.handles] return self @property def __UpperCAmelCase ( self ): return list(filter(lambda _a : len(list(x.state_dict().keys() ) ) > 0 , self.traced ) ) @dataclass class __lowerCAmelCase : __UpperCAmelCase : nn.Module __UpperCAmelCase : nn.Module __UpperCAmelCase : int = 1 __UpperCAmelCase : List = field(default_factory=__SCREAMING_SNAKE_CASE ) __UpperCAmelCase : List = field(default_factory=__SCREAMING_SNAKE_CASE ) __UpperCAmelCase : bool = True def __call__( self , _a ): __a = Tracker(self.dest )(_a ).parametrized __a = Tracker(self.src )(_a ).parametrized __a = list(filter(lambda _a : type(_a ) not in self.src_skip , _a ) ) __a = list(filter(lambda _a : type(_a ) not in self.dest_skip , _a ) ) if len(_a ) != len(_a ) and self.raise_if_mismatch: raise Exception( f'''Numbers of operations are different. Source module has {len(_a )} operations while''' f''' destination module has {len(_a )}.''' ) for dest_m, src_m in zip(_a , _a ): dest_m.load_state_dict(src_m.state_dict() ) if self.verbose == 1: print(f'''Transfered from={src_m} to={dest_m}''' ) class __lowerCAmelCase ( nn.Module ): def __init__( self , _a ): super().__init__() __a = [] # - get the stem feature_blocks.append(('''conv1''', model.stem) ) # - get all the feature blocks for k, v in model.trunk_output.named_children(): assert k.startswith('''block''' ), f'''Unexpected layer name {k}''' __a = len(_a ) + 1 feature_blocks.append((f'''res{block_index}''', v) ) __a = nn.ModuleDict(_a ) def __UpperCAmelCase ( self , _a ): return get_trunk_forward_outputs( _a , out_feat_keys=_a , feature_blocks=self._feature_blocks , ) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): def __UpperCAmelCase ( self , _a ): __a = x.split('''-''' ) return x_split[0] + x_split[1] + "_" + "".join(x_split[2:] ) def __getitem__( self , _a ): if x not in self: __a = self.convert_name_to_timm(_a ) __a = partial(lambda: (timm.create_model(_a , pretrained=_a ).eval(), None) ) else: __a = super().__getitem__(_a ) return val class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): def __getitem__( self , _a ): if "seer" in x and "in1k" not in x: __a = RegNetModel else: __a = RegNetForImageClassification return val def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Tuple[str, str]] ) -> Optional[Any]: for from_key, to_key in keys: __a = from_state_dict[from_key].clone() print(f'''Copied key={from_key} to={to_key}''' ) return to_state_dict def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Callable[[], nn.Module] , lowerCAmelCase__ : Callable[[], nn.Module] , lowerCAmelCase__ : RegNetConfig , lowerCAmelCase__ : Path , lowerCAmelCase__ : bool = True , ) -> List[Any]: print(f'''Converting {name}...''' ) with torch.no_grad(): __a , __a = from_model_func() __a = our_model_func(_UpperCamelCase ).eval() __a = ModuleTransfer(src=_UpperCamelCase , dest=_UpperCamelCase , raise_if_mismatch=_UpperCamelCase ) __a = torch.randn((1, 3, 224, 224) ) module_transfer(_UpperCamelCase ) if from_state_dict is not None: __a = [] # for seer - in1k finetuned we have to manually copy the head if "seer" in name and "in1k" in name: __a = [('''0.clf.0.weight''', '''classifier.1.weight'''), ('''0.clf.0.bias''', '''classifier.1.bias''')] __a = manually_copy_vissl_head(_UpperCamelCase , our_model.state_dict() , _UpperCamelCase ) our_model.load_state_dict(_UpperCamelCase ) __a = our_model(_UpperCamelCase , output_hidden_states=_UpperCamelCase ) __a = ( our_outputs.logits if isinstance(_UpperCamelCase , _UpperCamelCase ) else our_outputs.last_hidden_state ) __a = from_model(_UpperCamelCase ) __a = from_output[-1] if type(_UpperCamelCase ) is list else from_output # now since I don't want to use any config files, vissl seer model doesn't actually have an head, so let's just check the last hidden state if "seer" in name and "in1k" in name: __a = our_outputs.hidden_states[-1] assert torch.allclose(_UpperCamelCase , _UpperCamelCase ), "The model logits don't match the original one." if push_to_hub: our_model.push_to_hub( repo_path_or_name=save_directory / name , commit_message='''Add model''' , use_temp_dir=_UpperCamelCase , ) __a = 224 if '''seer''' not in name else 384 # we can use the convnext one __a = AutoImageProcessor.from_pretrained('''facebook/convnext-base-224-22k-1k''' , size=_UpperCamelCase ) image_processor.push_to_hub( repo_path_or_name=save_directory / name , commit_message='''Add image processor''' , use_temp_dir=_UpperCamelCase , ) print(f'''Pushed {name}''' ) def lowercase ( lowerCAmelCase__ : Path , lowerCAmelCase__ : str = None , lowerCAmelCase__ : bool = True ) -> Dict: __a = '''imagenet-1k-id2label.json''' __a = 1000 __a = (1, num_labels) __a = '''huggingface/label-files''' __a = num_labels __a = json.load(open(cached_download(hf_hub_url(_UpperCamelCase , _UpperCamelCase , repo_type='''dataset''' ) ) , '''r''' ) ) __a = {int(_UpperCamelCase ): v for k, v in idalabel.items()} __a = idalabel __a = {v: k for k, v in idalabel.items()} __a = partial(_UpperCamelCase , num_labels=_UpperCamelCase , idalabel=_UpperCamelCase , labelaid=_UpperCamelCase ) __a = { '''regnet-x-002''': ImageNetPreTrainedConfig( depths=[1, 1, 4, 7] , hidden_sizes=[24, 56, 152, 368] , groups_width=8 , layer_type='''x''' ), '''regnet-x-004''': ImageNetPreTrainedConfig( depths=[1, 2, 7, 12] , hidden_sizes=[32, 64, 160, 384] , groups_width=16 , layer_type='''x''' ), '''regnet-x-006''': ImageNetPreTrainedConfig( depths=[1, 3, 5, 7] , hidden_sizes=[48, 96, 240, 528] , groups_width=24 , layer_type='''x''' ), '''regnet-x-008''': ImageNetPreTrainedConfig( depths=[1, 3, 7, 5] , hidden_sizes=[64, 128, 288, 672] , groups_width=16 , layer_type='''x''' ), '''regnet-x-016''': ImageNetPreTrainedConfig( depths=[2, 4, 10, 2] , hidden_sizes=[72, 168, 408, 912] , groups_width=24 , layer_type='''x''' ), '''regnet-x-032''': ImageNetPreTrainedConfig( depths=[2, 6, 15, 2] , hidden_sizes=[96, 192, 432, 1008] , groups_width=48 , layer_type='''x''' ), '''regnet-x-040''': ImageNetPreTrainedConfig( depths=[2, 5, 14, 2] , hidden_sizes=[80, 240, 560, 1360] , groups_width=40 , layer_type='''x''' ), '''regnet-x-064''': ImageNetPreTrainedConfig( depths=[2, 4, 10, 1] , hidden_sizes=[168, 392, 784, 1624] , groups_width=56 , layer_type='''x''' ), '''regnet-x-080''': ImageNetPreTrainedConfig( depths=[2, 5, 15, 1] , hidden_sizes=[80, 240, 720, 1920] , groups_width=120 , layer_type='''x''' ), '''regnet-x-120''': ImageNetPreTrainedConfig( depths=[2, 5, 11, 1] , hidden_sizes=[224, 448, 896, 2240] , groups_width=112 , layer_type='''x''' ), '''regnet-x-160''': ImageNetPreTrainedConfig( depths=[2, 6, 13, 1] , hidden_sizes=[256, 512, 896, 2048] , groups_width=128 , layer_type='''x''' ), '''regnet-x-320''': ImageNetPreTrainedConfig( depths=[2, 7, 13, 1] , hidden_sizes=[336, 672, 1344, 2520] , groups_width=168 , layer_type='''x''' ), # y variant '''regnet-y-002''': ImageNetPreTrainedConfig(depths=[1, 1, 4, 7] , hidden_sizes=[24, 56, 152, 368] , groups_width=8 ), '''regnet-y-004''': ImageNetPreTrainedConfig( depths=[1, 3, 6, 6] , hidden_sizes=[48, 104, 208, 440] , groups_width=8 ), '''regnet-y-006''': ImageNetPreTrainedConfig( depths=[1, 3, 7, 4] , hidden_sizes=[48, 112, 256, 608] , groups_width=16 ), '''regnet-y-008''': ImageNetPreTrainedConfig( depths=[1, 3, 8, 2] , hidden_sizes=[64, 128, 320, 768] , groups_width=16 ), '''regnet-y-016''': ImageNetPreTrainedConfig( depths=[2, 6, 17, 2] , hidden_sizes=[48, 120, 336, 888] , groups_width=24 ), '''regnet-y-032''': ImageNetPreTrainedConfig( depths=[2, 5, 13, 1] , hidden_sizes=[72, 216, 576, 1512] , groups_width=24 ), '''regnet-y-040''': ImageNetPreTrainedConfig( depths=[2, 6, 12, 2] , hidden_sizes=[128, 192, 512, 1088] , groups_width=64 ), '''regnet-y-064''': ImageNetPreTrainedConfig( depths=[2, 7, 14, 2] , hidden_sizes=[144, 288, 576, 1296] , groups_width=72 ), '''regnet-y-080''': ImageNetPreTrainedConfig( depths=[2, 4, 10, 1] , hidden_sizes=[168, 448, 896, 2016] , groups_width=56 ), '''regnet-y-120''': ImageNetPreTrainedConfig( depths=[2, 5, 11, 1] , hidden_sizes=[224, 448, 896, 2240] , groups_width=112 ), '''regnet-y-160''': ImageNetPreTrainedConfig( depths=[2, 4, 11, 1] , hidden_sizes=[224, 448, 1232, 3024] , groups_width=112 ), '''regnet-y-320''': ImageNetPreTrainedConfig( depths=[2, 5, 12, 1] , hidden_sizes=[232, 696, 1392, 3712] , groups_width=232 ), # models created by SEER -> https://arxiv.org/abs/2202.08360 '''regnet-y-320-seer''': RegNetConfig(depths=[2, 5, 12, 1] , hidden_sizes=[232, 696, 1392, 3712] , groups_width=232 ), '''regnet-y-640-seer''': RegNetConfig(depths=[2, 5, 12, 1] , hidden_sizes=[328, 984, 1968, 4920] , groups_width=328 ), '''regnet-y-1280-seer''': RegNetConfig( depths=[2, 7, 17, 1] , hidden_sizes=[528, 1056, 2904, 7392] , groups_width=264 ), '''regnet-y-2560-seer''': RegNetConfig( depths=[3, 7, 16, 1] , hidden_sizes=[640, 1696, 2544, 5088] , groups_width=640 ), '''regnet-y-10b-seer''': ImageNetPreTrainedConfig( depths=[2, 7, 17, 1] , hidden_sizes=[2020, 4040, 11110, 28280] , groups_width=1010 ), # finetuned on imagenet '''regnet-y-320-seer-in1k''': ImageNetPreTrainedConfig( depths=[2, 5, 12, 1] , hidden_sizes=[232, 696, 1392, 3712] , groups_width=232 ), '''regnet-y-640-seer-in1k''': ImageNetPreTrainedConfig( depths=[2, 5, 12, 1] , hidden_sizes=[328, 984, 1968, 4920] , groups_width=328 ), '''regnet-y-1280-seer-in1k''': ImageNetPreTrainedConfig( depths=[2, 7, 17, 1] , hidden_sizes=[528, 1056, 2904, 7392] , groups_width=264 ), '''regnet-y-2560-seer-in1k''': ImageNetPreTrainedConfig( depths=[3, 7, 16, 1] , hidden_sizes=[640, 1696, 2544, 5088] , groups_width=640 ), '''regnet-y-10b-seer-in1k''': ImageNetPreTrainedConfig( depths=[2, 7, 17, 1] , hidden_sizes=[2020, 4040, 11110, 28280] , groups_width=1010 ), } __a = NameToOurModelFuncMap() __a = NameToFromModelFuncMap() # add seer weights logic def load_using_classy_vision(lowerCAmelCase__ : str , lowerCAmelCase__ : Callable[[], nn.Module] ) -> Tuple[nn.Module, Dict]: __a = torch.hub.load_state_dict_from_url(_UpperCamelCase , model_dir=str(_UpperCamelCase ) , map_location='''cpu''' ) __a = model_func() # check if we have a head, if yes add it __a = files['''classy_state_dict''']['''base_model''']['''model'''] __a = model_state_dict['''trunk'''] model.load_state_dict(_UpperCamelCase ) return model.eval(), model_state_dict["heads"] # pretrained __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_regnet32d/seer_regnet32gf_model_iteration244000.torch''' , lambda: FakeRegNetVisslWrapper(RegNetYaagf() ) , ) __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_regnet64/seer_regnet64gf_model_final_checkpoint_phase0.torch''' , lambda: FakeRegNetVisslWrapper(RegNetYaagf() ) , ) __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/swav_ig1b_regnet128Gf_cnstant_bs32_node16_sinkhorn10_proto16k_syncBN64_warmup8k/model_final_checkpoint_phase0.torch''' , lambda: FakeRegNetVisslWrapper(RegNetYaaagf() ) , ) __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_regnet10B/model_iteration124500_conso.torch''' , lambda: FakeRegNetVisslWrapper( RegNet(RegNetParams(depth=27 , group_width=1010 , w_a=1744 , w_a=620.83 , w_m=2.52 ) ) ) , ) # IN1K finetuned __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet32_finetuned_in1k_model_final_checkpoint_phase78.torch''' , lambda: FakeRegNetVisslWrapper(RegNetYaagf() ) , ) __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet64_finetuned_in1k_model_final_checkpoint_phase78.torch''' , lambda: FakeRegNetVisslWrapper(RegNetYaagf() ) , ) __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_regnet128_finetuned_in1k_model_final_checkpoint_phase78.torch''' , lambda: FakeRegNetVisslWrapper(RegNetYaaagf() ) , ) __a = partial( _UpperCamelCase , '''https://dl.fbaipublicfiles.com/vissl/model_zoo/seer_finetuned/seer_10b_finetuned_in1k_model_phase28_conso.torch''' , lambda: FakeRegNetVisslWrapper( RegNet(RegNetParams(depth=27 , group_width=1010 , w_a=1744 , w_a=620.83 , w_m=2.52 ) ) ) , ) if model_name: convert_weight_and_push( _UpperCamelCase , names_to_from_model_map[model_name] , names_to_ours_model_map[model_name] , names_to_config[model_name] , _UpperCamelCase , _UpperCamelCase , ) else: for model_name, config in names_to_config.items(): convert_weight_and_push( _UpperCamelCase , names_to_from_model_map[model_name] , names_to_ours_model_map[model_name] , _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , ) return config, expected_shape if __name__ == "__main__": lowercase_ = argparse.ArgumentParser() # Required parameters parser.add_argument( "--model_name", default=None, type=str, help=( "The name of the model you wish to convert, it must be one of the supported regnet* architecture," " currently: regnetx-*, regnety-*. If `None`, all of them will the converted." ), ) parser.add_argument( "--pytorch_dump_folder_path", default=None, type=Path, required=True, help="Path to the output PyTorch model directory.", ) parser.add_argument( "--push_to_hub", default=True, type=bool, required=False, help="If True, push model and image processor to the hub.", ) lowercase_ = parser.parse_args() lowercase_ = args.pytorch_dump_folder_path pytorch_dump_folder_path.mkdir(exist_ok=True, parents=True) convert_weights_and_push(pytorch_dump_folder_path, args.model_name, args.push_to_hub)
715
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_mask2former": [ "MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "Mask2FormerConfig", ], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["Mask2FormerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST", "Mask2FormerForUniversalSegmentation", "Mask2FormerModel", "Mask2FormerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_maskaformer import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskaFormerConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_maskaformer import MaskaFormerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_maskaformer import ( MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST, MaskaFormerForUniversalSegmentation, MaskaFormerModel, MaskaFormerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
65
0
"""simple docstring""" import numpy as np import torch from torch.utils.data import DataLoader from accelerate.utils.dataclasses import DistributedType class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a=2 , _a=3 , _a=64 , _a=None ): __a = np.random.default_rng(_a ) __a = length __a = rng.normal(size=(length,) ).astype(np.floataa ) __a = a * self.x + b + rng.normal(scale=0.1 , size=(length,) ).astype(np.floataa ) def __len__( self ): return self.length def __getitem__( self , _a ): return {"x": self.x[i], "y": self.y[i]} class __lowerCAmelCase ( torch.nn.Module ): '''simple docstring''' def __init__( self , _a=0 , _a=0 , _a=False ): super().__init__() __a = torch.nn.Parameter(torch.tensor([2, 3] ).float() ) __a = torch.nn.Parameter(torch.tensor([2, 3] ).float() ) __a = True def __UpperCAmelCase ( self , _a=None ): if self.first_batch: print(f'''Model dtype: {self.a.dtype}, {self.b.dtype}. Input dtype: {x.dtype}''' ) __a = False return x * self.a[0] + self.b[0] class __lowerCAmelCase ( torch.nn.Module ): '''simple docstring''' def __init__( self , _a=0 , _a=0 , _a=False ): super().__init__() __a = torch.nn.Parameter(torch.tensor(_a ).float() ) __a = torch.nn.Parameter(torch.tensor(_a ).float() ) __a = True def __UpperCAmelCase ( self , _a=None ): if self.first_batch: print(f'''Model dtype: {self.a.dtype}, {self.b.dtype}. Input dtype: {x.dtype}''' ) __a = False return x * self.a + self.b def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Tuple = 16 ) -> int: from datasets import load_dataset from transformers import AutoTokenizer __a = AutoTokenizer.from_pretrained('''bert-base-cased''' ) __a = {'''train''': '''tests/test_samples/MRPC/train.csv''', '''validation''': '''tests/test_samples/MRPC/dev.csv'''} __a = load_dataset('''csv''' , data_files=lowerCAmelCase__ ) __a = datasets['''train'''].unique('''label''' ) __a = {v: i for i, v in enumerate(lowerCAmelCase__ )} def tokenize_function(lowerCAmelCase__ : int ): # max_length=None => use the model max length (it's actually the default) __a = tokenizer( examples['''sentence1'''] , examples['''sentence2'''] , truncation=lowerCAmelCase__ , max_length=lowerCAmelCase__ , padding='''max_length''' ) if "label" in examples: __a = [label_to_id[l] for l in examples['''label''']] return outputs # Apply the method we just defined to all the examples in all the splits of the dataset __a = datasets.map( lowerCAmelCase__ , batched=lowerCAmelCase__ , remove_columns=['''sentence1''', '''sentence2''', '''label'''] , ) def collate_fn(lowerCAmelCase__ : List[Any] ): # On TPU it's best to pad everything to the same length or training will be very slow. if accelerator.distributed_type == DistributedType.TPU: return tokenizer.pad(lowerCAmelCase__ , padding='''max_length''' , max_length=128 , return_tensors='''pt''' ) return tokenizer.pad(lowerCAmelCase__ , padding='''longest''' , return_tensors='''pt''' ) # Instantiate dataloaders. __a = DataLoader(tokenized_datasets['''train'''] , shuffle=lowerCAmelCase__ , collate_fn=lowerCAmelCase__ , batch_size=2 ) __a = DataLoader(tokenized_datasets['''validation'''] , shuffle=lowerCAmelCase__ , collate_fn=lowerCAmelCase__ , batch_size=1 ) return train_dataloader, eval_dataloader
716
"""simple docstring""" from sklearn.metrics import fa_score import datasets lowercase_ = "\nThe F1 score is the harmonic mean of the precision and recall. It can be computed with the equation:\nF1 = 2 * (precision * recall) / (precision + recall)\n" lowercase_ = "\nArgs:\n predictions (`list` of `int`): Predicted labels.\n references (`list` of `int`): Ground truth labels.\n labels (`list` of `int`): The set of labels to include when `average` is not set to `'binary'`, and the order of the labels if `average` is `None`. Labels present in the data can be excluded, for example to calculate a multiclass average ignoring a majority negative class. Labels not present in the data will result in 0 components in a macro average. For multilabel targets, labels are column indices. By default, all labels in `predictions` and `references` are used in sorted order. Defaults to None.\n pos_label (`int`): The class to be considered the positive class, in the case where `average` is set to `binary`. Defaults to 1.\n average (`string`): This parameter is required for multiclass/multilabel targets. If set to `None`, the scores for each class are returned. Otherwise, this determines the type of averaging performed on the data. Defaults to `'binary'`.\n\n - 'binary': Only report results for the class specified by `pos_label`. This is applicable only if the classes found in `predictions` and `references` are binary.\n - 'micro': Calculate metrics globally by counting the total true positives, false negatives and false positives.\n - 'macro': Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.\n - 'weighted': Calculate metrics for each label, and find their average weighted by support (the number of true instances for each label). This alters `'macro'` to account for label imbalance. This option can result in an F-score that is not between precision and recall.\n - 'samples': Calculate metrics for each instance, and find their average (only meaningful for multilabel classification).\n sample_weight (`list` of `float`): Sample weights Defaults to None.\n\nReturns:\n f1 (`float` or `array` of `float`): F1 score or list of f1 scores, depending on the value passed to `average`. Minimum possible value is 0. Maximum possible value is 1. Higher f1 scores are better.\n\nExamples:\n\n Example 1-A simple binary example\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0])\n >>> print(results)\n {'f1': 0.5}\n\n Example 2-The same simple binary example as in Example 1, but with `pos_label` set to `0`.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], pos_label=0)\n >>> print(round(results['f1'], 2))\n 0.67\n\n Example 3-The same simple binary example as in Example 1, but with `sample_weight` included.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], sample_weight=[0.9, 0.5, 3.9, 1.2, 0.3])\n >>> print(round(results['f1'], 2))\n 0.35\n\n Example 4-A multiclass example, with different values for the `average` input.\n >>> predictions = [0, 2, 1, 0, 0, 1]\n >>> references = [0, 1, 2, 0, 1, 2]\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"macro\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"micro\")\n >>> print(round(results['f1'], 2))\n 0.33\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"weighted\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=None)\n >>> print(results)\n {'f1': array([0.8, 0. , 0. ])}\n" lowercase_ = "\n@article{scikit-learn,\n title={Scikit-learn: Machine Learning in {P}ython},\n author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.\n and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.\n and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and\n Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},\n journal={Journal of Machine Learning Research},\n volume={12},\n pages={2825--2830},\n year={2011}\n}\n" @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class __lowerCAmelCase ( datasets.Metric ): '''simple docstring''' def __UpperCAmelCase ( self ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { '''predictions''': datasets.Sequence(datasets.Value('''int32''' ) ), '''references''': datasets.Sequence(datasets.Value('''int32''' ) ), } if self.config_name == '''multilabel''' else { '''predictions''': datasets.Value('''int32''' ), '''references''': datasets.Value('''int32''' ), } ) , reference_urls=['''https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html'''] , ) def __UpperCAmelCase ( self , _a , _a , _a=None , _a=1 , _a="binary" , _a=None ): __a = fa_score( _a , _a , labels=_a , pos_label=_a , average=_a , sample_weight=_a ) return {"f1": float(_a ) if score.size == 1 else score}
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : list[int] ) -> List[Any]: if not numbers: return 0 if not isinstance(lowerCAmelCase__ , (list, tuple) ) or not all( isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for number in numbers ): raise ValueError('''numbers must be an iterable of integers''' ) __a = numbers[0] for i in range(1 , len(lowerCAmelCase__ ) ): # update the maximum and minimum subarray products __a = numbers[i] if number < 0: __a = min_till_now, max_till_now __a = max(lowerCAmelCase__ , max_till_now * number ) __a = min(lowerCAmelCase__ , min_till_now * number ) # update the maximum product found till now __a = max(lowerCAmelCase__ , lowerCAmelCase__ ) return max_prod
717
"""simple docstring""" from ..utils import DummyObject, requires_backends class __lowerCAmelCase ( metaclass=__SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = ['onnx'] def __init__( self , *_a , **_a ): requires_backends(self , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] )
65
0
"""simple docstring""" from collections import OrderedDict from ...utils import logging from .auto_factory import _BaseAutoModelClass, _LazyAutoMapping, auto_class_update from .configuration_auto import CONFIG_MAPPING_NAMES lowercase_ = logging.get_logger(__name__) lowercase_ = OrderedDict( [ # Base model mapping ("albert", "FlaxAlbertModel"), ("bart", "FlaxBartModel"), ("beit", "FlaxBeitModel"), ("bert", "FlaxBertModel"), ("big_bird", "FlaxBigBirdModel"), ("blenderbot", "FlaxBlenderbotModel"), ("blenderbot-small", "FlaxBlenderbotSmallModel"), ("clip", "FlaxCLIPModel"), ("distilbert", "FlaxDistilBertModel"), ("electra", "FlaxElectraModel"), ("gpt-sw3", "FlaxGPT2Model"), ("gpt2", "FlaxGPT2Model"), ("gpt_neo", "FlaxGPTNeoModel"), ("gptj", "FlaxGPTJModel"), ("longt5", "FlaxLongT5Model"), ("marian", "FlaxMarianModel"), ("mbart", "FlaxMBartModel"), ("mt5", "FlaxMT5Model"), ("opt", "FlaxOPTModel"), ("pegasus", "FlaxPegasusModel"), ("regnet", "FlaxRegNetModel"), ("resnet", "FlaxResNetModel"), ("roberta", "FlaxRobertaModel"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormModel"), ("roformer", "FlaxRoFormerModel"), ("t5", "FlaxT5Model"), ("vision-text-dual-encoder", "FlaxVisionTextDualEncoderModel"), ("vit", "FlaxViTModel"), ("wav2vec2", "FlaxWav2Vec2Model"), ("whisper", "FlaxWhisperModel"), ("xglm", "FlaxXGLMModel"), ("xlm-roberta", "FlaxXLMRobertaModel"), ] ) lowercase_ = OrderedDict( [ # Model for pre-training mapping ("albert", "FlaxAlbertForPreTraining"), ("bart", "FlaxBartForConditionalGeneration"), ("bert", "FlaxBertForPreTraining"), ("big_bird", "FlaxBigBirdForPreTraining"), ("electra", "FlaxElectraForPreTraining"), ("longt5", "FlaxLongT5ForConditionalGeneration"), ("mbart", "FlaxMBartForConditionalGeneration"), ("mt5", "FlaxMT5ForConditionalGeneration"), ("roberta", "FlaxRobertaForMaskedLM"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForMaskedLM"), ("roformer", "FlaxRoFormerForMaskedLM"), ("t5", "FlaxT5ForConditionalGeneration"), ("wav2vec2", "FlaxWav2Vec2ForPreTraining"), ("whisper", "FlaxWhisperForConditionalGeneration"), ("xlm-roberta", "FlaxXLMRobertaForMaskedLM"), ] ) lowercase_ = OrderedDict( [ # Model for Masked LM mapping ("albert", "FlaxAlbertForMaskedLM"), ("bart", "FlaxBartForConditionalGeneration"), ("bert", "FlaxBertForMaskedLM"), ("big_bird", "FlaxBigBirdForMaskedLM"), ("distilbert", "FlaxDistilBertForMaskedLM"), ("electra", "FlaxElectraForMaskedLM"), ("mbart", "FlaxMBartForConditionalGeneration"), ("roberta", "FlaxRobertaForMaskedLM"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForMaskedLM"), ("roformer", "FlaxRoFormerForMaskedLM"), ("xlm-roberta", "FlaxXLMRobertaForMaskedLM"), ] ) lowercase_ = OrderedDict( [ # Model for Seq2Seq Causal LM mapping ("bart", "FlaxBartForConditionalGeneration"), ("blenderbot", "FlaxBlenderbotForConditionalGeneration"), ("blenderbot-small", "FlaxBlenderbotSmallForConditionalGeneration"), ("encoder-decoder", "FlaxEncoderDecoderModel"), ("longt5", "FlaxLongT5ForConditionalGeneration"), ("marian", "FlaxMarianMTModel"), ("mbart", "FlaxMBartForConditionalGeneration"), ("mt5", "FlaxMT5ForConditionalGeneration"), ("pegasus", "FlaxPegasusForConditionalGeneration"), ("t5", "FlaxT5ForConditionalGeneration"), ] ) lowercase_ = OrderedDict( [ # Model for Image-classsification ("beit", "FlaxBeitForImageClassification"), ("regnet", "FlaxRegNetForImageClassification"), ("resnet", "FlaxResNetForImageClassification"), ("vit", "FlaxViTForImageClassification"), ] ) lowercase_ = OrderedDict( [ ("vision-encoder-decoder", "FlaxVisionEncoderDecoderModel"), ] ) lowercase_ = OrderedDict( [ # Model for Causal LM mapping ("bart", "FlaxBartForCausalLM"), ("bert", "FlaxBertForCausalLM"), ("big_bird", "FlaxBigBirdForCausalLM"), ("electra", "FlaxElectraForCausalLM"), ("gpt-sw3", "FlaxGPT2LMHeadModel"), ("gpt2", "FlaxGPT2LMHeadModel"), ("gpt_neo", "FlaxGPTNeoForCausalLM"), ("gptj", "FlaxGPTJForCausalLM"), ("opt", "FlaxOPTForCausalLM"), ("roberta", "FlaxRobertaForCausalLM"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForCausalLM"), ("xglm", "FlaxXGLMForCausalLM"), ("xlm-roberta", "FlaxXLMRobertaForCausalLM"), ] ) lowercase_ = OrderedDict( [ # Model for Sequence Classification mapping ("albert", "FlaxAlbertForSequenceClassification"), ("bart", "FlaxBartForSequenceClassification"), ("bert", "FlaxBertForSequenceClassification"), ("big_bird", "FlaxBigBirdForSequenceClassification"), ("distilbert", "FlaxDistilBertForSequenceClassification"), ("electra", "FlaxElectraForSequenceClassification"), ("mbart", "FlaxMBartForSequenceClassification"), ("roberta", "FlaxRobertaForSequenceClassification"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForSequenceClassification"), ("roformer", "FlaxRoFormerForSequenceClassification"), ("xlm-roberta", "FlaxXLMRobertaForSequenceClassification"), ] ) lowercase_ = OrderedDict( [ # Model for Question Answering mapping ("albert", "FlaxAlbertForQuestionAnswering"), ("bart", "FlaxBartForQuestionAnswering"), ("bert", "FlaxBertForQuestionAnswering"), ("big_bird", "FlaxBigBirdForQuestionAnswering"), ("distilbert", "FlaxDistilBertForQuestionAnswering"), ("electra", "FlaxElectraForQuestionAnswering"), ("mbart", "FlaxMBartForQuestionAnswering"), ("roberta", "FlaxRobertaForQuestionAnswering"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForQuestionAnswering"), ("roformer", "FlaxRoFormerForQuestionAnswering"), ("xlm-roberta", "FlaxXLMRobertaForQuestionAnswering"), ] ) lowercase_ = OrderedDict( [ # Model for Token Classification mapping ("albert", "FlaxAlbertForTokenClassification"), ("bert", "FlaxBertForTokenClassification"), ("big_bird", "FlaxBigBirdForTokenClassification"), ("distilbert", "FlaxDistilBertForTokenClassification"), ("electra", "FlaxElectraForTokenClassification"), ("roberta", "FlaxRobertaForTokenClassification"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForTokenClassification"), ("roformer", "FlaxRoFormerForTokenClassification"), ("xlm-roberta", "FlaxXLMRobertaForTokenClassification"), ] ) lowercase_ = OrderedDict( [ # Model for Multiple Choice mapping ("albert", "FlaxAlbertForMultipleChoice"), ("bert", "FlaxBertForMultipleChoice"), ("big_bird", "FlaxBigBirdForMultipleChoice"), ("distilbert", "FlaxDistilBertForMultipleChoice"), ("electra", "FlaxElectraForMultipleChoice"), ("roberta", "FlaxRobertaForMultipleChoice"), ("roberta-prelayernorm", "FlaxRobertaPreLayerNormForMultipleChoice"), ("roformer", "FlaxRoFormerForMultipleChoice"), ("xlm-roberta", "FlaxXLMRobertaForMultipleChoice"), ] ) lowercase_ = OrderedDict( [ ("bert", "FlaxBertForNextSentencePrediction"), ] ) lowercase_ = OrderedDict( [ ("speech-encoder-decoder", "FlaxSpeechEncoderDecoderModel"), ("whisper", "FlaxWhisperForConditionalGeneration"), ] ) lowercase_ = OrderedDict( [ ("whisper", "FlaxWhisperForAudioClassification"), ] ) lowercase_ = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_MAPPING_NAMES) lowercase_ = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_PRETRAINING_MAPPING_NAMES) lowercase_ = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_MASKED_LM_MAPPING_NAMES) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMES) lowercase_ = _LazyAutoMapping(CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_CAUSAL_LM_MAPPING_NAMES) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES ) lowercase_ = _LazyAutoMapping( CONFIG_MAPPING_NAMES, FLAX_MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMES ) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Dict = FLAX_MODEL_MAPPING lowercase_ = auto_class_update(FlaxAutoModel) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Dict = FLAX_MODEL_FOR_PRETRAINING_MAPPING lowercase_ = auto_class_update(FlaxAutoModelForPreTraining, head_doc="pretraining") class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Any = FLAX_MODEL_FOR_CAUSAL_LM_MAPPING lowercase_ = auto_class_update(FlaxAutoModelForCausalLM, head_doc="causal language modeling") class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : str = FLAX_MODEL_FOR_MASKED_LM_MAPPING lowercase_ = auto_class_update(FlaxAutoModelForMaskedLM, head_doc="masked language modeling") class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Tuple = FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING lowercase_ = auto_class_update( FlaxAutoModelForSeqaSeqLM, head_doc="sequence-to-sequence language modeling", checkpoint_for_example="t5-base" ) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Tuple = FLAX_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING lowercase_ = auto_class_update( FlaxAutoModelForSequenceClassification, head_doc="sequence classification" ) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : int = FLAX_MODEL_FOR_QUESTION_ANSWERING_MAPPING lowercase_ = auto_class_update(FlaxAutoModelForQuestionAnswering, head_doc="question answering") class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = FLAX_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING lowercase_ = auto_class_update( FlaxAutoModelForTokenClassification, head_doc="token classification" ) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Any = FLAX_MODEL_FOR_MULTIPLE_CHOICE_MAPPING lowercase_ = auto_class_update(FlaxAutoModelForMultipleChoice, head_doc="multiple choice") class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : List[str] = FLAX_MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING lowercase_ = auto_class_update( FlaxAutoModelForNextSentencePrediction, head_doc="next sentence prediction" ) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : Optional[int] = FLAX_MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING lowercase_ = auto_class_update( FlaxAutoModelForImageClassification, head_doc="image classification" ) class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : int = FLAX_MODEL_FOR_VISION_2_SEQ_MAPPING lowercase_ = auto_class_update(FlaxAutoModelForVisionaSeq, head_doc="vision-to-text modeling") class __lowerCAmelCase ( _BaseAutoModelClass ): '''simple docstring''' __UpperCAmelCase : int = FLAX_MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING lowercase_ = auto_class_update( FlaxAutoModelForSpeechSeqaSeq, head_doc="sequence-to-sequence speech-to-text modeling" )
718
"""simple docstring""" 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 lowercase_ = 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 __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=16 , _a=13 , _a=7 , _a=14 , _a=10 , _a=19 , _a=5 , _a=4 , _a=True , _a=16 , _a=2 , _a=4 , _a=4 , _a="gelu" , _a=0.1 , _a=0.1 , _a=[1, 2, 3, 4, 5] , _a=25 , _a=5 , ): __a = d_model __a = parent __a = batch_size __a = prediction_length __a = context_length __a = cardinality __a = num_time_features __a = lags_sequence __a = embedding_dimension __a = is_training __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = context_length __a = prediction_length + label_length __a = label_length __a = moving_average __a = autocorrelation_factor def __UpperCAmelCase ( self ): 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 __UpperCAmelCase ( self , _a ): __a = config.context_length + max(config.lags_sequence ) __a = ids_tensor([self.batch_size, 1] , config.cardinality[0] ) __a = floats_tensor([self.batch_size, _past_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, _past_length] ) __a = floats_tensor([self.batch_size, _past_length] ) > 0.5 # decoder inputs __a = floats_tensor([self.batch_size, config.prediction_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, config.prediction_length] ) __a = { '''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 __UpperCAmelCase ( self ): __a = self.get_config() __a = self.prepare_autoformer_inputs_dict(_a ) return config, inputs_dict def __UpperCAmelCase ( self ): __a , __a = self.prepare_config_and_inputs() return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = AutoformerModel(config=_a ).to(_a ).eval() __a = model(**_a ) __a = outputs.encoder_last_hidden_state __a = outputs.last_hidden_state with tempfile.TemporaryDirectory() as tmpdirname: __a = model.get_encoder() encoder.save_pretrained(_a ) __a = AutoformerEncoder.from_pretrained(_a ).to(_a ) __a , __a , __a , __a , __a = model.create_network_inputs(**_a ) __a , __a = model.decomposition_layer(transformer_inputs[:, : config.context_length, ...] ) __a = torch.cat( (transformer_inputs[:, : config.context_length, ...], feature[:, : config.context_length, ...]) , dim=-1 , ) __a = encoder(inputs_embeds=_a )[0] self.parent.assertTrue((encoder_last_hidden_state_a - encoder_last_hidden_state).abs().max().item() < 1E-3 ) __a = ( torch.mean(transformer_inputs[:, : config.context_length, ...] , dim=1 ) .unsqueeze(1 ) .repeat(1 , config.prediction_length , 1 ) ) __a = torch.zeros( [transformer_inputs.shape[0], config.prediction_length, transformer_inputs.shape[2]] , device=enc_input.device , ) __a = torch.cat( ( torch.cat((seasonal_input[:, -config.label_length :, ...], zeros) , dim=1 ), feature[:, config.context_length - config.label_length :, ...], ) , dim=-1 , ) __a = 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: __a = model.get_decoder() decoder.save_pretrained(_a ) __a = AutoformerDecoder.from_pretrained(_a ).to(_a ) __a = decoder( trend=_a , inputs_embeds=_a , encoder_hidden_states=_a , )[0] self.parent.assertTrue((last_hidden_state_a - last_hidden_state).abs().max().item() < 1E-3 ) @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Any = (AutoformerModel, AutoformerForPrediction) if is_torch_available() else () __UpperCAmelCase : int = (AutoformerForPrediction,) if is_torch_available() else () __UpperCAmelCase : Any = {'feature-extraction': AutoformerModel} if is_torch_available() else {} __UpperCAmelCase : int = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False __UpperCAmelCase : Any = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False def __UpperCAmelCase ( self ): __a = AutoformerModelTester(self ) __a = ConfigTester(self , config_class=_a , has_text_modality=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs() for model_class in self.all_model_classes: __a = model_class(_a ) with tempfile.TemporaryDirectory() as tmpdirname: model.save_pretrained(_a ) __a , __a = model_class.from_pretrained(_a , output_loading_info=_a ) self.assertEqual(info['''missing_keys'''] , [] ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.check_encoder_decoder_model_standalone(*_a ) @unittest.skip(reason='''Model has no tokens embeddings''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = inspect.signature(getattr(_a , '''forward''' ) ) # The main input is the name of the argument after `self` __a = list(model_signature.parameters.keys() )[1] self.assertEqual(AutoformerModel.main_input_name , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ) __a = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic __a = [*signature.parameters.keys()] __a = [ '''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(_a )] , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() __a = True __a = getattr(self.model_tester , '''seq_length''' , _a ) __a = getattr(self.model_tester , '''decoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''encoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''d_model''' , _a ) __a = getattr(self.model_tester , '''num_attention_heads''' , _a ) __a = d_model // num_attention_heads for model_class in self.all_model_classes: __a = True __a = False __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) # check that output_attentions also work using config del inputs_dict["output_attentions"] __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) self.assertListEqual( list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, encoder_seq_length, dim] , ) __a = len(_a ) __a = 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(_a , _a ) # decoder attentions __a = outputs.decoder_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = outputs.cross_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = True __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) self.assertEqual(out_len + 2 , len(_a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , 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 __UpperCAmelCase ( self ): super().test_retain_grad_hidden_states_attentions() def lowercase ( lowerCAmelCase__ : Optional[int]="train-batch.pt" ) -> List[str]: __a = hf_hub_download(repo_id='''hf-internal-testing/tourism-monthly-batch''' , filename=lowerCAmelCase__ , repo_type='''dataset''' ) __a = torch.load(lowerCAmelCase__ , map_location=lowerCAmelCase__ ) return batch @require_torch @slow class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = AutoformerModel.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch() with torch.no_grad(): __a = 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] __a = torch.Size( (64, model.config.prediction_length + model.config.label_length, model.config.feature_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[0.3593, -1.3398, 0.6330], [0.2279, 1.5396, -0.1792], [0.0450, 1.3225, -0.2335]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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 __a = torch.Size((64, model.config.context_length, model.config.d_model) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[-0.0734, -0.9036, 0.8358], [4.7186, 2.4113, 1.9581], [1.7953, 2.3558, 1.2970]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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'''] , ) __a = torch.Size((64, model.config.num_parallel_samples, model.config.prediction_length) ) self.assertEqual(outputs.sequences.shape , _a ) __a = torch.tensor([3130.6763, 4056.5293, 7053.0786] , device=_a ) __a = outputs.sequences.mean(dim=1 ) self.assertTrue(torch.allclose(mean_prediction[0, -3:] , _a , rtol=1E-1 ) )
65
0
"""simple docstring""" from collections import OrderedDict from typing import TYPE_CHECKING, Any, List, Mapping, Optional from packaging import version if TYPE_CHECKING: from ... import PreTrainedTokenizer, TensorType from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfigWithPast, PatchingSpec from ...utils import is_torch_available, logging lowercase_ : Optional[Any] = logging.get_logger(__name__) lowercase_ : Tuple = { """bigscience/bloom""": """https://huggingface.co/bigscience/bloom/resolve/main/config.json""", """bigscience/bloom-560m""": """https://huggingface.co/bigscience/bloom-560m/blob/main/config.json""", """bigscience/bloom-1b1""": """https://huggingface.co/bigscience/bloom-1b1/blob/main/config.json""", """bigscience/bloom-1b7""": """https://huggingface.co/bigscience/bloom-1b7/blob/main/config.json""", """bigscience/bloom-3b""": """https://huggingface.co/bigscience/bloom-3b/blob/main/config.json""", """bigscience/bloom-7b1""": """https://huggingface.co/bigscience/bloom-7b1/blob/main/config.json""", } class __lowerCAmelCase ( lowercase_ ): '''simple docstring''' __UpperCAmelCase : List[Any] = '''bloom''' __UpperCAmelCase : str = ['''past_key_values'''] __UpperCAmelCase : List[str] = { '''num_hidden_layers''': '''n_layer''', '''num_attention_heads''': '''n_head''', } def __init__( self , _a=250_880 , _a=64 , _a=2 , _a=8 , _a=1E-5 , _a=0.02 , _a=True , _a=1 , _a=2 , _a=False , _a=0.0 , _a=0.0 , _a=1 , _a=False , **_a , ): __a = vocab_size # Backward compatibility with n_embed kwarg __a = kwargs.pop('''n_embed''' , _a ) __a = hidden_size if n_embed is None else n_embed __a = n_layer __a = n_head __a = layer_norm_epsilon __a = initializer_range __a = use_cache __a = pretraining_tp __a = apply_residual_connection_post_layernorm __a = hidden_dropout __a = attention_dropout __a = bos_token_id __a = eos_token_id __a = slow_but_exact super().__init__(bos_token_id=_a , eos_token_id=_a , **_a ) class __lowerCAmelCase ( lowercase_ ): '''simple docstring''' __UpperCAmelCase : List[str] = version.parse('1.12' ) def __init__( self , _a , _a = "default" , _a = None , _a = False , ): super().__init__(_a , task=_a , patching_specs=_a , use_past=_a ) if not getattr(self._config , '''pad_token_id''' , _a ): # TODO: how to do that better? __a = 0 @property def __UpperCAmelCase ( self ): __a = OrderedDict({'''input_ids''': {0: '''batch''', 1: '''sequence'''}} ) if self.use_past: # BLOOM stores values on dynamic axis 2. For more details see: https://github.com/huggingface/transformers/pull/18344 self.fill_with_past_key_values_(_a , direction='''inputs''' , inverted_values_shape=_a ) __a = {0: '''batch''', 1: '''past_sequence + sequence'''} else: __a = {0: '''batch''', 1: '''sequence'''} return common_inputs @property def __UpperCAmelCase ( self ): return self._config.n_layer @property def __UpperCAmelCase ( self ): return self._config.n_head @property def __UpperCAmelCase ( self ): return 1E-3 def __UpperCAmelCase ( self , _a , _a = -1 , _a = -1 , _a = False , _a = None , ): __a = super(_a , self ).generate_dummy_inputs( _a , batch_size=_a , seq_length=_a , is_pair=_a , framework=_a ) # We need to order the input in the way they appears in the forward() __a = OrderedDict({'''input_ids''': common_inputs['''input_ids''']} ) # Need to add the past_keys if self.use_past: if not is_torch_available(): raise ValueError('''Cannot generate dummy past_keys inputs without PyTorch installed.''' ) else: import torch __a , __a = common_inputs['''input_ids'''].shape # Not using the same length for past_key_values __a = seqlen + 2 __a = self._config.hidden_size // self.num_attention_heads __a = ( batch * self.num_attention_heads, head_dim, past_key_values_length, ) __a = ( batch * self.num_attention_heads, past_key_values_length, head_dim, ) __a = [ (torch.zeros(_a ), torch.zeros(_a )) for _ in range(self.num_layers ) ] __a = common_inputs['''attention_mask'''] if self.use_past: __a = ordered_inputs['''attention_mask'''].dtype __a = torch.cat( [ordered_inputs['''attention_mask'''], torch.ones(_a , _a , dtype=_a )] , dim=1 ) return ordered_inputs @property def __UpperCAmelCase ( self ): return 13
719
"""simple docstring""" import inspect from typing import Optional, Union import numpy as np import PIL import torch from torch.nn import functional as F from torchvision import transforms from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DiffusionPipeline, DPMSolverMultistepScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, randn_tensor, ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> Any: if isinstance(lowerCAmelCase__ , torch.Tensor ): return image elif isinstance(lowerCAmelCase__ , PIL.Image.Image ): __a = [image] if isinstance(image[0] , PIL.Image.Image ): __a = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['''lanczos'''] ) )[None, :] for i in image] __a = np.concatenate(lowerCAmelCase__ , axis=0 ) __a = np.array(lowerCAmelCase__ ).astype(np.floataa ) / 2_55.0 __a = image.transpose(0 , 3 , 1 , 2 ) __a = 2.0 * image - 1.0 __a = torch.from_numpy(lowerCAmelCase__ ) elif isinstance(image[0] , torch.Tensor ): __a = torch.cat(lowerCAmelCase__ , dim=0 ) return image def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=0.99_95 ) -> int: if not isinstance(lowerCAmelCase__ , np.ndarray ): __a = True __a = va.device __a = va.cpu().numpy() __a = va.cpu().numpy() __a = np.sum(va * va / (np.linalg.norm(lowerCAmelCase__ ) * np.linalg.norm(lowerCAmelCase__ )) ) if np.abs(lowerCAmelCase__ ) > DOT_THRESHOLD: __a = (1 - t) * va + t * va else: __a = np.arccos(lowerCAmelCase__ ) __a = np.sin(lowerCAmelCase__ ) __a = theta_a * t __a = np.sin(lowerCAmelCase__ ) __a = np.sin(theta_a - theta_t ) / sin_theta_a __a = sin_theta_t / sin_theta_a __a = sa * va + sa * va if inputs_are_torch: __a = torch.from_numpy(lowerCAmelCase__ ).to(lowerCAmelCase__ ) return va def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> int: __a = F.normalize(lowerCAmelCase__ , dim=-1 ) __a = F.normalize(lowerCAmelCase__ , dim=-1 ) return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> List[str]: for param in model.parameters(): __a = value class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a , _a , _a , _a=None , _a=None , _a=None , ): super().__init__() self.register_modules( vae=_a , text_encoder=_a , clip_model=_a , tokenizer=_a , unet=_a , scheduler=_a , feature_extractor=_a , coca_model=_a , coca_tokenizer=_a , coca_transform=_a , ) __a = ( feature_extractor.size if isinstance(feature_extractor.size , _a ) else feature_extractor.size['''shortest_edge'''] ) __a = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std ) set_requires_grad(self.text_encoder , _a ) set_requires_grad(self.clip_model , _a ) def __UpperCAmelCase ( self , _a = "auto" ): if slice_size == "auto": # half the attention head size is usually a good trade-off between # speed and memory __a = self.unet.config.attention_head_dim // 2 self.unet.set_attention_slice(_a ) def __UpperCAmelCase ( self ): self.enable_attention_slicing(_a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self , _a , _a , _a ): # get the original timestep using init_timestep __a = min(int(num_inference_steps * strength ) , _a ) __a = max(num_inference_steps - init_timestep , 0 ) __a = self.scheduler.timesteps[t_start:] return timesteps, num_inference_steps - t_start def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a=None ): if not isinstance(_a , torch.Tensor ): raise ValueError(f'''`image` has to be of type `torch.Tensor` but is {type(_a )}''' ) __a = image.to(device=_a , dtype=_a ) if isinstance(_a , _a ): __a = [ self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_a ) ] __a = torch.cat(_a , dim=0 ) else: __a = self.vae.encode(_a ).latent_dist.sample(_a ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 0.1_8215 * init_latents __a = init_latents.repeat_interleave(_a , dim=0 ) __a = randn_tensor(init_latents.shape , generator=_a , device=_a , dtype=_a ) # get latents __a = self.scheduler.add_noise(_a , _a , _a ) __a = init_latents return latents def __UpperCAmelCase ( self , _a ): __a = self.coca_transform(_a ).unsqueeze(0 ) with torch.no_grad(), torch.cuda.amp.autocast(): __a = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) ) __a = self.coca_tokenizer.decode(generated[0].cpu().numpy() ) return generated.split('''<end_of_text>''' )[0].replace('''<start_of_text>''' , '''''' ).rstrip(''' .,''' ) def __UpperCAmelCase ( self , _a , _a ): __a = self.feature_extractor.preprocess(_a ) __a = torch.from_numpy(clip_image_input['''pixel_values'''][0] ).unsqueeze(0 ).to(self.device ).half() __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = image_embeddings_clip.repeat_interleave(_a , dim=0 ) return image_embeddings_clip @torch.enable_grad() def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , ): __a = latents.detach().requires_grad_() __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ): __a = self.scheduler.alphas_cumprod[timestep] __a = 1 - alpha_prod_t # compute predicted original sample from predicted noise also called # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf __a = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5 __a = torch.sqrt(_a ) __a = pred_original_sample * (fac) + latents * (1 - fac) elif isinstance(self.scheduler , _a ): __a = self.scheduler.sigmas[index] __a = latents - sigma * noise_pred else: raise ValueError(f'''scheduler type {type(self.scheduler )} not supported''' ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * sample __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = transforms.Resize(self.feature_extractor_size )(_a ) __a = self.normalize(_a ).to(latents.dtype ) __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = spherical_dist_loss(_a , _a ).mean() * clip_guidance_scale __a = -torch.autograd.grad(_a , _a )[0] if isinstance(self.scheduler , _a ): __a = latents.detach() + grads * (sigma**2) __a = noise_pred_original else: __a = noise_pred_original - torch.sqrt(_a ) * grads return noise_pred, latents @torch.no_grad() def __call__( self , _a , _a , _a = None , _a = None , _a = 512 , _a = 512 , _a = 0.6 , _a = 50 , _a = 7.5 , _a = 1 , _a = 0.0 , _a = 100 , _a = None , _a = "pil" , _a = True , _a = 0.8 , _a = 0.1 , _a = 0.1 , ): if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError(f'''You have passed {batch_size} batch_size, but only {len(_a )} generators.''' ) if height % 8 != 0 or width % 8 != 0: raise ValueError(f'''`height` and `width` have to be divisible by 8 but are {height} and {width}.''' ) if isinstance(_a , torch.Generator ) and batch_size > 1: __a = [generator] + [None] * (batch_size - 1) __a = [ ('''model''', self.coca_model is None), ('''tokenizer''', self.coca_tokenizer is None), ('''transform''', self.coca_transform is None), ] __a = [x[0] for x in coca_is_none if x[1]] __a = ''', '''.join(_a ) # generate prompts with coca model if prompt is None if content_prompt is None: if len(_a ): raise ValueError( f'''Content prompt is None and CoCa [{coca_is_none_str}] is None.''' f'''Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) if style_prompt is None: if len(_a ): raise ValueError( f'''Style prompt is None and CoCa [{coca_is_none_str}] is None.''' f''' Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) # get prompt text embeddings for content and style __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0] __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0] __a = slerp(_a , _a , _a ) # duplicate text embeddings for each generation per prompt __a = text_embeddings.repeat_interleave(_a , dim=0 ) # set timesteps __a = '''offset''' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() ) __a = {} if accepts_offset: __a = 1 self.scheduler.set_timesteps(_a , **_a ) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand self.scheduler.timesteps.to(self.device ) __a , __a = self.get_timesteps(_a , _a , self.device ) __a = timesteps[:1].repeat(_a ) # Preprocess image __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = slerp(_a , _a , _a ) if clip_guidance_scale > 0: __a = self.get_clip_image_embeddings(_a , _a ) __a = self.get_clip_image_embeddings(_a , _a ) __a = slerp( _a , _a , _a ) # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` # corresponds to doing no classifier free guidance. __a = guidance_scale > 1.0 # get unconditional embeddings for classifier free guidance if do_classifier_free_guidance: __a = content_text_input.input_ids.shape[-1] __a = self.tokenizer([''''''] , padding='''max_length''' , max_length=_a , return_tensors='''pt''' ) __a = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0] # duplicate unconditional embeddings for each generation per prompt __a = uncond_embeddings.repeat_interleave(_a , dim=0 ) # For classifier free guidance, we need to do two forward passes. # Here we concatenate the unconditional and text embeddings into a single batch # to avoid doing two forward passes __a = torch.cat([uncond_embeddings, text_embeddings] ) # get the initial random noise unless the user supplied it # Unlike in other pipelines, latents need to be generated in the target device # for 1-to-1 results reproducibility with the CompVis implementation. # However this currently doesn't work in `mps`. __a = (batch_size, self.unet.config.in_channels, height // 8, width // 8) __a = text_embeddings.dtype if latents is None: if self.device.type == "mps": # randn does not work reproducibly on mps __a = torch.randn(_a , generator=_a , device='''cpu''' , dtype=_a ).to( self.device ) else: __a = torch.randn(_a , generator=_a , device=self.device , dtype=_a ) else: if latents.shape != latents_shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {latents_shape}''' ) __a = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler __a = latents * self.scheduler.init_noise_sigma # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 # and should be between [0, 1] __a = '''eta''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) __a = {} if accepts_eta: __a = eta # check if the scheduler accepts generator __a = '''generator''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) if accepts_generator: __a = generator with self.progress_bar(total=_a ): for i, t in enumerate(_a ): # expand the latents if we are doing classifier free guidance __a = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample # perform classifier free guidance if do_classifier_free_guidance: __a , __a = noise_pred.chunk(2 ) __a = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) # perform clip guidance if clip_guidance_scale > 0: __a = ( text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings ) __a , __a = self.cond_fn( _a , _a , _a , _a , _a , _a , _a , ) # compute the previous noisy sample x_t -> x_t-1 __a = self.scheduler.step(_a , _a , _a , **_a ).prev_sample # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * latents __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image, None) return StableDiffusionPipelineOutput(images=_a , nsfw_content_detected=_a )
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: __a = abs(lowerCamelCase__ ) __a = 0 while n > 0: res += n % 10 n //= 10 return res def lowercase ( lowerCAmelCase__ : Any ) -> int: __a = abs(lowerCamelCase__ ) return n if n < 10 else n % 10 + sum_of_digits(n // 10 ) def lowercase ( lowerCAmelCase__ : List[str] ) -> int: return sum(int(lowerCamelCase__ ) for c in str(abs(lowerCamelCase__ ) ) ) def lowercase ( ) -> None: from collections.abc import Callable from timeit import timeit def benchmark_a_function(lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Union[str, Any] ) -> None: __a = f'''{func.__name__}({value})''' __a = timeit(f'''__main__.{call}''' , setup='''import __main__''' ) print(f'''{call:56} = {func(lowerCamelCase__ )} -- {timing:.4f} seconds''' ) for value in (262144, 1125899906842624, 1267650600228229401496703205376): for func in (sum_of_digits, sum_of_digits_recursion, sum_of_digits_compact): benchmark_a_function(lowerCamelCase__ , lowerCamelCase__ ) print() if __name__ == "__main__": import doctest doctest.testmod() benchmark()
720
"""simple docstring""" # tests directory-specific settings - this file is run automatically # by pytest before any tests are run import sys import warnings from os.path import abspath, dirname, join # allow having multiple repository checkouts and not needing to remember to rerun # 'pip install -e .[dev]' when switching between checkouts and running tests. lowercase_ = abspath(join(dirname(dirname(dirname(__file__))), "src")) sys.path.insert(1, git_repo_path) # silence FutureWarning warnings in tests since often we can't act on them until # they become normal warnings - i.e. the tests still need to test the current functionality warnings.simplefilter(action="ignore", category=FutureWarning) def lowercase ( lowerCAmelCase__ : List[Any] ) -> str: from transformers.testing_utils import pytest_addoption_shared pytest_addoption_shared(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Union[str, Any]: from transformers.testing_utils import pytest_terminal_summary_main __a = terminalreporter.config.getoption('''--make-reports''' ) if make_reports: pytest_terminal_summary_main(lowerCAmelCase__ , id=lowerCAmelCase__ )
65
0
"""simple docstring""" import argparse import json from pathlib import Path import requests import torch from huggingface_hub import cached_download, hf_hub_url from PIL import Image from transformers import DPTConfig, DPTForDepthEstimation, DPTForSemanticSegmentation, DPTImageProcessor from transformers.utils import logging logging.set_verbosity_info() lowercase_ = logging.get_logger(__name__) def lowercase ( lowerCAmelCase__ : str ) -> List[str]: __a = DPTConfig() if "large" in checkpoint_url: __a = 1024 __a = 4096 __a = 24 __a = 16 __a = [5, 11, 17, 23] __a = [256, 512, 1024, 1024] __a = (1, 384, 384) if "ade" in checkpoint_url: __a = True __a = 150 __a = '''huggingface/label-files''' __a = '''ade20k-id2label.json''' __a = json.load(open(cached_download(hf_hub_url(lowerCAmelCase__ , lowerCAmelCase__ , repo_type='''dataset''' ) ) , '''r''' ) ) __a = {int(lowerCAmelCase__ ): v for k, v in idalabel.items()} __a = idalabel __a = {v: k for k, v in idalabel.items()} __a = [1, 150, 480, 480] return config, expected_shape def lowercase ( lowerCAmelCase__ : List[str] ) -> Any: __a = ['''pretrained.model.head.weight''', '''pretrained.model.head.bias'''] for k in ignore_keys: state_dict.pop(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Any: if ( "pretrained.model" in name and "cls_token" not in name and "pos_embed" not in name and "patch_embed" not in name ): __a = name.replace('''pretrained.model''' , '''dpt.encoder''' ) if "pretrained.model" in name: __a = name.replace('''pretrained.model''' , '''dpt.embeddings''' ) if "patch_embed" in name: __a = name.replace('''patch_embed''' , '''patch_embeddings''' ) if "pos_embed" in name: __a = name.replace('''pos_embed''' , '''position_embeddings''' ) if "attn.proj" in name: __a = name.replace('''attn.proj''' , '''attention.output.dense''' ) if "proj" in name and "project" not in name: __a = name.replace('''proj''' , '''projection''' ) if "blocks" in name: __a = name.replace('''blocks''' , '''layer''' ) if "mlp.fc1" in name: __a = name.replace('''mlp.fc1''' , '''intermediate.dense''' ) if "mlp.fc2" in name: __a = name.replace('''mlp.fc2''' , '''output.dense''' ) if "norm1" in name: __a = name.replace('''norm1''' , '''layernorm_before''' ) if "norm2" in name: __a = name.replace('''norm2''' , '''layernorm_after''' ) if "scratch.output_conv" in name: __a = name.replace('''scratch.output_conv''' , '''head''' ) if "scratch" in name: __a = name.replace('''scratch''' , '''neck''' ) if "layer1_rn" in name: __a = name.replace('''layer1_rn''' , '''convs.0''' ) if "layer2_rn" in name: __a = name.replace('''layer2_rn''' , '''convs.1''' ) if "layer3_rn" in name: __a = name.replace('''layer3_rn''' , '''convs.2''' ) if "layer4_rn" in name: __a = name.replace('''layer4_rn''' , '''convs.3''' ) if "refinenet" in name: __a = int(name[len('''neck.refinenet''' ) : len('''neck.refinenet''' ) + 1] ) # tricky here: we need to map 4 to 0, 3 to 1, 2 to 2 and 1 to 3 __a = name.replace(f'''refinenet{layer_idx}''' , f'''fusion_stage.layers.{abs(layer_idx-4 )}''' ) if "out_conv" in name: __a = name.replace('''out_conv''' , '''projection''' ) if "resConfUnit1" in name: __a = name.replace('''resConfUnit1''' , '''residual_layer1''' ) if "resConfUnit2" in name: __a = name.replace('''resConfUnit2''' , '''residual_layer2''' ) if "conv1" in name: __a = name.replace('''conv1''' , '''convolution1''' ) if "conv2" in name: __a = name.replace('''conv2''' , '''convolution2''' ) # readout blocks if "pretrained.act_postprocess1.0.project.0" in name: __a = name.replace('''pretrained.act_postprocess1.0.project.0''' , '''neck.reassemble_stage.readout_projects.0.0''' ) if "pretrained.act_postprocess2.0.project.0" in name: __a = name.replace('''pretrained.act_postprocess2.0.project.0''' , '''neck.reassemble_stage.readout_projects.1.0''' ) if "pretrained.act_postprocess3.0.project.0" in name: __a = name.replace('''pretrained.act_postprocess3.0.project.0''' , '''neck.reassemble_stage.readout_projects.2.0''' ) if "pretrained.act_postprocess4.0.project.0" in name: __a = name.replace('''pretrained.act_postprocess4.0.project.0''' , '''neck.reassemble_stage.readout_projects.3.0''' ) # resize blocks if "pretrained.act_postprocess1.3" in name: __a = name.replace('''pretrained.act_postprocess1.3''' , '''neck.reassemble_stage.layers.0.projection''' ) if "pretrained.act_postprocess1.4" in name: __a = name.replace('''pretrained.act_postprocess1.4''' , '''neck.reassemble_stage.layers.0.resize''' ) if "pretrained.act_postprocess2.3" in name: __a = name.replace('''pretrained.act_postprocess2.3''' , '''neck.reassemble_stage.layers.1.projection''' ) if "pretrained.act_postprocess2.4" in name: __a = name.replace('''pretrained.act_postprocess2.4''' , '''neck.reassemble_stage.layers.1.resize''' ) if "pretrained.act_postprocess3.3" in name: __a = name.replace('''pretrained.act_postprocess3.3''' , '''neck.reassemble_stage.layers.2.projection''' ) if "pretrained.act_postprocess4.3" in name: __a = name.replace('''pretrained.act_postprocess4.3''' , '''neck.reassemble_stage.layers.3.projection''' ) if "pretrained.act_postprocess4.4" in name: __a = name.replace('''pretrained.act_postprocess4.4''' , '''neck.reassemble_stage.layers.3.resize''' ) if "pretrained" in name: __a = name.replace('''pretrained''' , '''dpt''' ) if "bn" in name: __a = name.replace('''bn''' , '''batch_norm''' ) if "head" in name: __a = name.replace('''head''' , '''head.head''' ) if "encoder.norm" in name: __a = name.replace('''encoder.norm''' , '''layernorm''' ) if "auxlayer" in name: __a = name.replace('''auxlayer''' , '''auxiliary_head.head''' ) return name def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : str ) -> Optional[int]: for i in range(config.num_hidden_layers ): # read in weights + bias of input projection layer (in timm, this is a single matrix + bias) __a = state_dict.pop(f'''dpt.encoder.layer.{i}.attn.qkv.weight''' ) __a = state_dict.pop(f'''dpt.encoder.layer.{i}.attn.qkv.bias''' ) # next, add query, keys and values (in that order) to the state dict __a = in_proj_weight[: config.hidden_size, :] __a = in_proj_bias[: config.hidden_size] __a = in_proj_weight[ config.hidden_size : config.hidden_size * 2, : ] __a = in_proj_bias[ config.hidden_size : config.hidden_size * 2 ] __a = in_proj_weight[ -config.hidden_size :, : ] __a = in_proj_bias[-config.hidden_size :] def lowercase ( ) -> Tuple: __a = '''http://images.cocodataset.org/val2017/000000039769.jpg''' __a = Image.open(requests.get(lowerCAmelCase__ , stream=lowerCAmelCase__ ).raw ) return im @torch.no_grad() def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : List[str] ) -> Optional[Any]: __a , __a = get_dpt_config(lowerCAmelCase__ ) # load original state_dict from URL __a = torch.hub.load_state_dict_from_url(lowerCAmelCase__ , map_location='''cpu''' ) # remove certain keys remove_ignore_keys_(lowerCAmelCase__ ) # rename keys for key in state_dict.copy().keys(): __a = state_dict.pop(lowerCAmelCase__ ) __a = val # read in qkv matrices read_in_q_k_v(lowerCAmelCase__ , lowerCAmelCase__ ) # load HuggingFace model __a = DPTForSemanticSegmentation(lowerCAmelCase__ ) if '''ade''' in checkpoint_url else DPTForDepthEstimation(lowerCAmelCase__ ) model.load_state_dict(lowerCAmelCase__ ) model.eval() # Check outputs on an image __a = 480 if '''ade''' in checkpoint_url else 384 __a = DPTImageProcessor(size=lowerCAmelCase__ ) __a = prepare_img() __a = image_processor(lowerCAmelCase__ , return_tensors='''pt''' ) # forward pass __a = model(**lowerCAmelCase__ ).logits if '''ade''' in checkpoint_url else model(**lowerCAmelCase__ ).predicted_depth # Assert logits __a = torch.tensor([[6.31_99, 6.36_29, 6.41_48], [6.38_50, 6.36_15, 6.41_66], [6.35_19, 6.31_76, 6.35_75]] ) if "ade" in checkpoint_url: __a = torch.tensor([[4.04_80, 4.24_20, 4.43_60], [4.31_24, 4.56_93, 4.82_61], [4.57_68, 4.89_65, 5.21_63]] ) assert outputs.shape == torch.Size(lowerCAmelCase__ ) assert ( torch.allclose(outputs[0, 0, :3, :3] , lowerCAmelCase__ , atol=1e-4 ) if "ade" in checkpoint_url else torch.allclose(outputs[0, :3, :3] , lowerCAmelCase__ ) ) Path(lowerCAmelCase__ ).mkdir(exist_ok=lowerCAmelCase__ ) print(f'''Saving model to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowerCAmelCase__ ) print(f'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowerCAmelCase__ ) if push_to_hub: print('''Pushing model to hub...''' ) model.push_to_hub( repo_path_or_name=Path(lowerCAmelCase__ , lowerCAmelCase__ ) , organization='''nielsr''' , commit_message='''Add model''' , use_temp_dir=lowerCAmelCase__ , ) image_processor.push_to_hub( repo_path_or_name=Path(lowerCAmelCase__ , lowerCAmelCase__ ) , organization='''nielsr''' , commit_message='''Add image processor''' , use_temp_dir=lowerCAmelCase__ , ) if __name__ == "__main__": lowercase_ = argparse.ArgumentParser() # Required parameters parser.add_argument( "--checkpoint_url", default="https://github.com/intel-isl/DPT/releases/download/1_0/dpt_large-midas-2f21e586.pt", type=str, help="URL of the original DPT checkpoint you\'d like to convert.", ) parser.add_argument( "--pytorch_dump_folder_path", default=None, type=str, required=True, help="Path to the output PyTorch model directory.", ) parser.add_argument( "--push_to_hub", action="store_true", ) parser.add_argument( "--model_name", default="dpt-large", type=str, help="Name of the model, in case you\'re pushing to the hub.", ) lowercase_ = parser.parse_args() convert_dpt_checkpoint(args.checkpoint_url, args.pytorch_dump_folder_path, args.push_to_hub, args.model_name)
721
"""simple docstring""" 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. lowercase_ = {"LayoutLMv2Config", "LayoutLMv3Config"} @is_pipeline_test class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING __UpperCAmelCase : Any = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING if model_mapping is not None: __UpperCAmelCase : Any = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP} if tf_model_mapping is not None: __UpperCAmelCase : Dict = { config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP } def __UpperCAmelCase ( self , _a , _a , _a ): __a = ZeroShotClassificationPipeline( model=_a , tokenizer=_a , candidate_labels=['''polics''', '''health'''] ) return classifier, ["Who are you voting for in 2020?", "My stomach hurts."] def __UpperCAmelCase ( self , _a , _a ): __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # No kwarg __a = classifier('''Who are you voting for in 2020?''' , ['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics, public health''' ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health'''] ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''This text is about {}''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # https://github.com/huggingface/transformers/issues/13846 __a = classifier(['''I am happy'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(1 ) ] , ) __a = classifier(['''I am happy''', '''I am sad'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(2 ) ] , ) with self.assertRaises(_a ): classifier('''''' , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier(_a , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels='''''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels=_a ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''Not formatting template''' , ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template=_a , ) self.run_entailment_id(_a ) def __UpperCAmelCase ( self , _a ): __a = zero_shot_classifier.model.config __a = config.labelaid __a = zero_shot_classifier.entailment_id __a = {'''LABEL_0''': 0, '''LABEL_1''': 1, '''LABEL_2''': 2} self.assertEqual(zero_shot_classifier.entailment_id , -1 ) __a = {'''entailment''': 0, '''neutral''': 1, '''contradiction''': 2} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 0, '''NON-ENTAIL''': 1} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 2, '''NEUTRAL''': 1, '''CONTR''': 0} self.assertEqual(zero_shot_classifier.entailment_id , 2 ) __a = original_labelaid self.assertEqual(_a , zero_shot_classifier.entailment_id ) @require_torch def __UpperCAmelCase ( self ): __a = 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 __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''pt''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @require_tf def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''tf''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @slow @require_torch def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''pt''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , ) @slow @require_tf def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''tf''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , )
65
0
"""simple docstring""" import argparse import json import os import tensorstore as ts import torch from flax import serialization from flax.traverse_util import flatten_dict, unflatten_dict from tensorflow.io import gfile from transformers.modeling_utils import dtype_byte_size from transformers.models.switch_transformers.convert_switch_transformers_original_flax_checkpoint_to_pytorch import ( rename_keys, ) from transformers.utils import WEIGHTS_INDEX_NAME, WEIGHTS_NAME from transformers.utils.hub import convert_file_size_to_int def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Dict: if flax_key_tuple[-1] == "kernel" and flax_tensor.ndim == 3: # expert layer __a = flax_key_tuple[:-1] + ('''weight''',) __a = torch.permute(lowerCAmelCase__ , (0, 2, 1) ) elif flax_key_tuple[-1] == "kernel" and ".".join(lowerCAmelCase__ ): # linear layer __a = flax_key_tuple[:-1] + ('''weight''',) __a = flax_tensor.T elif flax_key_tuple[-1] in ["scale", "embedding"]: __a = flax_key_tuple[:-1] + ('''weight''',) return flax_key_tuple, flax_tensor def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Dict ) -> str: if "metadata" in layer: __a = layer.split('''metadata''' ) __a = ''''''.join(split_layer[0] )[:-1] __a = [tuple(('''metadata''' + split_layer[1]).split('''/''' ) )] elif "kvstore" in layer: __a = layer.split('''kvstore''' ) __a = ''''''.join(split_layer[0] )[:-1] __a = [tuple(('''kvstore''' + split_layer[1]).split('''/''' ) )] else: __a = layer.split('''/''' ) __a = '''/'''.join(split_layer[:-1] ) __a = (split_layer[-1],) if "kvstore/path" in layer: __a = f'''{switch_checkpoint_path}/{checkpoint_info[layer]}''' elif "kvstore/driver" in layer: __a = '''file''' else: __a = checkpoint_info[layer] return curr_real_layer_name, split_layer, content def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Union[str, Any] ) -> Any: __a = rename_keys(lowerCAmelCase__ ) __a = {} for k, v in current_block.items(): __a = v __a = new_current_block torch.save(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Dict , lowerCAmelCase__ : int , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str = WEIGHTS_NAME ) -> Optional[Any]: __a = convert_file_size_to_int(lowerCAmelCase__ ) __a = [] __a = {} __a = 0 __a = 0 os.makedirs(lowerCAmelCase__ , exist_ok=lowerCAmelCase__ ) with gfile.GFile(switch_checkpoint_path + '''/checkpoint''' , '''rb''' ) as fp: __a = serialization.msgpack_restore(fp.read() )['''optimizer''']['''target'''] __a = flatten_dict(lowerCAmelCase__ , sep='''/''' ) __a = {} for layer in checkpoint_info.keys(): __a , __a , __a = get_key_and_tensorstore_dict( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) if curr_real_layer_name in all_layers: __a = content else: __a = {split_layer[-1]: content} for key in all_layers.keys(): # open tensorstore file __a = ts.open(unflatten_dict(all_layers[key] ) ).result().read().result() __a = torch.tensor(lowerCAmelCase__ ) __a = raw_weights.numel() * dtype_byte_size(raw_weights.dtype ) # use the renaming pattern from the small conversion scripts __a , __a = rename_base_flax_keys(tuple(key.split('''/''' ) ) , lowerCAmelCase__ ) __a = '''/'''.join(lowerCAmelCase__ ) # If this weight is going to tip up over the maximal size, we split. if current_block_size + weight_size > max_shard_size: __a = os.path.join( lowerCAmelCase__ , weights_name.replace('''.bin''' , f'''-{len(lowerCAmelCase__ )+1:05d}-of-???.bin''' ) ) rename_and_save_block(lowerCAmelCase__ , lowerCAmelCase__ ) sharded_state_dicts.append(current_block.keys() ) del current_block __a = {} __a = 0 __a = raw_weights.to(getattr(lowerCAmelCase__ , lowerCAmelCase__ ) ) current_block_size += weight_size total_size += weight_size # Add the last block __a = os.path.join(lowerCAmelCase__ , weights_name.replace('''.bin''' , f'''-{len(lowerCAmelCase__ )+1:05d}-of-???.bin''' ) ) rename_and_save_block(lowerCAmelCase__ , lowerCAmelCase__ ) sharded_state_dicts.append(current_block.keys() ) # If we only have one shard, we return it if len(lowerCAmelCase__ ) == 1: return {weights_name: sharded_state_dicts[0]}, None # Otherwise, let's build the index __a = {} __a = {} for idx, shard in enumerate(lowerCAmelCase__ ): __a = weights_name.replace( '''.bin''' , f'''-{idx+1:05d}-of-{len(lowerCAmelCase__ ):05d}.bin''' ) # len(sharded_state_dicts):05d} __a = os.path.join(lowerCAmelCase__ , weights_name.replace('''.bin''' , f'''-{idx+1:05d}-of-???.bin''' ) ) os.rename(lowerCAmelCase__ , os.path.join(lowerCAmelCase__ , lowerCAmelCase__ ) ) __a = shard for key in shard: __a = shard_file # Add the metadata __a = {'''total_size''': total_size} __a = {'''metadata''': metadata, '''weight_map''': weight_map} with open(os.path.join(lowerCAmelCase__ , lowerCAmelCase__ ) , '''w''' , encoding='''utf-8''' ) as f: __a = json.dumps(lowerCAmelCase__ , indent=2 , sort_keys=lowerCAmelCase__ ) + '''\n''' f.write(lowerCAmelCase__ ) return metadata, index if __name__ == "__main__": lowercase_ = argparse.ArgumentParser() # Required parameters parser.add_argument( "--switch_t5x_checkpoint_path", default="/mnt/disks/disk_switch/original_checkpoints/switch-xxl-128/checkpoint_634600", type=str, required=False, help="Path to a directory containing a folder per layer. Follows the original Google format.", ) parser.add_argument("--max_shard_size", default="10GB", required=False, help="Max shard size") parser.add_argument("--dtype", default="bfloat16", type=str, required=False, help="dtype of the saved model") parser.add_argument( "--pytorch_dump_folder_path", default="/mnt/disks/disk_switch/original_checkpoints/switch-xxl-128-converted", type=str, required=False, help="Path to the output pytorch model.", ) lowercase_ = parser.parse_args() shard_on_the_fly( args.switch_tax_checkpoint_path, args.pytorch_dump_folder_path, args.max_shard_size, args.dtype, ) def lowercase ( ) -> int: from transformers import SwitchTransformersConfig, SwitchTransformersForConditionalGeneration, TaTokenizer __a = SwitchTransformersConfig.from_pretrained('''google/switch-base-8''' ) config.save_pretrained('''/home/arthur_huggingface_co/transformers/switch_converted''' ) __a = SwitchTransformersForConditionalGeneration.from_pretrained( '''/home/arthur_huggingface_co/transformers/switch_converted''' , device_map='''auto''' ) __a = TaTokenizer.from_pretrained('''t5-small''' ) __a = '''A <extra_id_0> walks into a bar a orders a <extra_id_1> with <extra_id_2> pinch of <extra_id_3>.''' __a = tokenizer(lowerCAmelCase__ , return_tensors='''pt''' ).input_ids __a = model.generate(lowerCAmelCase__ , decoder_start_token_id=0 ) print(tokenizer.decode(out[0] ) )
700
"""simple docstring""" import pytest from datasets import inspect_metric, list_metrics, load_metric @pytest.fixture def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' , set() ) @pytest.fixture def lowercase ( lowerCAmelCase__ : Any ) -> Any: class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = metric_id class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Any = [MetricMock(__SCREAMING_SNAKE_CASE ) for metric_id in ['accuracy', 'mse', 'precision', 'codeparrot/apps_metric']] def __UpperCAmelCase ( self ): return self._metrics monkeypatch.setattr('''datasets.inspect.huggingface_hub''' , HfhMock() ) @pytest.mark.parametrize( '''func, args''' , [(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Tuple ) -> Optional[int]: if "tmp_path" in args: __a = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args ) with pytest.warns(lowerCAmelCase__ , match='''https://huggingface.co/docs/evaluate''' ): func(*lowerCAmelCase__ )
65
0
"""simple docstring""" from math import pi, sqrt, tan def lowercase ( lowerCAmelCase__ : float ) -> float: if side_length < 0: raise ValueError('''surface_area_cube() only accepts non-negative values''' ) return 6 * side_length**2 def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if length < 0 or breadth < 0 or height < 0: raise ValueError('''surface_area_cuboid() only accepts non-negative values''' ) return 2 * ((length * breadth) + (breadth * height) + (length * height)) def lowercase ( lowerCAmelCase__ : float ) -> float: if radius < 0: raise ValueError('''surface_area_sphere() only accepts non-negative values''' ) return 4 * pi * radius**2 def lowercase ( lowerCAmelCase__ : float ) -> float: if radius < 0: raise ValueError('''surface_area_hemisphere() only accepts non-negative values''' ) return 3 * pi * radius**2 def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if radius < 0 or height < 0: raise ValueError('''surface_area_cone() only accepts non-negative values''' ) return pi * radius * (radius + (height**2 + radius**2) ** 0.5) def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if radius_a < 0 or radius_a < 0 or height < 0: raise ValueError( '''surface_area_conical_frustum() only accepts non-negative values''' ) __a = (height**2 + (radius_a - radius_a) ** 2) ** 0.5 return pi * ((slant_height * (radius_a + radius_a)) + radius_a**2 + radius_a**2) def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if radius < 0 or height < 0: raise ValueError('''surface_area_cylinder() only accepts non-negative values''' ) return 2 * pi * radius * (height + radius) def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if torus_radius < 0 or tube_radius < 0: raise ValueError('''surface_area_torus() only accepts non-negative values''' ) if torus_radius < tube_radius: raise ValueError( '''surface_area_torus() does not support spindle or self intersecting tori''' ) return 4 * pow(lowerCAmelCase__ , 2 ) * torus_radius * tube_radius def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if length < 0 or width < 0: raise ValueError('''area_rectangle() only accepts non-negative values''' ) return length * width def lowercase ( lowerCAmelCase__ : float ) -> float: if side_length < 0: raise ValueError('''area_square() only accepts non-negative values''' ) return side_length**2 def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if base < 0 or height < 0: raise ValueError('''area_triangle() only accepts non-negative values''' ) return (base * height) / 2 def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if sidea < 0 or sidea < 0 or sidea < 0: raise ValueError('''area_triangle_three_sides() only accepts non-negative values''' ) elif sidea + sidea < sidea or sidea + sidea < sidea or sidea + sidea < sidea: raise ValueError('''Given three sides do not form a triangle''' ) __a = (sidea + sidea + sidea) / 2 __a = sqrt( semi_perimeter * (semi_perimeter - sidea) * (semi_perimeter - sidea) * (semi_perimeter - sidea) ) return area def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if base < 0 or height < 0: raise ValueError('''area_parallelogram() only accepts non-negative values''' ) return base * height def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if basea < 0 or basea < 0 or height < 0: raise ValueError('''area_trapezium() only accepts non-negative values''' ) return 1 / 2 * (basea + basea) * height def lowercase ( lowerCAmelCase__ : float ) -> float: if radius < 0: raise ValueError('''area_circle() only accepts non-negative values''' ) return pi * radius**2 def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if radius_x < 0 or radius_y < 0: raise ValueError('''area_ellipse() only accepts non-negative values''' ) return pi * radius_x * radius_y def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float ) -> float: if diagonal_a < 0 or diagonal_a < 0: raise ValueError('''area_rhombus() only accepts non-negative values''' ) return 1 / 2 * diagonal_a * diagonal_a def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : float ) -> float: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or sides < 3: raise ValueError( '''area_reg_polygon() only accepts integers greater than or \ equal to three as number of sides''' ) elif length < 0: raise ValueError( '''area_reg_polygon() only accepts non-negative values as \ length of a side''' ) return (sides * length**2) / (4 * tan(pi / sides )) return (sides * length**2) / (4 * tan(pi / sides )) if __name__ == "__main__": import doctest doctest.testmod(verbose=True) # verbose so we can see methods missing tests print("[DEMO] Areas of various geometric shapes: \n") print(F'''Rectangle: {area_rectangle(1_0, 2_0) = }''') print(F'''Square: {area_square(1_0) = }''') print(F'''Triangle: {area_triangle(1_0, 1_0) = }''') print(F'''Triangle: {area_triangle_three_sides(5, 1_2, 1_3) = }''') print(F'''Parallelogram: {area_parallelogram(1_0, 2_0) = }''') print(F'''Rhombus: {area_rhombus(1_0, 2_0) = }''') print(F'''Trapezium: {area_trapezium(1_0, 2_0, 3_0) = }''') print(F'''Circle: {area_circle(2_0) = }''') print(F'''Ellipse: {area_ellipse(1_0, 2_0) = }''') print("\nSurface Areas of various geometric shapes: \n") print(F'''Cube: {surface_area_cube(2_0) = }''') print(F'''Cuboid: {surface_area_cuboid(1_0, 2_0, 3_0) = }''') print(F'''Sphere: {surface_area_sphere(2_0) = }''') print(F'''Hemisphere: {surface_area_hemisphere(2_0) = }''') print(F'''Cone: {surface_area_cone(1_0, 2_0) = }''') print(F'''Conical Frustum: {surface_area_conical_frustum(1_0, 2_0, 3_0) = }''') print(F'''Cylinder: {surface_area_cylinder(1_0, 2_0) = }''') print(F'''Torus: {surface_area_torus(2_0, 1_0) = }''') print(F'''Equilateral Triangle: {area_reg_polygon(3, 1_0) = }''') print(F'''Square: {area_reg_polygon(4, 1_0) = }''') print(F'''Reqular Pentagon: {area_reg_polygon(5, 1_0) = }''')
701
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> int: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or number < 0: raise ValueError('''Input must be a non-negative integer''' ) __a = 0 while number: # This way we arrive at next set bit (next 1) instead of looping # through each bit and checking for 1s hence the # loop won't run 32 times it will only run the number of `1` times number &= number - 1 count += 1 return count if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import itertools import json import linecache import os import pickle import re import socket import string from collections import Counter from logging import getLogger from pathlib import Path from typing import Callable, Dict, Iterable, List import git import torch from torch.utils.data import Dataset from transformers import BartTokenizer, RagTokenizer, TaTokenizer def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : int , lowerCAmelCase__ : int , lowerCAmelCase__ : int , lowerCAmelCase__ : List[str]=True , lowerCAmelCase__ : Union[str, Any]="pt" ) -> int: __a = {'''add_prefix_space''': True} if isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) and not line.startswith(''' ''' ) else {} __a = padding_side return tokenizer( [line] , max_length=lowerCAmelCase__ , padding='''max_length''' if pad_to_max_length else None , truncation=lowerCAmelCase__ , return_tensors=lowerCAmelCase__ , add_special_tokens=lowerCAmelCase__ , **lowerCAmelCase__ , ) def lowercase ( lowerCAmelCase__ : Dict , lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Any=None , ) -> int: __a = input_ids.ne(lowerCAmelCase__ ).any(dim=0 ) if attention_mask is None: return input_ids[:, keep_column_mask] else: return (input_ids[:, keep_column_mask], attention_mask[:, keep_column_mask]) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a="train" , _a=None , _a=None , _a=None , _a="" , ): super().__init__() __a = Path(_a ).joinpath(type_path + '''.source''' ) __a = Path(_a ).joinpath(type_path + '''.target''' ) __a = self.get_char_lens(self.src_file ) __a = max_source_length __a = max_target_length assert min(self.src_lens ) > 0, f'''found empty line in {self.src_file}''' __a = tokenizer __a = prefix if n_obs is not None: __a = self.src_lens[:n_obs] __a = src_lang __a = tgt_lang def __len__( self ): return len(self.src_lens ) def __getitem__( self , _a ): __a = index + 1 # linecache starts at 1 __a = self.prefix + linecache.getline(str(self.src_file ) , _a ).rstrip('''\n''' ) __a = linecache.getline(str(self.tgt_file ) , _a ).rstrip('''\n''' ) assert source_line, f'''empty source line for index {index}''' assert tgt_line, f'''empty tgt line for index {index}''' # Need to add eos token manually for T5 if isinstance(self.tokenizer , _a ): source_line += self.tokenizer.eos_token tgt_line += self.tokenizer.eos_token # Pad source and target to the right __a = ( self.tokenizer.question_encoder if isinstance(self.tokenizer , _a ) else self.tokenizer ) __a = self.tokenizer.generator if isinstance(self.tokenizer , _a ) else self.tokenizer __a = encode_line(_a , _a , self.max_source_length , '''right''' ) __a = encode_line(_a , _a , self.max_target_length , '''right''' ) __a = source_inputs['''input_ids'''].squeeze() __a = target_inputs['''input_ids'''].squeeze() __a = source_inputs['''attention_mask'''].squeeze() return { "input_ids": source_ids, "attention_mask": src_mask, "decoder_input_ids": target_ids, } @staticmethod def __UpperCAmelCase ( _a ): return [len(_a ) for x in Path(_a ).open().readlines()] def __UpperCAmelCase ( self , _a ): __a = torch.stack([x['''input_ids'''] for x in batch] ) __a = torch.stack([x['''attention_mask'''] for x in batch] ) __a = torch.stack([x['''decoder_input_ids'''] for x in batch] ) __a = ( self.tokenizer.generator.pad_token_id if isinstance(self.tokenizer , _a ) else self.tokenizer.pad_token_id ) __a = ( self.tokenizer.question_encoder.pad_token_id if isinstance(self.tokenizer , _a ) else self.tokenizer.pad_token_id ) __a = trim_batch(_a , _a ) __a , __a = trim_batch(_a , _a , attention_mask=_a ) __a = { '''input_ids''': source_ids, '''attention_mask''': source_mask, '''decoder_input_ids''': y, } return batch lowercase_ : List[str] = getLogger(__name__) def lowercase ( lowerCAmelCase__ : List[List] ) -> int: return list(itertools.chain.from_iterable(lowerCAmelCase__ ) ) def lowercase ( lowerCAmelCase__ : str ) -> None: __a = get_git_info() save_json(lowerCAmelCase__ , os.path.join(lowerCAmelCase__ , '''git_log.json''' ) ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : int=4 , **lowerCAmelCase__ : Optional[int] ) -> str: with open(lowerCAmelCase__ , '''w''' ) as f: json.dump(lowerCAmelCase__ , lowerCAmelCase__ , indent=lowerCAmelCase__ , **lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Optional[Any] ) -> Dict: with open(lowerCAmelCase__ ) as f: return json.load(lowerCAmelCase__ ) def lowercase ( ) -> Optional[Any]: __a = git.Repo(search_parent_directories=lowerCAmelCase__ ) __a = { '''repo_id''': str(lowerCAmelCase__ ), '''repo_sha''': str(repo.head.object.hexsha ), '''repo_branch''': str(repo.active_branch ), '''hostname''': str(socket.gethostname() ), } return repo_infos def lowercase ( lowerCAmelCase__ : Callable , lowerCAmelCase__ : Iterable ) -> List: return list(map(lowerCAmelCase__ , lowerCAmelCase__ ) ) def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> Optional[int]: with open(lowerCAmelCase__ , '''wb''' ) as f: return pickle.dump(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : List[str] ) -> str: def remove_articles(lowerCAmelCase__ : Any ): return re.sub(r'''\b(a|an|the)\b''' , ''' ''' , lowerCAmelCase__ ) def white_space_fix(lowerCAmelCase__ : List[Any] ): return " ".join(text.split() ) def remove_punc(lowerCAmelCase__ : int ): __a = set(string.punctuation ) return "".join(ch for ch in text if ch not in exclude ) def lower(lowerCAmelCase__ : int ): return text.lower() return white_space_fix(remove_articles(remove_punc(lower(lowerCAmelCase__ ) ) ) ) def lowercase ( lowerCAmelCase__ : Dict , lowerCAmelCase__ : str ) -> Union[str, Any]: __a = normalize_answer(lowerCAmelCase__ ).split() __a = normalize_answer(lowerCAmelCase__ ).split() __a = Counter(lowerCAmelCase__ ) & Counter(lowerCAmelCase__ ) __a = sum(common.values() ) if num_same == 0: return 0 __a = 1.0 * num_same / len(lowerCAmelCase__ ) __a = 1.0 * num_same / len(lowerCAmelCase__ ) __a = (2 * precision * recall) / (precision + recall) return fa def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : int ) -> int: return normalize_answer(lowerCAmelCase__ ) == normalize_answer(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : List[str] ) -> Dict: assert len(lowerCAmelCase__ ) == len(lowerCAmelCase__ ) __a = 0 for hypo, pred in zip(lowerCAmelCase__ , lowerCAmelCase__ ): em += exact_match_score(lowerCAmelCase__ , lowerCAmelCase__ ) if len(lowerCAmelCase__ ) > 0: em /= len(lowerCAmelCase__ ) return {"em": em} def lowercase ( lowerCAmelCase__ : Optional[int] ) -> Tuple: return model_prefix.startswith('''rag''' ) def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Optional[Any]: __a = {p: p for p in extra_params} # T5 models don't have `dropout` param, they have `dropout_rate` instead __a = '''dropout_rate''' for p in extra_params: if getattr(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ): if not hasattr(lowerCAmelCase__ , lowerCAmelCase__ ) and not hasattr(lowerCAmelCase__ , equivalent_param[p] ): logger.info('''config doesn\'t have a `{}` attribute'''.format(lowerCAmelCase__ ) ) delattr(lowerCAmelCase__ , lowerCAmelCase__ ) continue __a = p if hasattr(lowerCAmelCase__ , lowerCAmelCase__ ) else equivalent_param[p] setattr(lowerCAmelCase__ , lowerCAmelCase__ , getattr(lowerCAmelCase__ , lowerCAmelCase__ ) ) delattr(lowerCAmelCase__ , lowerCAmelCase__ ) return hparams, config
702
"""simple docstring""" import pytest from datasets import Dataset, DatasetDict, Features, NamedSplit, Value from datasets.io.text import TextDatasetReader from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.split == split if split else "train" @pytest.mark.parametrize('''path_type''' , [str, list] ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict: if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = text_path elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [text_path] __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for split in splits: __a = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str: __a = tmp_path / '''cache''' # CSV file loses col_1 string dtype information: default now is "int64" instead of "string" __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict: if split: __a = {split: text_path} else: __a = '''train''' __a = {'''train''': text_path, '''test''': text_path} __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() )
65
0
import math def lowercase ( lowerCAmelCase__ : int ) -> list[int]: __a = [] __a = 2 __a = int(math.sqrt(lowerCAmelCase__ ) ) # Size of every segment __a = [True] * (end + 1) __a = [] while start <= end: if temp[start] is True: in_prime.append(lowerCAmelCase__ ) for i in range(start * start , end + 1 , lowerCAmelCase__ ): __a = False start += 1 prime += in_prime __a = end + 1 __a = min(2 * end , lowerCAmelCase__ ) while low <= n: __a = [True] * (high - low + 1) for each in in_prime: __a = math.floor(low / each ) * each if t < low: t += each for j in range(lowerCAmelCase__ , high + 1 , lowerCAmelCase__ ): __a = False for j in range(len(lowerCAmelCase__ ) ): if temp[j] is True: prime.append(j + low ) __a = high + 1 __a = min(high + end , lowerCAmelCase__ ) return prime print(sieve(1_0**6))
703
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> bool: return numa ^ numa < 0 if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" from ..utils import DummyObject, requires_backends class __lowerCAmelCase ( metaclass=__SCREAMING_SNAKE_CASE ): '''simple docstring''' snake_case : List[str] = ['onnx'] def __init__( self , *_a , **_a ): requires_backends(self , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] )
704
"""simple docstring""" import unittest from transformers import SPIECE_UNDERLINE from transformers.models.speechta import SpeechTaTokenizer from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow from transformers.tokenization_utils import AddedToken from ...test_tokenization_common import TokenizerTesterMixin lowercase_ = get_tests_dir("fixtures/test_sentencepiece_bpe_char.model") @require_sentencepiece @require_tokenizers class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : int = SpeechTaTokenizer __UpperCAmelCase : Tuple = False __UpperCAmelCase : Dict = True def __UpperCAmelCase ( self ): super().setUp() # We have a SentencePiece fixture for testing __a = SpeechTaTokenizer(_a ) __a = AddedToken('''<mask>''' , lstrip=_a , rstrip=_a ) __a = mask_token tokenizer.add_special_tokens({'''mask_token''': mask_token} ) tokenizer.add_tokens(['''<ctc_blank>'''] ) tokenizer.save_pretrained(self.tmpdirname ) def __UpperCAmelCase ( self , _a ): __a = '''this is a test''' __a = '''this is a test''' return input_text, output_text def __UpperCAmelCase ( self , _a , _a=False , _a=20 , _a=5 ): __a , __a = self.get_input_output_texts(_a ) __a = tokenizer.encode(_a , add_special_tokens=_a ) __a = tokenizer.decode(_a , clean_up_tokenization_spaces=_a ) return text, ids def __UpperCAmelCase ( self ): __a = '''<pad>''' __a = 1 self.assertEqual(self.get_tokenizer()._convert_token_to_id(_a ) , _a ) self.assertEqual(self.get_tokenizer()._convert_id_to_token(_a ) , _a ) def __UpperCAmelCase ( self ): __a = list(self.get_tokenizer().get_vocab().keys() ) self.assertEqual(vocab_keys[0] , '''<s>''' ) self.assertEqual(vocab_keys[1] , '''<pad>''' ) self.assertEqual(vocab_keys[-4] , '''œ''' ) self.assertEqual(vocab_keys[-2] , '''<mask>''' ) self.assertEqual(vocab_keys[-1] , '''<ctc_blank>''' ) self.assertEqual(len(_a ) , 81 ) def __UpperCAmelCase ( self ): self.assertEqual(self.get_tokenizer().vocab_size , 79 ) def __UpperCAmelCase ( self ): __a = self.get_tokenizers(do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) # We usually have added tokens from the start in tests because our vocab fixtures are # smaller than the original vocabs - let's not assert this # self.assertEqual(vocab_size, all_size) __a = ['''aaaaa bbbbbb''', '''cccccccccdddddddd'''] __a = tokenizer.add_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size + len(_a ) ) __a = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 4 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) __a = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''} __a = tokenizer.add_special_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size_a + len(_a ) ) __a = tokenizer.encode( '''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 6 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[0] , tokens[1] ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokens[-4] ) self.assertEqual(tokens[0] , tokenizer.eos_token_id ) self.assertEqual(tokens[-3] , tokenizer.pad_token_id ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tokenizer.tokenize('''This is a test''' ) # fmt: off self.assertListEqual(_a , [SPIECE_UNDERLINE, '''T''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''a''', SPIECE_UNDERLINE, '''t''', '''e''', '''s''', '''t'''] ) # fmt: on self.assertListEqual( tokenizer.convert_tokens_to_ids(_a ) , [4, 32, 11, 10, 12, 4, 10, 12, 4, 7, 4, 6, 5, 12, 6] , ) __a = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''92000''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) __a = tokenizer.convert_tokens_to_ids(_a ) # fmt: off self.assertListEqual(_a , [4, 30, 4, 20, 7, 12, 4, 25, 8, 13, 9, 4, 10, 9, 4, 3, 23, 4, 7, 9, 14, 4, 6, 11, 10, 12, 4, 10, 12, 4, 19, 7, 15, 12, 73, 26] ) # fmt: on __a = tokenizer.convert_ids_to_tokens(_a ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''<unk>''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) @slow def __UpperCAmelCase ( self ): # Use custom sequence because this tokenizer does not handle numbers. __a = [ '''Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides ''' '''general-purpose architectures (BERT, GPT, RoBERTa, XLM, DistilBert, XLNet...) for Natural ''' '''Language Understanding (NLU) and Natural Language Generation (NLG) with over thirty-two pretrained ''' '''models in one hundred plus languages and deep interoperability between Jax, PyTorch and TensorFlow.''', '''BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly ''' '''conditioning on both left and right context in all layers.''', '''The quick brown fox jumps over the lazy dog.''', ] # fmt: off __a = { '''input_ids''': [ [4, 32, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 64, 19, 8, 13, 18, 5, 13, 15, 22, 4, 28, 9, 8, 20, 9, 4, 7, 12, 4, 24, 22, 6, 8, 13, 17, 11, 39, 6, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 7, 9, 14, 4, 24, 22, 6, 8, 13, 17, 11, 39, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 39, 25, 5, 13, 6, 63, 4, 24, 13, 8, 27, 10, 14, 5, 12, 4, 21, 5, 9, 5, 13, 7, 15, 39, 24, 16, 13, 24, 8, 12, 5, 4, 7, 13, 17, 11, 10, 6, 5, 17, 6, 16, 13, 5, 12, 4, 64, 40, 47, 54, 32, 23, 4, 53, 49, 32, 23, 4, 54, 8, 40, 47, 54, 32, 7, 23, 4, 69, 52, 43, 23, 4, 51, 10, 12, 6, 10, 15, 40, 5, 13, 6, 23, 4, 69, 52, 48, 5, 6, 26, 26, 26, 63, 4, 19, 8, 13, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 61, 9, 14, 5, 13, 12, 6, 7, 9, 14, 10, 9, 21, 4, 64, 48, 52, 61, 63, 4, 7, 9, 14, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 53, 5, 9, 5, 13, 7, 6, 10, 8, 9, 4, 64, 48, 52, 53, 63, 4, 20, 10, 6, 11, 4, 8, 27, 5, 13, 4, 6, 11, 10, 13, 6, 22, 39, 6, 20, 8, 4, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 4, 18, 8, 14, 5, 15, 12, 4, 10, 9, 4, 8, 9, 5, 4, 11, 16, 9, 14, 13, 5, 14, 4, 24, 15, 16, 12, 4, 15, 7, 9, 21, 16, 7, 21, 5, 12, 4, 7, 9, 14, 4, 14, 5, 5, 24, 4, 10, 9, 6, 5, 13, 8, 24, 5, 13, 7, 25, 10, 15, 10, 6, 22, 4, 25, 5, 6, 20, 5, 5, 9, 4, 58, 7, 37, 23, 4, 49, 22, 32, 8, 13, 17, 11, 4, 7, 9, 14, 4, 32, 5, 9, 12, 8, 13, 55, 15, 8, 20, 26, 2], [4, 40, 47, 54, 32, 4, 10, 12, 4, 14, 5, 12, 10, 21, 9, 5, 14, 4, 6, 8, 4, 24, 13, 5, 39, 6, 13, 7, 10, 9, 4, 14, 5, 5, 24, 4, 25, 10, 14, 10, 13, 5, 17, 6, 10, 8, 9, 7, 15, 4, 13, 5, 24, 13, 5, 12, 5, 9, 6, 7, 6, 10, 8, 9, 12, 4, 19, 13, 8, 18, 4, 16, 9, 15, 7, 25, 5, 15, 5, 14, 4, 6, 5, 37, 6, 4, 25, 22, 4, 46, 8, 10, 9, 6, 15, 22, 4, 17, 8, 9, 14, 10, 6, 10, 8, 9, 10, 9, 21, 4, 8, 9, 4, 25, 8, 6, 11, 4, 15, 5, 19, 6, 4, 7, 9, 14, 4, 13, 10, 21, 11, 6, 4, 17, 8, 9, 6, 5, 37, 6, 4, 10, 9, 4, 7, 15, 15, 4, 15, 7, 22, 5, 13, 12, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [4, 32, 11, 5, 4, 45, 16, 10, 17, 28, 4, 25, 13, 8, 20, 9, 4, 19, 8, 37, 4, 46, 16, 18, 24, 12, 4, 8, 27, 5, 13, 4, 6, 11, 5, 4, 15, 7, 57, 22, 4, 14, 8, 21, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], ], '''attention_mask''': [ [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], ] } # fmt: on self.tokenizer_integration_test_util( expected_encoding=_a , model_name='''microsoft/speecht5_asr''' , revision='''c5ef64c71905caeccde0e4462ef3f9077224c524''' , sequences=_a , )
65
0
"""simple docstring""" from __future__ import annotations def lowercase ( lowerCAmelCase__ : float , lowerCAmelCase__ : float , lowerCAmelCase__ : float , ) -> tuple[str, float]: if (stress, tangential_force, area).count(0 ) != 1: raise ValueError('''You cannot supply more or less than 2 values''' ) elif stress < 0: raise ValueError('''Stress cannot be negative''' ) elif tangential_force < 0: raise ValueError('''Tangential Force cannot be negative''' ) elif area < 0: raise ValueError('''Area cannot be negative''' ) elif stress == 0: return ( "stress", tangential_force / area, ) elif tangential_force == 0: return ( "tangential_force", stress * area, ) else: return ( "area", tangential_force / stress, ) if __name__ == "__main__": import doctest doctest.testmod()
705
"""simple docstring""" from typing import List, Optional, Union import numpy as np from ....audio_utils import mel_filter_bank, optimal_fft_length, spectrogram, window_function from ....feature_extraction_sequence_utils import SequenceFeatureExtractor from ....feature_extraction_utils import BatchFeature from ....file_utils import PaddingStrategy, TensorType from ....utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Any = ['input_features', 'attention_mask'] def __init__( self , _a=80 , _a=16_000 , _a=0.0 , _a=10 , _a=25 , _a="hamming_window" , _a=3_2768.0 , _a=0.97 , _a=1.0 , _a=True , _a=True , _a=False , **_a , ): super().__init__(feature_size=_a , sampling_rate=_a , padding_value=_a , **_a ) __a = feature_size __a = sampling_rate __a = padding_value __a = hop_length __a = win_length __a = frame_signal_scale __a = preemphasis_coeff __a = mel_floor __a = normalize_means __a = normalize_vars __a = win_function __a = return_attention_mask __a = win_length * sampling_rate // 1_000 __a = hop_length * sampling_rate // 1_000 __a = optimal_fft_length(self.sample_size ) __a = (self.n_fft // 2) + 1 def __UpperCAmelCase ( self , _a ): if self.win_function == "hamming_window": __a = window_function(window_length=self.sample_size , name=self.win_function , periodic=_a ) else: __a = window_function(window_length=self.sample_size , name=self.win_function ) __a = mel_filter_bank( num_frequency_bins=self.n_freqs , num_mel_filters=self.feature_size , min_frequency=0.0 , max_frequency=self.sampling_rate / 2.0 , sampling_rate=self.sampling_rate , ) __a = spectrogram( one_waveform * self.frame_signal_scale , window=_a , frame_length=self.sample_size , hop_length=self.sample_stride , fft_length=self.n_fft , center=_a , preemphasis=self.preemphasis_coeff , mel_filters=_a , mel_floor=self.mel_floor , log_mel='''log''' , ) return msfc_features.T def __UpperCAmelCase ( self , _a , _a , _a ): # make sure we normalize float32 arrays if self.normalize_means: __a = x[:input_length].mean(axis=0 ) __a = np.subtract(_a , _a ) if self.normalize_vars: __a = x[:input_length].std(axis=0 ) __a = np.divide(_a , _a ) if input_length < x.shape[0]: __a = padding_value # make sure array is in float32 __a = x.astype(np.floataa ) return x def __UpperCAmelCase ( self , _a , _a = None ): __a = attention_mask.sum(-1 ) if attention_mask is not None else [x.shape[0] for x in input_features] return [self._normalize_one(_a , _a , self.padding_value ) for x, n in zip(_a , _a )] def __call__( self , _a , _a = False , _a = None , _a = False , _a = None , _a = None , _a = None , _a = None , **_a , ): if sampling_rate is not None: if sampling_rate != self.sampling_rate: raise ValueError( f'''The model corresponding to this feature extractor: {self} was trained using a sampling rate of''' f''' {self.sampling_rate}. Please make sure that the provided `raw_speech` input was sampled with''' f''' {self.sampling_rate} and not {sampling_rate}.''' ) else: logger.warning( '''It is strongly recommended to pass the ``sampling_rate`` argument to this function. ''' '''Failing to do so can result in silent errors that might be hard to debug.''' ) __a = isinstance(_a , np.ndarray ) and len(raw_speech.shape ) > 1 if is_batched_numpy and len(raw_speech.shape ) > 2: raise ValueError(f'''Only mono-channel audio is supported for input to {self}''' ) __a = is_batched_numpy or ( isinstance(_a , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) )) ) if is_batched: __a = [np.asarray(_a , dtype=np.floataa ) for speech in raw_speech] elif not is_batched and not isinstance(_a , np.ndarray ): __a = np.asarray(_a , dtype=np.floataa ) elif isinstance(_a , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ): __a = raw_speech.astype(np.floataa ) # always return batch if not is_batched: __a = [raw_speech] # extract fbank features __a = [self._extract_mfsc_features(_a ) for one_waveform in raw_speech] # convert into correct format for padding __a = BatchFeature({'''input_features''': features} ) __a = self.pad( _a , padding=_a , max_length=_a , truncation=_a , pad_to_multiple_of=_a , return_attention_mask=_a , **_a , ) # make sure list is in array format __a = padded_inputs.get('''input_features''' ) if isinstance(input_features[0] , _a ): __a = [np.asarray(_a , dtype=np.floataa ) for feature in input_features] __a = padded_inputs.get('''attention_mask''' ) if attention_mask is not None: __a = [np.asarray(_a , dtype=np.intaa ) for array in attention_mask] if self.normalize_means or self.normalize_vars: __a = ( np.array(_a , dtype=np.intaa ) if self._get_padding_strategies(_a , max_length=_a ) is not PaddingStrategy.DO_NOT_PAD and padding else None ) __a = self.normalize( padded_inputs['''input_features'''] , attention_mask=_a ) if return_tensors is not None: __a = padded_inputs.convert_to_tensors(_a ) return padded_inputs
65
0
from collections import deque from .hash_table import HashTable class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , *_a , **_a ): super().__init__(*_a , **_a ) def __UpperCAmelCase ( self , _a , _a ): __a = deque([] ) if self.values[key] is None else self.values[key] self.values[key].appendleft(_a ) __a = self.values[key] def __UpperCAmelCase ( self ): return ( sum(self.charge_factor - len(_a ) for slot in self.values ) / self.size_table * self.charge_factor ) def __UpperCAmelCase ( self , _a , _a=None ): if not ( len(self.values[key] ) == self.charge_factor and self.values.count(_a ) == 0 ): return key return super()._collision_resolution(_a , _a )
706
"""simple docstring""" import json import os import subprocess import unittest from ast import literal_eval import pytest from parameterized import parameterized, 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_model_parallelism.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, { 'framework': 'pytorch', 'script': 'run_glue.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, ] ) class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): 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=_a , ) assert hasattr(self , '''env''' ) def __UpperCAmelCase ( self , _a ): # configuration for running training on smdistributed Model Parallel __a = { '''enabled''': True, '''processes_per_host''': 8, } __a = { '''enabled''': True, '''parameters''': { '''microbatches''': 4, '''placement_strategy''': '''spread''', '''pipeline''': '''interleaved''', '''optimize''': '''speed''', '''partitions''': 4, '''ddp''': True, }, } __a = {'''smdistributed''': {'''modelparallel''': smp_options}, '''mpi''': mpi_options} __a = '''trainer''' if self.script == '''run_glue.py''' else '''smtrainer''' # 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}-{instance_count}-smp-{name_extension}''' , instance_count=_a , instance_type=self.instance_type , debugger_hook_config=_a , hyperparameters={ **self.env.hyperparameters, '''model_name_or_path''': self.model_name_or_path, '''max_steps''': 500, } , metric_definitions=self.env.metric_definitions , distribution=_a , py_version='''py36''' , ) def __UpperCAmelCase ( self , _a ): TrainingJobAnalytics(_a ).export_csv(f'''{self.env.test_path}/{job_name}_metrics.csv''' ) @parameterized.expand([(1,)] ) def __UpperCAmelCase ( self , _a ): # create estimator __a = self.create_estimator(_a ) # run training estimator.fit() # result dataframe __a = TrainingJobAnalytics(estimator.latest_training_job.name ).dataframe() # extract kpis __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_accuracy''']['''value'''] ) __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_loss''']['''value'''] ) # get train time from SageMaker job, this includes starting, preprocessing, stopping __a = ( Session().describe_training_job(estimator.latest_training_job.name ).get('''TrainingTimeInSeconds''' , 999_999 ) ) # 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} , _a )
65
0
"""simple docstring""" from string import ascii_uppercase lowercase_ = {str(ord(c) - 5_5): c for c in ascii_uppercase} def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> str: if isinstance(lowerCAmelCase__ , lowerCAmelCase__ ): raise TypeError('''int() can\'t convert non-string with explicit base''' ) if num < 0: raise ValueError('''parameter must be positive int''' ) if isinstance(lowerCAmelCase__ , lowerCAmelCase__ ): raise TypeError('''\'str\' object cannot be interpreted as an integer''' ) if isinstance(lowerCAmelCase__ , lowerCAmelCase__ ): raise TypeError('''\'float\' object cannot be interpreted as an integer''' ) if base in (0, 1): raise ValueError('''base must be >= 2''' ) if base > 36: raise ValueError('''base must be <= 36''' ) __a = '''''' __a = 0 __a = 0 while div != 1: __a , __a = divmod(lowerCAmelCase__ , lowerCAmelCase__ ) if base >= 11 and 9 < mod < 36: __a = ALPHABET_VALUES[str(lowerCAmelCase__ )] else: __a = str(lowerCAmelCase__ ) new_value += actual_value __a = num // base __a = div if div == 0: return str(new_value[::-1] ) elif div == 1: new_value += str(lowerCAmelCase__ ) return str(new_value[::-1] ) return new_value[::-1] if __name__ == "__main__": import doctest doctest.testmod() for base in range(2, 3_7): for num in range(1_0_0_0): assert int(decimal_to_any(num, base), base) == num, ( num, base, decimal_to_any(num, base), int(decimal_to_any(num, base), base), )
707
"""simple docstring""" import warnings from ..trainer import Trainer from ..utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a=None , **_a ): warnings.warn( '''`SageMakerTrainer` is deprecated and will be removed in v5 of Transformers. You can use `Trainer` ''' '''instead.''' , _a , ) super().__init__(args=_a , **_a )
65
0
"""simple docstring""" import argparse import json import requests import torch from huggingface_hub import hf_hub_download from PIL import Image from transformers import ViTImageProcessor, ViTMSNConfig, ViTMSNModel from transformers.image_utils import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD torch.set_grad_enabled(False) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : List[str]=False ) -> Dict: '''simple docstring''' __a = [] for i in range(config.num_hidden_layers ): # encoder layers: output projection, 2 feedforward neural networks and 2 layernorms rename_keys.append((f'''module.blocks.{i}.norm1.weight''', f'''vit.encoder.layer.{i}.layernorm_before.weight''') ) rename_keys.append((f'''module.blocks.{i}.norm1.bias''', f'''vit.encoder.layer.{i}.layernorm_before.bias''') ) rename_keys.append( (f'''module.blocks.{i}.attn.proj.weight''', f'''vit.encoder.layer.{i}.attention.output.dense.weight''') ) rename_keys.append((f'''module.blocks.{i}.attn.proj.bias''', f'''vit.encoder.layer.{i}.attention.output.dense.bias''') ) rename_keys.append((f'''module.blocks.{i}.norm2.weight''', f'''vit.encoder.layer.{i}.layernorm_after.weight''') ) rename_keys.append((f'''module.blocks.{i}.norm2.bias''', f'''vit.encoder.layer.{i}.layernorm_after.bias''') ) rename_keys.append((f'''module.blocks.{i}.mlp.fc1.weight''', f'''vit.encoder.layer.{i}.intermediate.dense.weight''') ) rename_keys.append((f'''module.blocks.{i}.mlp.fc1.bias''', f'''vit.encoder.layer.{i}.intermediate.dense.bias''') ) rename_keys.append((f'''module.blocks.{i}.mlp.fc2.weight''', f'''vit.encoder.layer.{i}.output.dense.weight''') ) rename_keys.append((f'''module.blocks.{i}.mlp.fc2.bias''', f'''vit.encoder.layer.{i}.output.dense.bias''') ) # projection layer + position embeddings rename_keys.extend( [ ('''module.cls_token''', '''vit.embeddings.cls_token'''), ('''module.patch_embed.proj.weight''', '''vit.embeddings.patch_embeddings.projection.weight'''), ('''module.patch_embed.proj.bias''', '''vit.embeddings.patch_embeddings.projection.bias'''), ('''module.pos_embed''', '''vit.embeddings.position_embeddings'''), ] ) if base_model: # layernorm + pooler rename_keys.extend( [ ('''module.norm.weight''', '''layernorm.weight'''), ('''module.norm.bias''', '''layernorm.bias'''), ] ) # if just the base model, we should remove "vit" from all keys that start with "vit" __a = [(pair[0], pair[1][4:]) if pair[1].startswith('''vit''' ) else pair for pair in rename_keys] else: # layernorm + classification head rename_keys.extend( [ ('''norm.weight''', '''vit.layernorm.weight'''), ('''norm.bias''', '''vit.layernorm.bias'''), ('''head.weight''', '''classifier.weight'''), ('''head.bias''', '''classifier.bias'''), ] ) return rename_keys def lowercase ( lowerCAmelCase__ : Dict , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int]=False ) -> List[Any]: '''simple docstring''' for i in range(config.num_hidden_layers ): if base_model: __a = '''''' else: __a = '''vit.''' # read in weights + bias of input projection layer (in timm, this is a single matrix + bias) __a = state_dict.pop(f'''module.blocks.{i}.attn.qkv.weight''' ) __a = state_dict.pop(f'''module.blocks.{i}.attn.qkv.bias''' ) # next, add query, keys and values (in that order) to the state dict __a = in_proj_weight[ : config.hidden_size, : ] __a = in_proj_bias[: config.hidden_size] __a = in_proj_weight[ config.hidden_size : config.hidden_size * 2, : ] __a = in_proj_bias[ config.hidden_size : config.hidden_size * 2 ] __a = in_proj_weight[ -config.hidden_size :, : ] __a = in_proj_bias[-config.hidden_size :] def lowercase ( lowerCAmelCase__ : Dict ) -> str: '''simple docstring''' __a = ['''head.weight''', '''head.bias'''] for k in ignore_keys: state_dict.pop(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : List[str] ) -> str: '''simple docstring''' __a = [ '''module.fc.fc1.weight''', '''module.fc.fc1.bias''', '''module.fc.bn1.weight''', '''module.fc.bn1.bias''', '''module.fc.bn1.running_mean''', '''module.fc.bn1.running_var''', '''module.fc.bn1.num_batches_tracked''', '''module.fc.fc2.weight''', '''module.fc.fc2.bias''', '''module.fc.bn2.weight''', '''module.fc.bn2.bias''', '''module.fc.bn2.running_mean''', '''module.fc.bn2.running_var''', '''module.fc.bn2.num_batches_tracked''', '''module.fc.fc3.weight''', '''module.fc.fc3.bias''', ] for k in ignore_keys: state_dict.pop(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : int , lowerCAmelCase__ : Any ) -> int: '''simple docstring''' __a = dct.pop(lowerCAmelCase__ ) __a = val def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> List[str]: '''simple docstring''' __a = ViTMSNConfig() __a = 1000 __a = '''datasets/huggingface/label-files''' __a = '''imagenet-1k-id2label.json''' __a = json.load(open(hf_hub_download(lowerCAmelCase__ , lowerCAmelCase__ ) , '''r''' ) ) __a = {int(lowerCAmelCase__ ): v for k, v in idalabel.items()} __a = idalabel __a = {v: k for k, v in idalabel.items()} if "s16" in checkpoint_url: __a = 384 __a = 1536 __a = 6 elif "l16" in checkpoint_url: __a = 1024 __a = 4096 __a = 24 __a = 16 __a = 0.1 elif "b4" in checkpoint_url: __a = 4 elif "l7" in checkpoint_url: __a = 7 __a = 1024 __a = 4096 __a = 24 __a = 16 __a = 0.1 __a = ViTMSNModel(lowerCAmelCase__ ) __a = torch.hub.load_state_dict_from_url(lowerCAmelCase__ , map_location='''cpu''' )['''target_encoder'''] __a = ViTImageProcessor(size=config.image_size ) remove_projection_head(lowerCAmelCase__ ) __a = create_rename_keys(lowerCAmelCase__ , base_model=lowerCAmelCase__ ) for src, dest in rename_keys: rename_key(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) read_in_q_k_v(lowerCAmelCase__ , lowerCAmelCase__ , base_model=lowerCAmelCase__ ) model.load_state_dict(lowerCAmelCase__ ) model.eval() __a = '''http://images.cocodataset.org/val2017/000000039769.jpg''' __a = Image.open(requests.get(lowerCAmelCase__ , stream=lowerCAmelCase__ ).raw ) __a = ViTImageProcessor( size=config.image_size , image_mean=lowerCAmelCase__ , image_std=lowerCAmelCase__ ) __a = image_processor(images=lowerCAmelCase__ , return_tensors='''pt''' ) # forward pass torch.manual_seed(2 ) __a = model(**lowerCAmelCase__ ) __a = outputs.last_hidden_state # The following Colab Notebook was used to generate these outputs: # https://colab.research.google.com/gist/sayakpaul/3672419a04f5997827503fd84079bdd1/scratchpad.ipynb if "s16" in checkpoint_url: __a = torch.tensor([[-1.09_15, -1.48_76, -1.18_09]] ) elif "b16" in checkpoint_url: __a = torch.tensor([[14.28_89, -18.90_45, 11.72_81]] ) elif "l16" in checkpoint_url: __a = torch.tensor([[41.50_28, -22.86_81, 45.64_75]] ) elif "b4" in checkpoint_url: __a = torch.tensor([[-4.38_68, 5.29_32, -0.41_37]] ) else: __a = torch.tensor([[-0.17_92, -0.64_65, 2.42_63]] ) # verify logits assert torch.allclose(last_hidden_state[:, 0, :3] , lowerCAmelCase__ , atol=1e-4 ) print(f'''Saving model to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowerCAmelCase__ ) print(f'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowerCAmelCase__ ) if __name__ == "__main__": lowercase_ = argparse.ArgumentParser() # Required parameters parser.add_argument( "--checkpoint_url", default="https://dl.fbaipublicfiles.com/msn/vits16_800ep.pth.tar", type=str, help="URL of the checkpoint you'd like to convert.", ) parser.add_argument( "--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory." ) lowercase_ = parser.parse_args() convert_vit_msn_checkpoint(args.checkpoint_url, args.pytorch_dump_folder_path)
708
"""simple docstring""" from typing import TYPE_CHECKING from ...file_utils import _LazyModule, is_tokenizers_available, is_torch_available from ...utils import OptionalDependencyNotAvailable lowercase_ = {"configuration_gpt_neox": ["GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXConfig"]} try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["GPTNeoXTokenizerFast"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST", "GPTNeoXForCausalLM", "GPTNeoXForQuestionAnswering", "GPTNeoXForSequenceClassification", "GPTNeoXForTokenClassification", "GPTNeoXLayer", "GPTNeoXModel", "GPTNeoXPreTrainedModel", ] if TYPE_CHECKING: from .configuration_gpt_neox import GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXConfig try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .tokenization_gpt_neox_fast import GPTNeoXTokenizerFast try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_gpt_neox import ( GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST, GPTNeoXForCausalLM, GPTNeoXForQuestionAnswering, GPTNeoXForSequenceClassification, GPTNeoXForTokenClassification, GPTNeoXLayer, GPTNeoXModel, GPTNeoXPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
65
0
"""simple docstring""" from collections.abc import Callable class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a = None ): # Stores actual heap items. __a = [] # Stores indexes of each item for supporting updates and deletion. __a = {} # Stores current size of heap. __a = 0 # Stores function used to evaluate the score of an item on which basis ordering # will be done. __a = key or (lambda _a : x) def __UpperCAmelCase ( self , _a ): return int((i - 1) / 2 ) if i > 0 else None def __UpperCAmelCase ( self , _a ): __a = int(2 * i + 1 ) return left if 0 < left < self.size else None def __UpperCAmelCase ( self , _a ): __a = int(2 * i + 2 ) return right if 0 < right < self.size else None def __UpperCAmelCase ( self , _a , _a ): __a , __a = ( self.pos_map[self.arr[j][0]], self.pos_map[self.arr[i][0]], ) # Then swap the items in the list. __a , __a = self.arr[j], self.arr[i] def __UpperCAmelCase ( self , _a , _a ): return self.arr[i][1] < self.arr[j][1] def __UpperCAmelCase ( self , _a ): __a = self._left(_a ) __a = self._right(_a ) __a = i if left is not None and not self._cmp(_a , _a ): __a = left if right is not None and not self._cmp(_a , _a ): __a = right return valid_parent def __UpperCAmelCase ( self , _a ): __a = self._parent(_a ) while parent is not None and not self._cmp(_a , _a ): self._swap(_a , _a ) __a , __a = parent, self._parent(_a ) def __UpperCAmelCase ( self , _a ): __a = self._get_valid_parent(_a ) while valid_parent != index: self._swap(_a , _a ) __a , __a = valid_parent, self._get_valid_parent(_a ) def __UpperCAmelCase ( self , _a , _a ): if item not in self.pos_map: return __a = self.pos_map[item] __a = [item, self.key(_a )] # Make sure heap is right in both up and down direction. # Ideally only one of them will make any change. self._heapify_up(_a ) self._heapify_down(_a ) def __UpperCAmelCase ( self , _a ): if item not in self.pos_map: return __a = self.pos_map[item] del self.pos_map[item] __a = self.arr[self.size - 1] __a = index self.size -= 1 # Make sure heap is right in both up and down direction. Ideally only one # of them will make any change- so no performance loss in calling both. if self.size > index: self._heapify_up(_a ) self._heapify_down(_a ) def __UpperCAmelCase ( self , _a , _a ): __a = len(self.arr ) if arr_len == self.size: self.arr.append([item, self.key(_a )] ) else: __a = [item, self.key(_a )] __a = self.size self.size += 1 self._heapify_up(self.size - 1 ) def __UpperCAmelCase ( self ): return self.arr[0] if self.size else None def __UpperCAmelCase ( self ): __a = self.get_top() if top_item_tuple: self.delete_item(top_item_tuple[0] ) return top_item_tuple def lowercase ( ) -> None: pass if __name__ == "__main__": import doctest doctest.testmod()
709
"""simple docstring""" import unittest from transformers import MraConfig, is_torch_available from transformers.testing_utils import require_torch, slow, torch_device from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask if is_torch_available(): import torch from transformers import ( MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, MraModel, ) from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=2 , _a=8 , _a=True , _a=True , _a=True , _a=True , _a=99 , _a=16 , _a=5 , _a=2 , _a=36 , _a="gelu" , _a=0.0 , _a=0.0 , _a=512 , _a=16 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ): __a = parent __a = batch_size __a = seq_length __a = is_training __a = use_input_mask __a = use_token_type_ids __a = use_labels __a = vocab_size __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = type_sequence_label_size __a = initializer_range __a = num_labels __a = num_choices __a = scope def __UpperCAmelCase ( self ): __a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) __a = None if self.use_input_mask: __a = random_attention_mask([self.batch_size, self.seq_length] ) __a = None if self.use_token_type_ids: __a = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size ) __a = None __a = None __a = None if self.use_labels: __a = ids_tensor([self.batch_size] , self.type_sequence_label_size ) __a = ids_tensor([self.batch_size, self.seq_length] , self.num_labels ) __a = ids_tensor([self.batch_size] , self.num_choices ) __a = self.get_config() return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels def __UpperCAmelCase ( self ): return MraConfig( 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=_a , initializer_range=self.initializer_range , ) def __UpperCAmelCase ( self ): __a = self.get_config() __a = 300 return config def __UpperCAmelCase ( self ): ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = self.prepare_config_and_inputs() __a = True __a = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] ) __a = 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraModel(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a ) __a = model(_a , token_type_ids=_a ) __a = model(_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , _a , _a , ): __a = True __a = MraModel(_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , encoder_attention_mask=_a , ) __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , ) __a = model(_a , attention_mask=_a , token_type_ids=_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForMaskedLM(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForQuestionAnswering(config=_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , start_positions=_a , end_positions=_a , ) 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForSequenceClassification(_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForTokenClassification(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_choices __a = MraForMultipleChoice(config=_a ) model.to(_a ) model.eval() __a = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = model( _a , attention_mask=_a , token_type_ids=_a , labels=_a , ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) ) def __UpperCAmelCase ( self ): __a = self.prepare_config_and_inputs() ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = config_and_inputs __a = {'''input_ids''': input_ids, '''token_type_ids''': token_type_ids, '''attention_mask''': input_mask} return config, inputs_dict @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = ( ( MraModel, MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, ) if is_torch_available() else () ) __UpperCAmelCase : str = False __UpperCAmelCase : Any = False __UpperCAmelCase : List[Any] = False __UpperCAmelCase : Optional[Any] = False __UpperCAmelCase : Dict = () def __UpperCAmelCase ( self ): __a = MraModelTester(self ) __a = ConfigTester(self , config_class=_a , hidden_size=37 ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() for type in ["absolute", "relative_key", "relative_key_query"]: __a = type self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_masked_lm(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_multiple_choice(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_question_answering(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_sequence_classification(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_token_classification(*_a ) @slow def __UpperCAmelCase ( self ): for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: __a = MraModel.from_pretrained(_a ) self.assertIsNotNone(_a ) @unittest.skip(reason='''MRA does not output attentions''' ) def __UpperCAmelCase ( self ): return @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @slow def __UpperCAmelCase ( self ): __a = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = torch.Size((1, 256, 768) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[-0.0140, 0.0830, -0.0381], [0.1546, 0.1402, 0.0220], [0.1162, 0.0851, 0.0165]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 256, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[9.2595, -3.6038, 11.8819], [9.3869, -3.2693, 11.0956], [11.8524, -3.4938, 13.1210]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' ) __a = torch.arange(4_096 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 4_096, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[5.4789, -2.3564, 7.5064], [7.9067, -1.3369, 9.9668], [9.0712, -1.8106, 7.0380]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) )
65
0
"""simple docstring""" import dataclasses import json import warnings from dataclasses import dataclass, field from time import time from typing import List from ..utils import logging lowercase_ = logging.get_logger(__name__) def lowercase ( lowerCAmelCase__ : Union[str, Any]=None , lowerCAmelCase__ : Any=None ) -> int: return field(default_factory=lambda: default , metadata=lowerCAmelCase__ ) @dataclass class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : List[str] = list_field( default=[] , metadata={ 'help': ( 'Model checkpoints to be provided to the AutoModel classes. Leave blank to benchmark the base version' ' of all available models' ) } , ) __UpperCAmelCase : List[int] = list_field( default=[8] , metadata={'help': 'List of batch sizes for which memory and time performance will be evaluated'} ) __UpperCAmelCase : List[int] = list_field( default=[8, 3_2, 1_2_8, 5_1_2] , metadata={'help': 'List of sequence lengths for which memory and time performance will be evaluated'} , ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Whether to benchmark inference of model. Inference can be disabled via --no-inference.'} , ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Whether to run on available cuda devices. Cuda can be disabled via --no-cuda.'} , ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Whether to run on available tpu devices. TPU can be disabled via --no-tpu.'} ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Use FP16 to accelerate inference.'} ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Benchmark training of model'} ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Verbose memory tracing'} ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Whether to perform speed measurements. Speed measurements can be disabled via --no-speed.'} , ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={ 'help': 'Whether to perform memory measurements. Memory measurements can be disabled via --no-memory' } , ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Trace memory line by line'} ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Save result to a CSV file'} ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Save all print statements in a log file'} ) __UpperCAmelCase : bool = field(default=__SCREAMING_SNAKE_CASE , metadata={'help': 'Whether to print environment information'} ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={ 'help': ( 'Whether to use multiprocessing for memory and speed measurement. It is highly recommended to use' ' multiprocessing for accurate CPU and GPU memory measurements. This option should only be disabled' ' for debugging / testing and on TPU.' ) } , ) __UpperCAmelCase : str = field( default=f'inference_time_{round(time() )}.csv' , metadata={'help': 'CSV filename used if saving time results to csv.'} , ) __UpperCAmelCase : str = field( default=f'inference_memory_{round(time() )}.csv' , metadata={'help': 'CSV filename used if saving memory results to csv.'} , ) __UpperCAmelCase : str = field( default=f'train_time_{round(time() )}.csv' , metadata={'help': 'CSV filename used if saving time results to csv for training.'} , ) __UpperCAmelCase : str = field( default=f'train_memory_{round(time() )}.csv' , metadata={'help': 'CSV filename used if saving memory results to csv for training.'} , ) __UpperCAmelCase : str = field( default=f'env_info_{round(time() )}.csv' , metadata={'help': 'CSV filename used if saving environment information.'} , ) __UpperCAmelCase : str = field( default=f'log_{round(time() )}.csv' , metadata={'help': 'Log filename used if print statements are saved in log.'} , ) __UpperCAmelCase : int = field(default=3 , metadata={'help': 'Times an experiment will be run.'} ) __UpperCAmelCase : bool = field( default=__SCREAMING_SNAKE_CASE , metadata={ 'help': ( 'Instead of loading the model as defined in `config.architectures` if exists, just load the pretrain' ' model weights.' ) } , ) def __UpperCAmelCase ( self ): warnings.warn( f'''The class {self.__class__} is deprecated. Hugging Face Benchmarking utils''' ''' are deprecated in general and it is advised to use external Benchmarking libraries ''' ''' to benchmark Transformer models.''' , _a , ) def __UpperCAmelCase ( self ): return json.dumps(dataclasses.asdict(self ) , indent=2 ) @property def __UpperCAmelCase ( self ): if len(self.models ) <= 0: raise ValueError( '''Please make sure you provide at least one model name / model identifier, *e.g.* `--models''' ''' bert-base-cased` or `args.models = [\'bert-base-cased\'].''' ) return self.models @property def __UpperCAmelCase ( self ): if not self.multi_process: return False elif self.is_tpu: logger.info('''Multiprocessing is currently not possible on TPU.''' ) return False else: return True
710
"""simple docstring""" from __future__ import annotations def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> list[str]: if partitions <= 0: raise ValueError('''partitions must be a positive number!''' ) if partitions > number_of_bytes: raise ValueError('''partitions can not > number_of_bytes!''' ) __a = number_of_bytes // partitions __a = [] for i in range(lowerCAmelCase__ ): __a = i * bytes_per_partition + 1 __a = ( number_of_bytes if i == partitions - 1 else (i + 1) * bytes_per_partition ) allocation_list.append(f'''{start_bytes}-{end_bytes}''' ) return allocation_list if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" lowercase_ = range(2, 2_0 + 1) lowercase_ = [1_0**k for k in range(ks[-1] + 1)] lowercase_ = {} def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : List[str] ) -> List[Any]: __a = sum(a_i[j] for j in range(lowerCAmelCase__ , len(lowerCAmelCase__ ) ) ) __a = sum(a_i[j] * base[j] for j in range(min(len(lowerCAmelCase__ ) , lowerCAmelCase__ ) ) ) __a , __a = 0, 0 __a = n - i __a = memo.get(lowerCAmelCase__ ) if sub_memo is not None: __a = sub_memo.get(lowerCAmelCase__ ) if jumps is not None and len(lowerCAmelCase__ ) > 0: # find and make the largest jump without going over __a = -1 for _k in range(len(lowerCAmelCase__ ) - 1 , -1 , -1 ): if jumps[_k][2] <= k and jumps[_k][1] <= max_dn: __a = _k break if max_jump >= 0: __a , __a , __a = jumps[max_jump] # since the difference between jumps is cached, add c __a = diff + c for j in range(min(lowerCAmelCase__ , len(lowerCAmelCase__ ) ) ): __a , __a = divmod(lowerCAmelCase__ , 10 ) if new_c > 0: add(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) else: __a = [] else: __a = {c: []} __a = sub_memo if dn >= max_dn or c + diff >= base[k]: return diff, dn if k > ks[0]: while True: # keep doing smaller jumps __a , __a = next_term(lowerCAmelCase__ , k - 1 , i + dn , lowerCAmelCase__ ) diff += _diff dn += terms_jumped if dn >= max_dn or c + diff >= base[k]: break else: # would be too small a jump, just compute sequential terms instead __a , __a = compute(lowerCAmelCase__ , lowerCAmelCase__ , i + dn , lowerCAmelCase__ ) diff += _diff dn += terms_jumped __a = sub_memo[c] # keep jumps sorted by # of terms skipped __a = 0 while j < len(lowerCAmelCase__ ): if jumps[j][1] > dn: break j += 1 # cache the jump for this value digitsum(b) and c sub_memo[c].insert(lowerCAmelCase__ , (diff, dn, k) ) return (diff, dn) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> Tuple: if i >= n: return 0, i if k > len(lowerCAmelCase__ ): a_i.extend([0 for _ in range(k - len(lowerCAmelCase__ ) )] ) # note: a_i -> b * 10^k + c # ds_b -> digitsum(b) # ds_c -> digitsum(c) __a = i __a , __a , __a = 0, 0, 0 for j in range(len(lowerCAmelCase__ ) ): if j >= k: ds_b += a_i[j] else: ds_c += a_i[j] while i < n: i += 1 __a = ds_c + ds_b diff += addend __a = 0 for j in range(lowerCAmelCase__ ): __a = a_i[j] + addend __a , __a = divmod(lowerCAmelCase__ , 10 ) ds_c += a_i[j] if addend > 0: break if addend > 0: add(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) return diff, i - start_i def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : int , lowerCAmelCase__ : List[Any] ) -> Optional[int]: for j in range(lowerCAmelCase__ , len(lowerCAmelCase__ ) ): __a = digits[j] + addend if s >= 10: __a , __a = divmod(lowerCAmelCase__ , 10 ) __a = addend // 10 + quotient else: __a = s __a = addend // 10 if addend == 0: break while addend > 0: __a , __a = divmod(lowerCAmelCase__ , 10 ) digits.append(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int = 10**15 ) -> int: __a = [1] __a = 1 __a = 0 while True: __a , __a = next_term(lowerCAmelCase__ , 20 , i + dn , lowerCAmelCase__ ) dn += terms_jumped if dn == n - i: break __a = 0 for j in range(len(lowerCAmelCase__ ) ): a_n += digits[j] * 10**j return a_n if __name__ == "__main__": print(F'''{solution() = }''')
711
"""simple docstring""" def lowercase ( lowerCAmelCase__ : list , lowerCAmelCase__ : list , lowerCAmelCase__ : int ) -> list: __a = len(lowerCAmelCase__ ) __a = [[0] * n for i in range(lowerCAmelCase__ )] for i in range(lowerCAmelCase__ ): __a = y_points[i] for i in range(2 , lowerCAmelCase__ ): for j in range(lowerCAmelCase__ , lowerCAmelCase__ ): __a = ( (xa - x_points[j - i + 1]) * q[j][i - 1] - (xa - x_points[j]) * q[j - 1][i - 1] ) / (x_points[j] - x_points[j - i + 1]) return [q[n - 1][n - 1], q] if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import os from shutil import copyfile from typing import List, Optional, Tuple from ...tokenization_utils import AddedToken from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import is_sentencepiece_available, logging if is_sentencepiece_available(): from .tokenization_albert import AlbertTokenizer else: lowercase_ = None lowercase_ = logging.get_logger(__name__) lowercase_ = {"vocab_file": "spiece.model", "tokenizer_file": "tokenizer.json"} lowercase_ = { "vocab_file": { "albert-base-v1": "https://huggingface.co/albert-base-v1/resolve/main/spiece.model", "albert-large-v1": "https://huggingface.co/albert-large-v1/resolve/main/spiece.model", "albert-xlarge-v1": "https://huggingface.co/albert-xlarge-v1/resolve/main/spiece.model", "albert-xxlarge-v1": "https://huggingface.co/albert-xxlarge-v1/resolve/main/spiece.model", "albert-base-v2": "https://huggingface.co/albert-base-v2/resolve/main/spiece.model", "albert-large-v2": "https://huggingface.co/albert-large-v2/resolve/main/spiece.model", "albert-xlarge-v2": "https://huggingface.co/albert-xlarge-v2/resolve/main/spiece.model", "albert-xxlarge-v2": "https://huggingface.co/albert-xxlarge-v2/resolve/main/spiece.model", }, "tokenizer_file": { "albert-base-v1": "https://huggingface.co/albert-base-v1/resolve/main/tokenizer.json", "albert-large-v1": "https://huggingface.co/albert-large-v1/resolve/main/tokenizer.json", "albert-xlarge-v1": "https://huggingface.co/albert-xlarge-v1/resolve/main/tokenizer.json", "albert-xxlarge-v1": "https://huggingface.co/albert-xxlarge-v1/resolve/main/tokenizer.json", "albert-base-v2": "https://huggingface.co/albert-base-v2/resolve/main/tokenizer.json", "albert-large-v2": "https://huggingface.co/albert-large-v2/resolve/main/tokenizer.json", "albert-xlarge-v2": "https://huggingface.co/albert-xlarge-v2/resolve/main/tokenizer.json", "albert-xxlarge-v2": "https://huggingface.co/albert-xxlarge-v2/resolve/main/tokenizer.json", }, } lowercase_ = { "albert-base-v1": 5_1_2, "albert-large-v1": 5_1_2, "albert-xlarge-v1": 5_1_2, "albert-xxlarge-v1": 5_1_2, "albert-base-v2": 5_1_2, "albert-large-v2": 5_1_2, "albert-xlarge-v2": 5_1_2, "albert-xxlarge-v2": 5_1_2, } lowercase_ = "▁" class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = VOCAB_FILES_NAMES __UpperCAmelCase : Tuple = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : Optional[Any] = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : int = AlbertTokenizer def __init__( self , _a=None , _a=None , _a=True , _a=True , _a=False , _a="[CLS]" , _a="[SEP]" , _a="<unk>" , _a="[SEP]" , _a="<pad>" , _a="[CLS]" , _a="[MASK]" , **_a , ): # Mask token behave like a normal word, i.e. include the space before it and # is included in the raw text, there should be a match in a non-normalized sentence. __a = ( AddedToken(_a , lstrip=_a , rstrip=_a , normalized=_a ) if isinstance(_a , _a ) else mask_token ) super().__init__( _a , tokenizer_file=_a , do_lower_case=_a , remove_space=_a , keep_accents=_a , bos_token=_a , eos_token=_a , unk_token=_a , sep_token=_a , pad_token=_a , cls_token=_a , mask_token=_a , **_a , ) __a = do_lower_case __a = remove_space __a = keep_accents __a = vocab_file __a = False if not self.vocab_file else True def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [self.cls_token_id] if token_ids_a is None: return cls + token_ids_a + sep return cls + token_ids_a + sep + token_ids_a + sep def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [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 __UpperCAmelCase ( self , _a , _a = None ): if not self.can_save_slow_tokenizer: raise ValueError( '''Your fast tokenizer does not have the necessary information to save the vocabulary for a slow ''' '''tokenizer.''' ) if not os.path.isdir(_a ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] ) if os.path.abspath(self.vocab_file ) != os.path.abspath(_a ): copyfile(self.vocab_file , _a ) return (out_vocab_file,)
712
"""simple docstring""" import json import os from functools import lru_cache from typing import List, Optional, Tuple import regex as re from ...tokenization_utils import AddedToken, PreTrainedTokenizer from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = {"vocab_file": "vocab.json", "merges_file": "merges.txt"} lowercase_ = { "vocab_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/vocab.json", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/vocab.json" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/vocab.json" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/vocab.json" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/vocab.json" ), }, "merges_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/merges.txt", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/merges.txt" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/merges.txt" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/merges.txt" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/merges.txt" ), }, } lowercase_ = { "allenai/longformer-base-4096": 4_0_9_6, "allenai/longformer-large-4096": 4_0_9_6, "allenai/longformer-large-4096-finetuned-triviaqa": 4_0_9_6, "allenai/longformer-base-4096-extra.pos.embd.only": 4_0_9_6, "allenai/longformer-large-4096-extra.pos.embd.only": 4_0_9_6, } @lru_cache() # Copied from transformers.models.roberta.tokenization_roberta.bytes_to_unicode def lowercase ( ) -> Union[str, Any]: __a = ( list(range(ord('''!''' ) , ord('''~''' ) + 1 ) ) + list(range(ord('''¡''' ) , ord('''¬''' ) + 1 ) ) + list(range(ord('''®''' ) , ord('''ÿ''' ) + 1 ) ) ) __a = bs[:] __a = 0 for b in range(2**8 ): if b not in bs: bs.append(lowerCAmelCase__ ) cs.append(2**8 + n ) n += 1 __a = [chr(lowerCAmelCase__ ) for n in cs] return dict(zip(lowerCAmelCase__ , lowerCAmelCase__ ) ) def lowercase ( lowerCAmelCase__ : Tuple ) -> Tuple: __a = set() __a = word[0] for char in word[1:]: pairs.add((prev_char, char) ) __a = char return pairs class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Optional[int] = VOCAB_FILES_NAMES __UpperCAmelCase : int = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : Dict = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : Union[str, Any] = ['input_ids', 'attention_mask'] def __init__( self , _a , _a , _a="replace" , _a="<s>" , _a="</s>" , _a="</s>" , _a="<s>" , _a="<unk>" , _a="<pad>" , _a="<mask>" , _a=False , **_a , ): __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else bos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else eos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else sep_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else cls_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else unk_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else pad_token # Mask token behave like a normal word, i.e. include the space before it __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else mask_token super().__init__( errors=_a , bos_token=_a , eos_token=_a , unk_token=_a , sep_token=_a , cls_token=_a , pad_token=_a , mask_token=_a , add_prefix_space=_a , **_a , ) with open(_a , encoding='''utf-8''' ) as vocab_handle: __a = json.load(_a ) __a = {v: k for k, v in self.encoder.items()} __a = errors # how to handle errors in decoding __a = bytes_to_unicode() __a = {v: k for k, v in self.byte_encoder.items()} with open(_a , encoding='''utf-8''' ) as merges_handle: __a = merges_handle.read().split('''\n''' )[1:-1] __a = [tuple(merge.split() ) for merge in bpe_merges] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = {} __a = add_prefix_space # Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions __a = re.compile(R'''\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+''' ) @property def __UpperCAmelCase ( self ): return len(self.encoder ) def __UpperCAmelCase ( self ): return dict(self.encoder , **self.added_tokens_encoder ) def __UpperCAmelCase ( self , _a ): if token in self.cache: return self.cache[token] __a = tuple(_a ) __a = get_pairs(_a ) if not pairs: return token while True: __a = min(_a , key=lambda _a : self.bpe_ranks.get(_a , float('''inf''' ) ) ) if bigram not in self.bpe_ranks: break __a , __a = bigram __a = [] __a = 0 while i < len(_a ): try: __a = word.index(_a , _a ) except ValueError: new_word.extend(word[i:] ) break else: new_word.extend(word[i:j] ) __a = j if word[i] == first and i < len(_a ) - 1 and word[i + 1] == second: new_word.append(first + second ) i += 2 else: new_word.append(word[i] ) i += 1 __a = tuple(_a ) __a = new_word if len(_a ) == 1: break else: __a = get_pairs(_a ) __a = ''' '''.join(_a ) __a = word return word def __UpperCAmelCase ( self , _a ): __a = [] for token in re.findall(self.pat , _a ): __a = ''''''.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(_a ).split(''' ''' ) ) return bpe_tokens def __UpperCAmelCase ( self , _a ): return self.encoder.get(_a , self.encoder.get(self.unk_token ) ) def __UpperCAmelCase ( self , _a ): return self.decoder.get(_a ) def __UpperCAmelCase ( self , _a ): __a = ''''''.join(_a ) __a = bytearray([self.byte_decoder[c] for c in text] ).decode('''utf-8''' , errors=self.errors ) return text def __UpperCAmelCase ( self , _a , _a = None ): if not os.path.isdir(_a ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] ) __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] ) with open(_a , '''w''' , encoding='''utf-8''' ) as f: f.write(json.dumps(self.encoder , indent=2 , sort_keys=_a , ensure_ascii=_a ) + '''\n''' ) __a = 0 with open(_a , '''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 _a : 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!''' ) __a = token_index writer.write(''' '''.join(_a ) + '''\n''' ) index += 1 return vocab_file, merge_file def __UpperCAmelCase ( self , _a , _a = None ): if token_ids_a is None: return [self.cls_token_id] + token_ids_a + [self.sep_token_id] __a = [self.cls_token_id] __a = [self.sep_token_id] return cls + token_ids_a + sep + sep + token_ids_a + sep def __UpperCAmelCase ( self , _a , _a = None , _a = False ): if already_has_special_tokens: return super().get_special_tokens_mask( token_ids_a=_a , token_ids_a=_a , already_has_special_tokens=_a ) if token_ids_a is None: return [1] + ([0] * len(_a )) + [1] return [1] + ([0] * len(_a )) + [1, 1] + ([0] * len(_a )) + [1] def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [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 __UpperCAmelCase ( self , _a , _a=False , **_a ): __a = kwargs.pop('''add_prefix_space''' , self.add_prefix_space ) if (is_split_into_words or add_prefix_space) and (len(_a ) > 0 and not text[0].isspace()): __a = ''' ''' + text return (text, kwargs)
65
0
"""simple docstring""" import argparse import json import os import evaluate import torch from datasets import load_dataset from torch.optim import AdamW from torch.utils.data import DataLoader from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed from accelerate import Accelerator, DistributedType from accelerate.utils.deepspeed import DummyOptim, DummyScheduler lowercase_ = 1_6 lowercase_ = 3_2 def lowercase ( lowerCAmelCase__ : Accelerator , lowerCAmelCase__ : int = 16 , lowerCAmelCase__ : str = "bert-base-cased" ) -> Dict: __a = AutoTokenizer.from_pretrained(lowerCAmelCase__ ) __a = load_dataset('''glue''' , '''mrpc''' ) def tokenize_function(lowerCAmelCase__ : Union[str, Any] ): # max_length=None => use the model max length (it's actually the default) __a = tokenizer(examples['''sentence1'''] , examples['''sentence2'''] , truncation=lowerCAmelCase__ , max_length=lowerCAmelCase__ ) return outputs # Apply the method we just defined to all the examples in all the splits of the dataset __a = datasets.map( lowerCAmelCase__ , batched=lowerCAmelCase__ , remove_columns=['''idx''', '''sentence1''', '''sentence2'''] , load_from_cache_file=lowerCAmelCase__ ) # We also rename the 'label' column to 'labels' which is the expected name for labels by the models of the # transformers library __a = tokenized_datasets.rename_column('''label''' , '''labels''' ) def collate_fn(lowerCAmelCase__ : Optional[Any] ): # On TPU it's best to pad everything to the same length or training will be very slow. if accelerator.distributed_type == DistributedType.TPU: return tokenizer.pad(lowerCAmelCase__ , padding='''max_length''' , max_length=128 , return_tensors='''pt''' ) return tokenizer.pad(lowerCAmelCase__ , padding='''longest''' , return_tensors='''pt''' ) # Instantiate dataloaders. __a = DataLoader( tokenized_datasets['''train'''] , shuffle=lowerCAmelCase__ , collate_fn=lowerCAmelCase__ , batch_size=lowerCAmelCase__ ) __a = DataLoader( tokenized_datasets['''validation'''] , shuffle=lowerCAmelCase__ , collate_fn=lowerCAmelCase__ , batch_size=lowerCAmelCase__ ) return train_dataloader, eval_dataloader def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> Any: # Initialize accelerator __a = Accelerator() # Sample hyper-parameters for learning rate, batch size, seed and a few other HPs __a = config['''lr'''] __a = int(config['''num_epochs'''] ) __a = int(config['''seed'''] ) __a = int(config['''batch_size'''] ) __a = args.model_name_or_path set_seed(lowerCAmelCase__ ) __a , __a = get_dataloaders(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) # Instantiate the model (we build the model here so that the seed also control new weights initialization) __a = AutoModelForSequenceClassification.from_pretrained(lowerCAmelCase__ , return_dict=lowerCAmelCase__ ) # Instantiate optimizer __a = ( AdamW if accelerator.state.deepspeed_plugin is None or '''optimizer''' not in accelerator.state.deepspeed_plugin.deepspeed_config else DummyOptim ) __a = optimizer_cls(params=model.parameters() , lr=lowerCAmelCase__ ) if accelerator.state.deepspeed_plugin is not None: __a = accelerator.state.deepspeed_plugin.deepspeed_config[ '''gradient_accumulation_steps''' ] else: __a = 1 __a = (len(lowerCAmelCase__ ) * num_epochs) // gradient_accumulation_steps # Instantiate scheduler if ( accelerator.state.deepspeed_plugin is None or "scheduler" not in accelerator.state.deepspeed_plugin.deepspeed_config ): __a = get_linear_schedule_with_warmup( optimizer=lowerCAmelCase__ , num_warmup_steps=0 , num_training_steps=lowerCAmelCase__ , ) else: __a = DummyScheduler(lowerCAmelCase__ , total_num_steps=lowerCAmelCase__ , warmup_num_steps=0 ) # Prepare everything # There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the # prepare method. __a , __a , __a , __a , __a = accelerator.prepare( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) # We need to keep track of how many total steps we have iterated over __a = 0 # We also need to keep track of the stating epoch so files are named properly __a = 0 # Now we train the model __a = evaluate.load('''glue''' , '''mrpc''' ) __a = 0 __a = {} for epoch in range(lowerCAmelCase__ , lowerCAmelCase__ ): model.train() for step, batch in enumerate(lowerCAmelCase__ ): __a = model(**lowerCAmelCase__ ) __a = outputs.loss __a = loss / gradient_accumulation_steps accelerator.backward(lowerCAmelCase__ ) if step % gradient_accumulation_steps == 0: optimizer.step() lr_scheduler.step() optimizer.zero_grad() overall_step += 1 model.eval() __a = 0 for step, batch in enumerate(lowerCAmelCase__ ): # We could avoid this line since we set the accelerator with `device_placement=True`. batch.to(accelerator.device ) with torch.no_grad(): __a = model(**lowerCAmelCase__ ) __a = outputs.logits.argmax(dim=-1 ) # It is slightly faster to call this once, than multiple times __a , __a = accelerator.gather( (predictions, batch['''labels''']) ) # If we are in a multiprocess environment, the last batch has duplicates if accelerator.use_distributed: if step == len(lowerCAmelCase__ ) - 1: __a = predictions[: len(eval_dataloader.dataset ) - samples_seen] __a = references[: len(eval_dataloader.dataset ) - samples_seen] else: samples_seen += references.shape[0] metric.add_batch( predictions=lowerCAmelCase__ , references=lowerCAmelCase__ , ) __a = metric.compute() # Use accelerator.print to print only on the main process. accelerator.print(f'''epoch {epoch}:''' , lowerCAmelCase__ ) __a = eval_metric['''accuracy'''] if best_performance < eval_metric["accuracy"]: __a = eval_metric['''accuracy'''] if args.performance_lower_bound is not None: assert ( args.performance_lower_bound <= best_performance ), f'''Best performance metric {best_performance} is lower than the lower bound {args.performance_lower_bound}''' accelerator.wait_for_everyone() if accelerator.is_main_process: with open(os.path.join(args.output_dir , '''all_results.json''' ) , '''w''' ) as f: json.dump(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( ) -> List[Any]: __a = argparse.ArgumentParser(description='''Simple example of training script tracking peak GPU memory usage.''' ) parser.add_argument( '''--model_name_or_path''' , type=lowerCAmelCase__ , default='''bert-base-cased''' , help='''Path to pretrained model or model identifier from huggingface.co/models.''' , required=lowerCAmelCase__ , ) parser.add_argument( '''--output_dir''' , type=lowerCAmelCase__ , default='''.''' , help='''Optional save directory where all checkpoint folders will be stored. Default is the current working directory.''' , ) parser.add_argument( '''--performance_lower_bound''' , type=lowerCAmelCase__ , default=lowerCAmelCase__ , help='''Optional lower bound for the performance metric. If set, the training will throw error when the performance metric drops below this value.''' , ) parser.add_argument( '''--num_epochs''' , type=lowerCAmelCase__ , default=3 , help='''Number of train epochs.''' , ) __a = parser.parse_args() __a = {'''lr''': 2e-5, '''num_epochs''': args.num_epochs, '''seed''': 42, '''batch_size''': 16} training_function(lowerCAmelCase__ , lowerCAmelCase__ ) if __name__ == "__main__": main()
713
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "unc-nlp/lxmert-base-uncased": "https://huggingface.co/unc-nlp/lxmert-base-uncased/resolve/main/config.json", } class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = 'lxmert' __UpperCAmelCase : str = {} def __init__( self , _a=30_522 , _a=768 , _a=12 , _a=9_500 , _a=1_600 , _a=400 , _a=3_072 , _a="gelu" , _a=0.1 , _a=0.1 , _a=512 , _a=2 , _a=0.02 , _a=1E-12 , _a=9 , _a=5 , _a=5 , _a=2_048 , _a=4 , _a=6.67 , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , **_a , ): __a = vocab_size __a = hidden_size __a = num_attention_heads __a = hidden_act __a = intermediate_size __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = initializer_range __a = layer_norm_eps __a = num_qa_labels __a = num_object_labels __a = num_attr_labels __a = l_layers __a = x_layers __a = r_layers __a = visual_feat_dim __a = visual_pos_dim __a = visual_loss_normalizer __a = task_matched __a = task_mask_lm __a = task_obj_predict __a = task_qa __a = visual_obj_loss __a = visual_attr_loss __a = visual_feat_loss __a = {'''vision''': r_layers, '''cross_encoder''': x_layers, '''language''': l_layers} super().__init__(**_a )
65
0
"""simple docstring""" import copy from dataclasses import dataclass, field from typing import ClassVar, Dict from ..features import ClassLabel, Features, Value from .base import TaskTemplate @dataclass(frozen=__SCREAMING_SNAKE_CASE ) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = field(default='text-classification' , metadata={'include_in_asdict_even_if_is_default': True} ) __UpperCAmelCase : ClassVar[Features] = Features({'text': Value('string' )} ) __UpperCAmelCase : ClassVar[Features] = Features({'labels': ClassLabel} ) __UpperCAmelCase : str = "text" __UpperCAmelCase : str = "labels" def __UpperCAmelCase ( self , _a ): if self.label_column not in features: raise ValueError(f'''Column {self.label_column} is not present in features.''' ) if not isinstance(features[self.label_column] , _a ): raise ValueError(f'''Column {self.label_column} is not a ClassLabel.''' ) __a = copy.deepcopy(self ) __a = self.label_schema.copy() __a = features[self.label_column] __a = label_schema return task_template @property def __UpperCAmelCase ( self ): return { self.text_column: "text", self.label_column: "labels", }
714
"""simple docstring""" import itertools import math def lowercase ( lowerCAmelCase__ : int ) -> bool: 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(lowerCAmelCase__ ) + 1 ) , 6 ): if number % i == 0 or number % (i + 2) == 0: return False return True def lowercase ( ) -> int: __a = 2 while True: if is_prime(lowerCAmelCase__ ): yield num num += 1 def lowercase ( lowerCAmelCase__ : int = 10001 ) -> int: return next(itertools.islice(prime_generator() , nth - 1 , lowerCAmelCase__ ) ) if __name__ == "__main__": print(F'''{solution() = }''')
65
0
"""simple docstring""" class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): pass class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): pass class __lowerCAmelCase : def __init__( self ): __a = [ [], [], [], ] def __UpperCAmelCase ( self , _a , _a ): try: if len(self.queues[priority] ) >= 100: raise OverflowError('''Maximum queue size is 100''' ) self.queues[priority].append(_a ) except IndexError: raise ValueError('''Valid priorities are 0, 1, and 2''' ) def __UpperCAmelCase ( self ): for queue in self.queues: if queue: return queue.pop(0 ) raise UnderFlowError('''All queues are empty''' ) def __str__( self ): return "\n".join(f'''Priority {i}: {q}''' for i, q in enumerate(self.queues ) ) class __lowerCAmelCase : def __init__( self ): __a = [] def __UpperCAmelCase ( self , _a ): if len(self.queue ) == 100: raise OverFlowError('''Maximum queue size is 100''' ) self.queue.append(_a ) def __UpperCAmelCase ( self ): if not self.queue: raise UnderFlowError('''The queue is empty''' ) else: __a = min(self.queue ) self.queue.remove(_a ) return data def __str__( self ): return str(self.queue ) def lowercase ( ) -> Tuple: __a = FixedPriorityQueue() fpq.enqueue(0 , 10 ) fpq.enqueue(1 , 70 ) fpq.enqueue(0 , 100 ) fpq.enqueue(2 , 1 ) fpq.enqueue(2 , 5 ) fpq.enqueue(1 , 7 ) fpq.enqueue(2 , 4 ) fpq.enqueue(1 , 64 ) fpq.enqueue(0 , 128 ) print(lowerCAmelCase__ ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(lowerCAmelCase__ ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(fpq.dequeue() ) print(fpq.dequeue() ) def lowercase ( ) -> List[str]: __a = ElementPriorityQueue() epq.enqueue(10 ) epq.enqueue(70 ) epq.enqueue(100 ) epq.enqueue(1 ) epq.enqueue(5 ) epq.enqueue(7 ) epq.enqueue(4 ) epq.enqueue(64 ) epq.enqueue(128 ) print(lowerCAmelCase__ ) print(epq.dequeue() ) print(epq.dequeue() ) print(epq.dequeue() ) print(epq.dequeue() ) print(epq.dequeue() ) print(lowerCAmelCase__ ) print(epq.dequeue() ) print(epq.dequeue() ) print(epq.dequeue() ) print(epq.dequeue() ) print(epq.dequeue() ) if __name__ == "__main__": fixed_priority_queue() element_priority_queue()
715
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_mask2former": [ "MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "Mask2FormerConfig", ], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["Mask2FormerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST", "Mask2FormerForUniversalSegmentation", "Mask2FormerModel", "Mask2FormerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_maskaformer import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskaFormerConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_maskaformer import MaskaFormerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_maskaformer import ( MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST, MaskaFormerForUniversalSegmentation, MaskaFormerModel, MaskaFormerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
65
0
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "caidas/swin2sr-classicalsr-x2-64": ( "https://huggingface.co/caidas/swin2sr-classicalsr-x2-64/resolve/main/config.json" ), } class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = 'swin2sr' __UpperCAmelCase : Tuple = { 'hidden_size': 'embed_dim', 'num_attention_heads': 'num_heads', 'num_hidden_layers': 'num_layers', } def __init__( self , _a=64 , _a=1 , _a=3 , _a=180 , _a=[6, 6, 6, 6, 6, 6] , _a=[6, 6, 6, 6, 6, 6] , _a=8 , _a=2.0 , _a=True , _a=0.0 , _a=0.0 , _a=0.1 , _a="gelu" , _a=False , _a=0.02 , _a=1E-5 , _a=2 , _a=1.0 , _a="1conv" , _a="pixelshuffle" , **_a , ): super().__init__(**_a ) __a = image_size __a = patch_size __a = num_channels __a = embed_dim __a = depths __a = len(_a ) __a = num_heads __a = window_size __a = mlp_ratio __a = qkv_bias __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = drop_path_rate __a = hidden_act __a = use_absolute_embeddings __a = layer_norm_eps __a = initializer_range __a = upscale __a = img_range __a = resi_connection __a = upsampler
716
"""simple docstring""" from sklearn.metrics import fa_score import datasets lowercase_ = "\nThe F1 score is the harmonic mean of the precision and recall. It can be computed with the equation:\nF1 = 2 * (precision * recall) / (precision + recall)\n" lowercase_ = "\nArgs:\n predictions (`list` of `int`): Predicted labels.\n references (`list` of `int`): Ground truth labels.\n labels (`list` of `int`): The set of labels to include when `average` is not set to `'binary'`, and the order of the labels if `average` is `None`. Labels present in the data can be excluded, for example to calculate a multiclass average ignoring a majority negative class. Labels not present in the data will result in 0 components in a macro average. For multilabel targets, labels are column indices. By default, all labels in `predictions` and `references` are used in sorted order. Defaults to None.\n pos_label (`int`): The class to be considered the positive class, in the case where `average` is set to `binary`. Defaults to 1.\n average (`string`): This parameter is required for multiclass/multilabel targets. If set to `None`, the scores for each class are returned. Otherwise, this determines the type of averaging performed on the data. Defaults to `'binary'`.\n\n - 'binary': Only report results for the class specified by `pos_label`. This is applicable only if the classes found in `predictions` and `references` are binary.\n - 'micro': Calculate metrics globally by counting the total true positives, false negatives and false positives.\n - 'macro': Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.\n - 'weighted': Calculate metrics for each label, and find their average weighted by support (the number of true instances for each label). This alters `'macro'` to account for label imbalance. This option can result in an F-score that is not between precision and recall.\n - 'samples': Calculate metrics for each instance, and find their average (only meaningful for multilabel classification).\n sample_weight (`list` of `float`): Sample weights Defaults to None.\n\nReturns:\n f1 (`float` or `array` of `float`): F1 score or list of f1 scores, depending on the value passed to `average`. Minimum possible value is 0. Maximum possible value is 1. Higher f1 scores are better.\n\nExamples:\n\n Example 1-A simple binary example\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0])\n >>> print(results)\n {'f1': 0.5}\n\n Example 2-The same simple binary example as in Example 1, but with `pos_label` set to `0`.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], pos_label=0)\n >>> print(round(results['f1'], 2))\n 0.67\n\n Example 3-The same simple binary example as in Example 1, but with `sample_weight` included.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], sample_weight=[0.9, 0.5, 3.9, 1.2, 0.3])\n >>> print(round(results['f1'], 2))\n 0.35\n\n Example 4-A multiclass example, with different values for the `average` input.\n >>> predictions = [0, 2, 1, 0, 0, 1]\n >>> references = [0, 1, 2, 0, 1, 2]\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"macro\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"micro\")\n >>> print(round(results['f1'], 2))\n 0.33\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"weighted\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=None)\n >>> print(results)\n {'f1': array([0.8, 0. , 0. ])}\n" lowercase_ = "\n@article{scikit-learn,\n title={Scikit-learn: Machine Learning in {P}ython},\n author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.\n and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.\n and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and\n Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},\n journal={Journal of Machine Learning Research},\n volume={12},\n pages={2825--2830},\n year={2011}\n}\n" @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class __lowerCAmelCase ( datasets.Metric ): '''simple docstring''' def __UpperCAmelCase ( self ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { '''predictions''': datasets.Sequence(datasets.Value('''int32''' ) ), '''references''': datasets.Sequence(datasets.Value('''int32''' ) ), } if self.config_name == '''multilabel''' else { '''predictions''': datasets.Value('''int32''' ), '''references''': datasets.Value('''int32''' ), } ) , reference_urls=['''https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html'''] , ) def __UpperCAmelCase ( self , _a , _a , _a=None , _a=1 , _a="binary" , _a=None ): __a = fa_score( _a , _a , labels=_a , pos_label=_a , average=_a , sample_weight=_a ) return {"f1": float(_a ) if score.size == 1 else score}
65
0
"""simple docstring""" 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. lowercase_ = {"LayoutLMv2Config", "LayoutLMv3Config"} @is_pipeline_test class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING __UpperCAmelCase : Any = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING if model_mapping is not None: __UpperCAmelCase : Any = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP} if tf_model_mapping is not None: __UpperCAmelCase : Dict = { config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP } def __UpperCAmelCase ( self , _a , _a , _a ): __a = ZeroShotClassificationPipeline( model=_a , tokenizer=_a , candidate_labels=['''polics''', '''health'''] ) return classifier, ["Who are you voting for in 2020?", "My stomach hurts."] def __UpperCAmelCase ( self , _a , _a ): __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # No kwarg __a = classifier('''Who are you voting for in 2020?''' , ['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics, public health''' ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health'''] ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''This text is about {}''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # https://github.com/huggingface/transformers/issues/13846 __a = classifier(['''I am happy'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(1 ) ] , ) __a = classifier(['''I am happy''', '''I am sad'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(2 ) ] , ) with self.assertRaises(_a ): classifier('''''' , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier(_a , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels='''''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels=_a ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''Not formatting template''' , ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template=_a , ) self.run_entailment_id(_a ) def __UpperCAmelCase ( self , _a ): __a = zero_shot_classifier.model.config __a = config.labelaid __a = zero_shot_classifier.entailment_id __a = {'''LABEL_0''': 0, '''LABEL_1''': 1, '''LABEL_2''': 2} self.assertEqual(zero_shot_classifier.entailment_id , -1 ) __a = {'''entailment''': 0, '''neutral''': 1, '''contradiction''': 2} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 0, '''NON-ENTAIL''': 1} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 2, '''NEUTRAL''': 1, '''CONTR''': 0} self.assertEqual(zero_shot_classifier.entailment_id , 2 ) __a = original_labelaid self.assertEqual(_a , zero_shot_classifier.entailment_id ) @require_torch def __UpperCAmelCase ( self ): __a = 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 __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''pt''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @require_tf def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''tf''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @slow @require_torch def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''pt''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , ) @slow @require_tf def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''tf''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , )
717
"""simple docstring""" from ..utils import DummyObject, requires_backends class __lowerCAmelCase ( metaclass=__SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = ['onnx'] def __init__( self , *_a , **_a ): requires_backends(self , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] )
65
0
"""simple docstring""" class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): # we need a list not a string, so do something to change the type __a = arr.split(''',''' ) def __UpperCAmelCase ( self ): __a = [int(self.array[0] )] * len(self.array ) __a = [int(self.array[0] )] * len(self.array ) for i in range(1 , len(self.array ) ): __a = max( int(self.array[i] ) + sum_value[i - 1] , int(self.array[i] ) ) __a = max(sum_value[i] , rear[i - 1] ) return rear[len(self.array ) - 1] if __name__ == "__main__": lowercase_ = input("please input some numbers:") lowercase_ = SubArray(whole_array) lowercase_ = array.solve_sub_array() print(("the results is:", re))
718
"""simple docstring""" 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 lowercase_ = 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 __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=16 , _a=13 , _a=7 , _a=14 , _a=10 , _a=19 , _a=5 , _a=4 , _a=True , _a=16 , _a=2 , _a=4 , _a=4 , _a="gelu" , _a=0.1 , _a=0.1 , _a=[1, 2, 3, 4, 5] , _a=25 , _a=5 , ): __a = d_model __a = parent __a = batch_size __a = prediction_length __a = context_length __a = cardinality __a = num_time_features __a = lags_sequence __a = embedding_dimension __a = is_training __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = context_length __a = prediction_length + label_length __a = label_length __a = moving_average __a = autocorrelation_factor def __UpperCAmelCase ( self ): 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 __UpperCAmelCase ( self , _a ): __a = config.context_length + max(config.lags_sequence ) __a = ids_tensor([self.batch_size, 1] , config.cardinality[0] ) __a = floats_tensor([self.batch_size, _past_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, _past_length] ) __a = floats_tensor([self.batch_size, _past_length] ) > 0.5 # decoder inputs __a = floats_tensor([self.batch_size, config.prediction_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, config.prediction_length] ) __a = { '''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 __UpperCAmelCase ( self ): __a = self.get_config() __a = self.prepare_autoformer_inputs_dict(_a ) return config, inputs_dict def __UpperCAmelCase ( self ): __a , __a = self.prepare_config_and_inputs() return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = AutoformerModel(config=_a ).to(_a ).eval() __a = model(**_a ) __a = outputs.encoder_last_hidden_state __a = outputs.last_hidden_state with tempfile.TemporaryDirectory() as tmpdirname: __a = model.get_encoder() encoder.save_pretrained(_a ) __a = AutoformerEncoder.from_pretrained(_a ).to(_a ) __a , __a , __a , __a , __a = model.create_network_inputs(**_a ) __a , __a = model.decomposition_layer(transformer_inputs[:, : config.context_length, ...] ) __a = torch.cat( (transformer_inputs[:, : config.context_length, ...], feature[:, : config.context_length, ...]) , dim=-1 , ) __a = encoder(inputs_embeds=_a )[0] self.parent.assertTrue((encoder_last_hidden_state_a - encoder_last_hidden_state).abs().max().item() < 1E-3 ) __a = ( torch.mean(transformer_inputs[:, : config.context_length, ...] , dim=1 ) .unsqueeze(1 ) .repeat(1 , config.prediction_length , 1 ) ) __a = torch.zeros( [transformer_inputs.shape[0], config.prediction_length, transformer_inputs.shape[2]] , device=enc_input.device , ) __a = torch.cat( ( torch.cat((seasonal_input[:, -config.label_length :, ...], zeros) , dim=1 ), feature[:, config.context_length - config.label_length :, ...], ) , dim=-1 , ) __a = 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: __a = model.get_decoder() decoder.save_pretrained(_a ) __a = AutoformerDecoder.from_pretrained(_a ).to(_a ) __a = decoder( trend=_a , inputs_embeds=_a , encoder_hidden_states=_a , )[0] self.parent.assertTrue((last_hidden_state_a - last_hidden_state).abs().max().item() < 1E-3 ) @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Any = (AutoformerModel, AutoformerForPrediction) if is_torch_available() else () __UpperCAmelCase : int = (AutoformerForPrediction,) if is_torch_available() else () __UpperCAmelCase : Any = {'feature-extraction': AutoformerModel} if is_torch_available() else {} __UpperCAmelCase : int = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False __UpperCAmelCase : Any = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False def __UpperCAmelCase ( self ): __a = AutoformerModelTester(self ) __a = ConfigTester(self , config_class=_a , has_text_modality=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs() for model_class in self.all_model_classes: __a = model_class(_a ) with tempfile.TemporaryDirectory() as tmpdirname: model.save_pretrained(_a ) __a , __a = model_class.from_pretrained(_a , output_loading_info=_a ) self.assertEqual(info['''missing_keys'''] , [] ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.check_encoder_decoder_model_standalone(*_a ) @unittest.skip(reason='''Model has no tokens embeddings''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = inspect.signature(getattr(_a , '''forward''' ) ) # The main input is the name of the argument after `self` __a = list(model_signature.parameters.keys() )[1] self.assertEqual(AutoformerModel.main_input_name , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ) __a = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic __a = [*signature.parameters.keys()] __a = [ '''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(_a )] , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() __a = True __a = getattr(self.model_tester , '''seq_length''' , _a ) __a = getattr(self.model_tester , '''decoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''encoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''d_model''' , _a ) __a = getattr(self.model_tester , '''num_attention_heads''' , _a ) __a = d_model // num_attention_heads for model_class in self.all_model_classes: __a = True __a = False __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) # check that output_attentions also work using config del inputs_dict["output_attentions"] __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) self.assertListEqual( list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, encoder_seq_length, dim] , ) __a = len(_a ) __a = 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(_a , _a ) # decoder attentions __a = outputs.decoder_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = outputs.cross_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = True __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) self.assertEqual(out_len + 2 , len(_a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , 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 __UpperCAmelCase ( self ): super().test_retain_grad_hidden_states_attentions() def lowercase ( lowerCAmelCase__ : Optional[int]="train-batch.pt" ) -> List[str]: __a = hf_hub_download(repo_id='''hf-internal-testing/tourism-monthly-batch''' , filename=lowerCAmelCase__ , repo_type='''dataset''' ) __a = torch.load(lowerCAmelCase__ , map_location=lowerCAmelCase__ ) return batch @require_torch @slow class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = AutoformerModel.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch() with torch.no_grad(): __a = 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] __a = torch.Size( (64, model.config.prediction_length + model.config.label_length, model.config.feature_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[0.3593, -1.3398, 0.6330], [0.2279, 1.5396, -0.1792], [0.0450, 1.3225, -0.2335]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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 __a = torch.Size((64, model.config.context_length, model.config.d_model) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[-0.0734, -0.9036, 0.8358], [4.7186, 2.4113, 1.9581], [1.7953, 2.3558, 1.2970]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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'''] , ) __a = torch.Size((64, model.config.num_parallel_samples, model.config.prediction_length) ) self.assertEqual(outputs.sequences.shape , _a ) __a = torch.tensor([3130.6763, 4056.5293, 7053.0786] , device=_a ) __a = outputs.sequences.mean(dim=1 ) self.assertTrue(torch.allclose(mean_prediction[0, -3:] , _a , rtol=1E-1 ) )
65
0
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import ( OptionalDependencyNotAvailable, _LazyModule, is_sentencepiece_available, is_speech_available, is_torch_available, ) lowercase_ : Optional[int] = { "configuration_trocr": ["TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP", "TrOCRConfig"], "processing_trocr": ["TrOCRProcessor"], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ : str = [ "TROCR_PRETRAINED_MODEL_ARCHIVE_LIST", "TrOCRForCausalLM", "TrOCRPreTrainedModel", ] if TYPE_CHECKING: from .configuration_trocr import TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP, TrOCRConfig from .processing_trocr import TrOCRProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_trocr import TROCR_PRETRAINED_MODEL_ARCHIVE_LIST, TrOCRForCausalLM, TrOCRPreTrainedModel else: import sys lowercase_ : Any = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
719
"""simple docstring""" import inspect from typing import Optional, Union import numpy as np import PIL import torch from torch.nn import functional as F from torchvision import transforms from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DiffusionPipeline, DPMSolverMultistepScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, randn_tensor, ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> Any: if isinstance(lowerCAmelCase__ , torch.Tensor ): return image elif isinstance(lowerCAmelCase__ , PIL.Image.Image ): __a = [image] if isinstance(image[0] , PIL.Image.Image ): __a = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['''lanczos'''] ) )[None, :] for i in image] __a = np.concatenate(lowerCAmelCase__ , axis=0 ) __a = np.array(lowerCAmelCase__ ).astype(np.floataa ) / 2_55.0 __a = image.transpose(0 , 3 , 1 , 2 ) __a = 2.0 * image - 1.0 __a = torch.from_numpy(lowerCAmelCase__ ) elif isinstance(image[0] , torch.Tensor ): __a = torch.cat(lowerCAmelCase__ , dim=0 ) return image def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=0.99_95 ) -> int: if not isinstance(lowerCAmelCase__ , np.ndarray ): __a = True __a = va.device __a = va.cpu().numpy() __a = va.cpu().numpy() __a = np.sum(va * va / (np.linalg.norm(lowerCAmelCase__ ) * np.linalg.norm(lowerCAmelCase__ )) ) if np.abs(lowerCAmelCase__ ) > DOT_THRESHOLD: __a = (1 - t) * va + t * va else: __a = np.arccos(lowerCAmelCase__ ) __a = np.sin(lowerCAmelCase__ ) __a = theta_a * t __a = np.sin(lowerCAmelCase__ ) __a = np.sin(theta_a - theta_t ) / sin_theta_a __a = sin_theta_t / sin_theta_a __a = sa * va + sa * va if inputs_are_torch: __a = torch.from_numpy(lowerCAmelCase__ ).to(lowerCAmelCase__ ) return va def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> int: __a = F.normalize(lowerCAmelCase__ , dim=-1 ) __a = F.normalize(lowerCAmelCase__ , dim=-1 ) return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> List[str]: for param in model.parameters(): __a = value class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a , _a , _a , _a=None , _a=None , _a=None , ): super().__init__() self.register_modules( vae=_a , text_encoder=_a , clip_model=_a , tokenizer=_a , unet=_a , scheduler=_a , feature_extractor=_a , coca_model=_a , coca_tokenizer=_a , coca_transform=_a , ) __a = ( feature_extractor.size if isinstance(feature_extractor.size , _a ) else feature_extractor.size['''shortest_edge'''] ) __a = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std ) set_requires_grad(self.text_encoder , _a ) set_requires_grad(self.clip_model , _a ) def __UpperCAmelCase ( self , _a = "auto" ): if slice_size == "auto": # half the attention head size is usually a good trade-off between # speed and memory __a = self.unet.config.attention_head_dim // 2 self.unet.set_attention_slice(_a ) def __UpperCAmelCase ( self ): self.enable_attention_slicing(_a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self , _a , _a , _a ): # get the original timestep using init_timestep __a = min(int(num_inference_steps * strength ) , _a ) __a = max(num_inference_steps - init_timestep , 0 ) __a = self.scheduler.timesteps[t_start:] return timesteps, num_inference_steps - t_start def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a=None ): if not isinstance(_a , torch.Tensor ): raise ValueError(f'''`image` has to be of type `torch.Tensor` but is {type(_a )}''' ) __a = image.to(device=_a , dtype=_a ) if isinstance(_a , _a ): __a = [ self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_a ) ] __a = torch.cat(_a , dim=0 ) else: __a = self.vae.encode(_a ).latent_dist.sample(_a ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 0.1_8215 * init_latents __a = init_latents.repeat_interleave(_a , dim=0 ) __a = randn_tensor(init_latents.shape , generator=_a , device=_a , dtype=_a ) # get latents __a = self.scheduler.add_noise(_a , _a , _a ) __a = init_latents return latents def __UpperCAmelCase ( self , _a ): __a = self.coca_transform(_a ).unsqueeze(0 ) with torch.no_grad(), torch.cuda.amp.autocast(): __a = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) ) __a = self.coca_tokenizer.decode(generated[0].cpu().numpy() ) return generated.split('''<end_of_text>''' )[0].replace('''<start_of_text>''' , '''''' ).rstrip(''' .,''' ) def __UpperCAmelCase ( self , _a , _a ): __a = self.feature_extractor.preprocess(_a ) __a = torch.from_numpy(clip_image_input['''pixel_values'''][0] ).unsqueeze(0 ).to(self.device ).half() __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = image_embeddings_clip.repeat_interleave(_a , dim=0 ) return image_embeddings_clip @torch.enable_grad() def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , ): __a = latents.detach().requires_grad_() __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ): __a = self.scheduler.alphas_cumprod[timestep] __a = 1 - alpha_prod_t # compute predicted original sample from predicted noise also called # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf __a = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5 __a = torch.sqrt(_a ) __a = pred_original_sample * (fac) + latents * (1 - fac) elif isinstance(self.scheduler , _a ): __a = self.scheduler.sigmas[index] __a = latents - sigma * noise_pred else: raise ValueError(f'''scheduler type {type(self.scheduler )} not supported''' ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * sample __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = transforms.Resize(self.feature_extractor_size )(_a ) __a = self.normalize(_a ).to(latents.dtype ) __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = spherical_dist_loss(_a , _a ).mean() * clip_guidance_scale __a = -torch.autograd.grad(_a , _a )[0] if isinstance(self.scheduler , _a ): __a = latents.detach() + grads * (sigma**2) __a = noise_pred_original else: __a = noise_pred_original - torch.sqrt(_a ) * grads return noise_pred, latents @torch.no_grad() def __call__( self , _a , _a , _a = None , _a = None , _a = 512 , _a = 512 , _a = 0.6 , _a = 50 , _a = 7.5 , _a = 1 , _a = 0.0 , _a = 100 , _a = None , _a = "pil" , _a = True , _a = 0.8 , _a = 0.1 , _a = 0.1 , ): if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError(f'''You have passed {batch_size} batch_size, but only {len(_a )} generators.''' ) if height % 8 != 0 or width % 8 != 0: raise ValueError(f'''`height` and `width` have to be divisible by 8 but are {height} and {width}.''' ) if isinstance(_a , torch.Generator ) and batch_size > 1: __a = [generator] + [None] * (batch_size - 1) __a = [ ('''model''', self.coca_model is None), ('''tokenizer''', self.coca_tokenizer is None), ('''transform''', self.coca_transform is None), ] __a = [x[0] for x in coca_is_none if x[1]] __a = ''', '''.join(_a ) # generate prompts with coca model if prompt is None if content_prompt is None: if len(_a ): raise ValueError( f'''Content prompt is None and CoCa [{coca_is_none_str}] is None.''' f'''Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) if style_prompt is None: if len(_a ): raise ValueError( f'''Style prompt is None and CoCa [{coca_is_none_str}] is None.''' f''' Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) # get prompt text embeddings for content and style __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0] __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0] __a = slerp(_a , _a , _a ) # duplicate text embeddings for each generation per prompt __a = text_embeddings.repeat_interleave(_a , dim=0 ) # set timesteps __a = '''offset''' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() ) __a = {} if accepts_offset: __a = 1 self.scheduler.set_timesteps(_a , **_a ) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand self.scheduler.timesteps.to(self.device ) __a , __a = self.get_timesteps(_a , _a , self.device ) __a = timesteps[:1].repeat(_a ) # Preprocess image __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = slerp(_a , _a , _a ) if clip_guidance_scale > 0: __a = self.get_clip_image_embeddings(_a , _a ) __a = self.get_clip_image_embeddings(_a , _a ) __a = slerp( _a , _a , _a ) # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` # corresponds to doing no classifier free guidance. __a = guidance_scale > 1.0 # get unconditional embeddings for classifier free guidance if do_classifier_free_guidance: __a = content_text_input.input_ids.shape[-1] __a = self.tokenizer([''''''] , padding='''max_length''' , max_length=_a , return_tensors='''pt''' ) __a = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0] # duplicate unconditional embeddings for each generation per prompt __a = uncond_embeddings.repeat_interleave(_a , dim=0 ) # For classifier free guidance, we need to do two forward passes. # Here we concatenate the unconditional and text embeddings into a single batch # to avoid doing two forward passes __a = torch.cat([uncond_embeddings, text_embeddings] ) # get the initial random noise unless the user supplied it # Unlike in other pipelines, latents need to be generated in the target device # for 1-to-1 results reproducibility with the CompVis implementation. # However this currently doesn't work in `mps`. __a = (batch_size, self.unet.config.in_channels, height // 8, width // 8) __a = text_embeddings.dtype if latents is None: if self.device.type == "mps": # randn does not work reproducibly on mps __a = torch.randn(_a , generator=_a , device='''cpu''' , dtype=_a ).to( self.device ) else: __a = torch.randn(_a , generator=_a , device=self.device , dtype=_a ) else: if latents.shape != latents_shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {latents_shape}''' ) __a = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler __a = latents * self.scheduler.init_noise_sigma # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 # and should be between [0, 1] __a = '''eta''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) __a = {} if accepts_eta: __a = eta # check if the scheduler accepts generator __a = '''generator''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) if accepts_generator: __a = generator with self.progress_bar(total=_a ): for i, t in enumerate(_a ): # expand the latents if we are doing classifier free guidance __a = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample # perform classifier free guidance if do_classifier_free_guidance: __a , __a = noise_pred.chunk(2 ) __a = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) # perform clip guidance if clip_guidance_scale > 0: __a = ( text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings ) __a , __a = self.cond_fn( _a , _a , _a , _a , _a , _a , _a , ) # compute the previous noisy sample x_t -> x_t-1 __a = self.scheduler.step(_a , _a , _a , **_a ).prev_sample # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * latents __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image, None) return StableDiffusionPipelineOutput(images=_a , nsfw_content_detected=_a )
65
0
"""simple docstring""" from __future__ import annotations def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> list[str]: if partitions <= 0: raise ValueError('''partitions must be a positive number!''' ) if partitions > number_of_bytes: raise ValueError('''partitions can not > number_of_bytes!''' ) __a = number_of_bytes // partitions __a = [] for i in range(lowerCAmelCase__ ): __a = i * bytes_per_partition + 1 __a = ( number_of_bytes if i == partitions - 1 else (i + 1) * bytes_per_partition ) allocation_list.append(f'''{start_bytes}-{end_bytes}''' ) return allocation_list if __name__ == "__main__": import doctest doctest.testmod()
720
"""simple docstring""" # tests directory-specific settings - this file is run automatically # by pytest before any tests are run import sys import warnings from os.path import abspath, dirname, join # allow having multiple repository checkouts and not needing to remember to rerun # 'pip install -e .[dev]' when switching between checkouts and running tests. lowercase_ = abspath(join(dirname(dirname(dirname(__file__))), "src")) sys.path.insert(1, git_repo_path) # silence FutureWarning warnings in tests since often we can't act on them until # they become normal warnings - i.e. the tests still need to test the current functionality warnings.simplefilter(action="ignore", category=FutureWarning) def lowercase ( lowerCAmelCase__ : List[Any] ) -> str: from transformers.testing_utils import pytest_addoption_shared pytest_addoption_shared(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Union[str, Any]: from transformers.testing_utils import pytest_terminal_summary_main __a = terminalreporter.config.getoption('''--make-reports''' ) if make_reports: pytest_terminal_summary_main(lowerCAmelCase__ , id=lowerCAmelCase__ )
65
0
"""simple docstring""" from typing import Dict from .base import GenericTensor, Pipeline class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __UpperCAmelCase ( self , _a=None , _a=None , _a=None , **_a ): if tokenize_kwargs is None: __a = {} if truncation is not None: if "truncation" in tokenize_kwargs: raise ValueError( '''truncation parameter defined twice (given as keyword argument as well as in tokenize_kwargs)''' ) __a = truncation __a = tokenize_kwargs __a = {} if return_tensors is not None: __a = return_tensors return preprocess_params, {}, postprocess_params def __UpperCAmelCase ( self , _a , **_a ): __a = self.framework __a = self.tokenizer(_a , return_tensors=_a , **_a ) return model_inputs def __UpperCAmelCase ( self , _a ): __a = self.model(**_a ) return model_outputs def __UpperCAmelCase ( self , _a , _a=False ): # [0] is the first available tensor, logits or last_hidden_state. if return_tensors: return model_outputs[0] if self.framework == "pt": return model_outputs[0].tolist() elif self.framework == "tf": return model_outputs[0].numpy().tolist() def __call__( self , *_a , **_a ): return super().__call__(*_a , **_a )
721
"""simple docstring""" 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. lowercase_ = {"LayoutLMv2Config", "LayoutLMv3Config"} @is_pipeline_test class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING __UpperCAmelCase : Any = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING if model_mapping is not None: __UpperCAmelCase : Any = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP} if tf_model_mapping is not None: __UpperCAmelCase : Dict = { config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP } def __UpperCAmelCase ( self , _a , _a , _a ): __a = ZeroShotClassificationPipeline( model=_a , tokenizer=_a , candidate_labels=['''polics''', '''health'''] ) return classifier, ["Who are you voting for in 2020?", "My stomach hurts."] def __UpperCAmelCase ( self , _a , _a ): __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # No kwarg __a = classifier('''Who are you voting for in 2020?''' , ['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics, public health''' ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health'''] ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''This text is about {}''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # https://github.com/huggingface/transformers/issues/13846 __a = classifier(['''I am happy'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(1 ) ] , ) __a = classifier(['''I am happy''', '''I am sad'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(2 ) ] , ) with self.assertRaises(_a ): classifier('''''' , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier(_a , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels='''''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels=_a ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''Not formatting template''' , ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template=_a , ) self.run_entailment_id(_a ) def __UpperCAmelCase ( self , _a ): __a = zero_shot_classifier.model.config __a = config.labelaid __a = zero_shot_classifier.entailment_id __a = {'''LABEL_0''': 0, '''LABEL_1''': 1, '''LABEL_2''': 2} self.assertEqual(zero_shot_classifier.entailment_id , -1 ) __a = {'''entailment''': 0, '''neutral''': 1, '''contradiction''': 2} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 0, '''NON-ENTAIL''': 1} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 2, '''NEUTRAL''': 1, '''CONTR''': 0} self.assertEqual(zero_shot_classifier.entailment_id , 2 ) __a = original_labelaid self.assertEqual(_a , zero_shot_classifier.entailment_id ) @require_torch def __UpperCAmelCase ( self ): __a = 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 __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''pt''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @require_tf def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''tf''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @slow @require_torch def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''pt''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , ) @slow @require_tf def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''tf''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , )
65
0
"""simple docstring""" import unittest from transformers import is_torch_available from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device if is_torch_available(): from transformers import AutoModelForSeqaSeqLM, AutoTokenizer @require_torch @require_sentencepiece @require_tokenizers class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @slow def __UpperCAmelCase ( self ): __a = AutoModelForSeqaSeqLM.from_pretrained('''google/mt5-small''' , return_dict=_a ).to(_a ) __a = AutoTokenizer.from_pretrained('''google/mt5-small''' ) __a = tokenizer('''Hello there''' , return_tensors='''pt''' ).input_ids __a = tokenizer('''Hi I am''' , return_tensors='''pt''' ).input_ids __a = model(input_ids.to(_a ) , labels=labels.to(_a ) ).loss __a = -(labels.shape[-1] * loss.item()) __a = -84.9127 self.assertTrue(abs(mtf_score - EXPECTED_SCORE ) < 1E-4 )
700
"""simple docstring""" import pytest from datasets import inspect_metric, list_metrics, load_metric @pytest.fixture def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' , set() ) @pytest.fixture def lowercase ( lowerCAmelCase__ : Any ) -> Any: class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = metric_id class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Any = [MetricMock(__SCREAMING_SNAKE_CASE ) for metric_id in ['accuracy', 'mse', 'precision', 'codeparrot/apps_metric']] def __UpperCAmelCase ( self ): return self._metrics monkeypatch.setattr('''datasets.inspect.huggingface_hub''' , HfhMock() ) @pytest.mark.parametrize( '''func, args''' , [(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Tuple ) -> Optional[int]: if "tmp_path" in args: __a = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args ) with pytest.warns(lowerCAmelCase__ , match='''https://huggingface.co/docs/evaluate''' ): func(*lowerCAmelCase__ )
65
0
"""simple docstring""" import json import os import unittest from typing import Tuple from transformers import WavaVecaPhonemeCTCTokenizer from transformers.models.wavaveca.tokenization_wavaveca import VOCAB_FILES_NAMES from transformers.models.wavaveca_phoneme.tokenization_wavaveca_phoneme import WavaVecaPhonemeCTCTokenizerOutput from transformers.testing_utils import require_phonemizer from ...test_tokenization_common import TokenizerTesterMixin @require_phonemizer class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Union[str, Any] = WavaVecaPhonemeCTCTokenizer __UpperCAmelCase : Any = False def __UpperCAmelCase ( self ): super().setUp() __a = ( '''<s> <pad> </s> <unk> n s t ə l a i k d m ɛ ɾ e ɪ p o ɐ z ð f j v b ɹ ʁ ʊ iː r w ʌ u ɡ æ aɪ ʃ h ɔ ɑː ''' '''ŋ ɚ eɪ β uː y ɑ̃ oʊ ᵻ eː θ aʊ ts oː ɔ̃ ɣ ɜ ɑ dʒ əl x ɜː ç ʒ tʃ ɔː ɑːɹ ɛ̃ ʎ ɔːɹ ʋ aː ɕ œ ø oːɹ ɲ yː ''' '''ʔ iə i5 s. tɕ ?? nʲ ɛː œ̃ ɭ ɔø ʑ tʲ ɨ ɛɹ ts. rʲ ɪɹ ɭʲ i.5 ɔɪ q sʲ u5 ʊɹ iɜ a5 iɛ5 øː ʕ ja əɜ th ɑ5 ''' '''oɪ dʲ ə5 tɕh ts.h mʲ ɯ dʑ vʲ e̞ tʃʲ ei5 o5 onɡ5 ɑu5 iɑ5 ai5 aɪɚ kh ə1 ʐ i2 ʉ ħ t[ aɪə ʲ ju ə2 u2 oɜ ''' '''pː iɛɜ ou5 y5 uɜ tː uo5 d[ uoɜ tsh ɑɜ ɵ i̪5 uei5 ɟ aɜ ɑɨ i.ɜ eʊ o2 ɐ̃ ä pʲ kʲ n̩ ɒ ph ɑu2 uɨ əɪ ɫ ɬ ''' '''yɜ bʲ ɑ2 s̪ aiɜ χ ɐ̃ʊ̃ 1 ə4 yæɜ a2 ɨː t̪ iouɜ ũ onɡɜ aɨ iɛ2 ɔɨ ɑuɜ o̞ ei2 iou2 c kː y2 ɖ oe dˤ yɛɜ ''' '''əʊ S ɡʲ onɡ2 u" eiɜ ʈ ɯᵝ iou5 dZ r̝̊ i.2 tS s^ ʝ yə5 iɑɜ uə5 pf ɨu iɑ2 ou2 ər2 fʲ ai2 r̝ uəɜ ɳ əɨ ''' '''ua5 uɪ ɽ bː yu5 uo2 yɛ5 l̩ ɻ ərɜ ʂ i̪2 ouɜ uaɜ a. a.ː yæ5 dː r̩ ee ɪu ər5 i̪ ɜ æi u: i.ː t^ o1 ɪ^ ''' '''ai ueiɜ æː ɛɪ eə i. ɴ ie ua2 ɑ1 o4 tʃː o: ɑ: u1 N i̪1 au yæ2 u. qː yəɜ y: kʰ tʃʰ iʊ sx õ uo tʰ ''' '''uai5 bʰ u.ː uə2 ʊə d^ s̪ː yiɜ dʰ r. oe: i1 ɟː yu2 nʲʲ i̪4 uei2 tsʲ ɸ ĩ ɑ4 t̪ː eɑ u4 e: tsː ʈʰ ɡʰ ''' '''ɯɯ dʒʲ ʂʲ X ɵː uaiɜ tɕʲ ã t^ː ẽː yɛ2 cː i.1 ɛʊ dˤdˤ dʒː i4 ɡː yi ɕʲ ɟʰ pʰ dʑʲ yuɜ ua1 ua4 æiː ɐɐ ''' '''ui iou1 ʊː a1 iou4 cʰ iɛ1 yə2 ɖʰ ẽ ʒʲ ää ər4 iːː ɪː iɑ1 ər1 œː øi ɪuː cʰcʰ əː1 iː1 ũ kʰː o̞o̞ xʲ ''' '''ou1 iɛ4 e̞e̞ y1 dzː dʲʲ dʰː ɯᵝɯᵝ lː uo1 i.4 i: yɛ5ʲ a4''' ).split(''' ''' ) __a = dict(zip(_a , range(len(_a ) ) ) ) __a = {'''pad_token''': '''<pad>''', '''unk_token''': '''<unk>''', '''bos_token''': '''<s>''', '''eos_token''': '''</s>'''} __a = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''vocab_file'''] ) with open(self.vocab_file , '''w''' , encoding='''utf-8''' ) as fp: fp.write(json.dumps(_a ) + '''\n''' ) def __UpperCAmelCase ( self , _a , _a=False , _a=20 , _a=5 ): __a = [(i, tokenizer.decode([i] , clean_up_tokenization_spaces=_a )) for i in range(len(_a ) )] __a = list(filter(lambda _a : [t[0]] == tokenizer.encode(t[1] , do_phonemize=_a ) , _a ) ) if max_length is not None and len(_a ) > max_length: __a = toks[:max_length] if min_length is not None and len(_a ) < min_length and len(_a ) > 0: while len(_a ) < min_length: __a = toks + toks # toks_str = [t[1] for t in toks] __a = [t[0] for t in toks] # Ensure consistency __a = tokenizer.decode(_a , clean_up_tokenization_spaces=_a ) if " " not in output_txt and len(_a ) > 1: __a = ( tokenizer.decode([toks_ids[0]] , clean_up_tokenization_spaces=_a ) + ''' ''' + tokenizer.decode(toks_ids[1:] , clean_up_tokenization_spaces=_a ) ) if with_prefix_space: __a = ''' ''' + output_txt __a = tokenizer.encode(_a , add_special_tokens=_a ) return output_txt, output_ids def __UpperCAmelCase ( self , **_a ): kwargs.update(self.special_tokens_map ) return WavaVecaPhonemeCTCTokenizer.from_pretrained(self.tmpdirname , **_a ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) # check adding a single token tokenizer.add_tokens('''xxx''' ) __a = tokenizer('''m xxx ɪ''' , do_phonemize=_a ).input_ids self.assertEqual(_a , [13, 392, 17] ) # xxx should be last token tokenizer.add_tokens(['''aaa''', '''bbb''', '''ccc'''] ) __a = tokenizer('''m aaa ɪ ccc''' , do_phonemize=_a ).input_ids self.assertEqual(_a , [13, 393, 17, 395] ) # aaa and ccc should be after xxx and 2 after aaa __a = tokenizer('''maɪ c''' , do_phonemize=_a ).input_ids self.assertEqual(_a , [3, 200] ) # mai should be <unk> (=3) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) self.assertEqual(_a , '''h ə l oʊ h aʊ ɑːɹ j uː''' ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) self.assertEqual(tokenizer(_a ).input_ids , tokenizer(_a , do_phonemize=_a ).input_ids ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) __a = tokenizer.decode(tokenizer(_a ).input_ids ) self.assertEqual(_a , _a ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) __a = [ [11, 5, 15, tokenizer.pad_token_id, 15, 8, 98], [24, 22, 5, 24, 22, 5, 77], ] __a = tokenizer.decode(sample_ids[0] ) __a = tokenizer.batch_decode(_a ) self.assertEqual(_a , batch_tokens[0] ) self.assertEqual(_a , ['''k s ɾ ɾ l ɭʲ''', '''j ð s j ð s oːɹ'''] ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained( '''facebook/wav2vec2-lv-60-espeak-cv-ft''' , word_delimiter_token='''|''' ) tokenizer.add_tokens('''|''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) self.assertEqual(_a , '''h ə l oʊ | h aʊ | ɑːɹ | j uː |''' ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained( '''facebook/wav2vec2-lv-60-espeak-cv-ft''' , word_delimiter_token='''|''' ) tokenizer.add_tokens('''|''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) self.assertEqual(tokenizer(_a ).input_ids , tokenizer(_a , do_phonemize=_a ).input_ids ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained( '''facebook/wav2vec2-lv-60-espeak-cv-ft''' , word_delimiter_token='''|''' ) tokenizer.add_tokens('''|''' ) # fmt: off __a = [ [11, 5, 15, tokenizer.pad_token_id, tokenizer.word_delimiter_token_id, 15, 8, tokenizer.word_delimiter_token_id, 98], [tokenizer.word_delimiter_token_id, 24, 22, tokenizer.word_delimiter_token_id, 5, 24, 22, 5, 77], ] # fmt: on # decode with word_del_token filter __a = tokenizer.decode(sample_ids[0] ) __a = tokenizer.batch_decode(_a ) self.assertEqual(_a , batch_tokens[0] ) self.assertEqual(_a , ['''k s ɾ ɾ l ɭʲ''', '''j ð s j ð s oːɹ'''] ) # decode with no word_del_token filter __a = tokenizer.decode(sample_ids[0] , filter_word_delimiter_token=_a ) __a = tokenizer.batch_decode(_a , filter_word_delimiter_token=_a ) self.assertEqual(_a , batch_tokens[0] ) self.assertEqual(_a , ['''k s ɾ | ɾ l | ɭʲ''', '''| j ð | s j ð s oːɹ'''] ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained( '''facebook/wav2vec2-lv-60-espeak-cv-ft''' , word_delimiter_token='''|''' ) tokenizer.add_tokens('''|''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) __a = tokenizer.decode(tokenizer(_a ).input_ids , filter_word_delimiter_token=_a ) self.assertEqual(_a , _a ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained( '''facebook/wav2vec2-lv-60-espeak-cv-ft''' , word_delimiter_token='''|''' ) tokenizer.add_tokens('''|''' ) __a = '''Hello how are you''' __a = tokenizer.phonemize(_a , phonemizer_lang='''en-us''' ) __a = tokenizer.decode(tokenizer(_a ).input_ids , filter_word_delimiter_token=_a ) self.assertEqual(''' '''.join([p.strip() for p in phonemes.split(''' |''' )] ).strip() , _a ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained( '''facebook/wav2vec2-lv-60-espeak-cv-ft''' , word_delimiter_token=_a ) __a = '''Hello how are you''' __a = tokenizer(_a , phonemizer_lang='''en-us''' ).input_ids __a = tokenizer(_a , phonemizer_lang='''fr-fr''' ).input_ids self.assertNotEqual(_a , _a ) __a = tokenizer.decode(_a ) __a = tokenizer.decode(_a ) self.assertEqual(_a , '''h ə l oʊ h aʊ ɑːɹ j uː''' ) self.assertEqual(_a , '''ɛ l o h aʊ a ʁ j u''' ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) __a = '''Hello how Are you''' __a = '''hello how are you''' __a = tokenizer(_a ).input_ids __a = tokenizer(_a ).input_ids self.assertEqual(_a , _a ) def __UpperCAmelCase ( self ): __a = self.tokenizer_class.from_pretrained('''facebook/wav2vec2-lv-60-espeak-cv-ft''' ) tokenizer.add_tokens(['''!''', '''?'''] ) tokenizer.add_special_tokens({'''cls_token''': '''$$$'''} ) # fmt: off __a = [ [11, 5, 15, tokenizer.pad_token_id, 15, 8, 98, 392, 392, 393, 392, 392, 393, 394, 394], [24, 22, 5, 24, 22, 5, 77, tokenizer.pad_token_id, 394, 394], ] # fmt: on __a = tokenizer.batch_decode(_a ) self.assertEqual(_a , ['''k s ɾ ɾ l ɭʲ!?!? $$$''', '''j ð s j ð s oːɹ $$$'''] ) @staticmethod def __UpperCAmelCase ( _a , _a ): __a = [d[key] for d in offsets] return retrieved_list def __UpperCAmelCase ( self ): __a = self.get_tokenizer(word_delimiter_token='''|''' ) tokenizer.add_tokens('''|''' ) # fmt: off # ksssɾɾ|ɾɾ<pad>ɾɾ|<pad>ɾlll|ɭʲ -> k s ɾ ɾ | ɾ l | ɭʲ" __a = [11, 5, 5, 5, 15, 15, tokenizer.pad_token_id, 15, 15, tokenizer.word_delimiter_token_id, tokenizer.pad_token_id, 15, 8, 8, 8, tokenizer.word_delimiter_token_id, 98] # fmt: on __a = tokenizer.decode(_a , output_char_offsets=_a , filter_word_delimiter_token=_a ) # check Wav2Vec2CTCTokenizerOutput keys for char self.assertEqual(len(outputs.keys() ) , 2 ) self.assertTrue('''text''' in outputs ) self.assertTrue('''char_offsets''' in outputs ) self.assertTrue(isinstance(_a , _a ) ) # check that order of chars is correct and identical for both outputs self.assertEqual(''' '''.join(self.get_from_offsets(outputs['''char_offsets'''] , '''char''' ) ) , outputs.text ) self.assertListEqual( self.get_from_offsets(outputs['''char_offsets'''] , '''char''' ) , ['''k''', '''s''', '''ɾ''', '''ɾ''', '''|''', '''ɾ''', '''l''', '''|''', '''ɭʲ'''] ) # check that offsets are actually correct for char # 0-1 is 11, 1-4 is 5, 4-6 is first 15, 6-7 is <pad> (thus not shown), 7-9 is second 15, 9-10 is word_delimiter_token, # 10-11 is <pad> (thus not shown), 11-12 is third 15, 12-15 is 8, 15-16 is word_delimiter_token, 16-17 is 98 self.assertListEqual( self.get_from_offsets(outputs['''char_offsets'''] , '''start_offset''' ) , [0, 1, 4, 7, 9, 11, 12, 15, 16] ) self.assertListEqual( self.get_from_offsets(outputs['''char_offsets'''] , '''end_offset''' ) , [1, 4, 6, 9, 10, 12, 15, 16, 17] ) def __UpperCAmelCase ( self ): __a = self.get_tokenizer(word_delimiter_token='''|''' ) def check_list_tuples_equal(_a , _a ): self.assertTrue(isinstance(_a , _a ) ) self.assertTrue(isinstance(outputs_list[0] , _a ) ) # transform list to ModelOutput __a = WavaVecaPhonemeCTCTokenizerOutput( {k: [d[k] for d in outputs_list] for k in outputs_list[0]} ) self.assertListEqual(outputs_batch['''text'''] , outputs_batch_a['''text'''] ) def recursive_check(_a , _a ): if isinstance(_a , _a ): [recursive_check(_a , _a ) for la, la in zip(_a , _a )] self.assertEqual(_a , _a ) if "char_offsets" in outputs_batch: recursive_check(outputs_batch['''char_offsets'''] , outputs_batch_a['''char_offsets'''] ) # fmt: off __a = [ [11, 5, 15, tokenizer.pad_token_id, 15, 4, 8, 98, 32, 32, 32, 32, 4, 33, tokenizer.word_delimiter_token_id, 32, 32, 33, 34, 34], [24, 22, 5, tokenizer.word_delimiter_token_id, tokenizer.word_delimiter_token_id, 24, 22, 22, 22, 4, 5, 77, tokenizer.pad_token_id, 22, 22, 4, 34, 34, 34, 34], ] # fmt: on # We assume that `decode` works as expected. All we will check now is # the output type is correct and the output is identical to `decode` # char __a = tokenizer.batch_decode(_a , output_char_offsets=_a ) __a = [tokenizer.decode(_a , output_char_offsets=_a ) for ids in sample_ids] check_list_tuples_equal(_a , _a ) @unittest.skip('''Wav2Vec2PhonemeTokenizer always lower cases letters to correctly map to phonemes''' ) def __UpperCAmelCase ( self ): pass @unittest.skip('''Wav2Vec2PhonemeTokenizer always puts spaces between phonemes''' ) def __UpperCAmelCase ( self ): pass @unittest.skip('''encodes to text to ids, but decodes ids to phonemes -> not possible to have internal consistency''' ) def __UpperCAmelCase ( self ): pass @unittest.skip('''Wav2Vec2PhonemeModel has no max model length => no testing''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = self.get_tokenizers(do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) # We usually have added tokens from the start in tests because our vocab fixtures are # smaller than the original vocabs - let's not assert this # self.assertEqual(vocab_size, all_size) __a = ['''aaaaa bbbbbb''', '''cccccccccdddddddd'''] __a = tokenizer.add_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size + len(_a ) ) __a = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 4 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) __a = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''} __a = tokenizer.add_special_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size_a + len(_a ) ) __a = tokenizer.encode( '''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 6 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[0] , tokens[1] ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokens[-4] ) self.assertEqual(tokens[0] , tokenizer.eos_token_id ) self.assertEqual(tokens[-3] , tokenizer.pad_token_id ) @unittest.skip('''The tokenizer shouldn\'t be used to encode input IDs (except for labels), only to decode.''' ) def __UpperCAmelCase ( self ): pass @unittest.skip('''The tokenizer shouldn\'t be used to encode input IDs (except for labels), only to decode.''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): # The default common tokenizer tests assumes that the output of `convert_tokens_to_string` is a string which # is not the case for Wav2Vec2PhonemeCTCTokenizer. __a = self.get_tokenizers(fast=_a , do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = ['''ð''', '''ɪ''', '''s''', '''ɪ''', '''z''', '''ɐ''', '''t''', '''ɛ''', '''k''', '''s''', '''t'''] __a = tokenizer.convert_tokens_to_string(_a ) self.assertIsInstance(output['''text'''] , _a )
701
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> int: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or number < 0: raise ValueError('''Input must be a non-negative integer''' ) __a = 0 while number: # This way we arrive at next set bit (next 1) instead of looping # through each bit and checking for 1s hence the # loop won't run 32 times it will only run the number of `1` times number &= number - 1 count += 1 return count if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import os def lowercase ( ) -> List[Any]: __a = os.path.join(os.path.dirname(lowerCAmelCase__ ) , '''num.txt''' ) with open(lowerCAmelCase__ ) as file_hand: return str(sum(int(lowerCAmelCase__ ) for line in file_hand ) )[:10] if __name__ == "__main__": print(solution())
702
"""simple docstring""" import pytest from datasets import Dataset, DatasetDict, Features, NamedSplit, Value from datasets.io.text import TextDatasetReader from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.split == split if split else "train" @pytest.mark.parametrize('''path_type''' , [str, list] ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict: if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = text_path elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [text_path] __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for split in splits: __a = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str: __a = tmp_path / '''cache''' # CSV file loses col_1 string dtype information: default now is "int64" instead of "string" __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict: if split: __a = {split: text_path} else: __a = '''train''' __a = {'''train''': text_path, '''test''': text_path} __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() )
65
0
import unittest from transformers import JukeboxTokenizer from transformers.testing_utils import require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : List[str] = JukeboxTokenizer __UpperCAmelCase : Dict = { 'artist': 'Zac Brown Band', 'genres': 'Country', 'lyrics': 'I met a traveller from an antique land,\n Who said "Two vast and trunkless legs of stone\n Stand in the desert. . . . Near them, on the sand,\n Half sunk a shattered visage lies, whose frown,\n And wrinkled lip, and sneer of cold command,\n Tell that its sculptor well those passions read\n Which yet survive, stamped on these lifeless things,\n The hand that mocked them, and the heart that fed;\n And on the pedestal, these words appear:\n My name is Ozymandias, King of Kings;\n Look on my Works, ye Mighty, and despair!\n Nothing beside remains. Round the decay\n Of that colossal Wreck, boundless and bare\n The lone and level sands stretch far away\n ', } @require_torch def __UpperCAmelCase ( self ): import torch __a = JukeboxTokenizer.from_pretrained('''openai/jukebox-1b-lyrics''' ) __a = tokenizer(**self.metas )['''input_ids'''] # fmt: off __a = [ torch.tensor([[ 0, 0, 0, 7_169, 507, 9, 76, 39, 31, 46, 76, 27, 76, 46, 44, 27, 48, 31, 38, 38, 31, 44, 76, 32, 44, 41, 39, 76, 27, 40, 76, 27, 40, 46, 35, 43, 47, 31, 76, 38, 27, 40, 30, 64, 78, 76, 76, 76, 76, 76, 76, 76, 76, 23, 34, 41, 76, 45, 27, 35, 30, 76, 71, 20, 49, 41, 76, 48, 27, 45, 46, 76, 27, 40, 30, 76, 46, 44, 47, 40, 37, 38, 31, 45, 45, 76, 38, 31, 33, 45, 76, 41, 32, 76, 45, 46, 41, 40, 31, 78, 76, 76, 76, 76, 76, 76, 76, 76, 19, 46, 27, 40, 30, 76, 35, 40, 76, 46, 34, 31, 76, 30, 31, 45, 31, 44, 46, 63, 76, 63, 76, 63, 76, 63, 76, 14, 31, 27, 44, 76, 46, 34, 31, 39, 64, 76, 41, 40, 76, 46, 34, 31, 76, 45, 27, 40, 30, 64, 78, 76, 76, 76, 76, 76, 76, 76, 76, 8, 27, 38, 32, 76, 45, 47, 40, 37, 76, 27, 76, 45, 34, 27, 46, 46, 31, 44, 31, 30, 76, 48, 35, 45, 27, 33, 31, 76, 38, 35, 31, 45, 64, 76, 49, 34, 41, 45, 31, 76, 32, 44, 41, 49, 40, 64, 78, 76, 76, 76, 76, 76, 76, 76, 76, 1, 40, 30, 76, 49, 44, 35, 40, 37, 38, 31, 30, 76, 38, 35, 42, 64, 76, 27, 40, 30, 76, 45, 40, 31, 31, 44, 76, 41, 32, 76, 29, 41, 38, 30, 76, 29, 41, 39, 39, 27, 40, 30, 64, 78, 76, 76, 76, 76, 76, 76, 76, 76, 20, 31, 38, 38, 76, 46, 34, 27, 46, 76, 35, 46, 45, 76, 45, 29, 47, 38, 42, 46, 41, 44, 76, 49, 31, 38, 38, 76, 46, 34, 41, 45, 31, 76, 42, 27, 45, 45, 35, 41, 40, 45, 76, 44, 31, 27, 30, 78, 76, 76, 76, 76, 76, 76, 76, 76, 23, 34, 35, 29, 34, 76, 51, 31, 46, 76, 45, 47, 44, 48, 35, 48, 31, 64, 76, 45, 46, 27, 39, 42, 31, 30, 76, 41, 40, 76, 46, 34, 31, 45, 31, 76, 38, 35, 32, 31, 38, 31, 45, 45, 76, 46, 34, 35, 40, 33, 45, 64, 78, 76, 76, 76, 76, 76, 76, 76, 76, 20, 34, 31, 76, 34, 27, 40, 30, 76, 46, 34, 27, 46, 76, 39, 41, 29, 37, 31, 30, 76, 46, 34, 31, 39, 64, 76, 27, 40, 30, 76, 46, 34, 31, 76, 34, 31, 27, 44, 46, 76, 46, 34, 27, 46, 76, 32, 31, 30, 66, 78, 76, 76, 76, 76, 76, 76, 76, 76, 1, 40, 30, 76, 41, 40, 76, 46, 34, 31, 76, 42, 31, 30, 31, 45, 46, 27, 38, 64, 76, 46, 34, 31, 45, 31, 76, 49, 41, 44, 30, 45, 76, 27, 42, 42, 31, 27, 44, 65, 78, 76, 76, 76, 76, 76, 76, 76, 76, 13, 51, 76, 40, 27, 39, 31, 76, 35, 45, 76, 15, 52, 51, 39, 27, 40, 30, 35, 27, 45, 64, 76, 11, 35, 40, 33, 76, 41, 32, 76, 11, 35, 40, 33, 45, 66, 78, 76, 76, 76, 76, 76, 76, 76, 76, 12, 41, 41, 37, 76, 41, 40, 76, 39, 51, 76, 23, 41, 44, 37, 45, 64, 76, 51, 31, 76, 13, 35, 33, 34, 46, 51, 64, 76, 27, 40, 30, 76, 30, 31, 45, 42, 27, 35, 44, 67, 78, 76, 76, 76, 76, 76, 76, 76, 76, 14, 41, 46, 34, 35, 40, 33, 76, 28, 31, 45, 35, 30, 31, 76, 44, 31, 39, 27, 35, 40, 45, 63, 76, 18, 41, 47, 40, 30, 76, 46, 34, 31, 76, 30, 31, 29, 27, 51, 78, 76, 76, 76, 76, 76, 76, 76, 76, 15, 32, 76, 46, 34, 27, 46, 76, 29, 41, 38, 41, 45, 45, 27, 38, 76, 23, 44, 31, 29, 37, 64, 76, 28, 41, 47, 40, 30, 38, 31, 45, 45, 76, 27, 40, 30, 76, 28, 27, 44, 31, 78, 76, 76, 76, 76, 76, 76, 76, 76, 20, 34, 31, 76, 38, 41, 40, 31, 76, 27, 40, 30, 76, 38, 31, 48, 31, 38, 76, 45, 27, 40, 30, 45, 76, 45, 46, 44, 31, 46, 29, 34, 76, 32, 27, 44, 76, 27, 49, 27, 51, 78, 76, 76, 76, 76, 76, 76, 76, 76]] ), torch.tensor([[0, 0, 0, 1_069, 11]] ), torch.tensor([[0, 0, 0, 1_069, 11]] ), ] # fmt: on self.assertTrue(torch.allclose(tokens[0] , EXPECTED_OUTPUT[0] ) ) self.assertTrue(torch.allclose(tokens[1] , EXPECTED_OUTPUT[1] ) ) self.assertTrue(torch.allclose(tokens[2] , EXPECTED_OUTPUT[2] ) ) @require_torch def __UpperCAmelCase ( self ): import torch __a = JukeboxTokenizer.from_pretrained('''openai/jukebox-5b-lyrics''' ) __a = tokenizer(**self.metas )['''input_ids'''] # fmt: off __a = [ torch.tensor([[ 0, 0, 0, 1_069, 11, -1, -1, -1, -1, 9, 77, 39, 31, 46, 77, 27, 77, 46, 44, 27, 48, 31, 38, 38, 31, 44, 77, 32, 44, 41, 39, 77, 27, 40, 77, 27, 40, 46, 35, 43, 47, 31, 77, 38, 27, 40, 30, 64, 79, 77, 77, 77, 77, 77, 77, 77, 77, 23, 34, 41, 77, 45, 27, 35, 30, 77, 72, 20, 49, 41, 77, 48, 27, 45, 46, 77, 27, 40, 30, 77, 46, 44, 47, 40, 37, 38, 31, 45, 45, 77, 38, 31, 33, 45, 77, 41, 32, 77, 45, 46, 41, 40, 31, 79, 77, 77, 77, 77, 77, 77, 77, 77, 19, 46, 27, 40, 30, 77, 35, 40, 77, 46, 34, 31, 77, 30, 31, 45, 31, 44, 46, 63, 77, 63, 77, 63, 77, 63, 77, 14, 31, 27, 44, 77, 46, 34, 31, 39, 64, 77, 41, 40, 77, 46, 34, 31, 77, 45, 27, 40, 30, 64, 79, 77, 77, 77, 77, 77, 77, 77, 77, 8, 27, 38, 32, 77, 45, 47, 40, 37, 77, 27, 77, 45, 34, 27, 46, 46, 31, 44, 31, 30, 77, 48, 35, 45, 27, 33, 31, 77, 38, 35, 31, 45, 64, 77, 49, 34, 41, 45, 31, 77, 32, 44, 41, 49, 40, 64, 79, 77, 77, 77, 77, 77, 77, 77, 77, 1, 40, 30, 77, 49, 44, 35, 40, 37, 38, 31, 30, 77, 38, 35, 42, 64, 77, 27, 40, 30, 77, 45, 40, 31, 31, 44, 77, 41, 32, 77, 29, 41, 38, 30, 77, 29, 41, 39, 39, 27, 40, 30, 64, 79, 77, 77, 77, 77, 77, 77, 77, 77, 20, 31, 38, 38, 77, 46, 34, 27, 46, 77, 35, 46, 45, 77, 45, 29, 47, 38, 42, 46, 41, 44, 77, 49, 31, 38, 38, 77, 46, 34, 41, 45, 31, 77, 42, 27, 45, 45, 35, 41, 40, 45, 77, 44, 31, 27, 30, 79, 77, 77, 77, 77, 77, 77, 77, 77, 23, 34, 35, 29, 34, 77, 51, 31, 46, 77, 45, 47, 44, 48, 35, 48, 31, 64, 77, 45, 46, 27, 39, 42, 31, 30, 77, 41, 40, 77, 46, 34, 31, 45, 31, 77, 38, 35, 32, 31, 38, 31, 45, 45, 77, 46, 34, 35, 40, 33, 45, 64, 79, 77, 77, 77, 77, 77, 77, 77, 77, 20, 34, 31, 77, 34, 27, 40, 30, 77, 46, 34, 27, 46, 77, 39, 41, 29, 37, 31, 30, 77, 46, 34, 31, 39, 64, 77, 27, 40, 30, 77, 46, 34, 31, 77, 34, 31, 27, 44, 46, 77, 46, 34, 27, 46, 77, 32, 31, 30, 66, 79, 77, 77, 77, 77, 77, 77, 77, 77, 1, 40, 30, 77, 41, 40, 77, 46, 34, 31, 77, 42, 31, 30, 31, 45, 46, 27, 38, 64, 77, 46, 34, 31, 45, 31, 77, 49, 41, 44, 30, 45, 77, 27, 42, 42, 31, 27, 44, 65, 79, 77, 77, 77, 77, 77, 77, 77, 77, 13, 51, 77, 40, 27, 39, 31, 77, 35, 45, 77, 15, 52, 51, 39, 27, 40, 30, 35, 27, 45, 64, 77, 11, 35, 40, 33, 77, 41, 32, 77, 11, 35, 40, 33, 45, 66, 79, 77, 77, 77, 77, 77, 77, 77, 77, 12, 41, 41, 37, 77, 41, 40, 77, 39, 51, 77, 23, 41, 44, 37, 45, 64, 77, 51, 31, 77, 13, 35, 33, 34, 46, 51, 64, 77, 27, 40, 30, 77, 30, 31, 45, 42, 27, 35, 44, 67, 79, 77, 77, 77, 77, 77, 77, 77, 77, 14, 41, 46, 34, 35, 40, 33, 77, 28, 31, 45, 35, 30, 31, 77, 44, 31, 39, 27, 35, 40, 45, 63, 77, 18, 41, 47, 40, 30, 77, 46, 34, 31, 77, 30, 31, 29, 27, 51, 79, 77, 77, 77, 77, 77, 77, 77, 77, 15, 32, 77, 46, 34, 27, 46, 77, 29, 41, 38, 41, 45, 45, 27, 38, 77, 23, 44, 31, 29, 37, 64, 77, 28, 41, 47, 40, 30, 38, 31, 45, 45, 77, 27, 40, 30, 77, 28, 27, 44, 31, 79, 77, 77, 77, 77, 77, 77, 77, 77, 20, 34, 31, 77, 38, 41, 40, 31, 77, 27, 40, 30, 77, 38, 31, 48, 31, 38, 77, 45, 27, 40, 30, 45, 77, 45, 46, 44, 31, 46, 29, 34, 77, 32, 27, 44, 77, 27, 49, 27, 51, 79, 77, 77, 77, 77, 77, 77, 77, 77]] ), torch.tensor([[0, 0, 0, 1_069, 11, -1, -1, -1, -1]] ), torch.tensor([[0, 0, 0, 1_069, 11, -1, -1, -1, -1]] ), ] # fmt: on self.assertTrue(torch.allclose(tokens[0] , EXPECTED_OUTPUT[0] ) ) self.assertTrue(torch.allclose(tokens[1] , EXPECTED_OUTPUT[1] ) ) self.assertTrue(torch.allclose(tokens[2] , EXPECTED_OUTPUT[2] ) )
703
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> bool: return numa ^ numa < 0 if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import argparse import logging import os from datetime import datetime import numpy as np import torch from torch import nn from torch.utils.data import DataLoader, RandomSampler, TensorDataset from tqdm import tqdm from transformers import GPTaLMHeadModel lowercase_ = logging.getLogger(__name__) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> str: # save results if os.path.exists(lowerCAmelCase__ ): if os.path.exists(os.path.join(lowerCAmelCase__ , '''config.json''' ) ) and os.path.isfile( os.path.join(lowerCAmelCase__ , '''config.json''' ) ): os.remove(os.path.join(lowerCAmelCase__ , '''config.json''' ) ) if os.path.exists(os.path.join(lowerCAmelCase__ , '''pytorch_model.bin''' ) ) and os.path.isfile( os.path.join(lowerCAmelCase__ , '''pytorch_model.bin''' ) ): os.remove(os.path.join(lowerCAmelCase__ , '''pytorch_model.bin''' ) ) else: os.makedirs(lowerCAmelCase__ ) model.save_pretrained(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : str=False ) -> Optional[Any]: __a = 2 if unlogit: __a = torch.pow(lowerCAmelCase__ , lowerCAmelCase__ ) __a = p * torch.log(lowerCAmelCase__ ) __a = 0 return -plogp.sum(dim=-1 ) def lowercase ( lowerCAmelCase__ : Optional[int] ) -> List[Any]: logger.info('''lv, h >\t''' + '''\t'''.join(f'''{x + 1}''' for x in range(len(lowerCAmelCase__ ) ) ) ) for row in range(len(lowerCAmelCase__ ) ): if tensor.dtype != torch.long: logger.info(f'''layer {row + 1}:\t''' + '''\t'''.join(f'''{x:.5f}''' for x in tensor[row].cpu().data ) ) else: logger.info(f'''layer {row + 1}:\t''' + '''\t'''.join(f'''{x:d}''' for x in tensor[row].cpu().data ) ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=True , lowerCAmelCase__ : int=True , lowerCAmelCase__ : str=None , lowerCAmelCase__ : List[Any]=False ) -> Optional[Any]: __a , __a = model.config.num_hidden_layers, model.config.num_attention_heads __a = torch.zeros(lowerCAmelCase__ , lowerCAmelCase__ ).to(args.device ) __a = torch.zeros(lowerCAmelCase__ , lowerCAmelCase__ ).to(args.device ) if head_mask is None: __a = torch.ones(lowerCAmelCase__ , lowerCAmelCase__ ).to(args.device ) head_mask.requires_grad_(requires_grad=lowerCAmelCase__ ) # If actually pruned attention multi-head, set head mask to None to avoid shape mismatch if actually_pruned: __a = None __a = 0.0 __a = 0.0 for step, inputs in enumerate(tqdm(lowerCAmelCase__ , desc='''Iteration''' , disable=args.local_rank not in [-1, 0] ) ): __a = tuple(t.to(args.device ) for t in inputs ) ((__a ) , ) = inputs # Do a forward pass (not with torch.no_grad() since we need gradients for importance score - see below) __a = model(lowerCAmelCase__ , labels=lowerCAmelCase__ , head_mask=lowerCAmelCase__ ) # (loss), lm_logits, presents, (all hidden_states), (attentions) __a , __a , __a = ( outputs[0], outputs[1], outputs[-1], ) # Loss and logits are the first, attention the last loss.backward() # Backpropagate to populate the gradients in the head mask total_loss += loss.detach().cpu().numpy() if compute_entropy: for layer, attn in enumerate(lowerCAmelCase__ ): __a = entropy(attn.detach() , lowerCAmelCase__ ) attn_entropy[layer] += masked_entropy.sum(-1 ).sum(0 ).sum(0 ).detach() if compute_importance: head_importance += head_mask.grad.abs().detach() tot_tokens += torch.ones_like(lowerCAmelCase__ ).float().detach().sum().data # Normalize attn_entropy /= tot_tokens head_importance /= tot_tokens # Layerwise importance normalization if not args.dont_normalize_importance_by_layer: __a = 2 __a = torch.pow(torch.pow(lowerCAmelCase__ , lowerCAmelCase__ ).sum(-1 ) , 1 / exponent ) head_importance /= norm_by_layer.unsqueeze(-1 ) + 1e-2_0 if not args.dont_normalize_global_importance: __a = (head_importance - head_importance.min()) / (head_importance.max() - head_importance.min()) # Print matrices if compute_entropy: logger.info('''Attention entropies''' ) print_ad_tensor(lowerCAmelCase__ ) if compute_importance: logger.info('''Head importance scores''' ) print_ad_tensor(lowerCAmelCase__ ) logger.info('''Head ranked by importance scores''' ) __a = torch.zeros(head_importance.numel() , dtype=torch.long , device=args.device ) __a = torch.arange( head_importance.numel() , device=args.device ) __a = head_ranks.view_as(lowerCAmelCase__ ) print_ad_tensor(lowerCAmelCase__ ) return attn_entropy, head_importance, total_loss def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Union[str, Any] ) -> str: __a , __a , __a = compute_heads_importance(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , compute_entropy=lowerCAmelCase__ ) __a = 1 / loss # instead of downsteam score use the LM loss logger.info('''Pruning: original score: %f, threshold: %f''' , lowerCAmelCase__ , original_score * args.masking_threshold ) __a = torch.ones_like(lowerCAmelCase__ ) __a = max(1 , int(new_head_mask.numel() * args.masking_amount ) ) __a = original_score while current_score >= original_score * args.masking_threshold: __a = new_head_mask.clone().detach() # save current head mask # heads from least important to most - keep only not-masked heads __a = float('''Inf''' ) __a = head_importance.view(-1 ).sort()[1] if len(lowerCAmelCase__ ) <= num_to_mask: print('''BREAK BY num_to_mask''' ) break # mask heads __a = current_heads_to_mask[:num_to_mask] logger.info('''Heads to mask: %s''' , str(current_heads_to_mask.tolist() ) ) __a = new_head_mask.view(-1 ) __a = 0.0 __a = new_head_mask.view_as(lowerCAmelCase__ ) __a = new_head_mask.clone().detach() print_ad_tensor(lowerCAmelCase__ ) # Compute metric and head importance again __a , __a , __a = compute_heads_importance( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , compute_entropy=lowerCAmelCase__ , head_mask=lowerCAmelCase__ ) __a = 1 / loss logger.info( '''Masking: current score: %f, remaining heads %d (%.1f percents)''' , lowerCAmelCase__ , new_head_mask.sum() , new_head_mask.sum() / new_head_mask.numel() * 100 , ) logger.info('''Final head mask''' ) print_ad_tensor(lowerCAmelCase__ ) np.save(os.path.join(args.output_dir , '''head_mask.npy''' ) , head_mask.detach().cpu().numpy() ) return head_mask def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Any ) -> Tuple: __a = datetime.now() __a , __a , __a = compute_heads_importance( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , compute_entropy=lowerCAmelCase__ , compute_importance=lowerCAmelCase__ , head_mask=lowerCAmelCase__ ) __a = 1 / loss __a = datetime.now() - before_time __a = sum(p.numel() for p in model.parameters() ) __a = { layer: (1 - head_mask[layer].long()).nonzero().squeeze().tolist() for layer in range(len(lowerCAmelCase__ ) ) } for k, v in heads_to_prune.items(): if isinstance(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [ v, ] assert sum(len(lowerCAmelCase__ ) for h in heads_to_prune.values() ) == (1 - head_mask.long()).sum().item() model.prune_heads(lowerCAmelCase__ ) __a = sum(p.numel() for p in model.parameters() ) __a = datetime.now() __a , __a , __a = compute_heads_importance( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , compute_entropy=lowerCAmelCase__ , compute_importance=lowerCAmelCase__ , head_mask=lowerCAmelCase__ , actually_pruned=lowerCAmelCase__ , ) __a = 1 / loss __a = datetime.now() - before_time logger.info( '''Pruning: original num of params: %.2e, after pruning %.2e (%.1f percents)''' , lowerCAmelCase__ , lowerCAmelCase__ , pruned_num_params / original_num_params * 100 , ) logger.info('''Pruning: score with masking: %f score with pruning: %f''' , lowerCAmelCase__ , lowerCAmelCase__ ) logger.info('''Pruning: speed ratio (original timing / new timing): %f percents''' , original_time / new_time * 100 ) save_model(lowerCAmelCase__ , args.output_dir ) def lowercase ( ) -> Optional[int]: __a = argparse.ArgumentParser() # Required parameters parser.add_argument( '''--data_dir''' , default=lowerCAmelCase__ , type=lowerCAmelCase__ , required=lowerCAmelCase__ , help='''The input data dir. Should contain the .tsv files (or other data files) for the task.''' , ) parser.add_argument( '''--model_name_or_path''' , default=lowerCAmelCase__ , type=lowerCAmelCase__ , required=lowerCAmelCase__ , help='''Path to pretrained model or model identifier from huggingface.co/models''' , ) parser.add_argument( '''--output_dir''' , default=lowerCAmelCase__ , type=lowerCAmelCase__ , required=lowerCAmelCase__ , help='''The output directory where the model predictions and checkpoints will be written.''' , ) # Other parameters parser.add_argument( '''--config_name''' , default='''''' , type=lowerCAmelCase__ , help='''Pretrained config name or path if not the same as model_name_or_path''' , ) parser.add_argument( '''--tokenizer_name''' , default='''''' , type=lowerCAmelCase__ , help='''Pretrained tokenizer name or path if not the same as model_name_or_path''' , ) parser.add_argument( '''--cache_dir''' , default=lowerCAmelCase__ , type=lowerCAmelCase__ , help='''Where do you want to store the pre-trained models downloaded from s3''' , ) parser.add_argument( '''--data_subset''' , type=lowerCAmelCase__ , default=-1 , help='''If > 0: limit the data to a subset of data_subset instances.''' ) parser.add_argument( '''--overwrite_output_dir''' , action='''store_true''' , help='''Whether to overwrite data in output directory''' ) parser.add_argument( '''--overwrite_cache''' , action='''store_true''' , help='''Overwrite the cached training and evaluation sets''' ) parser.add_argument( '''--dont_normalize_importance_by_layer''' , action='''store_true''' , help='''Don\'t normalize importance score by layers''' ) parser.add_argument( '''--dont_normalize_global_importance''' , action='''store_true''' , help='''Don\'t normalize all importance scores between 0 and 1''' , ) parser.add_argument( '''--try_masking''' , action='''store_true''' , help='''Whether to try to mask head until a threshold of accuracy.''' ) parser.add_argument( '''--masking_threshold''' , default=0.9 , type=lowerCAmelCase__ , help='''masking threshold in term of metrics (stop masking when metric < threshold * original metric value).''' , ) parser.add_argument( '''--masking_amount''' , default=0.1 , type=lowerCAmelCase__ , help='''Amount to heads to masking at each masking step.''' ) parser.add_argument('''--metric_name''' , default='''acc''' , type=lowerCAmelCase__ , help='''Metric to use for head masking.''' ) parser.add_argument( '''--max_seq_length''' , default=128 , type=lowerCAmelCase__ , help=( '''The maximum total input sequence length after WordPiece tokenization. \n''' '''Sequences longer than this will be truncated, sequences shorter padded.''' ) , ) parser.add_argument('''--batch_size''' , default=1 , type=lowerCAmelCase__ , help='''Batch size.''' ) parser.add_argument('''--seed''' , type=lowerCAmelCase__ , default=42 ) parser.add_argument('''--local_rank''' , type=lowerCAmelCase__ , default=-1 , help='''local_rank for distributed training on gpus''' ) parser.add_argument('''--no_cuda''' , action='''store_true''' , help='''Whether not to use CUDA when available''' ) parser.add_argument('''--server_ip''' , type=lowerCAmelCase__ , default='''''' , help='''Can be used for distant debugging.''' ) parser.add_argument('''--server_port''' , type=lowerCAmelCase__ , default='''''' , help='''Can be used for distant debugging.''' ) __a = parser.parse_args() if args.server_ip and args.server_port: # Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script import ptvsd print('''Waiting for debugger attach''' ) ptvsd.enable_attach(address=(args.server_ip, args.server_port) , redirect_output=lowerCAmelCase__ ) ptvsd.wait_for_attach() # Setup devices and distributed training if args.local_rank == -1 or args.no_cuda: __a = torch.device('''cuda''' if torch.cuda.is_available() and not args.no_cuda else '''cpu''' ) __a = 0 if args.no_cuda else torch.cuda.device_count() else: torch.cuda.set_device(args.local_rank ) __a = torch.device('''cuda''' , args.local_rank ) __a = 1 torch.distributed.init_process_group(backend='''nccl''' ) # Initializes the distributed backend # Setup logging logging.basicConfig(level=logging.INFO if args.local_rank in [-1, 0] else logging.WARN ) logger.info('''device: {} n_gpu: {}, distributed: {}'''.format(args.device , args.n_gpu , bool(args.local_rank != -1 ) ) ) __a = GPTaLMHeadModel.from_pretrained(args.model_name_or_path ) # Distributed and parallel training model.to(args.device ) if args.local_rank != -1: __a = nn.parallel.DistributedDataParallel( lowerCAmelCase__ , device_ids=[args.local_rank] , output_device=args.local_rank , find_unused_parameters=lowerCAmelCase__ ) elif args.n_gpu > 1: __a = nn.DataParallel(lowerCAmelCase__ ) # Print/save training arguments os.makedirs(args.output_dir , exist_ok=lowerCAmelCase__ ) torch.save(lowerCAmelCase__ , os.path.join(args.output_dir , '''run_args.bin''' ) ) logger.info('''Training/evaluation parameters %s''' , lowerCAmelCase__ ) # Prepare dataset __a = np.concatenate( [ np.loadtxt(args.data_dir , dtype=np.intaa ), ] ) __a = (torch.from_numpy(lowerCAmelCase__ ),) __a = TensorDataset(*lowerCAmelCase__ ) __a = RandomSampler(lowerCAmelCase__ ) __a = DataLoader(lowerCAmelCase__ , sampler=lowerCAmelCase__ , batch_size=args.batch_size ) # Compute head entropy and importance score compute_heads_importance(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) # Try head masking (set heads to zero until the score goes under a threshole) # and head pruning (remove masked heads and see the effect on the network) if args.try_masking and args.masking_threshold > 0.0 and args.masking_threshold < 1.0: __a = mask_heads(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) prune_heads(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) if __name__ == "__main__": main()
704
"""simple docstring""" import unittest from transformers import SPIECE_UNDERLINE from transformers.models.speechta import SpeechTaTokenizer from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow from transformers.tokenization_utils import AddedToken from ...test_tokenization_common import TokenizerTesterMixin lowercase_ = get_tests_dir("fixtures/test_sentencepiece_bpe_char.model") @require_sentencepiece @require_tokenizers class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : int = SpeechTaTokenizer __UpperCAmelCase : Tuple = False __UpperCAmelCase : Dict = True def __UpperCAmelCase ( self ): super().setUp() # We have a SentencePiece fixture for testing __a = SpeechTaTokenizer(_a ) __a = AddedToken('''<mask>''' , lstrip=_a , rstrip=_a ) __a = mask_token tokenizer.add_special_tokens({'''mask_token''': mask_token} ) tokenizer.add_tokens(['''<ctc_blank>'''] ) tokenizer.save_pretrained(self.tmpdirname ) def __UpperCAmelCase ( self , _a ): __a = '''this is a test''' __a = '''this is a test''' return input_text, output_text def __UpperCAmelCase ( self , _a , _a=False , _a=20 , _a=5 ): __a , __a = self.get_input_output_texts(_a ) __a = tokenizer.encode(_a , add_special_tokens=_a ) __a = tokenizer.decode(_a , clean_up_tokenization_spaces=_a ) return text, ids def __UpperCAmelCase ( self ): __a = '''<pad>''' __a = 1 self.assertEqual(self.get_tokenizer()._convert_token_to_id(_a ) , _a ) self.assertEqual(self.get_tokenizer()._convert_id_to_token(_a ) , _a ) def __UpperCAmelCase ( self ): __a = list(self.get_tokenizer().get_vocab().keys() ) self.assertEqual(vocab_keys[0] , '''<s>''' ) self.assertEqual(vocab_keys[1] , '''<pad>''' ) self.assertEqual(vocab_keys[-4] , '''œ''' ) self.assertEqual(vocab_keys[-2] , '''<mask>''' ) self.assertEqual(vocab_keys[-1] , '''<ctc_blank>''' ) self.assertEqual(len(_a ) , 81 ) def __UpperCAmelCase ( self ): self.assertEqual(self.get_tokenizer().vocab_size , 79 ) def __UpperCAmelCase ( self ): __a = self.get_tokenizers(do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) # We usually have added tokens from the start in tests because our vocab fixtures are # smaller than the original vocabs - let's not assert this # self.assertEqual(vocab_size, all_size) __a = ['''aaaaa bbbbbb''', '''cccccccccdddddddd'''] __a = tokenizer.add_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size + len(_a ) ) __a = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 4 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) __a = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''} __a = tokenizer.add_special_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size_a + len(_a ) ) __a = tokenizer.encode( '''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 6 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[0] , tokens[1] ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokens[-4] ) self.assertEqual(tokens[0] , tokenizer.eos_token_id ) self.assertEqual(tokens[-3] , tokenizer.pad_token_id ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tokenizer.tokenize('''This is a test''' ) # fmt: off self.assertListEqual(_a , [SPIECE_UNDERLINE, '''T''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''a''', SPIECE_UNDERLINE, '''t''', '''e''', '''s''', '''t'''] ) # fmt: on self.assertListEqual( tokenizer.convert_tokens_to_ids(_a ) , [4, 32, 11, 10, 12, 4, 10, 12, 4, 7, 4, 6, 5, 12, 6] , ) __a = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''92000''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) __a = tokenizer.convert_tokens_to_ids(_a ) # fmt: off self.assertListEqual(_a , [4, 30, 4, 20, 7, 12, 4, 25, 8, 13, 9, 4, 10, 9, 4, 3, 23, 4, 7, 9, 14, 4, 6, 11, 10, 12, 4, 10, 12, 4, 19, 7, 15, 12, 73, 26] ) # fmt: on __a = tokenizer.convert_ids_to_tokens(_a ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''<unk>''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) @slow def __UpperCAmelCase ( self ): # Use custom sequence because this tokenizer does not handle numbers. __a = [ '''Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides ''' '''general-purpose architectures (BERT, GPT, RoBERTa, XLM, DistilBert, XLNet...) for Natural ''' '''Language Understanding (NLU) and Natural Language Generation (NLG) with over thirty-two pretrained ''' '''models in one hundred plus languages and deep interoperability between Jax, PyTorch and TensorFlow.''', '''BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly ''' '''conditioning on both left and right context in all layers.''', '''The quick brown fox jumps over the lazy dog.''', ] # fmt: off __a = { '''input_ids''': [ [4, 32, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 64, 19, 8, 13, 18, 5, 13, 15, 22, 4, 28, 9, 8, 20, 9, 4, 7, 12, 4, 24, 22, 6, 8, 13, 17, 11, 39, 6, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 7, 9, 14, 4, 24, 22, 6, 8, 13, 17, 11, 39, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 39, 25, 5, 13, 6, 63, 4, 24, 13, 8, 27, 10, 14, 5, 12, 4, 21, 5, 9, 5, 13, 7, 15, 39, 24, 16, 13, 24, 8, 12, 5, 4, 7, 13, 17, 11, 10, 6, 5, 17, 6, 16, 13, 5, 12, 4, 64, 40, 47, 54, 32, 23, 4, 53, 49, 32, 23, 4, 54, 8, 40, 47, 54, 32, 7, 23, 4, 69, 52, 43, 23, 4, 51, 10, 12, 6, 10, 15, 40, 5, 13, 6, 23, 4, 69, 52, 48, 5, 6, 26, 26, 26, 63, 4, 19, 8, 13, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 61, 9, 14, 5, 13, 12, 6, 7, 9, 14, 10, 9, 21, 4, 64, 48, 52, 61, 63, 4, 7, 9, 14, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 53, 5, 9, 5, 13, 7, 6, 10, 8, 9, 4, 64, 48, 52, 53, 63, 4, 20, 10, 6, 11, 4, 8, 27, 5, 13, 4, 6, 11, 10, 13, 6, 22, 39, 6, 20, 8, 4, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 4, 18, 8, 14, 5, 15, 12, 4, 10, 9, 4, 8, 9, 5, 4, 11, 16, 9, 14, 13, 5, 14, 4, 24, 15, 16, 12, 4, 15, 7, 9, 21, 16, 7, 21, 5, 12, 4, 7, 9, 14, 4, 14, 5, 5, 24, 4, 10, 9, 6, 5, 13, 8, 24, 5, 13, 7, 25, 10, 15, 10, 6, 22, 4, 25, 5, 6, 20, 5, 5, 9, 4, 58, 7, 37, 23, 4, 49, 22, 32, 8, 13, 17, 11, 4, 7, 9, 14, 4, 32, 5, 9, 12, 8, 13, 55, 15, 8, 20, 26, 2], [4, 40, 47, 54, 32, 4, 10, 12, 4, 14, 5, 12, 10, 21, 9, 5, 14, 4, 6, 8, 4, 24, 13, 5, 39, 6, 13, 7, 10, 9, 4, 14, 5, 5, 24, 4, 25, 10, 14, 10, 13, 5, 17, 6, 10, 8, 9, 7, 15, 4, 13, 5, 24, 13, 5, 12, 5, 9, 6, 7, 6, 10, 8, 9, 12, 4, 19, 13, 8, 18, 4, 16, 9, 15, 7, 25, 5, 15, 5, 14, 4, 6, 5, 37, 6, 4, 25, 22, 4, 46, 8, 10, 9, 6, 15, 22, 4, 17, 8, 9, 14, 10, 6, 10, 8, 9, 10, 9, 21, 4, 8, 9, 4, 25, 8, 6, 11, 4, 15, 5, 19, 6, 4, 7, 9, 14, 4, 13, 10, 21, 11, 6, 4, 17, 8, 9, 6, 5, 37, 6, 4, 10, 9, 4, 7, 15, 15, 4, 15, 7, 22, 5, 13, 12, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [4, 32, 11, 5, 4, 45, 16, 10, 17, 28, 4, 25, 13, 8, 20, 9, 4, 19, 8, 37, 4, 46, 16, 18, 24, 12, 4, 8, 27, 5, 13, 4, 6, 11, 5, 4, 15, 7, 57, 22, 4, 14, 8, 21, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], ], '''attention_mask''': [ [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], ] } # fmt: on self.tokenizer_integration_test_util( expected_encoding=_a , model_name='''microsoft/speecht5_asr''' , revision='''c5ef64c71905caeccde0e4462ef3f9077224c524''' , sequences=_a , )
65
0
"""simple docstring""" import copy from dataclasses import dataclass, field from typing import ClassVar, Dict from ..features import ClassLabel, Features, Image from .base import TaskTemplate @dataclass(frozen=__SCREAMING_SNAKE_CASE ) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = field(default='image-classification' , metadata={'include_in_asdict_even_if_is_default': True} ) __UpperCAmelCase : ClassVar[Features] = Features({'image': Image()} ) __UpperCAmelCase : ClassVar[Features] = Features({'labels': ClassLabel} ) __UpperCAmelCase : str = "image" __UpperCAmelCase : str = "labels" def __UpperCAmelCase ( self , _a ): if self.label_column not in features: raise ValueError(f'''Column {self.label_column} is not present in features.''' ) if not isinstance(features[self.label_column] , _a ): raise ValueError(f'''Column {self.label_column} is not a ClassLabel.''' ) __a = copy.deepcopy(self ) __a = self.label_schema.copy() __a = features[self.label_column] __a = label_schema return task_template @property def __UpperCAmelCase ( self ): return { self.image_column: "image", self.label_column: "labels", }
705
"""simple docstring""" from typing import List, Optional, Union import numpy as np from ....audio_utils import mel_filter_bank, optimal_fft_length, spectrogram, window_function from ....feature_extraction_sequence_utils import SequenceFeatureExtractor from ....feature_extraction_utils import BatchFeature from ....file_utils import PaddingStrategy, TensorType from ....utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Any = ['input_features', 'attention_mask'] def __init__( self , _a=80 , _a=16_000 , _a=0.0 , _a=10 , _a=25 , _a="hamming_window" , _a=3_2768.0 , _a=0.97 , _a=1.0 , _a=True , _a=True , _a=False , **_a , ): super().__init__(feature_size=_a , sampling_rate=_a , padding_value=_a , **_a ) __a = feature_size __a = sampling_rate __a = padding_value __a = hop_length __a = win_length __a = frame_signal_scale __a = preemphasis_coeff __a = mel_floor __a = normalize_means __a = normalize_vars __a = win_function __a = return_attention_mask __a = win_length * sampling_rate // 1_000 __a = hop_length * sampling_rate // 1_000 __a = optimal_fft_length(self.sample_size ) __a = (self.n_fft // 2) + 1 def __UpperCAmelCase ( self , _a ): if self.win_function == "hamming_window": __a = window_function(window_length=self.sample_size , name=self.win_function , periodic=_a ) else: __a = window_function(window_length=self.sample_size , name=self.win_function ) __a = mel_filter_bank( num_frequency_bins=self.n_freqs , num_mel_filters=self.feature_size , min_frequency=0.0 , max_frequency=self.sampling_rate / 2.0 , sampling_rate=self.sampling_rate , ) __a = spectrogram( one_waveform * self.frame_signal_scale , window=_a , frame_length=self.sample_size , hop_length=self.sample_stride , fft_length=self.n_fft , center=_a , preemphasis=self.preemphasis_coeff , mel_filters=_a , mel_floor=self.mel_floor , log_mel='''log''' , ) return msfc_features.T def __UpperCAmelCase ( self , _a , _a , _a ): # make sure we normalize float32 arrays if self.normalize_means: __a = x[:input_length].mean(axis=0 ) __a = np.subtract(_a , _a ) if self.normalize_vars: __a = x[:input_length].std(axis=0 ) __a = np.divide(_a , _a ) if input_length < x.shape[0]: __a = padding_value # make sure array is in float32 __a = x.astype(np.floataa ) return x def __UpperCAmelCase ( self , _a , _a = None ): __a = attention_mask.sum(-1 ) if attention_mask is not None else [x.shape[0] for x in input_features] return [self._normalize_one(_a , _a , self.padding_value ) for x, n in zip(_a , _a )] def __call__( self , _a , _a = False , _a = None , _a = False , _a = None , _a = None , _a = None , _a = None , **_a , ): if sampling_rate is not None: if sampling_rate != self.sampling_rate: raise ValueError( f'''The model corresponding to this feature extractor: {self} was trained using a sampling rate of''' f''' {self.sampling_rate}. Please make sure that the provided `raw_speech` input was sampled with''' f''' {self.sampling_rate} and not {sampling_rate}.''' ) else: logger.warning( '''It is strongly recommended to pass the ``sampling_rate`` argument to this function. ''' '''Failing to do so can result in silent errors that might be hard to debug.''' ) __a = isinstance(_a , np.ndarray ) and len(raw_speech.shape ) > 1 if is_batched_numpy and len(raw_speech.shape ) > 2: raise ValueError(f'''Only mono-channel audio is supported for input to {self}''' ) __a = is_batched_numpy or ( isinstance(_a , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) )) ) if is_batched: __a = [np.asarray(_a , dtype=np.floataa ) for speech in raw_speech] elif not is_batched and not isinstance(_a , np.ndarray ): __a = np.asarray(_a , dtype=np.floataa ) elif isinstance(_a , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ): __a = raw_speech.astype(np.floataa ) # always return batch if not is_batched: __a = [raw_speech] # extract fbank features __a = [self._extract_mfsc_features(_a ) for one_waveform in raw_speech] # convert into correct format for padding __a = BatchFeature({'''input_features''': features} ) __a = self.pad( _a , padding=_a , max_length=_a , truncation=_a , pad_to_multiple_of=_a , return_attention_mask=_a , **_a , ) # make sure list is in array format __a = padded_inputs.get('''input_features''' ) if isinstance(input_features[0] , _a ): __a = [np.asarray(_a , dtype=np.floataa ) for feature in input_features] __a = padded_inputs.get('''attention_mask''' ) if attention_mask is not None: __a = [np.asarray(_a , dtype=np.intaa ) for array in attention_mask] if self.normalize_means or self.normalize_vars: __a = ( np.array(_a , dtype=np.intaa ) if self._get_padding_strategies(_a , max_length=_a ) is not PaddingStrategy.DO_NOT_PAD and padding else None ) __a = self.normalize( padded_inputs['''input_features'''] , attention_mask=_a ) if return_tensors is not None: __a = padded_inputs.convert_to_tensors(_a ) return padded_inputs
65
0
from math import isqrt def lowercase ( lowerCAmelCase__ : int ) -> bool: return all(number % divisor != 0 for divisor in range(2 , isqrt(lowerCAmelCase__ ) + 1 ) ) def lowercase ( lowerCAmelCase__ : int = 10**6 ) -> int: __a = 0 __a = 1 __a = 7 while prime_candidate < max_prime: primes_count += is_prime(lowerCAmelCase__ ) cube_index += 1 prime_candidate += 6 * cube_index return primes_count if __name__ == "__main__": print(F'''{solution() = }''')
706
"""simple docstring""" import json import os import subprocess import unittest from ast import literal_eval import pytest from parameterized import parameterized, 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_model_parallelism.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, { 'framework': 'pytorch', 'script': 'run_glue.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, ] ) class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): 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=_a , ) assert hasattr(self , '''env''' ) def __UpperCAmelCase ( self , _a ): # configuration for running training on smdistributed Model Parallel __a = { '''enabled''': True, '''processes_per_host''': 8, } __a = { '''enabled''': True, '''parameters''': { '''microbatches''': 4, '''placement_strategy''': '''spread''', '''pipeline''': '''interleaved''', '''optimize''': '''speed''', '''partitions''': 4, '''ddp''': True, }, } __a = {'''smdistributed''': {'''modelparallel''': smp_options}, '''mpi''': mpi_options} __a = '''trainer''' if self.script == '''run_glue.py''' else '''smtrainer''' # 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}-{instance_count}-smp-{name_extension}''' , instance_count=_a , instance_type=self.instance_type , debugger_hook_config=_a , hyperparameters={ **self.env.hyperparameters, '''model_name_or_path''': self.model_name_or_path, '''max_steps''': 500, } , metric_definitions=self.env.metric_definitions , distribution=_a , py_version='''py36''' , ) def __UpperCAmelCase ( self , _a ): TrainingJobAnalytics(_a ).export_csv(f'''{self.env.test_path}/{job_name}_metrics.csv''' ) @parameterized.expand([(1,)] ) def __UpperCAmelCase ( self , _a ): # create estimator __a = self.create_estimator(_a ) # run training estimator.fit() # result dataframe __a = TrainingJobAnalytics(estimator.latest_training_job.name ).dataframe() # extract kpis __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_accuracy''']['''value'''] ) __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_loss''']['''value'''] ) # get train time from SageMaker job, this includes starting, preprocessing, stopping __a = ( Session().describe_training_job(estimator.latest_training_job.name ).get('''TrainingTimeInSeconds''' , 999_999 ) ) # 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} , _a )
65
0
"""simple docstring""" from ...processing_utils import ProcessorMixin class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Tuple = 'SpeechT5FeatureExtractor' __UpperCAmelCase : int = 'SpeechT5Tokenizer' def __init__( self , _a , _a ): super().__init__(_a , _a ) def __call__( self , *_a , **_a ): __a = kwargs.pop('''audio''' , _a ) __a = kwargs.pop('''text''' , _a ) __a = kwargs.pop('''text_target''' , _a ) __a = kwargs.pop('''audio_target''' , _a ) __a = kwargs.pop('''sampling_rate''' , _a ) if audio is not None and text is not None: raise ValueError( '''Cannot process both `audio` and `text` inputs. Did you mean `audio_target` or `text_target`?''' ) if audio_target is not None and text_target is not None: raise ValueError( '''Cannot process both `audio_target` and `text_target` inputs. Did you mean `audio` or `text`?''' ) if audio is None and audio_target is None and text is None and text_target is None: raise ValueError( '''You need to specify either an `audio`, `audio_target`, `text`, or `text_target` input to process.''' ) if audio is not None: __a = self.feature_extractor(_a , *_a , sampling_rate=_a , **_a ) elif text is not None: __a = self.tokenizer(_a , **_a ) else: __a = None if audio_target is not None: __a = self.feature_extractor(audio_target=_a , *_a , sampling_rate=_a , **_a ) __a = targets['''input_values'''] elif text_target is not None: __a = self.tokenizer(_a , **_a ) __a = targets['''input_ids'''] else: __a = None if inputs is None: return targets if targets is not None: __a = labels __a = targets.get('''attention_mask''' ) if decoder_attention_mask is not None: __a = decoder_attention_mask return inputs def __UpperCAmelCase ( self , *_a , **_a ): __a = kwargs.pop('''input_values''' , _a ) __a = kwargs.pop('''input_ids''' , _a ) __a = kwargs.pop('''labels''' , _a ) if input_values is not None and input_ids is not None: raise ValueError('''Cannot process both `input_values` and `input_ids` inputs.''' ) if input_values is None and input_ids is None and labels is None: raise ValueError( '''You need to specify either an `input_values`, `input_ids`, or `labels` input to be padded.''' ) if input_values is not None: __a = self.feature_extractor.pad(_a , *_a , **_a ) elif input_ids is not None: __a = self.tokenizer.pad(_a , **_a ) else: __a = None if labels is not None: if "input_ids" in labels or (isinstance(_a , _a ) and "input_ids" in labels[0]): __a = self.tokenizer.pad(_a , **_a ) __a = targets['''input_ids'''] else: __a = self.feature_extractor.feature_size __a = self.feature_extractor.num_mel_bins __a = self.feature_extractor.pad(_a , *_a , **_a ) __a = feature_size_hack __a = targets['''input_values'''] else: __a = None if inputs is None: return targets if targets is not None: __a = labels __a = targets.get('''attention_mask''' ) if decoder_attention_mask is not None: __a = decoder_attention_mask return inputs def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.batch_decode(*_a , **_a ) def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.decode(*_a , **_a )
707
"""simple docstring""" import warnings from ..trainer import Trainer from ..utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a=None , **_a ): warnings.warn( '''`SageMakerTrainer` is deprecated and will be removed in v5 of Transformers. You can use `Trainer` ''' '''instead.''' , _a , ) super().__init__(args=_a , **_a )
65
0
"""simple docstring""" import sys import warnings from os.path import abspath, dirname, join # allow having multiple repository checkouts and not needing to remember to rerun # 'pip install -e .[dev]' when switching between checkouts and running tests. lowercase_ = abspath(join(dirname(dirname(dirname(__file__))), "src")) sys.path.insert(1, git_repo_path) # silence FutureWarning warnings in tests since often we can't act on them until # they become normal warnings - i.e. the tests still need to test the current functionality warnings.simplefilter(action="ignore", category=FutureWarning) def lowercase ( lowerCAmelCase__ : List[Any] ) -> str: '''simple docstring''' from transformers.testing_utils import pytest_addoption_shared pytest_addoption_shared(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Union[str, Any]: '''simple docstring''' from transformers.testing_utils import pytest_terminal_summary_main __a = terminalreporter.config.getoption('''--make-reports''' ) if make_reports: pytest_terminal_summary_main(lowerCAmelCase__ , id=lowerCAmelCase__ )
708
"""simple docstring""" from typing import TYPE_CHECKING from ...file_utils import _LazyModule, is_tokenizers_available, is_torch_available from ...utils import OptionalDependencyNotAvailable lowercase_ = {"configuration_gpt_neox": ["GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXConfig"]} try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["GPTNeoXTokenizerFast"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST", "GPTNeoXForCausalLM", "GPTNeoXForQuestionAnswering", "GPTNeoXForSequenceClassification", "GPTNeoXForTokenClassification", "GPTNeoXLayer", "GPTNeoXModel", "GPTNeoXPreTrainedModel", ] if TYPE_CHECKING: from .configuration_gpt_neox import GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXConfig try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .tokenization_gpt_neox_fast import GPTNeoXTokenizerFast try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_gpt_neox import ( GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST, GPTNeoXForCausalLM, GPTNeoXForQuestionAnswering, GPTNeoXForSequenceClassification, GPTNeoXForTokenClassification, GPTNeoXLayer, GPTNeoXModel, GPTNeoXPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
65
0
"""simple docstring""" from typing import List, Optional, Tuple, Union import torch from ...schedulers import DDIMScheduler from ...utils import randn_tensor from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a ): super().__init__() # make sure scheduler can always be converted to DDIM __a = DDIMScheduler.from_config(scheduler.config ) self.register_modules(unet=_a , scheduler=_a ) @torch.no_grad() def __call__( self , _a = 1 , _a = None , _a = 0.0 , _a = 50 , _a = None , _a = "pil" , _a = True , ): # Sample gaussian noise to begin loop if isinstance(self.unet.config.sample_size , _a ): __a = ( batch_size, self.unet.config.in_channels, self.unet.config.sample_size, self.unet.config.sample_size, ) else: __a = (batch_size, self.unet.config.in_channels, *self.unet.config.sample_size) if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError( f'''You have passed a list of generators of length {len(_a )}, but requested an effective batch''' f''' size of {batch_size}. Make sure the batch size matches the length of the generators.''' ) __a = randn_tensor(_a , generator=_a , device=self.device , dtype=self.unet.dtype ) # set step values self.scheduler.set_timesteps(_a ) for t in self.progress_bar(self.scheduler.timesteps ): # 1. predict noise model_output __a = self.unet(_a , _a ).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 __a = self.scheduler.step( _a , _a , _a , eta=_a , use_clipped_model_output=_a , generator=_a ).prev_sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image,) return ImagePipelineOutput(images=_a )
709
"""simple docstring""" import unittest from transformers import MraConfig, is_torch_available from transformers.testing_utils import require_torch, slow, torch_device from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask if is_torch_available(): import torch from transformers import ( MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, MraModel, ) from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=2 , _a=8 , _a=True , _a=True , _a=True , _a=True , _a=99 , _a=16 , _a=5 , _a=2 , _a=36 , _a="gelu" , _a=0.0 , _a=0.0 , _a=512 , _a=16 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ): __a = parent __a = batch_size __a = seq_length __a = is_training __a = use_input_mask __a = use_token_type_ids __a = use_labels __a = vocab_size __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = type_sequence_label_size __a = initializer_range __a = num_labels __a = num_choices __a = scope def __UpperCAmelCase ( self ): __a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) __a = None if self.use_input_mask: __a = random_attention_mask([self.batch_size, self.seq_length] ) __a = None if self.use_token_type_ids: __a = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size ) __a = None __a = None __a = None if self.use_labels: __a = ids_tensor([self.batch_size] , self.type_sequence_label_size ) __a = ids_tensor([self.batch_size, self.seq_length] , self.num_labels ) __a = ids_tensor([self.batch_size] , self.num_choices ) __a = self.get_config() return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels def __UpperCAmelCase ( self ): return MraConfig( 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=_a , initializer_range=self.initializer_range , ) def __UpperCAmelCase ( self ): __a = self.get_config() __a = 300 return config def __UpperCAmelCase ( self ): ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = self.prepare_config_and_inputs() __a = True __a = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] ) __a = 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraModel(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a ) __a = model(_a , token_type_ids=_a ) __a = model(_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , _a , _a , ): __a = True __a = MraModel(_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , encoder_attention_mask=_a , ) __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , ) __a = model(_a , attention_mask=_a , token_type_ids=_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForMaskedLM(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForQuestionAnswering(config=_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , start_positions=_a , end_positions=_a , ) 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForSequenceClassification(_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForTokenClassification(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_choices __a = MraForMultipleChoice(config=_a ) model.to(_a ) model.eval() __a = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = model( _a , attention_mask=_a , token_type_ids=_a , labels=_a , ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) ) def __UpperCAmelCase ( self ): __a = self.prepare_config_and_inputs() ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = config_and_inputs __a = {'''input_ids''': input_ids, '''token_type_ids''': token_type_ids, '''attention_mask''': input_mask} return config, inputs_dict @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = ( ( MraModel, MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, ) if is_torch_available() else () ) __UpperCAmelCase : str = False __UpperCAmelCase : Any = False __UpperCAmelCase : List[Any] = False __UpperCAmelCase : Optional[Any] = False __UpperCAmelCase : Dict = () def __UpperCAmelCase ( self ): __a = MraModelTester(self ) __a = ConfigTester(self , config_class=_a , hidden_size=37 ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() for type in ["absolute", "relative_key", "relative_key_query"]: __a = type self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_masked_lm(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_multiple_choice(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_question_answering(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_sequence_classification(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_token_classification(*_a ) @slow def __UpperCAmelCase ( self ): for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: __a = MraModel.from_pretrained(_a ) self.assertIsNotNone(_a ) @unittest.skip(reason='''MRA does not output attentions''' ) def __UpperCAmelCase ( self ): return @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @slow def __UpperCAmelCase ( self ): __a = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = torch.Size((1, 256, 768) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[-0.0140, 0.0830, -0.0381], [0.1546, 0.1402, 0.0220], [0.1162, 0.0851, 0.0165]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 256, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[9.2595, -3.6038, 11.8819], [9.3869, -3.2693, 11.0956], [11.8524, -3.4938, 13.1210]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' ) __a = torch.arange(4_096 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 4_096, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[5.4789, -2.3564, 7.5064], [7.9067, -1.3369, 9.9668], [9.0712, -1.8106, 7.0380]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) )
65
0
"""simple docstring""" from itertools import permutations def lowercase ( lowerCAmelCase__ : tuple ) -> bool: if num[3] % 2 != 0: return False if (num[2] + num[3] + num[4]) % 3 != 0: return False if num[5] % 5 != 0: return False __a = [7, 11, 13, 17] for i, test in enumerate(lowerCAmelCase__ ): if (num[i + 4] * 100 + num[i + 5] * 10 + num[i + 6]) % test != 0: return False return True def lowercase ( lowerCAmelCase__ : int = 10 ) -> int: return sum( int(''''''.join(map(lowerCAmelCase__ , lowerCAmelCase__ ) ) ) for num in permutations(range(lowerCAmelCase__ ) ) if is_substring_divisible(lowerCAmelCase__ ) ) if __name__ == "__main__": print(F'''{solution() = }''')
710
"""simple docstring""" from __future__ import annotations def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> list[str]: if partitions <= 0: raise ValueError('''partitions must be a positive number!''' ) if partitions > number_of_bytes: raise ValueError('''partitions can not > number_of_bytes!''' ) __a = number_of_bytes // partitions __a = [] for i in range(lowerCAmelCase__ ): __a = i * bytes_per_partition + 1 __a = ( number_of_bytes if i == partitions - 1 else (i + 1) * bytes_per_partition ) allocation_list.append(f'''{start_bytes}-{end_bytes}''' ) return allocation_list if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import warnings from ...utils import logging from .image_processing_beit import BeitImageProcessor lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , *_a , **_a ): warnings.warn( '''The class BeitFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please''' ''' use BeitImageProcessor instead.''' , _a , ) super().__init__(*_a , **_a )
711
"""simple docstring""" def lowercase ( lowerCAmelCase__ : list , lowerCAmelCase__ : list , lowerCAmelCase__ : int ) -> list: __a = len(lowerCAmelCase__ ) __a = [[0] * n for i in range(lowerCAmelCase__ )] for i in range(lowerCAmelCase__ ): __a = y_points[i] for i in range(2 , lowerCAmelCase__ ): for j in range(lowerCAmelCase__ , lowerCAmelCase__ ): __a = ( (xa - x_points[j - i + 1]) * q[j][i - 1] - (xa - x_points[j]) * q[j - 1][i - 1] ) / (x_points[j] - x_points[j - i + 1]) return [q[n - 1][n - 1], q] if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import os import unittest from transformers import LxmertTokenizer, LxmertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES from transformers.testing_utils import require_tokenizers from ...test_tokenization_common import TokenizerTesterMixin @require_tokenizers class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : List[Any] = LxmertTokenizer __UpperCAmelCase : int = LxmertTokenizerFast __UpperCAmelCase : Any = True __UpperCAmelCase : str = True def __UpperCAmelCase ( self ): super().setUp() __a = [ '''[UNK]''', '''[CLS]''', '''[SEP]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''', ] __a = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''vocab_file'''] ) with open(self.vocab_file , '''w''' , encoding='''utf-8''' ) as vocab_writer: vocab_writer.write(''''''.join([x + '''\n''' for x in vocab_tokens] ) ) def __UpperCAmelCase ( self , _a ): __a = '''UNwant\u00E9d,running''' __a = '''unwanted, running''' return input_text, output_text def __UpperCAmelCase ( self ): __a = self.tokenizer_class(self.vocab_file ) __a = tokenizer.tokenize('''UNwant\u00E9d,running''' ) self.assertListEqual(_a , ['''un''', '''##want''', '''##ed''', ''',''', '''runn''', '''##ing'''] ) self.assertListEqual(tokenizer.convert_tokens_to_ids(_a ) , [7, 4, 5, 10, 8, 9] ) def __UpperCAmelCase ( self ): if not self.test_rust_tokenizer: return __a = self.get_tokenizer() __a = self.get_rust_tokenizer() __a = '''I was born in 92000, and this is falsé.''' __a = tokenizer.tokenize(_a ) __a = rust_tokenizer.tokenize(_a ) self.assertListEqual(_a , _a ) __a = tokenizer.encode(_a , add_special_tokens=_a ) __a = rust_tokenizer.encode(_a , add_special_tokens=_a ) self.assertListEqual(_a , _a ) __a = self.get_rust_tokenizer() __a = tokenizer.encode(_a ) __a = rust_tokenizer.encode(_a ) self.assertListEqual(_a , _a )
712
"""simple docstring""" import json import os from functools import lru_cache from typing import List, Optional, Tuple import regex as re from ...tokenization_utils import AddedToken, PreTrainedTokenizer from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = {"vocab_file": "vocab.json", "merges_file": "merges.txt"} lowercase_ = { "vocab_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/vocab.json", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/vocab.json" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/vocab.json" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/vocab.json" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/vocab.json" ), }, "merges_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/merges.txt", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/merges.txt" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/merges.txt" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/merges.txt" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/merges.txt" ), }, } lowercase_ = { "allenai/longformer-base-4096": 4_0_9_6, "allenai/longformer-large-4096": 4_0_9_6, "allenai/longformer-large-4096-finetuned-triviaqa": 4_0_9_6, "allenai/longformer-base-4096-extra.pos.embd.only": 4_0_9_6, "allenai/longformer-large-4096-extra.pos.embd.only": 4_0_9_6, } @lru_cache() # Copied from transformers.models.roberta.tokenization_roberta.bytes_to_unicode def lowercase ( ) -> Union[str, Any]: __a = ( list(range(ord('''!''' ) , ord('''~''' ) + 1 ) ) + list(range(ord('''¡''' ) , ord('''¬''' ) + 1 ) ) + list(range(ord('''®''' ) , ord('''ÿ''' ) + 1 ) ) ) __a = bs[:] __a = 0 for b in range(2**8 ): if b not in bs: bs.append(lowerCAmelCase__ ) cs.append(2**8 + n ) n += 1 __a = [chr(lowerCAmelCase__ ) for n in cs] return dict(zip(lowerCAmelCase__ , lowerCAmelCase__ ) ) def lowercase ( lowerCAmelCase__ : Tuple ) -> Tuple: __a = set() __a = word[0] for char in word[1:]: pairs.add((prev_char, char) ) __a = char return pairs class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Optional[int] = VOCAB_FILES_NAMES __UpperCAmelCase : int = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : Dict = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : Union[str, Any] = ['input_ids', 'attention_mask'] def __init__( self , _a , _a , _a="replace" , _a="<s>" , _a="</s>" , _a="</s>" , _a="<s>" , _a="<unk>" , _a="<pad>" , _a="<mask>" , _a=False , **_a , ): __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else bos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else eos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else sep_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else cls_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else unk_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else pad_token # Mask token behave like a normal word, i.e. include the space before it __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else mask_token super().__init__( errors=_a , bos_token=_a , eos_token=_a , unk_token=_a , sep_token=_a , cls_token=_a , pad_token=_a , mask_token=_a , add_prefix_space=_a , **_a , ) with open(_a , encoding='''utf-8''' ) as vocab_handle: __a = json.load(_a ) __a = {v: k for k, v in self.encoder.items()} __a = errors # how to handle errors in decoding __a = bytes_to_unicode() __a = {v: k for k, v in self.byte_encoder.items()} with open(_a , encoding='''utf-8''' ) as merges_handle: __a = merges_handle.read().split('''\n''' )[1:-1] __a = [tuple(merge.split() ) for merge in bpe_merges] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = {} __a = add_prefix_space # Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions __a = re.compile(R'''\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+''' ) @property def __UpperCAmelCase ( self ): return len(self.encoder ) def __UpperCAmelCase ( self ): return dict(self.encoder , **self.added_tokens_encoder ) def __UpperCAmelCase ( self , _a ): if token in self.cache: return self.cache[token] __a = tuple(_a ) __a = get_pairs(_a ) if not pairs: return token while True: __a = min(_a , key=lambda _a : self.bpe_ranks.get(_a , float('''inf''' ) ) ) if bigram not in self.bpe_ranks: break __a , __a = bigram __a = [] __a = 0 while i < len(_a ): try: __a = word.index(_a , _a ) except ValueError: new_word.extend(word[i:] ) break else: new_word.extend(word[i:j] ) __a = j if word[i] == first and i < len(_a ) - 1 and word[i + 1] == second: new_word.append(first + second ) i += 2 else: new_word.append(word[i] ) i += 1 __a = tuple(_a ) __a = new_word if len(_a ) == 1: break else: __a = get_pairs(_a ) __a = ''' '''.join(_a ) __a = word return word def __UpperCAmelCase ( self , _a ): __a = [] for token in re.findall(self.pat , _a ): __a = ''''''.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(_a ).split(''' ''' ) ) return bpe_tokens def __UpperCAmelCase ( self , _a ): return self.encoder.get(_a , self.encoder.get(self.unk_token ) ) def __UpperCAmelCase ( self , _a ): return self.decoder.get(_a ) def __UpperCAmelCase ( self , _a ): __a = ''''''.join(_a ) __a = bytearray([self.byte_decoder[c] for c in text] ).decode('''utf-8''' , errors=self.errors ) return text def __UpperCAmelCase ( self , _a , _a = None ): if not os.path.isdir(_a ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] ) __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] ) with open(_a , '''w''' , encoding='''utf-8''' ) as f: f.write(json.dumps(self.encoder , indent=2 , sort_keys=_a , ensure_ascii=_a ) + '''\n''' ) __a = 0 with open(_a , '''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 _a : 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!''' ) __a = token_index writer.write(''' '''.join(_a ) + '''\n''' ) index += 1 return vocab_file, merge_file def __UpperCAmelCase ( self , _a , _a = None ): if token_ids_a is None: return [self.cls_token_id] + token_ids_a + [self.sep_token_id] __a = [self.cls_token_id] __a = [self.sep_token_id] return cls + token_ids_a + sep + sep + token_ids_a + sep def __UpperCAmelCase ( self , _a , _a = None , _a = False ): if already_has_special_tokens: return super().get_special_tokens_mask( token_ids_a=_a , token_ids_a=_a , already_has_special_tokens=_a ) if token_ids_a is None: return [1] + ([0] * len(_a )) + [1] return [1] + ([0] * len(_a )) + [1, 1] + ([0] * len(_a )) + [1] def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [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 __UpperCAmelCase ( self , _a , _a=False , **_a ): __a = kwargs.pop('''add_prefix_space''' , self.add_prefix_space ) if (is_split_into_words or add_prefix_space) and (len(_a ) > 0 and not text[0].isspace()): __a = ''' ''' + text return (text, kwargs)
65
0
"""simple docstring""" import multiprocessing import time from arguments import PretokenizationArguments from datasets import load_dataset from transformers import AutoTokenizer, HfArgumentParser def lowercase ( lowerCAmelCase__ : str ) -> List[str]: __a = {} __a = tokenizer(example['''content'''] , truncation=lowerCAmelCase__ )['''input_ids'''] __a = len(example['''content'''] ) / len(output['''input_ids'''] ) return output lowercase_ = HfArgumentParser(PretokenizationArguments) lowercase_ = parser.parse_args() if args.num_workers is None: lowercase_ = multiprocessing.cpu_count() lowercase_ = AutoTokenizer.from_pretrained(args.tokenizer_dir) lowercase_ = time.time() lowercase_ = load_dataset(args.dataset_name, split="train") print(F'''Dataset loaded in {time.time()-t_start:.2f}s''') lowercase_ = time.time() lowercase_ = 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''') lowercase_ = time.time() ds.push_to_hub(args.tokenized_data_repo) print(F'''Data pushed to the hub in {time.time()-t_start:.2f}s''')
713
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "unc-nlp/lxmert-base-uncased": "https://huggingface.co/unc-nlp/lxmert-base-uncased/resolve/main/config.json", } class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = 'lxmert' __UpperCAmelCase : str = {} def __init__( self , _a=30_522 , _a=768 , _a=12 , _a=9_500 , _a=1_600 , _a=400 , _a=3_072 , _a="gelu" , _a=0.1 , _a=0.1 , _a=512 , _a=2 , _a=0.02 , _a=1E-12 , _a=9 , _a=5 , _a=5 , _a=2_048 , _a=4 , _a=6.67 , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , **_a , ): __a = vocab_size __a = hidden_size __a = num_attention_heads __a = hidden_act __a = intermediate_size __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = initializer_range __a = layer_norm_eps __a = num_qa_labels __a = num_object_labels __a = num_attr_labels __a = l_layers __a = x_layers __a = r_layers __a = visual_feat_dim __a = visual_pos_dim __a = visual_loss_normalizer __a = task_matched __a = task_mask_lm __a = task_obj_predict __a = task_qa __a = visual_obj_loss __a = visual_attr_loss __a = visual_feat_loss __a = {'''vision''': r_layers, '''cross_encoder''': x_layers, '''language''': l_layers} super().__init__(**_a )
65
0
"""simple docstring""" import os import re import shutil import sys import tempfile import unittest import black lowercase_ = os.path.abspath(os.path.dirname(os.path.dirname(os.path.dirname(__file__)))) sys.path.append(os.path.join(git_repo_path, "utils")) import check_copies # noqa: E402 # This is the reference code that will be used in the tests. # If DDPMSchedulerOutput is changed in scheduling_ddpm.py, this code needs to be manually updated. lowercase_ = " \"\"\"\n Output class for the scheduler's step function output.\n\n Args:\n prev_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):\n Computed sample (x_{t-1}) of previous timestep. `prev_sample` should be used as next model input in the\n denoising loop.\n pred_original_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):\n The predicted denoised sample (x_{0}) based on the model output from the current timestep.\n `pred_original_sample` can be used to preview progress or for guidance.\n \"\"\"\n\n prev_sample: torch.FloatTensor\n pred_original_sample: Optional[torch.FloatTensor] = None\n" class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = tempfile.mkdtemp() os.makedirs(os.path.join(self.diffusers_dir , '''schedulers/''' ) ) __a = self.diffusers_dir shutil.copy( os.path.join(_a , '''src/diffusers/schedulers/scheduling_ddpm.py''' ) , os.path.join(self.diffusers_dir , '''schedulers/scheduling_ddpm.py''' ) , ) def __UpperCAmelCase ( self ): __a = '''src/diffusers''' shutil.rmtree(self.diffusers_dir ) def __UpperCAmelCase ( self , _a , _a , _a , _a=None ): __a = comment + f'''\nclass {class_name}(nn.Module):\n''' + class_code if overwrite_result is not None: __a = comment + f'''\nclass {class_name}(nn.Module):\n''' + overwrite_result __a = black.Mode(target_versions={black.TargetVersion.PYaa} , line_length=119 ) __a = black.format_str(_a , mode=_a ) __a = os.path.join(self.diffusers_dir , '''new_code.py''' ) with open(_a , '''w''' , newline='''\n''' ) as f: f.write(_a ) if overwrite_result is None: self.assertTrue(len(check_copies.is_copy_consistent(_a ) ) == 0 ) else: check_copies.is_copy_consistent(f.name , overwrite=_a ) with open(_a , '''r''' ) as f: self.assertTrue(f.read() , _a ) def __UpperCAmelCase ( self ): __a = check_copies.find_code_in_diffusers('''schedulers.scheduling_ddpm.DDPMSchedulerOutput''' ) self.assertEqual(_a , _a ) def __UpperCAmelCase ( self ): # Base copy consistency self.check_copy_consistency( '''# Copied from diffusers.schedulers.scheduling_ddpm.DDPMSchedulerOutput''' , '''DDPMSchedulerOutput''' , REFERENCE_CODE + '''\n''' , ) # With no empty line at the end self.check_copy_consistency( '''# Copied from diffusers.schedulers.scheduling_ddpm.DDPMSchedulerOutput''' , '''DDPMSchedulerOutput''' , _a , ) # Copy consistency with rename self.check_copy_consistency( '''# Copied from diffusers.schedulers.scheduling_ddpm.DDPMSchedulerOutput with DDPM->Test''' , '''TestSchedulerOutput''' , re.sub('''DDPM''' , '''Test''' , _a ) , ) # Copy consistency with a really long name __a = '''TestClassWithAReallyLongNameBecauseSomePeopleLikeThatForSomeReason''' self.check_copy_consistency( f'''# Copied from diffusers.schedulers.scheduling_ddpm.DDPMSchedulerOutput with DDPM->{long_class_name}''' , f'''{long_class_name}SchedulerOutput''' , re.sub('''Bert''' , _a , _a ) , ) # Copy consistency with overwrite self.check_copy_consistency( '''# Copied from diffusers.schedulers.scheduling_ddpm.DDPMSchedulerOutput with DDPM->Test''' , '''TestSchedulerOutput''' , _a , overwrite_result=re.sub('''DDPM''' , '''Test''' , _a ) , )
714
"""simple docstring""" import itertools import math def lowercase ( lowerCAmelCase__ : int ) -> bool: 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(lowerCAmelCase__ ) + 1 ) , 6 ): if number % i == 0 or number % (i + 2) == 0: return False return True def lowercase ( ) -> int: __a = 2 while True: if is_prime(lowerCAmelCase__ ): yield num num += 1 def lowercase ( lowerCAmelCase__ : int = 10001 ) -> int: return next(itertools.islice(prime_generator() , nth - 1 , lowerCAmelCase__ ) ) if __name__ == "__main__": print(F'''{solution() = }''')
65
0
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_mask2former": [ "MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "Mask2FormerConfig", ], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["Mask2FormerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST", "Mask2FormerForUniversalSegmentation", "Mask2FormerModel", "Mask2FormerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_maskaformer import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskaFormerConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_maskaformer import MaskaFormerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_maskaformer import ( MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST, MaskaFormerForUniversalSegmentation, MaskaFormerModel, MaskaFormerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
715
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_mask2former": [ "MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "Mask2FormerConfig", ], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["Mask2FormerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST", "Mask2FormerForUniversalSegmentation", "Mask2FormerModel", "Mask2FormerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_maskaformer import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskaFormerConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_maskaformer import MaskaFormerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_maskaformer import ( MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST, MaskaFormerForUniversalSegmentation, MaskaFormerModel, MaskaFormerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
65
0
"""simple docstring""" from __future__ import annotations import unittest from transformers import BlenderbotSmallConfig, BlenderbotSmallTokenizer, is_tf_available from transformers.testing_utils import require_tf, require_tokenizers, slow from transformers.utils import cached_property from ...test_configuration_common import ConfigTester from ...test_modeling_tf_common import TFModelTesterMixin, ids_tensor from ...test_pipeline_mixin import PipelineTesterMixin if is_tf_available(): import tensorflow as tf from transformers import TFAutoModelForSeqaSeqLM, TFBlenderbotSmallForConditionalGeneration, TFBlenderbotSmallModel @require_tf class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : str = BlenderbotSmallConfig __UpperCAmelCase : int = {} __UpperCAmelCase : List[str] = 'gelu' def __init__( self , _a , _a=13 , _a=7 , _a=True , _a=False , _a=99 , _a=32 , _a=2 , _a=4 , _a=37 , _a=0.1 , _a=0.1 , _a=20 , _a=2 , _a=1 , _a=0 , ): __a = parent __a = batch_size __a = seq_length __a = is_training __a = use_labels __a = vocab_size __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = eos_token_id __a = pad_token_id __a = bos_token_id def __UpperCAmelCase ( self ): __a = ids_tensor([self.batch_size, self.seq_length - 1] , self.vocab_size ) __a = tf.expand_dims(tf.constant([self.eos_token_id] * self.batch_size ) , 1 ) __a = tf.concat([input_ids, eos_tensor] , axis=1 ) __a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) __a = self.config_cls( vocab_size=self.vocab_size , d_model=self.hidden_size , encoder_layers=self.num_hidden_layers , decoder_layers=self.num_hidden_layers , encoder_attention_heads=self.num_attention_heads , decoder_attention_heads=self.num_attention_heads , encoder_ffn_dim=self.intermediate_size , decoder_ffn_dim=self.intermediate_size , dropout=self.hidden_dropout_prob , attention_dropout=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , eos_token_ids=[2] , bos_token_id=self.bos_token_id , pad_token_id=self.pad_token_id , decoder_start_token_id=self.pad_token_id , **self.config_updates , ) __a = prepare_blenderbot_small_inputs_dict(_a , _a , _a ) return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = TFBlenderbotSmallModel(config=_a ).get_decoder() __a = inputs_dict['''input_ids'''] __a = input_ids[:1, :] __a = inputs_dict['''attention_mask'''][:1, :] __a = inputs_dict['''head_mask'''] __a = 1 # first forward pass __a = model(_a , attention_mask=_a , head_mask=_a , use_cache=_a ) __a , __a = outputs.to_tuple() # create hypothetical next token and extent to next_input_ids __a = ids_tensor((self.batch_size, 3) , config.vocab_size ) __a = tf.cast(ids_tensor((self.batch_size, 3) , 2 ) , tf.inta ) # append to next input_ids and __a = tf.concat([input_ids, next_tokens] , axis=-1 ) __a = tf.concat([attention_mask, next_attn_mask] , axis=-1 ) __a = model(_a , attention_mask=_a )[0] __a = model(_a , attention_mask=_a , past_key_values=_a )[0] self.parent.assertEqual(next_tokens.shape[1] , output_from_past.shape[1] ) # select random slice __a = int(ids_tensor((1,) , output_from_past.shape[-1] ) ) __a = output_from_no_past[:, -3:, random_slice_idx] __a = output_from_past[:, :, random_slice_idx] # test that outputs are equal for slice tf.debugging.assert_near(_a , _a , rtol=1E-3 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Any=None , lowerCAmelCase__ : str=None , lowerCAmelCase__ : Dict=None , lowerCAmelCase__ : Optional[int]=None , lowerCAmelCase__ : str=None , ) -> Dict: if attention_mask is None: __a = tf.cast(tf.math.not_equal(lowerCAmelCase__ , config.pad_token_id ) , tf.inta ) if decoder_attention_mask is None: __a = tf.concat( [ tf.ones(decoder_input_ids[:, :1].shape , dtype=tf.inta ), tf.cast(tf.math.not_equal(decoder_input_ids[:, 1:] , config.pad_token_id ) , tf.inta ), ] , axis=-1 , ) if head_mask is None: __a = tf.ones((config.encoder_layers, config.encoder_attention_heads) ) if decoder_head_mask is None: __a = tf.ones((config.decoder_layers, config.decoder_attention_heads) ) if cross_attn_head_mask is None: __a = tf.ones((config.decoder_layers, config.decoder_attention_heads) ) return { "input_ids": input_ids, "decoder_input_ids": decoder_input_ids, "attention_mask": attention_mask, "decoder_attention_mask": decoder_attention_mask, "head_mask": head_mask, "decoder_head_mask": decoder_head_mask, "cross_attn_head_mask": cross_attn_head_mask, } @require_tf class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : int = ( (TFBlenderbotSmallForConditionalGeneration, TFBlenderbotSmallModel) if is_tf_available() else () ) __UpperCAmelCase : str = (TFBlenderbotSmallForConditionalGeneration,) if is_tf_available() else () __UpperCAmelCase : Optional[Any] = ( { 'conversational': TFBlenderbotSmallForConditionalGeneration, 'feature-extraction': TFBlenderbotSmallModel, 'summarization': TFBlenderbotSmallForConditionalGeneration, 'text2text-generation': TFBlenderbotSmallForConditionalGeneration, 'translation': TFBlenderbotSmallForConditionalGeneration, } if is_tf_available() else {} ) __UpperCAmelCase : Any = True __UpperCAmelCase : Any = False __UpperCAmelCase : int = False def __UpperCAmelCase ( self ): __a = TFBlenderbotSmallModelTester(self ) __a = ConfigTester(self , config_class=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.check_decoder_model_past_large_inputs(*_a ) @require_tokenizers @require_tf class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : List[Any] = [ 'Social anxiety\nWow, I am never shy. Do you have anxiety?\nYes. I end up sweating and blushing and feel like ' ' i\'m going to throw up.\nand why is that?' ] __UpperCAmelCase : int = 'facebook/blenderbot_small-90M' @cached_property def __UpperCAmelCase ( self ): # use "old" tokenizer here because of bug when downloading new tokenizer return BlenderbotSmallTokenizer.from_pretrained('''facebook/blenderbot-90M''' ) @cached_property def __UpperCAmelCase ( self ): __a = TFAutoModelForSeqaSeqLM.from_pretrained(self.model_name ) return model @slow def __UpperCAmelCase ( self ): __a = self.tokenizer(self.src_text , return_tensors='''tf''' ) __a = self.model.generate( model_inputs.input_ids , attention_mask=model_inputs.attention_mask , num_beams=2 , use_cache=_a , ) __a = self.tokenizer.batch_decode(generated_ids.numpy() , skip_special_tokens=_a )[0] assert generated_words in ( "i don't know. i just feel like i'm going to throw up. it's not fun.", "i'm not sure. i just feel like i've been feeling like i have to be in a certain place", "i'm not sure. i just feel like i've been in a bad situation.", )
716
"""simple docstring""" from sklearn.metrics import fa_score import datasets lowercase_ = "\nThe F1 score is the harmonic mean of the precision and recall. It can be computed with the equation:\nF1 = 2 * (precision * recall) / (precision + recall)\n" lowercase_ = "\nArgs:\n predictions (`list` of `int`): Predicted labels.\n references (`list` of `int`): Ground truth labels.\n labels (`list` of `int`): The set of labels to include when `average` is not set to `'binary'`, and the order of the labels if `average` is `None`. Labels present in the data can be excluded, for example to calculate a multiclass average ignoring a majority negative class. Labels not present in the data will result in 0 components in a macro average. For multilabel targets, labels are column indices. By default, all labels in `predictions` and `references` are used in sorted order. Defaults to None.\n pos_label (`int`): The class to be considered the positive class, in the case where `average` is set to `binary`. Defaults to 1.\n average (`string`): This parameter is required for multiclass/multilabel targets. If set to `None`, the scores for each class are returned. Otherwise, this determines the type of averaging performed on the data. Defaults to `'binary'`.\n\n - 'binary': Only report results for the class specified by `pos_label`. This is applicable only if the classes found in `predictions` and `references` are binary.\n - 'micro': Calculate metrics globally by counting the total true positives, false negatives and false positives.\n - 'macro': Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.\n - 'weighted': Calculate metrics for each label, and find their average weighted by support (the number of true instances for each label). This alters `'macro'` to account for label imbalance. This option can result in an F-score that is not between precision and recall.\n - 'samples': Calculate metrics for each instance, and find their average (only meaningful for multilabel classification).\n sample_weight (`list` of `float`): Sample weights Defaults to None.\n\nReturns:\n f1 (`float` or `array` of `float`): F1 score or list of f1 scores, depending on the value passed to `average`. Minimum possible value is 0. Maximum possible value is 1. Higher f1 scores are better.\n\nExamples:\n\n Example 1-A simple binary example\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0])\n >>> print(results)\n {'f1': 0.5}\n\n Example 2-The same simple binary example as in Example 1, but with `pos_label` set to `0`.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], pos_label=0)\n >>> print(round(results['f1'], 2))\n 0.67\n\n Example 3-The same simple binary example as in Example 1, but with `sample_weight` included.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], sample_weight=[0.9, 0.5, 3.9, 1.2, 0.3])\n >>> print(round(results['f1'], 2))\n 0.35\n\n Example 4-A multiclass example, with different values for the `average` input.\n >>> predictions = [0, 2, 1, 0, 0, 1]\n >>> references = [0, 1, 2, 0, 1, 2]\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"macro\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"micro\")\n >>> print(round(results['f1'], 2))\n 0.33\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"weighted\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=None)\n >>> print(results)\n {'f1': array([0.8, 0. , 0. ])}\n" lowercase_ = "\n@article{scikit-learn,\n title={Scikit-learn: Machine Learning in {P}ython},\n author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.\n and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.\n and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and\n Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},\n journal={Journal of Machine Learning Research},\n volume={12},\n pages={2825--2830},\n year={2011}\n}\n" @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class __lowerCAmelCase ( datasets.Metric ): '''simple docstring''' def __UpperCAmelCase ( self ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { '''predictions''': datasets.Sequence(datasets.Value('''int32''' ) ), '''references''': datasets.Sequence(datasets.Value('''int32''' ) ), } if self.config_name == '''multilabel''' else { '''predictions''': datasets.Value('''int32''' ), '''references''': datasets.Value('''int32''' ), } ) , reference_urls=['''https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html'''] , ) def __UpperCAmelCase ( self , _a , _a , _a=None , _a=1 , _a="binary" , _a=None ): __a = fa_score( _a , _a , labels=_a , pos_label=_a , average=_a , sample_weight=_a ) return {"f1": float(_a ) if score.size == 1 else score}
65
0
"""simple docstring""" import json import os from typing import Dict, List, Optional, Tuple import regex as re from ...tokenization_utils import PreTrainedTokenizer from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "vocab_file": "vocab.json", "merges_file": "merges.txt", "tokenizer_config_file": "tokenizer_config.json", } lowercase_ = { "vocab_file": { "facebook/blenderbot_small-90M": "https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/vocab.json" }, "merges_file": { "facebook/blenderbot_small-90M": "https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/merges.txt" }, "tokenizer_config_file": { "facebook/blenderbot_small-90M": ( "https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/tokenizer_config.json" ) }, } lowercase_ = {"facebook/blenderbot_small-90M": 5_1_2} def lowercase ( lowerCAmelCase__ : str ) -> int: __a = set() __a = word[0] for char in word[1:]: pairs.add((prev_char, char) ) __a = char __a = set(lowerCAmelCase__ ) return pairs class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : int = VOCAB_FILES_NAMES __UpperCAmelCase : Union[str, Any] = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : Optional[Any] = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : str = ['input_ids', 'attention_mask'] def __init__( self , _a , _a , _a="__start__" , _a="__end__" , _a="__unk__" , _a="__null__" , **_a , ): super().__init__(unk_token=_a , bos_token=_a , eos_token=_a , pad_token=_a , **_a ) with open(_a , encoding='''utf-8''' ) as vocab_handle: __a = json.load(_a ) __a = {v: k for k, v in self.encoder.items()} with open(_a , encoding='''utf-8''' ) as merges_handle: __a = merges_handle.read().split('''\n''' )[1:-1] __a = [tuple(merge.split() ) for merge in merges] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = {} @property def __UpperCAmelCase ( self ): return len(self.encoder ) def __UpperCAmelCase ( self ): return dict(self.encoder , **self.added_tokens_encoder ) def __UpperCAmelCase ( self , _a ): if token in self.cache: return self.cache[token] __a = re.sub('''([.,!?()])''' , R''' \1''' , _a ) __a = re.sub('''(\')''' , R''' \1 ''' , _a ) __a = re.sub(R'''\s{2,}''' , ''' ''' , _a ) if "\n" in token: __a = token.replace('''\n''' , ''' __newln__''' ) __a = token.split(''' ''' ) __a = [] for token in tokens: if not len(_a ): continue __a = token.lower() __a = tuple(_a ) __a = tuple(list(word[:-1] ) + [word[-1] + '''</w>'''] ) __a = get_pairs(_a ) if not pairs: words.append(_a ) continue while True: __a = min(_a , key=lambda _a : self.bpe_ranks.get(_a , float('''inf''' ) ) ) if bigram not in self.bpe_ranks: break __a , __a = bigram __a = [] __a = 0 while i < len(_a ): try: __a = word.index(_a , _a ) new_word.extend(word[i:j] ) __a = j except ValueError: new_word.extend(word[i:] ) break if word[i] == first and i < len(_a ) - 1 and word[i + 1] == second: new_word.append(first + second ) i += 2 else: new_word.append(word[i] ) i += 1 __a = tuple(_a ) __a = new_word if len(_a ) == 1: break else: __a = get_pairs(_a ) __a = '''@@ '''.join(_a ) __a = word[:-4] __a = word words.append(_a ) return " ".join(_a ) def __UpperCAmelCase ( self , _a ): __a = [] __a = re.findall(R'''\S+\n?''' , _a ) for token in words: split_tokens.extend(list(self.bpe(_a ).split(''' ''' ) ) ) return split_tokens def __UpperCAmelCase ( self , _a ): __a = token.lower() return self.encoder.get(_a , self.encoder.get(self.unk_token ) ) def __UpperCAmelCase ( self , _a ): return self.decoder.get(_a , self.unk_token ) def __UpperCAmelCase ( self , _a ): __a = ''' '''.join(_a ).replace('''@@ ''' , '''''' ).strip() return out_string def __UpperCAmelCase ( self , _a , _a = None ): if not os.path.isdir(_a ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] ) __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] ) with open(_a , '''w''' , encoding='''utf-8''' ) as f: f.write(json.dumps(self.encoder , indent=2 , sort_keys=_a , ensure_ascii=_a ) + '''\n''' ) __a = 0 with open(_a , '''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 _a : 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!''' ) __a = token_index writer.write(''' '''.join(_a ) + '''\n''' ) index += 1 return vocab_file, merge_file
717
"""simple docstring""" from ..utils import DummyObject, requires_backends class __lowerCAmelCase ( metaclass=__SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = ['onnx'] def __init__( self , *_a , **_a ): requires_backends(self , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] )
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> int: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or number < 0: raise ValueError('''Input must be a non-negative integer''' ) __a = 0 while number: # This way we arrive at next set bit (next 1) instead of looping # through each bit and checking for 1s hence the # loop won't run 32 times it will only run the number of `1` times number &= number - 1 count += 1 return count if __name__ == "__main__": import doctest doctest.testmod()
718
"""simple docstring""" 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 lowercase_ = 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 __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=16 , _a=13 , _a=7 , _a=14 , _a=10 , _a=19 , _a=5 , _a=4 , _a=True , _a=16 , _a=2 , _a=4 , _a=4 , _a="gelu" , _a=0.1 , _a=0.1 , _a=[1, 2, 3, 4, 5] , _a=25 , _a=5 , ): __a = d_model __a = parent __a = batch_size __a = prediction_length __a = context_length __a = cardinality __a = num_time_features __a = lags_sequence __a = embedding_dimension __a = is_training __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = context_length __a = prediction_length + label_length __a = label_length __a = moving_average __a = autocorrelation_factor def __UpperCAmelCase ( self ): 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 __UpperCAmelCase ( self , _a ): __a = config.context_length + max(config.lags_sequence ) __a = ids_tensor([self.batch_size, 1] , config.cardinality[0] ) __a = floats_tensor([self.batch_size, _past_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, _past_length] ) __a = floats_tensor([self.batch_size, _past_length] ) > 0.5 # decoder inputs __a = floats_tensor([self.batch_size, config.prediction_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, config.prediction_length] ) __a = { '''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 __UpperCAmelCase ( self ): __a = self.get_config() __a = self.prepare_autoformer_inputs_dict(_a ) return config, inputs_dict def __UpperCAmelCase ( self ): __a , __a = self.prepare_config_and_inputs() return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = AutoformerModel(config=_a ).to(_a ).eval() __a = model(**_a ) __a = outputs.encoder_last_hidden_state __a = outputs.last_hidden_state with tempfile.TemporaryDirectory() as tmpdirname: __a = model.get_encoder() encoder.save_pretrained(_a ) __a = AutoformerEncoder.from_pretrained(_a ).to(_a ) __a , __a , __a , __a , __a = model.create_network_inputs(**_a ) __a , __a = model.decomposition_layer(transformer_inputs[:, : config.context_length, ...] ) __a = torch.cat( (transformer_inputs[:, : config.context_length, ...], feature[:, : config.context_length, ...]) , dim=-1 , ) __a = encoder(inputs_embeds=_a )[0] self.parent.assertTrue((encoder_last_hidden_state_a - encoder_last_hidden_state).abs().max().item() < 1E-3 ) __a = ( torch.mean(transformer_inputs[:, : config.context_length, ...] , dim=1 ) .unsqueeze(1 ) .repeat(1 , config.prediction_length , 1 ) ) __a = torch.zeros( [transformer_inputs.shape[0], config.prediction_length, transformer_inputs.shape[2]] , device=enc_input.device , ) __a = torch.cat( ( torch.cat((seasonal_input[:, -config.label_length :, ...], zeros) , dim=1 ), feature[:, config.context_length - config.label_length :, ...], ) , dim=-1 , ) __a = 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: __a = model.get_decoder() decoder.save_pretrained(_a ) __a = AutoformerDecoder.from_pretrained(_a ).to(_a ) __a = decoder( trend=_a , inputs_embeds=_a , encoder_hidden_states=_a , )[0] self.parent.assertTrue((last_hidden_state_a - last_hidden_state).abs().max().item() < 1E-3 ) @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Any = (AutoformerModel, AutoformerForPrediction) if is_torch_available() else () __UpperCAmelCase : int = (AutoformerForPrediction,) if is_torch_available() else () __UpperCAmelCase : Any = {'feature-extraction': AutoformerModel} if is_torch_available() else {} __UpperCAmelCase : int = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False __UpperCAmelCase : Any = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False def __UpperCAmelCase ( self ): __a = AutoformerModelTester(self ) __a = ConfigTester(self , config_class=_a , has_text_modality=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs() for model_class in self.all_model_classes: __a = model_class(_a ) with tempfile.TemporaryDirectory() as tmpdirname: model.save_pretrained(_a ) __a , __a = model_class.from_pretrained(_a , output_loading_info=_a ) self.assertEqual(info['''missing_keys'''] , [] ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.check_encoder_decoder_model_standalone(*_a ) @unittest.skip(reason='''Model has no tokens embeddings''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = inspect.signature(getattr(_a , '''forward''' ) ) # The main input is the name of the argument after `self` __a = list(model_signature.parameters.keys() )[1] self.assertEqual(AutoformerModel.main_input_name , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ) __a = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic __a = [*signature.parameters.keys()] __a = [ '''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(_a )] , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() __a = True __a = getattr(self.model_tester , '''seq_length''' , _a ) __a = getattr(self.model_tester , '''decoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''encoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''d_model''' , _a ) __a = getattr(self.model_tester , '''num_attention_heads''' , _a ) __a = d_model // num_attention_heads for model_class in self.all_model_classes: __a = True __a = False __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) # check that output_attentions also work using config del inputs_dict["output_attentions"] __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) self.assertListEqual( list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, encoder_seq_length, dim] , ) __a = len(_a ) __a = 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(_a , _a ) # decoder attentions __a = outputs.decoder_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = outputs.cross_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = True __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) self.assertEqual(out_len + 2 , len(_a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , 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 __UpperCAmelCase ( self ): super().test_retain_grad_hidden_states_attentions() def lowercase ( lowerCAmelCase__ : Optional[int]="train-batch.pt" ) -> List[str]: __a = hf_hub_download(repo_id='''hf-internal-testing/tourism-monthly-batch''' , filename=lowerCAmelCase__ , repo_type='''dataset''' ) __a = torch.load(lowerCAmelCase__ , map_location=lowerCAmelCase__ ) return batch @require_torch @slow class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = AutoformerModel.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch() with torch.no_grad(): __a = 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] __a = torch.Size( (64, model.config.prediction_length + model.config.label_length, model.config.feature_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[0.3593, -1.3398, 0.6330], [0.2279, 1.5396, -0.1792], [0.0450, 1.3225, -0.2335]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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 __a = torch.Size((64, model.config.context_length, model.config.d_model) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[-0.0734, -0.9036, 0.8358], [4.7186, 2.4113, 1.9581], [1.7953, 2.3558, 1.2970]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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'''] , ) __a = torch.Size((64, model.config.num_parallel_samples, model.config.prediction_length) ) self.assertEqual(outputs.sequences.shape , _a ) __a = torch.tensor([3130.6763, 4056.5293, 7053.0786] , device=_a ) __a = outputs.sequences.mean(dim=1 ) self.assertTrue(torch.allclose(mean_prediction[0, -3:] , _a , rtol=1E-1 ) )
65
0
"""simple docstring""" import json import os import unittest from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer from ...test_tokenization_common import TokenizerTesterMixin class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : str = CTRLTokenizer __UpperCAmelCase : Any = False __UpperCAmelCase : Optional[int] = False def __UpperCAmelCase ( self ): super().setUp() # Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt __a = ['''adapt''', '''re@@''', '''a@@''', '''apt''', '''c@@''', '''t''', '''<unk>'''] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = ['''#version: 0.2''', '''a p''', '''ap t</w>''', '''r e''', '''a d''', '''ad apt</w>''', ''''''] __a = {'''unk_token''': '''<unk>'''} __a = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''vocab_file'''] ) __a = 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(_a ) + '''\n''' ) with open(self.merges_file , '''w''' , encoding='''utf-8''' ) as fp: fp.write('''\n'''.join(_a ) ) def __UpperCAmelCase ( self , **_a ): kwargs.update(self.special_tokens_map ) return CTRLTokenizer.from_pretrained(self.tmpdirname , **_a ) def __UpperCAmelCase ( self , _a ): __a = '''adapt react readapt apt''' __a = '''adapt react readapt apt''' return input_text, output_text def __UpperCAmelCase ( self ): __a = CTRLTokenizer(self.vocab_file , self.merges_file , **self.special_tokens_map ) __a = '''adapt react readapt apt''' __a = '''adapt re@@ a@@ c@@ t re@@ adapt apt'''.split() __a = tokenizer.tokenize(_a ) self.assertListEqual(_a , _a ) __a = tokens + [tokenizer.unk_token] __a = [0, 1, 2, 4, 5, 1, 0, 3, 6] self.assertListEqual(tokenizer.convert_tokens_to_ids(_a ) , _a )
719
"""simple docstring""" import inspect from typing import Optional, Union import numpy as np import PIL import torch from torch.nn import functional as F from torchvision import transforms from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DiffusionPipeline, DPMSolverMultistepScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, randn_tensor, ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> Any: if isinstance(lowerCAmelCase__ , torch.Tensor ): return image elif isinstance(lowerCAmelCase__ , PIL.Image.Image ): __a = [image] if isinstance(image[0] , PIL.Image.Image ): __a = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['''lanczos'''] ) )[None, :] for i in image] __a = np.concatenate(lowerCAmelCase__ , axis=0 ) __a = np.array(lowerCAmelCase__ ).astype(np.floataa ) / 2_55.0 __a = image.transpose(0 , 3 , 1 , 2 ) __a = 2.0 * image - 1.0 __a = torch.from_numpy(lowerCAmelCase__ ) elif isinstance(image[0] , torch.Tensor ): __a = torch.cat(lowerCAmelCase__ , dim=0 ) return image def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=0.99_95 ) -> int: if not isinstance(lowerCAmelCase__ , np.ndarray ): __a = True __a = va.device __a = va.cpu().numpy() __a = va.cpu().numpy() __a = np.sum(va * va / (np.linalg.norm(lowerCAmelCase__ ) * np.linalg.norm(lowerCAmelCase__ )) ) if np.abs(lowerCAmelCase__ ) > DOT_THRESHOLD: __a = (1 - t) * va + t * va else: __a = np.arccos(lowerCAmelCase__ ) __a = np.sin(lowerCAmelCase__ ) __a = theta_a * t __a = np.sin(lowerCAmelCase__ ) __a = np.sin(theta_a - theta_t ) / sin_theta_a __a = sin_theta_t / sin_theta_a __a = sa * va + sa * va if inputs_are_torch: __a = torch.from_numpy(lowerCAmelCase__ ).to(lowerCAmelCase__ ) return va def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> int: __a = F.normalize(lowerCAmelCase__ , dim=-1 ) __a = F.normalize(lowerCAmelCase__ , dim=-1 ) return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> List[str]: for param in model.parameters(): __a = value class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a , _a , _a , _a=None , _a=None , _a=None , ): super().__init__() self.register_modules( vae=_a , text_encoder=_a , clip_model=_a , tokenizer=_a , unet=_a , scheduler=_a , feature_extractor=_a , coca_model=_a , coca_tokenizer=_a , coca_transform=_a , ) __a = ( feature_extractor.size if isinstance(feature_extractor.size , _a ) else feature_extractor.size['''shortest_edge'''] ) __a = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std ) set_requires_grad(self.text_encoder , _a ) set_requires_grad(self.clip_model , _a ) def __UpperCAmelCase ( self , _a = "auto" ): if slice_size == "auto": # half the attention head size is usually a good trade-off between # speed and memory __a = self.unet.config.attention_head_dim // 2 self.unet.set_attention_slice(_a ) def __UpperCAmelCase ( self ): self.enable_attention_slicing(_a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self , _a , _a , _a ): # get the original timestep using init_timestep __a = min(int(num_inference_steps * strength ) , _a ) __a = max(num_inference_steps - init_timestep , 0 ) __a = self.scheduler.timesteps[t_start:] return timesteps, num_inference_steps - t_start def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a=None ): if not isinstance(_a , torch.Tensor ): raise ValueError(f'''`image` has to be of type `torch.Tensor` but is {type(_a )}''' ) __a = image.to(device=_a , dtype=_a ) if isinstance(_a , _a ): __a = [ self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_a ) ] __a = torch.cat(_a , dim=0 ) else: __a = self.vae.encode(_a ).latent_dist.sample(_a ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 0.1_8215 * init_latents __a = init_latents.repeat_interleave(_a , dim=0 ) __a = randn_tensor(init_latents.shape , generator=_a , device=_a , dtype=_a ) # get latents __a = self.scheduler.add_noise(_a , _a , _a ) __a = init_latents return latents def __UpperCAmelCase ( self , _a ): __a = self.coca_transform(_a ).unsqueeze(0 ) with torch.no_grad(), torch.cuda.amp.autocast(): __a = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) ) __a = self.coca_tokenizer.decode(generated[0].cpu().numpy() ) return generated.split('''<end_of_text>''' )[0].replace('''<start_of_text>''' , '''''' ).rstrip(''' .,''' ) def __UpperCAmelCase ( self , _a , _a ): __a = self.feature_extractor.preprocess(_a ) __a = torch.from_numpy(clip_image_input['''pixel_values'''][0] ).unsqueeze(0 ).to(self.device ).half() __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = image_embeddings_clip.repeat_interleave(_a , dim=0 ) return image_embeddings_clip @torch.enable_grad() def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , ): __a = latents.detach().requires_grad_() __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ): __a = self.scheduler.alphas_cumprod[timestep] __a = 1 - alpha_prod_t # compute predicted original sample from predicted noise also called # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf __a = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5 __a = torch.sqrt(_a ) __a = pred_original_sample * (fac) + latents * (1 - fac) elif isinstance(self.scheduler , _a ): __a = self.scheduler.sigmas[index] __a = latents - sigma * noise_pred else: raise ValueError(f'''scheduler type {type(self.scheduler )} not supported''' ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * sample __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = transforms.Resize(self.feature_extractor_size )(_a ) __a = self.normalize(_a ).to(latents.dtype ) __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = spherical_dist_loss(_a , _a ).mean() * clip_guidance_scale __a = -torch.autograd.grad(_a , _a )[0] if isinstance(self.scheduler , _a ): __a = latents.detach() + grads * (sigma**2) __a = noise_pred_original else: __a = noise_pred_original - torch.sqrt(_a ) * grads return noise_pred, latents @torch.no_grad() def __call__( self , _a , _a , _a = None , _a = None , _a = 512 , _a = 512 , _a = 0.6 , _a = 50 , _a = 7.5 , _a = 1 , _a = 0.0 , _a = 100 , _a = None , _a = "pil" , _a = True , _a = 0.8 , _a = 0.1 , _a = 0.1 , ): if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError(f'''You have passed {batch_size} batch_size, but only {len(_a )} generators.''' ) if height % 8 != 0 or width % 8 != 0: raise ValueError(f'''`height` and `width` have to be divisible by 8 but are {height} and {width}.''' ) if isinstance(_a , torch.Generator ) and batch_size > 1: __a = [generator] + [None] * (batch_size - 1) __a = [ ('''model''', self.coca_model is None), ('''tokenizer''', self.coca_tokenizer is None), ('''transform''', self.coca_transform is None), ] __a = [x[0] for x in coca_is_none if x[1]] __a = ''', '''.join(_a ) # generate prompts with coca model if prompt is None if content_prompt is None: if len(_a ): raise ValueError( f'''Content prompt is None and CoCa [{coca_is_none_str}] is None.''' f'''Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) if style_prompt is None: if len(_a ): raise ValueError( f'''Style prompt is None and CoCa [{coca_is_none_str}] is None.''' f''' Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) # get prompt text embeddings for content and style __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0] __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0] __a = slerp(_a , _a , _a ) # duplicate text embeddings for each generation per prompt __a = text_embeddings.repeat_interleave(_a , dim=0 ) # set timesteps __a = '''offset''' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() ) __a = {} if accepts_offset: __a = 1 self.scheduler.set_timesteps(_a , **_a ) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand self.scheduler.timesteps.to(self.device ) __a , __a = self.get_timesteps(_a , _a , self.device ) __a = timesteps[:1].repeat(_a ) # Preprocess image __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = slerp(_a , _a , _a ) if clip_guidance_scale > 0: __a = self.get_clip_image_embeddings(_a , _a ) __a = self.get_clip_image_embeddings(_a , _a ) __a = slerp( _a , _a , _a ) # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` # corresponds to doing no classifier free guidance. __a = guidance_scale > 1.0 # get unconditional embeddings for classifier free guidance if do_classifier_free_guidance: __a = content_text_input.input_ids.shape[-1] __a = self.tokenizer([''''''] , padding='''max_length''' , max_length=_a , return_tensors='''pt''' ) __a = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0] # duplicate unconditional embeddings for each generation per prompt __a = uncond_embeddings.repeat_interleave(_a , dim=0 ) # For classifier free guidance, we need to do two forward passes. # Here we concatenate the unconditional and text embeddings into a single batch # to avoid doing two forward passes __a = torch.cat([uncond_embeddings, text_embeddings] ) # get the initial random noise unless the user supplied it # Unlike in other pipelines, latents need to be generated in the target device # for 1-to-1 results reproducibility with the CompVis implementation. # However this currently doesn't work in `mps`. __a = (batch_size, self.unet.config.in_channels, height // 8, width // 8) __a = text_embeddings.dtype if latents is None: if self.device.type == "mps": # randn does not work reproducibly on mps __a = torch.randn(_a , generator=_a , device='''cpu''' , dtype=_a ).to( self.device ) else: __a = torch.randn(_a , generator=_a , device=self.device , dtype=_a ) else: if latents.shape != latents_shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {latents_shape}''' ) __a = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler __a = latents * self.scheduler.init_noise_sigma # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 # and should be between [0, 1] __a = '''eta''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) __a = {} if accepts_eta: __a = eta # check if the scheduler accepts generator __a = '''generator''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) if accepts_generator: __a = generator with self.progress_bar(total=_a ): for i, t in enumerate(_a ): # expand the latents if we are doing classifier free guidance __a = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample # perform classifier free guidance if do_classifier_free_guidance: __a , __a = noise_pred.chunk(2 ) __a = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) # perform clip guidance if clip_guidance_scale > 0: __a = ( text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings ) __a , __a = self.cond_fn( _a , _a , _a , _a , _a , _a , _a , ) # compute the previous noisy sample x_t -> x_t-1 __a = self.scheduler.step(_a , _a , _a , **_a ).prev_sample # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * latents __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image, None) return StableDiffusionPipelineOutput(images=_a , nsfw_content_detected=_a )
65
0
"""simple docstring""" import os from dataclasses import dataclass, field from io import BytesIO from typing import TYPE_CHECKING, Any, ClassVar, Dict, Optional, Union import numpy as np import pyarrow as pa from .. import config from ..download.streaming_download_manager import xopen, xsplitext from ..table import array_cast from ..utils.py_utils import no_op_if_value_is_null, string_to_dict if TYPE_CHECKING: from .features import FeatureType lowercase_ , lowercase_ , lowercase_ = False, False, False @dataclass class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Optional[int] = None __UpperCAmelCase : bool = True __UpperCAmelCase : bool = True __UpperCAmelCase : Optional[str] = None # Automatically constructed __UpperCAmelCase : ClassVar[str] = "dict" __UpperCAmelCase : ClassVar[Any] = pa.struct({'bytes': pa.binary(), 'path': pa.string()} ) __UpperCAmelCase : str = field(default='Audio' , init=__SCREAMING_SNAKE_CASE , repr=__SCREAMING_SNAKE_CASE ) def __call__( self ): return self.pa_type def __UpperCAmelCase ( self , _a ): try: import soundfile as sf # soundfile is a dependency of librosa, needed to decode audio files. except ImportError as err: raise ImportError('''To support encoding audio data, please install \'soundfile\'.''' ) from err if isinstance(_a , _a ): return {"bytes": None, "path": value} elif isinstance(_a , _a ): return {"bytes": value, "path": None} elif "array" in value: # convert the audio array to wav bytes __a = BytesIO() sf.write(_a , value['''array'''] , value['''sampling_rate'''] , format='''wav''' ) return {"bytes": buffer.getvalue(), "path": None} elif value.get('''path''' ) is not None and os.path.isfile(value['''path'''] ): # we set "bytes": None to not duplicate the data if they're already available locally if value["path"].endswith('''pcm''' ): # "PCM" only has raw audio bytes if value.get('''sampling_rate''' ) is None: # At least, If you want to convert "PCM-byte" to "WAV-byte", you have to know sampling rate raise KeyError('''To use PCM files, please specify a \'sampling_rate\' in Audio object''' ) if value.get('''bytes''' ): # If we already had PCM-byte, we don`t have to make "read file, make bytes" (just use it!) __a = np.frombuffer(value['''bytes'''] , dtype=np.intaa ).astype(np.floataa ) / 32_767 else: __a = np.memmap(value['''path'''] , dtype='''h''' , mode='''r''' ).astype(np.floataa ) / 32_767 __a = BytesIO(bytes() ) sf.write(_a , _a , value['''sampling_rate'''] , format='''wav''' ) return {"bytes": buffer.getvalue(), "path": None} else: return {"bytes": None, "path": value.get('''path''' )} elif value.get('''bytes''' ) is not None or value.get('''path''' ) is not None: # store the audio bytes, and path is used to infer the audio format using the file extension return {"bytes": value.get('''bytes''' ), "path": value.get('''path''' )} else: raise ValueError( f'''An audio sample should have one of \'path\' or \'bytes\' but they are missing or None in {value}.''' ) def __UpperCAmelCase ( self , _a , _a = None ): if not self.decode: raise RuntimeError('''Decoding is disabled for this feature. Please use Audio(decode=True) instead.''' ) __a , __a = (value['''path'''], BytesIO(value['''bytes'''] )) if value['''bytes'''] is not None else (value['''path'''], None) if path is None and file is None: raise ValueError(f'''An audio sample should have one of \'path\' or \'bytes\' but both are None in {value}.''' ) try: import librosa import soundfile as sf except ImportError as err: raise ImportError('''To support decoding audio files, please install \'librosa\' and \'soundfile\'.''' ) from err __a = xsplitext(_a )[1][1:].lower() if path is not None else None if not config.IS_OPUS_SUPPORTED and audio_format == "opus": raise RuntimeError( '''Decoding \'opus\' files requires system library \'libsndfile\'>=1.0.31, ''' '''You can try to update `soundfile` python library: `pip install "soundfile>=0.12.1"`. ''' ) elif not config.IS_MP3_SUPPORTED and audio_format == "mp3": raise RuntimeError( '''Decoding \'mp3\' files requires system library \'libsndfile\'>=1.1.0, ''' '''You can try to update `soundfile` python library: `pip install "soundfile>=0.12.1"`. ''' ) if file is None: __a = token_per_repo_id or {} __a = path.split('''::''' )[-1] try: __a = string_to_dict(_a , config.HUB_DATASETS_URL )['''repo_id'''] __a = token_per_repo_id[repo_id] except (ValueError, KeyError): __a = None with xopen(_a , '''rb''' , use_auth_token=_a ) as f: __a , __a = sf.read(_a ) else: __a , __a = sf.read(_a ) __a = array.T if self.mono: __a = librosa.to_mono(_a ) if self.sampling_rate and self.sampling_rate != sampling_rate: __a = librosa.resample(_a , orig_sr=_a , target_sr=self.sampling_rate ) __a = self.sampling_rate return {"path": path, "array": array, "sampling_rate": sampling_rate} def __UpperCAmelCase ( self ): from .features import Value if self.decode: raise ValueError('''Cannot flatten a decoded Audio feature.''' ) return { "bytes": Value('''binary''' ), "path": Value('''string''' ), } def __UpperCAmelCase ( self , _a ): if pa.types.is_string(storage.type ): __a = pa.array([None] * len(_a ) , type=pa.binary() ) __a = pa.StructArray.from_arrays([bytes_array, storage] , ['''bytes''', '''path'''] , mask=storage.is_null() ) elif pa.types.is_binary(storage.type ): __a = pa.array([None] * len(_a ) , type=pa.string() ) __a = pa.StructArray.from_arrays([storage, path_array] , ['''bytes''', '''path'''] , mask=storage.is_null() ) elif pa.types.is_struct(storage.type ) and storage.type.get_all_field_indices('''array''' ): __a = pa.array([Audio().encode_example(_a ) if x is not None else None for x in storage.to_pylist()] ) elif pa.types.is_struct(storage.type ): if storage.type.get_field_index('''bytes''' ) >= 0: __a = storage.field('''bytes''' ) else: __a = pa.array([None] * len(_a ) , type=pa.binary() ) if storage.type.get_field_index('''path''' ) >= 0: __a = storage.field('''path''' ) else: __a = pa.array([None] * len(_a ) , type=pa.string() ) __a = pa.StructArray.from_arrays([bytes_array, path_array] , ['''bytes''', '''path'''] , mask=storage.is_null() ) return array_cast(_a , self.pa_type ) def __UpperCAmelCase ( self , _a ): @no_op_if_value_is_null def path_to_bytes(_a ): with xopen(_a , '''rb''' ) as f: __a = f.read() return bytes_ __a = pa.array( [ (path_to_bytes(x['''path'''] ) if x['''bytes'''] is None else x['''bytes''']) if x is not None else None for x in storage.to_pylist() ] , type=pa.binary() , ) __a = pa.array( [os.path.basename(_a ) if path is not None else None for path in storage.field('''path''' ).to_pylist()] , type=pa.string() , ) __a = pa.StructArray.from_arrays([bytes_array, path_array] , ['''bytes''', '''path'''] , mask=bytes_array.is_null() ) return array_cast(_a , self.pa_type )
720
"""simple docstring""" # tests directory-specific settings - this file is run automatically # by pytest before any tests are run import sys import warnings from os.path import abspath, dirname, join # allow having multiple repository checkouts and not needing to remember to rerun # 'pip install -e .[dev]' when switching between checkouts and running tests. lowercase_ = abspath(join(dirname(dirname(dirname(__file__))), "src")) sys.path.insert(1, git_repo_path) # silence FutureWarning warnings in tests since often we can't act on them until # they become normal warnings - i.e. the tests still need to test the current functionality warnings.simplefilter(action="ignore", category=FutureWarning) def lowercase ( lowerCAmelCase__ : List[Any] ) -> str: from transformers.testing_utils import pytest_addoption_shared pytest_addoption_shared(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Union[str, Any]: from transformers.testing_utils import pytest_terminal_summary_main __a = terminalreporter.config.getoption('''--make-reports''' ) if make_reports: pytest_terminal_summary_main(lowerCAmelCase__ , id=lowerCAmelCase__ )
65
0
"""simple docstring""" import copy from typing import Any, Dict, List, Optional, Union import numpy as np import torch from ...audio_utils import mel_filter_bank, spectrogram, window_function from ...feature_extraction_sequence_utils import SequenceFeatureExtractor from ...feature_extraction_utils import BatchFeature from ...utils import TensorType, logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Any = ['input_features', 'is_longer'] def __init__( self , _a=64 , _a=48_000 , _a=480 , _a=10 , _a=1_024 , _a=0.0 , _a=False , _a = 0 , _a = 14_000 , _a = None , _a = "fusion" , _a = "repeatpad" , **_a , ): super().__init__( feature_size=_a , sampling_rate=_a , padding_value=_a , return_attention_mask=_a , **_a , ) __a = top_db __a = truncation __a = padding __a = fft_window_size __a = (fft_window_size >> 1) + 1 __a = hop_length __a = max_length_s __a = max_length_s * sampling_rate __a = sampling_rate __a = frequency_min __a = frequency_max __a = mel_filter_bank( num_frequency_bins=self.nb_frequency_bins , num_mel_filters=_a , min_frequency=_a , max_frequency=_a , sampling_rate=_a , norm=_a , mel_scale='''htk''' , ) __a = mel_filter_bank( num_frequency_bins=self.nb_frequency_bins , num_mel_filters=_a , min_frequency=_a , max_frequency=_a , sampling_rate=_a , norm='''slaney''' , mel_scale='''slaney''' , ) def __UpperCAmelCase ( self ): __a = copy.deepcopy(self.__dict__ ) __a = self.__class__.__name__ if "mel_filters" in output: del output["mel_filters"] if "mel_filters_slaney" in output: del output["mel_filters_slaney"] return output def __UpperCAmelCase ( self , _a , _a = None ): __a = spectrogram( _a , window_function(self.fft_window_size , '''hann''' ) , frame_length=self.fft_window_size , hop_length=self.hop_length , power=2.0 , mel_filters=_a , log_mel='''dB''' , ) return log_mel_spectrogram.T def __UpperCAmelCase ( self , _a , _a , _a ): __a = np.array_split(list(range(0 , total_frames - chunk_frames + 1 ) ) , 3 ) if len(ranges[1] ) == 0: # if the audio is too short, we just use the first chunk __a = [0] if len(ranges[2] ) == 0: # if the audio is too short, we just use the first chunk __a = [0] # randomly choose index for each part __a = np.random.choice(ranges[0] ) __a = np.random.choice(ranges[1] ) __a = np.random.choice(ranges[2] ) __a = mel[idx_front : idx_front + chunk_frames, :] __a = mel[idx_middle : idx_middle + chunk_frames, :] __a = mel[idx_back : idx_back + chunk_frames, :] __a = torch.tensor(mel[None, None, :] ) __a = torch.nn.functional.interpolate( _a , size=[chunk_frames, 64] , mode='''bilinear''' , align_corners=_a ) __a = mel_shrink[0][0].numpy() __a = np.stack([mel_shrink, mel_chunk_front, mel_chunk_middle, mel_chunk_back] , axis=0 ) return mel_fusion def __UpperCAmelCase ( self , _a , _a , _a , _a ): if waveform.shape[0] > max_length: if truncation == "rand_trunc": __a = True # random crop to max_length (for compatibility) -> this should be handled by self.pad __a = len(_a ) - max_length __a = np.random.randint(0 , overflow + 1 ) __a = waveform[idx : idx + max_length] __a = self._np_extract_fbank_features(_a , self.mel_filters_slaney )[None, :] elif truncation == "fusion": __a = self._np_extract_fbank_features(_a , self.mel_filters ) __a = max_length // self.hop_length + 1 # the +1 related to how the spectrogram is computed __a = mel.shape[0] if chunk_frames == total_frames: # there is a corner case where the audio length is larger than max_length but smaller than max_length+hop_length. # In this case, we just use the whole audio. __a = np.stack([mel, mel, mel, mel] , axis=0 ) __a = False else: __a = self._random_mel_fusion(_a , _a , _a ) __a = True else: raise NotImplementedError(f'''data_truncating {truncation} not implemented''' ) else: __a = False # only use repeat as a new possible value for padding. you repeat the audio before applying the usual max_length padding if waveform.shape[0] < max_length: if padding == "repeat": __a = int(max_length / len(_a ) ) __a = np.stack(np.tile(_a , n_repeat + 1 ) )[:max_length] if padding == "repeatpad": __a = int(max_length / len(_a ) ) __a = np.stack(np.tile(_a , _a ) ) __a = np.pad(_a , (0, max_length - waveform.shape[0]) , mode='''constant''' , constant_values=0 ) if truncation == "fusion": __a = self._np_extract_fbank_features(_a , self.mel_filters ) __a = np.stack([input_mel, input_mel, input_mel, input_mel] , axis=0 ) else: __a = self._np_extract_fbank_features(_a , self.mel_filters_slaney )[None, :] return input_mel, longer def __call__( self , _a , _a = None , _a = None , _a = None , _a = None , _a = None , **_a , ): __a = truncation if truncation is not None else self.truncation __a = padding if padding else self.padding if sampling_rate is not None: if sampling_rate != self.sampling_rate: raise ValueError( f'''The model corresponding to this feature extractor: {self.__class__.__name__} was trained using a''' f''' sampling rate of {self.sampling_rate}. Please make sure that the provided `raw_speech` input''' f''' was sampled with {self.sampling_rate} and not {sampling_rate}.''' ) else: logger.warning( '''It is strongly recommended to pass the `sampling_rate` argument to this function. ''' '''Failing to do so can result in silent errors that might be hard to debug.''' ) __a = isinstance(_a , np.ndarray ) and len(raw_speech.shape ) > 1 if is_batched_numpy and len(raw_speech.shape ) > 2: raise ValueError(f'''Only mono-channel audio is supported for input to {self}''' ) __a = is_batched_numpy or ( isinstance(_a , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) )) ) if is_batched: __a = [np.asarray(_a , dtype=np.floataa ) for speech in raw_speech] elif not is_batched and not isinstance(_a , np.ndarray ): __a = np.asarray(_a , dtype=np.floataa ) elif isinstance(_a , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ): __a = raw_speech.astype(np.floataa ) # always return batch if not is_batched: __a = [np.asarray(_a )] # convert to mel spectrogram, truncate and pad if needed. __a = [ self._get_input_mel(_a , max_length if max_length else self.nb_max_samples , _a , _a ) for waveform in raw_speech ] __a = [] __a = [] for mel, longer in padded_inputs: input_mel.append(_a ) is_longer.append(_a ) if truncation == "fusion" and sum(_a ) == 0: # if no audio is longer than 10s, then randomly select one audio to be longer __a = np.random.randint(0 , len(_a ) ) __a = True if isinstance(input_mel[0] , _a ): __a = [np.asarray(_a , dtype=np.floataa ) for feature in input_mel] # is_longer is a list of bool __a = [[longer] for longer in is_longer] __a = {'''input_features''': input_mel, '''is_longer''': is_longer} __a = BatchFeature(_a ) if return_tensors is not None: __a = input_features.convert_to_tensors(_a ) return input_features
721
"""simple docstring""" 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. lowercase_ = {"LayoutLMv2Config", "LayoutLMv3Config"} @is_pipeline_test class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING __UpperCAmelCase : Any = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING if model_mapping is not None: __UpperCAmelCase : Any = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP} if tf_model_mapping is not None: __UpperCAmelCase : Dict = { config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP } def __UpperCAmelCase ( self , _a , _a , _a ): __a = ZeroShotClassificationPipeline( model=_a , tokenizer=_a , candidate_labels=['''polics''', '''health'''] ) return classifier, ["Who are you voting for in 2020?", "My stomach hurts."] def __UpperCAmelCase ( self , _a , _a ): __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # No kwarg __a = classifier('''Who are you voting for in 2020?''' , ['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics, public health''' ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health'''] ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''This text is about {}''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # https://github.com/huggingface/transformers/issues/13846 __a = classifier(['''I am happy'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(1 ) ] , ) __a = classifier(['''I am happy''', '''I am sad'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(2 ) ] , ) with self.assertRaises(_a ): classifier('''''' , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier(_a , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels='''''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels=_a ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''Not formatting template''' , ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template=_a , ) self.run_entailment_id(_a ) def __UpperCAmelCase ( self , _a ): __a = zero_shot_classifier.model.config __a = config.labelaid __a = zero_shot_classifier.entailment_id __a = {'''LABEL_0''': 0, '''LABEL_1''': 1, '''LABEL_2''': 2} self.assertEqual(zero_shot_classifier.entailment_id , -1 ) __a = {'''entailment''': 0, '''neutral''': 1, '''contradiction''': 2} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 0, '''NON-ENTAIL''': 1} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 2, '''NEUTRAL''': 1, '''CONTR''': 0} self.assertEqual(zero_shot_classifier.entailment_id , 2 ) __a = original_labelaid self.assertEqual(_a , zero_shot_classifier.entailment_id ) @require_torch def __UpperCAmelCase ( self ): __a = 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 __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''pt''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @require_tf def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''tf''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @slow @require_torch def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''pt''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , ) @slow @require_tf def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''tf''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , )
65
0
"""simple docstring""" import json import os import shutil import tempfile import unittest import numpy as np import pytest from transformers import CLIPTokenizer, CLIPTokenizerFast from transformers.models.clip.tokenization_clip import VOCAB_FILES_NAMES from transformers.testing_utils import require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import CLIPSegProcessor, ViTImageProcessor @require_vision class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = tempfile.mkdtemp() # fmt: off __a = ['''l''', '''o''', '''w''', '''e''', '''r''', '''s''', '''t''', '''i''', '''d''', '''n''', '''lo''', '''l</w>''', '''w</w>''', '''r</w>''', '''t</w>''', '''low</w>''', '''er</w>''', '''lowest</w>''', '''newer</w>''', '''wider''', '''<unk>''', '''<|startoftext|>''', '''<|endoftext|>'''] # fmt: on __a = dict(zip(_a , range(len(_a ) ) ) ) __a = ['''#version: 0.2''', '''l o''', '''lo w</w>''', '''e r</w>''', ''''''] __a = {'''unk_token''': '''<unk>'''} __a = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''vocab_file'''] ) __a = 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(_a ) + '''\n''' ) with open(self.merges_file , '''w''' , encoding='''utf-8''' ) as fp: fp.write('''\n'''.join(_a ) ) __a = { '''do_resize''': True, '''size''': 20, '''do_center_crop''': True, '''crop_size''': 18, '''do_normalize''': True, '''image_mean''': [0.4814_5466, 0.457_8275, 0.4082_1073], '''image_std''': [0.2686_2954, 0.2613_0258, 0.2757_7711], } __a = os.path.join(self.tmpdirname , _a ) with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp: json.dump(_a , _a ) def __UpperCAmelCase ( self , **_a ): return CLIPTokenizer.from_pretrained(self.tmpdirname , **_a ) def __UpperCAmelCase ( self , **_a ): return CLIPTokenizerFast.from_pretrained(self.tmpdirname , **_a ) def __UpperCAmelCase ( self , **_a ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a ) def __UpperCAmelCase ( self ): shutil.rmtree(self.tmpdirname ) def __UpperCAmelCase ( self ): __a = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] __a = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs] return image_inputs def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = self.get_rust_tokenizer() __a = self.get_image_processor() __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) processor_slow.save_pretrained(self.tmpdirname ) __a = CLIPSegProcessor.from_pretrained(self.tmpdirname , use_fast=_a ) __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) processor_fast.save_pretrained(self.tmpdirname ) __a = CLIPSegProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor_slow.tokenizer.get_vocab() , tokenizer_slow.get_vocab() ) self.assertEqual(processor_fast.tokenizer.get_vocab() , tokenizer_fast.get_vocab() ) self.assertEqual(tokenizer_slow.get_vocab() , tokenizer_fast.get_vocab() ) self.assertIsInstance(processor_slow.tokenizer , _a ) self.assertIsInstance(processor_fast.tokenizer , _a ) self.assertEqual(processor_slow.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertEqual(processor_fast.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor_slow.image_processor , _a ) self.assertIsInstance(processor_fast.image_processor , _a ) def __UpperCAmelCase ( self ): __a = CLIPSegProcessor(tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) __a = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' ) __a = self.get_image_processor(do_normalize=_a , padding_value=1.0 ) __a = CLIPSegProcessor.from_pretrained( self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , _a ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , _a ) def __UpperCAmelCase ( self ): __a = self.get_image_processor() __a = self.get_tokenizer() __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) __a = self.prepare_image_inputs() __a = image_processor(_a , return_tensors='''np''' ) __a = processor(images=_a , return_tensors='''np''' ) for key in input_feat_extract.keys(): self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1E-2 ) def __UpperCAmelCase ( self ): __a = self.get_image_processor() __a = self.get_tokenizer() __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) __a = '''lower newer''' __a = processor(text=_a ) __a = tokenizer(_a ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def __UpperCAmelCase ( self ): __a = self.get_image_processor() __a = self.get_tokenizer() __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) __a = '''lower newer''' __a = self.prepare_image_inputs() __a = processor(text=_a , images=_a ) self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''attention_mask''', '''pixel_values'''] ) # test if it raises when no input is passed with pytest.raises(_a ): processor() def __UpperCAmelCase ( self ): __a = self.get_image_processor() __a = self.get_tokenizer() __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) __a = self.prepare_image_inputs() __a = self.prepare_image_inputs() __a = processor(images=_a , visual_prompt=_a ) self.assertListEqual(list(inputs.keys() ) , ['''pixel_values''', '''conditional_pixel_values'''] ) # test if it raises when no input is passed with pytest.raises(_a ): processor() def __UpperCAmelCase ( self ): __a = self.get_image_processor() __a = self.get_tokenizer() __a = CLIPSegProcessor(tokenizer=_a , image_processor=_a ) __a = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] __a = processor.batch_decode(_a ) __a = tokenizer.batch_decode(_a ) self.assertListEqual(_a , _a )
700
"""simple docstring""" import pytest from datasets import inspect_metric, list_metrics, load_metric @pytest.fixture def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' , set() ) @pytest.fixture def lowercase ( lowerCAmelCase__ : Any ) -> Any: class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = metric_id class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Any = [MetricMock(__SCREAMING_SNAKE_CASE ) for metric_id in ['accuracy', 'mse', 'precision', 'codeparrot/apps_metric']] def __UpperCAmelCase ( self ): return self._metrics monkeypatch.setattr('''datasets.inspect.huggingface_hub''' , HfhMock() ) @pytest.mark.parametrize( '''func, args''' , [(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Tuple ) -> Optional[int]: if "tmp_path" in args: __a = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args ) with pytest.warns(lowerCAmelCase__ , match='''https://huggingface.co/docs/evaluate''' ): func(*lowerCAmelCase__ )
65
0
"""simple docstring""" import argparse import os # New Code # import evaluate import torch from datasets import load_dataset from torch.optim import AdamW from torch.utils.data import DataLoader from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed from accelerate import Accelerator, DistributedType from accelerate.utils import find_executable_batch_size ######################################################################## # This is a fully working simple example to use Accelerate, # specifically showcasing how to ensure out-of-memory errors never # interrupt training, and builds off the `nlp_example.py` script. # # This example trains a Bert base model on GLUE MRPC # in any of the following settings (with the same script): # - single CPU or single GPU # - multi GPUS (using PyTorch distributed mode) # - (multi) TPUs # - fp16 (mixed-precision) or fp32 (normal precision) # # New additions from the base script can be found quickly by # looking for the # New Code # tags # # To run it in each of these various modes, follow the instructions # in the readme for examples: # https://github.com/huggingface/accelerate/tree/main/examples # ######################################################################## lowercase_ = 1_6 lowercase_ = 3_2 def lowercase ( lowerCAmelCase__ : Accelerator , lowerCAmelCase__ : int = 16 ) -> Dict: __a = AutoTokenizer.from_pretrained('''bert-base-cased''' ) __a = load_dataset('''glue''' , '''mrpc''' ) def tokenize_function(lowerCAmelCase__ : Dict ): # max_length=None => use the model max length (it's actually the default) __a = tokenizer(examples['''sentence1'''] , examples['''sentence2'''] , truncation=lowerCAmelCase__ , max_length=lowerCAmelCase__ ) return outputs # Apply the method we just defined to all the examples in all the splits of the dataset # starting with the main process first: with accelerator.main_process_first(): __a = datasets.map( lowerCAmelCase__ , batched=lowerCAmelCase__ , remove_columns=['''idx''', '''sentence1''', '''sentence2'''] , ) # We also rename the 'label' column to 'labels' which is the expected name for labels by the models of the # transformers library __a = tokenized_datasets.rename_column('''label''' , '''labels''' ) def collate_fn(lowerCAmelCase__ : int ): # On TPU it's best to pad everything to the same length or training will be very slow. __a = 128 if accelerator.distributed_type == DistributedType.TPU else None # When using mixed precision we want round multiples of 8/16 if accelerator.mixed_precision == "fp8": __a = 16 elif accelerator.mixed_precision != "no": __a = 8 else: __a = None return tokenizer.pad( lowerCAmelCase__ , padding='''longest''' , max_length=lowerCAmelCase__ , pad_to_multiple_of=lowerCAmelCase__ , return_tensors='''pt''' , ) # Instantiate dataloaders. __a = DataLoader( tokenized_datasets['''train'''] , shuffle=lowerCAmelCase__ , collate_fn=lowerCAmelCase__ , batch_size=lowerCAmelCase__ ) __a = DataLoader( tokenized_datasets['''validation'''] , shuffle=lowerCAmelCase__ , collate_fn=lowerCAmelCase__ , batch_size=lowerCAmelCase__ ) return train_dataloader, eval_dataloader # For testing only if os.environ.get("TESTING_MOCKED_DATALOADERS", None) == "1": from accelerate.test_utils.training import mocked_dataloaders lowercase_ = mocked_dataloaders # noqa: F811 def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[Any] ) -> Optional[int]: # For testing only if os.environ.get('''TESTING_MOCKED_DATALOADERS''' , lowerCAmelCase__ ) == "1": __a = 2 # Initialize accelerator __a = Accelerator(cpu=args.cpu , mixed_precision=args.mixed_precision ) # Sample hyper-parameters for learning rate, batch size, seed and a few other HPs __a = config['''lr'''] __a = int(config['''num_epochs'''] ) __a = int(config['''seed'''] ) __a = int(config['''batch_size'''] ) __a = evaluate.load('''glue''' , '''mrpc''' ) # New Code # # We now can define an inner training loop function. It should take a batch size as the only parameter, # and build the dataloaders in there. # It also gets our decorator @find_executable_batch_size(starting_batch_size=lowerCAmelCase__ ) def inner_training_loop(lowerCAmelCase__ : List[Any] ): # And now just move everything below under this function # We need to bring in the Accelerator object from earlier nonlocal accelerator # And reset all of its attributes that could hold onto any memory: accelerator.free_memory() # Then we can declare the model, optimizer, and everything else: set_seed(lowerCAmelCase__ ) # Instantiate the model (we build the model here so that the seed also control new weights initialization) __a = AutoModelForSequenceClassification.from_pretrained('''bert-base-cased''' , return_dict=lowerCAmelCase__ ) # We could avoid this line since the accelerator is set with `device_placement=True` (default value). # Note that if you are placing tensors on devices manually, this line absolutely needs to be before the optimizer # creation otherwise training will not work on TPU (`accelerate` will kindly throw an error to make us aware of that). __a = model.to(accelerator.device ) # Instantiate optimizer __a = AdamW(params=model.parameters() , lr=lowerCAmelCase__ ) __a , __a = get_dataloaders(lowerCAmelCase__ , lowerCAmelCase__ ) # Instantiate scheduler __a = get_linear_schedule_with_warmup( optimizer=lowerCAmelCase__ , num_warmup_steps=100 , num_training_steps=(len(lowerCAmelCase__ ) * num_epochs) , ) # Prepare everything # There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the # prepare method. __a , __a , __a , __a , __a = accelerator.prepare( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) # Now we train the model for epoch in range(lowerCAmelCase__ ): model.train() for step, batch in enumerate(lowerCAmelCase__ ): # We could avoid this line since we set the accelerator with `device_placement=True`. batch.to(accelerator.device ) __a = model(**lowerCAmelCase__ ) __a = outputs.loss accelerator.backward(lowerCAmelCase__ ) optimizer.step() lr_scheduler.step() optimizer.zero_grad() model.eval() for step, batch in enumerate(lowerCAmelCase__ ): # We could avoid this line since we set the accelerator with `device_placement=True`. batch.to(accelerator.device ) with torch.no_grad(): __a = model(**lowerCAmelCase__ ) __a = outputs.logits.argmax(dim=-1 ) __a , __a = accelerator.gather_for_metrics((predictions, batch['''labels''']) ) metric.add_batch( predictions=lowerCAmelCase__ , references=lowerCAmelCase__ , ) __a = metric.compute() # Use accelerator.print to print only on the main process. accelerator.print(f'''epoch {epoch}:''' , lowerCAmelCase__ ) # New Code # # And call it at the end with no arguments # Note: You could also refactor this outside of your training loop function inner_training_loop() def lowercase ( ) -> str: __a = argparse.ArgumentParser(description='''Simple example of training script.''' ) parser.add_argument( '''--mixed_precision''' , type=lowerCAmelCase__ , default=lowerCAmelCase__ , choices=['''no''', '''fp16''', '''bf16''', '''fp8'''] , help='''Whether to use mixed precision. Choose''' '''between fp16 and bf16 (bfloat16). Bf16 requires PyTorch >= 1.10.''' '''and an Nvidia Ampere GPU.''' , ) parser.add_argument('''--cpu''' , action='''store_true''' , help='''If passed, will train on the CPU.''' ) __a = parser.parse_args() __a = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16} training_function(lowerCAmelCase__ , lowerCAmelCase__ ) if __name__ == "__main__": main()
701
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> int: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or number < 0: raise ValueError('''Input must be a non-negative integer''' ) __a = 0 while number: # This way we arrive at next set bit (next 1) instead of looping # through each bit and checking for 1s hence the # loop won't run 32 times it will only run the number of `1` times number &= number - 1 count += 1 return count if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> bool: __a = (1 + 24 * n) ** 0.5 return ((1 + root) / 6) % 1 == 0 def lowercase ( lowerCAmelCase__ : int = 5000 ) -> int: __a = [(i * (3 * i - 1)) // 2 for i in range(1 , lowerCAmelCase__ )] for i, pentagonal_i in enumerate(lowerCAmelCase__ ): for j in range(lowerCAmelCase__ , len(lowerCAmelCase__ ) ): __a = pentagonal_nums[j] __a = pentagonal_i + pentagonal_j __a = pentagonal_j - pentagonal_i if is_pentagonal(lowerCAmelCase__ ) and is_pentagonal(lowerCAmelCase__ ): return b return -1 if __name__ == "__main__": print(F'''{solution() = }''')
702
"""simple docstring""" import pytest from datasets import Dataset, DatasetDict, Features, NamedSplit, Value from datasets.io.text import TextDatasetReader from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.split == split if split else "train" @pytest.mark.parametrize('''path_type''' , [str, list] ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict: if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = text_path elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [text_path] __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for split in splits: __a = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str: __a = tmp_path / '''cache''' # CSV file loses col_1 string dtype information: default now is "int64" instead of "string" __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict: if split: __a = {split: text_path} else: __a = '''train''' __a = {'''train''': text_path, '''test''': text_path} __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() )
65
0
def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> int: return int((input_a, input_a).count(0 ) != 0 ) def lowercase ( ) -> None: assert nand_gate(0 , 0 ) == 1 assert nand_gate(0 , 1 ) == 1 assert nand_gate(1 , 0 ) == 1 assert nand_gate(1 , 1 ) == 0 if __name__ == "__main__": print(nand_gate(0, 0)) print(nand_gate(0, 1)) print(nand_gate(1, 0)) print(nand_gate(1, 1))
703
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> bool: return numa ^ numa < 0 if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" from ...utils import ( OptionalDependencyNotAvailable, is_torch_available, is_transformers_available, is_transformers_version, ) try: if not (is_transformers_available() and is_torch_available() and is_transformers_version(">=", "4.25.0")): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: from ...utils.dummy_torch_and_transformers_objects import UnCLIPImageVariationPipeline, UnCLIPPipeline else: from .pipeline_unclip import UnCLIPPipeline from .pipeline_unclip_image_variation import UnCLIPImageVariationPipeline from .text_proj import UnCLIPTextProjModel
704
"""simple docstring""" import unittest from transformers import SPIECE_UNDERLINE from transformers.models.speechta import SpeechTaTokenizer from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow from transformers.tokenization_utils import AddedToken from ...test_tokenization_common import TokenizerTesterMixin lowercase_ = get_tests_dir("fixtures/test_sentencepiece_bpe_char.model") @require_sentencepiece @require_tokenizers class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : int = SpeechTaTokenizer __UpperCAmelCase : Tuple = False __UpperCAmelCase : Dict = True def __UpperCAmelCase ( self ): super().setUp() # We have a SentencePiece fixture for testing __a = SpeechTaTokenizer(_a ) __a = AddedToken('''<mask>''' , lstrip=_a , rstrip=_a ) __a = mask_token tokenizer.add_special_tokens({'''mask_token''': mask_token} ) tokenizer.add_tokens(['''<ctc_blank>'''] ) tokenizer.save_pretrained(self.tmpdirname ) def __UpperCAmelCase ( self , _a ): __a = '''this is a test''' __a = '''this is a test''' return input_text, output_text def __UpperCAmelCase ( self , _a , _a=False , _a=20 , _a=5 ): __a , __a = self.get_input_output_texts(_a ) __a = tokenizer.encode(_a , add_special_tokens=_a ) __a = tokenizer.decode(_a , clean_up_tokenization_spaces=_a ) return text, ids def __UpperCAmelCase ( self ): __a = '''<pad>''' __a = 1 self.assertEqual(self.get_tokenizer()._convert_token_to_id(_a ) , _a ) self.assertEqual(self.get_tokenizer()._convert_id_to_token(_a ) , _a ) def __UpperCAmelCase ( self ): __a = list(self.get_tokenizer().get_vocab().keys() ) self.assertEqual(vocab_keys[0] , '''<s>''' ) self.assertEqual(vocab_keys[1] , '''<pad>''' ) self.assertEqual(vocab_keys[-4] , '''œ''' ) self.assertEqual(vocab_keys[-2] , '''<mask>''' ) self.assertEqual(vocab_keys[-1] , '''<ctc_blank>''' ) self.assertEqual(len(_a ) , 81 ) def __UpperCAmelCase ( self ): self.assertEqual(self.get_tokenizer().vocab_size , 79 ) def __UpperCAmelCase ( self ): __a = self.get_tokenizers(do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) # We usually have added tokens from the start in tests because our vocab fixtures are # smaller than the original vocabs - let's not assert this # self.assertEqual(vocab_size, all_size) __a = ['''aaaaa bbbbbb''', '''cccccccccdddddddd'''] __a = tokenizer.add_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size + len(_a ) ) __a = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 4 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) __a = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''} __a = tokenizer.add_special_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size_a + len(_a ) ) __a = tokenizer.encode( '''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 6 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[0] , tokens[1] ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokens[-4] ) self.assertEqual(tokens[0] , tokenizer.eos_token_id ) self.assertEqual(tokens[-3] , tokenizer.pad_token_id ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tokenizer.tokenize('''This is a test''' ) # fmt: off self.assertListEqual(_a , [SPIECE_UNDERLINE, '''T''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''a''', SPIECE_UNDERLINE, '''t''', '''e''', '''s''', '''t'''] ) # fmt: on self.assertListEqual( tokenizer.convert_tokens_to_ids(_a ) , [4, 32, 11, 10, 12, 4, 10, 12, 4, 7, 4, 6, 5, 12, 6] , ) __a = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''92000''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) __a = tokenizer.convert_tokens_to_ids(_a ) # fmt: off self.assertListEqual(_a , [4, 30, 4, 20, 7, 12, 4, 25, 8, 13, 9, 4, 10, 9, 4, 3, 23, 4, 7, 9, 14, 4, 6, 11, 10, 12, 4, 10, 12, 4, 19, 7, 15, 12, 73, 26] ) # fmt: on __a = tokenizer.convert_ids_to_tokens(_a ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''<unk>''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) @slow def __UpperCAmelCase ( self ): # Use custom sequence because this tokenizer does not handle numbers. __a = [ '''Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides ''' '''general-purpose architectures (BERT, GPT, RoBERTa, XLM, DistilBert, XLNet...) for Natural ''' '''Language Understanding (NLU) and Natural Language Generation (NLG) with over thirty-two pretrained ''' '''models in one hundred plus languages and deep interoperability between Jax, PyTorch and TensorFlow.''', '''BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly ''' '''conditioning on both left and right context in all layers.''', '''The quick brown fox jumps over the lazy dog.''', ] # fmt: off __a = { '''input_ids''': [ [4, 32, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 64, 19, 8, 13, 18, 5, 13, 15, 22, 4, 28, 9, 8, 20, 9, 4, 7, 12, 4, 24, 22, 6, 8, 13, 17, 11, 39, 6, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 7, 9, 14, 4, 24, 22, 6, 8, 13, 17, 11, 39, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 39, 25, 5, 13, 6, 63, 4, 24, 13, 8, 27, 10, 14, 5, 12, 4, 21, 5, 9, 5, 13, 7, 15, 39, 24, 16, 13, 24, 8, 12, 5, 4, 7, 13, 17, 11, 10, 6, 5, 17, 6, 16, 13, 5, 12, 4, 64, 40, 47, 54, 32, 23, 4, 53, 49, 32, 23, 4, 54, 8, 40, 47, 54, 32, 7, 23, 4, 69, 52, 43, 23, 4, 51, 10, 12, 6, 10, 15, 40, 5, 13, 6, 23, 4, 69, 52, 48, 5, 6, 26, 26, 26, 63, 4, 19, 8, 13, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 61, 9, 14, 5, 13, 12, 6, 7, 9, 14, 10, 9, 21, 4, 64, 48, 52, 61, 63, 4, 7, 9, 14, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 53, 5, 9, 5, 13, 7, 6, 10, 8, 9, 4, 64, 48, 52, 53, 63, 4, 20, 10, 6, 11, 4, 8, 27, 5, 13, 4, 6, 11, 10, 13, 6, 22, 39, 6, 20, 8, 4, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 4, 18, 8, 14, 5, 15, 12, 4, 10, 9, 4, 8, 9, 5, 4, 11, 16, 9, 14, 13, 5, 14, 4, 24, 15, 16, 12, 4, 15, 7, 9, 21, 16, 7, 21, 5, 12, 4, 7, 9, 14, 4, 14, 5, 5, 24, 4, 10, 9, 6, 5, 13, 8, 24, 5, 13, 7, 25, 10, 15, 10, 6, 22, 4, 25, 5, 6, 20, 5, 5, 9, 4, 58, 7, 37, 23, 4, 49, 22, 32, 8, 13, 17, 11, 4, 7, 9, 14, 4, 32, 5, 9, 12, 8, 13, 55, 15, 8, 20, 26, 2], [4, 40, 47, 54, 32, 4, 10, 12, 4, 14, 5, 12, 10, 21, 9, 5, 14, 4, 6, 8, 4, 24, 13, 5, 39, 6, 13, 7, 10, 9, 4, 14, 5, 5, 24, 4, 25, 10, 14, 10, 13, 5, 17, 6, 10, 8, 9, 7, 15, 4, 13, 5, 24, 13, 5, 12, 5, 9, 6, 7, 6, 10, 8, 9, 12, 4, 19, 13, 8, 18, 4, 16, 9, 15, 7, 25, 5, 15, 5, 14, 4, 6, 5, 37, 6, 4, 25, 22, 4, 46, 8, 10, 9, 6, 15, 22, 4, 17, 8, 9, 14, 10, 6, 10, 8, 9, 10, 9, 21, 4, 8, 9, 4, 25, 8, 6, 11, 4, 15, 5, 19, 6, 4, 7, 9, 14, 4, 13, 10, 21, 11, 6, 4, 17, 8, 9, 6, 5, 37, 6, 4, 10, 9, 4, 7, 15, 15, 4, 15, 7, 22, 5, 13, 12, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [4, 32, 11, 5, 4, 45, 16, 10, 17, 28, 4, 25, 13, 8, 20, 9, 4, 19, 8, 37, 4, 46, 16, 18, 24, 12, 4, 8, 27, 5, 13, 4, 6, 11, 5, 4, 15, 7, 57, 22, 4, 14, 8, 21, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], ], '''attention_mask''': [ [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], ] } # fmt: on self.tokenizer_integration_test_util( expected_encoding=_a , model_name='''microsoft/speecht5_asr''' , revision='''c5ef64c71905caeccde0e4462ef3f9077224c524''' , sequences=_a , )
65
0
"""simple docstring""" import requests def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : str ) -> None: __a = {'''Content-Type''': '''application/json'''} __a = requests.post(lowerCAmelCase__ , json={'''text''': message_body} , headers=lowerCAmelCase__ ) if response.status_code != 200: __a = ( '''Request to slack returned an error ''' f'''{response.status_code}, the response is:\n{response.text}''' ) raise ValueError(lowerCAmelCase__ ) if __name__ == "__main__": # Set the slack url to the one provided by Slack when you create the webhook at # https://my.slack.com/services/new/incoming-webhook/ send_slack_message("<YOUR MESSAGE BODY>", "<SLACK CHANNEL URL>")
705
"""simple docstring""" from typing import List, Optional, Union import numpy as np from ....audio_utils import mel_filter_bank, optimal_fft_length, spectrogram, window_function from ....feature_extraction_sequence_utils import SequenceFeatureExtractor from ....feature_extraction_utils import BatchFeature from ....file_utils import PaddingStrategy, TensorType from ....utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Any = ['input_features', 'attention_mask'] def __init__( self , _a=80 , _a=16_000 , _a=0.0 , _a=10 , _a=25 , _a="hamming_window" , _a=3_2768.0 , _a=0.97 , _a=1.0 , _a=True , _a=True , _a=False , **_a , ): super().__init__(feature_size=_a , sampling_rate=_a , padding_value=_a , **_a ) __a = feature_size __a = sampling_rate __a = padding_value __a = hop_length __a = win_length __a = frame_signal_scale __a = preemphasis_coeff __a = mel_floor __a = normalize_means __a = normalize_vars __a = win_function __a = return_attention_mask __a = win_length * sampling_rate // 1_000 __a = hop_length * sampling_rate // 1_000 __a = optimal_fft_length(self.sample_size ) __a = (self.n_fft // 2) + 1 def __UpperCAmelCase ( self , _a ): if self.win_function == "hamming_window": __a = window_function(window_length=self.sample_size , name=self.win_function , periodic=_a ) else: __a = window_function(window_length=self.sample_size , name=self.win_function ) __a = mel_filter_bank( num_frequency_bins=self.n_freqs , num_mel_filters=self.feature_size , min_frequency=0.0 , max_frequency=self.sampling_rate / 2.0 , sampling_rate=self.sampling_rate , ) __a = spectrogram( one_waveform * self.frame_signal_scale , window=_a , frame_length=self.sample_size , hop_length=self.sample_stride , fft_length=self.n_fft , center=_a , preemphasis=self.preemphasis_coeff , mel_filters=_a , mel_floor=self.mel_floor , log_mel='''log''' , ) return msfc_features.T def __UpperCAmelCase ( self , _a , _a , _a ): # make sure we normalize float32 arrays if self.normalize_means: __a = x[:input_length].mean(axis=0 ) __a = np.subtract(_a , _a ) if self.normalize_vars: __a = x[:input_length].std(axis=0 ) __a = np.divide(_a , _a ) if input_length < x.shape[0]: __a = padding_value # make sure array is in float32 __a = x.astype(np.floataa ) return x def __UpperCAmelCase ( self , _a , _a = None ): __a = attention_mask.sum(-1 ) if attention_mask is not None else [x.shape[0] for x in input_features] return [self._normalize_one(_a , _a , self.padding_value ) for x, n in zip(_a , _a )] def __call__( self , _a , _a = False , _a = None , _a = False , _a = None , _a = None , _a = None , _a = None , **_a , ): if sampling_rate is not None: if sampling_rate != self.sampling_rate: raise ValueError( f'''The model corresponding to this feature extractor: {self} was trained using a sampling rate of''' f''' {self.sampling_rate}. Please make sure that the provided `raw_speech` input was sampled with''' f''' {self.sampling_rate} and not {sampling_rate}.''' ) else: logger.warning( '''It is strongly recommended to pass the ``sampling_rate`` argument to this function. ''' '''Failing to do so can result in silent errors that might be hard to debug.''' ) __a = isinstance(_a , np.ndarray ) and len(raw_speech.shape ) > 1 if is_batched_numpy and len(raw_speech.shape ) > 2: raise ValueError(f'''Only mono-channel audio is supported for input to {self}''' ) __a = is_batched_numpy or ( isinstance(_a , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) )) ) if is_batched: __a = [np.asarray(_a , dtype=np.floataa ) for speech in raw_speech] elif not is_batched and not isinstance(_a , np.ndarray ): __a = np.asarray(_a , dtype=np.floataa ) elif isinstance(_a , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ): __a = raw_speech.astype(np.floataa ) # always return batch if not is_batched: __a = [raw_speech] # extract fbank features __a = [self._extract_mfsc_features(_a ) for one_waveform in raw_speech] # convert into correct format for padding __a = BatchFeature({'''input_features''': features} ) __a = self.pad( _a , padding=_a , max_length=_a , truncation=_a , pad_to_multiple_of=_a , return_attention_mask=_a , **_a , ) # make sure list is in array format __a = padded_inputs.get('''input_features''' ) if isinstance(input_features[0] , _a ): __a = [np.asarray(_a , dtype=np.floataa ) for feature in input_features] __a = padded_inputs.get('''attention_mask''' ) if attention_mask is not None: __a = [np.asarray(_a , dtype=np.intaa ) for array in attention_mask] if self.normalize_means or self.normalize_vars: __a = ( np.array(_a , dtype=np.intaa ) if self._get_padding_strategies(_a , max_length=_a ) is not PaddingStrategy.DO_NOT_PAD and padding else None ) __a = self.normalize( padded_inputs['''input_features'''] , attention_mask=_a ) if return_tensors is not None: __a = padded_inputs.convert_to_tensors(_a ) return padded_inputs
65
0
import tempfile import unittest from pathlib import Path from shutil import copyfile from transformers import BatchEncoding, MarianTokenizer from transformers.testing_utils import get_tests_dir, require_sentencepiece, slow from transformers.utils import is_sentencepiece_available, is_tf_available, is_torch_available if is_sentencepiece_available(): from transformers.models.marian.tokenization_marian import VOCAB_FILES_NAMES, save_json from ...test_tokenization_common import TokenizerTesterMixin lowercase_ = get_tests_dir("fixtures/test_sentencepiece.model") lowercase_ = {"target_lang": "fi", "source_lang": "en"} lowercase_ = ">>zh<<" lowercase_ = "Helsinki-NLP/" if is_torch_available(): lowercase_ = "pt" elif is_tf_available(): lowercase_ = "tf" else: lowercase_ = "jax" @require_sentencepiece class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : str = MarianTokenizer __UpperCAmelCase : List[str] = False __UpperCAmelCase : List[str] = True def __UpperCAmelCase ( self ): super().setUp() __a = ['''</s>''', '''<unk>''', '''▁This''', '''▁is''', '''▁a''', '''▁t''', '''est''', '''\u0120''', '''<pad>'''] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = Path(self.tmpdirname ) save_json(_a , save_dir / VOCAB_FILES_NAMES['''vocab'''] ) save_json(_a , save_dir / VOCAB_FILES_NAMES['''tokenizer_config_file'''] ) if not (save_dir / VOCAB_FILES_NAMES["source_spm"]).exists(): copyfile(_a , save_dir / VOCAB_FILES_NAMES['''source_spm'''] ) copyfile(_a , save_dir / VOCAB_FILES_NAMES['''target_spm'''] ) __a = MarianTokenizer.from_pretrained(self.tmpdirname ) tokenizer.save_pretrained(self.tmpdirname ) def __UpperCAmelCase ( self , **_a ): return MarianTokenizer.from_pretrained(self.tmpdirname , **_a ) def __UpperCAmelCase ( self , _a ): return ( "This is a test", "This is a test", ) def __UpperCAmelCase ( self ): __a = '''</s>''' __a = 0 self.assertEqual(self.get_tokenizer()._convert_token_to_id(_a ) , _a ) self.assertEqual(self.get_tokenizer()._convert_id_to_token(_a ) , _a ) def __UpperCAmelCase ( self ): __a = list(self.get_tokenizer().get_vocab().keys() ) self.assertEqual(vocab_keys[0] , '''</s>''' ) self.assertEqual(vocab_keys[1] , '''<unk>''' ) self.assertEqual(vocab_keys[-1] , '''<pad>''' ) self.assertEqual(len(_a ) , 9 ) def __UpperCAmelCase ( self ): self.assertEqual(self.get_tokenizer().vocab_size , 9 ) def __UpperCAmelCase ( self ): __a = MarianTokenizer.from_pretrained(f'''{ORG_NAME}opus-mt-en-de''' ) __a = en_de_tokenizer(['''I am a small frog'''] , return_tensors=_a ) self.assertIsInstance(_a , _a ) __a = [38, 121, 14, 697, 38_848, 0] self.assertListEqual(_a , batch.input_ids[0] ) __a = tempfile.mkdtemp() en_de_tokenizer.save_pretrained(_a ) __a = [x.name for x in Path(_a ).glob('''*''' )] self.assertIn('''source.spm''' , _a ) MarianTokenizer.from_pretrained(_a ) def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tok( ['''I am a small frog''' * 1_000, '''I am a small frog'''] , padding=_a , truncation=_a , return_tensors=_a ) self.assertIsInstance(_a , _a ) self.assertEqual(batch.input_ids.shape , (2, 512) ) def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tok(['''I am a tiny frog''', '''I am a small frog'''] , padding=_a , return_tensors=_a ) self.assertIsInstance(_a , _a ) self.assertEqual(batch_smaller.input_ids.shape , (2, 10) ) @slow def __UpperCAmelCase ( self ): # fmt: off __a = {'''input_ids''': [[43_495, 462, 20, 42_164, 1_369, 52, 464, 132, 1_703, 492, 13, 7_491, 38_999, 6, 8, 464, 132, 1_703, 492, 13, 4_669, 37_867, 13, 7_525, 27, 1_593, 988, 13, 33_972, 7_029, 6, 20, 8_251, 383, 2, 270, 5_866, 3_788, 2, 2_353, 8_251, 12_338, 2, 13_958, 387, 2, 3_629, 6_953, 188, 2_900, 2, 13_958, 8_011, 11_501, 23, 8_460, 4_073, 34_009, 20, 435, 11_439, 27, 8, 8_460, 4_073, 6_004, 20, 9_988, 375, 27, 33, 266, 1_945, 1_076, 1_350, 37_867, 3_288, 5, 577, 1_076, 4_374, 8, 5_082, 5, 26_453, 257, 556, 403, 2, 242, 132, 383, 316, 492, 8, 10_767, 6, 316, 304, 4_239, 3, 0], [148, 15_722, 19, 1_839, 12, 1_350, 13, 22_327, 5_082, 5_418, 47_567, 35_938, 59, 318, 19_552, 108, 2_183, 54, 14_976, 4_835, 32, 547, 1_114, 8, 315, 2_417, 5, 92, 19_088, 3, 0, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100], [36, 6_395, 12_570, 39_147, 11_597, 6, 266, 4, 45_405, 7_296, 3, 0, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100, 58_100]], '''attention_mask''': [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]} # noqa: E501 # fmt: on self.tokenizer_integration_test_util( expected_encoding=_a , model_name='''Helsinki-NLP/opus-mt-en-de''' , revision='''1a8c2263da11e68e50938f97e10cd57820bd504c''' , decode_kwargs={'''use_source_tokenizer''': True} , ) def __UpperCAmelCase ( self ): __a = MarianTokenizer.from_pretrained('''hf-internal-testing/test-marian-two-vocabs''' ) __a = '''Tämä on testi''' __a = '''This is a test''' __a = [76, 7, 2_047, 2] __a = [69, 12, 11, 940, 2] __a = tokenizer(_a ).input_ids self.assertListEqual(_a , _a ) __a = tokenizer(text_target=_a ).input_ids self.assertListEqual(_a , _a ) __a = tokenizer.decode(_a , skip_special_tokens=_a ) self.assertEqual(_a , _a )
706
"""simple docstring""" import json import os import subprocess import unittest from ast import literal_eval import pytest from parameterized import parameterized, 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_model_parallelism.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, { 'framework': 'pytorch', 'script': 'run_glue.py', 'model_name_or_path': 'roberta-large', 'instance_type': 'ml.p3dn.24xlarge', 'results': {'train_runtime': 1_6_0_0, 'eval_accuracy': 0.3, 'eval_loss': 1.2}, }, ] ) class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): 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=_a , ) assert hasattr(self , '''env''' ) def __UpperCAmelCase ( self , _a ): # configuration for running training on smdistributed Model Parallel __a = { '''enabled''': True, '''processes_per_host''': 8, } __a = { '''enabled''': True, '''parameters''': { '''microbatches''': 4, '''placement_strategy''': '''spread''', '''pipeline''': '''interleaved''', '''optimize''': '''speed''', '''partitions''': 4, '''ddp''': True, }, } __a = {'''smdistributed''': {'''modelparallel''': smp_options}, '''mpi''': mpi_options} __a = '''trainer''' if self.script == '''run_glue.py''' else '''smtrainer''' # 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}-{instance_count}-smp-{name_extension}''' , instance_count=_a , instance_type=self.instance_type , debugger_hook_config=_a , hyperparameters={ **self.env.hyperparameters, '''model_name_or_path''': self.model_name_or_path, '''max_steps''': 500, } , metric_definitions=self.env.metric_definitions , distribution=_a , py_version='''py36''' , ) def __UpperCAmelCase ( self , _a ): TrainingJobAnalytics(_a ).export_csv(f'''{self.env.test_path}/{job_name}_metrics.csv''' ) @parameterized.expand([(1,)] ) def __UpperCAmelCase ( self , _a ): # create estimator __a = self.create_estimator(_a ) # run training estimator.fit() # result dataframe __a = TrainingJobAnalytics(estimator.latest_training_job.name ).dataframe() # extract kpis __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_accuracy''']['''value'''] ) __a = list(result_metrics_df[result_metrics_df.metric_name == '''eval_loss''']['''value'''] ) # get train time from SageMaker job, this includes starting, preprocessing, stopping __a = ( Session().describe_training_job(estimator.latest_training_job.name ).get('''TrainingTimeInSeconds''' , 999_999 ) ) # 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} , _a )
65
0
"""simple docstring""" import os from typing import Optional import fsspec from fsspec.archive import AbstractArchiveFileSystem from fsspec.utils import DEFAULT_BLOCK_SIZE class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Optional[int] = '' __UpperCAmelCase : str = ( None # protocol passed in prefix to the url. ex: "gzip", for gzip://file.txt::http://foo.bar/file.txt.gz ) __UpperCAmelCase : str = None # compression type in fsspec. ex: "gzip" __UpperCAmelCase : str = None # extension of the filename to strip. ex: "".gz" to get file.txt from file.txt.gz def __init__( self , _a = "" , _a = None , _a = None , **_a ): super().__init__(self , **_a ) # always open as "rb" since fsspec can then use the TextIOWrapper to make it work for "r" mode __a = fsspec.open( _a , mode='''rb''' , protocol=_a , compression=self.compression , client_kwargs={ '''requote_redirect_url''': False, # see https://github.com/huggingface/datasets/pull/5459 '''trust_env''': True, # Enable reading proxy env variables. **(target_options or {}).pop('''client_kwargs''' , {} ), # To avoid issues if it was already passed. } , **(target_options or {}) , ) __a = os.path.basename(self.file.path.split('''::''' )[0] ) __a = ( self.compressed_name[: self.compressed_name.rindex('''.''' )] if '''.''' in self.compressed_name else self.compressed_name ) __a = None @classmethod def __UpperCAmelCase ( cls , _a ): # compressed file paths are always relative to the archive root return super()._strip_protocol(_a ).lstrip('''/''' ) def __UpperCAmelCase ( self ): if self.dir_cache is None: __a = {**self.file.fs.info(self.file.path ), '''name''': self.uncompressed_name} __a = {f['''name''']: f} def __UpperCAmelCase ( self , _a ): return self.file.open().read() def __UpperCAmelCase ( self , _a , _a = "rb" , _a=None , _a=True , _a=None , **_a , ): __a = self._strip_protocol(_a ) if mode != "rb": raise ValueError(f'''Tried to read with mode {mode} on file {self.file.path} opened with mode \'rb\'''' ) return self.file.open() class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : int = 'bz2' __UpperCAmelCase : Optional[Any] = 'bz2' __UpperCAmelCase : List[Any] = '.bz2' class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Tuple = 'gzip' __UpperCAmelCase : int = 'gzip' __UpperCAmelCase : str = '.gz' class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Union[str, Any] = 'lz4' __UpperCAmelCase : str = 'lz4' __UpperCAmelCase : List[str] = '.lz4' class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = 'xz' __UpperCAmelCase : Optional[int] = 'xz' __UpperCAmelCase : int = '.xz' class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = 'zstd' __UpperCAmelCase : List[str] = 'zstd' __UpperCAmelCase : Any = '.zst' def __init__( self , _a , _a = "rb" , _a = None , _a = None , _a = DEFAULT_BLOCK_SIZE , **_a , ): super().__init__( fo=_a , mode=_a , target_protocol=_a , target_options=_a , block_size=_a , **_a , ) # We need to wrap the zstd decompressor to avoid this error in fsspec==2021.7.0 and zstandard==0.15.2: # # File "/Users/user/.virtualenvs/hf-datasets/lib/python3.7/site-packages/fsspec/core.py", line 145, in open # out.close = close # AttributeError: 'zstd.ZstdDecompressionReader' object attribute 'close' is read-only # # see https://github.com/intake/filesystem_spec/issues/725 __a = self.file.__enter__ class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = file_ def __enter__( self ): self._file.__enter__() return self def __exit__( self , *_a , **_a ): self._file.__exit__(*_a , **_a ) def __iter__( self ): return iter(self._file ) def __UpperCAmelCase ( self ): return next(self._file ) def __getattr__( self , _a ): return getattr(self._file , _a ) def fixed_enter(*_a , **_a ): return WrappedFile(_enter(*_a , **_a ) ) __a = fixed_enter
707
"""simple docstring""" import warnings from ..trainer import Trainer from ..utils import logging lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a=None , **_a ): warnings.warn( '''`SageMakerTrainer` is deprecated and will be removed in v5 of Transformers. You can use `Trainer` ''' '''instead.''' , _a , ) super().__init__(args=_a , **_a )
65
0
"""simple docstring""" from __future__ import annotations from collections.abc import Callable def lowercase ( lowerCAmelCase__ : Callable[[int | float], int | float] , lowerCAmelCase__ : int | float , lowerCAmelCase__ : int | float , lowerCAmelCase__ : int = 100 , ) -> float: '''simple docstring''' __a = x_start __a = fnc(lowerCAmelCase__ ) __a = 0.0 for _ in range(lowerCAmelCase__ ): # Approximates small segments of curve as linear and solve # for trapezoidal area __a = (x_end - x_start) / steps + xa __a = fnc(lowerCAmelCase__ ) area += abs(fxa + fxa ) * (xa - xa) / 2 # Increment step __a = xa __a = fxa return area if __name__ == "__main__": def lowercase ( lowerCAmelCase__ : int ) -> Optional[int]: '''simple docstring''' return x**3 + x**2 print("f(x) = x^3 + x^2") print("The area between the curve, x = -5, x = 5 and the x axis is:") lowercase_ = 1_0 while i <= 1_0_0_0_0_0: print(F'''with {i} steps: {trapezoidal_area(f, -5, 5, i)}''') i *= 1_0
708
"""simple docstring""" from typing import TYPE_CHECKING from ...file_utils import _LazyModule, is_tokenizers_available, is_torch_available from ...utils import OptionalDependencyNotAvailable lowercase_ = {"configuration_gpt_neox": ["GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoXConfig"]} try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["GPTNeoXTokenizerFast"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST", "GPTNeoXForCausalLM", "GPTNeoXForQuestionAnswering", "GPTNeoXForSequenceClassification", "GPTNeoXForTokenClassification", "GPTNeoXLayer", "GPTNeoXModel", "GPTNeoXPreTrainedModel", ] if TYPE_CHECKING: from .configuration_gpt_neox import GPT_NEOX_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoXConfig try: if not is_tokenizers_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .tokenization_gpt_neox_fast import GPTNeoXTokenizerFast try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_gpt_neox import ( GPT_NEOX_PRETRAINED_MODEL_ARCHIVE_LIST, GPTNeoXForCausalLM, GPTNeoXForQuestionAnswering, GPTNeoXForSequenceClassification, GPTNeoXForTokenClassification, GPTNeoXLayer, GPTNeoXModel, GPTNeoXPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
65
0
"""simple docstring""" import inspect from typing import Optional, Union import numpy as np import PIL import torch from torch.nn import functional as F from torchvision import transforms from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DiffusionPipeline, DPMSolverMultistepScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, randn_tensor, ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> Any: if isinstance(lowerCAmelCase__ , torch.Tensor ): return image elif isinstance(lowerCAmelCase__ , PIL.Image.Image ): __a = [image] if isinstance(image[0] , PIL.Image.Image ): __a = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['''lanczos'''] ) )[None, :] for i in image] __a = np.concatenate(lowerCAmelCase__ , axis=0 ) __a = np.array(lowerCAmelCase__ ).astype(np.floataa ) / 255.0 __a = image.transpose(0 , 3 , 1 , 2 ) __a = 2.0 * image - 1.0 __a = torch.from_numpy(lowerCAmelCase__ ) elif isinstance(image[0] , torch.Tensor ): __a = torch.cat(lowerCAmelCase__ , dim=0 ) return image def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=0.99_95 ) -> int: if not isinstance(lowerCAmelCase__ , np.ndarray ): __a = True __a = va.device __a = va.cpu().numpy() __a = va.cpu().numpy() __a = np.sum(va * va / (np.linalg.norm(lowerCAmelCase__ ) * np.linalg.norm(lowerCAmelCase__ )) ) if np.abs(lowerCAmelCase__ ) > DOT_THRESHOLD: __a = (1 - t) * va + t * va else: __a = np.arccos(lowerCAmelCase__ ) __a = np.sin(lowerCAmelCase__ ) __a = theta_a * t __a = np.sin(lowerCAmelCase__ ) __a = np.sin(theta_a - theta_t ) / sin_theta_a __a = sin_theta_t / sin_theta_a __a = sa * va + sa * va if inputs_are_torch: __a = torch.from_numpy(lowerCAmelCase__ ).to(lowerCAmelCase__ ) return va def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> int: __a = F.normalize(lowerCAmelCase__ , dim=-1 ) __a = F.normalize(lowerCAmelCase__ , dim=-1 ) return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> List[str]: for param in model.parameters(): __a = value class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a , _a , _a , _a=None , _a=None , _a=None , ): super().__init__() self.register_modules( vae=_a , text_encoder=_a , clip_model=_a , tokenizer=_a , unet=_a , scheduler=_a , feature_extractor=_a , coca_model=_a , coca_tokenizer=_a , coca_transform=_a , ) __a = ( feature_extractor.size if isinstance(feature_extractor.size , _a ) else feature_extractor.size['''shortest_edge'''] ) __a = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std ) set_requires_grad(self.text_encoder , _a ) set_requires_grad(self.clip_model , _a ) def __UpperCAmelCase ( self , _a = "auto" ): if slice_size == "auto": # half the attention head size is usually a good trade-off between # speed and memory __a = self.unet.config.attention_head_dim // 2 self.unet.set_attention_slice(_a ) def __UpperCAmelCase ( self ): self.enable_attention_slicing(_a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self , _a , _a , _a ): # get the original timestep using init_timestep __a = min(int(num_inference_steps * strength ) , _a ) __a = max(num_inference_steps - init_timestep , 0 ) __a = self.scheduler.timesteps[t_start:] return timesteps, num_inference_steps - t_start def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a=None ): if not isinstance(_a , torch.Tensor ): raise ValueError(f'''`image` has to be of type `torch.Tensor` but is {type(_a )}''' ) __a = image.to(device=_a , dtype=_a ) if isinstance(_a , _a ): __a = [ self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_a ) ] __a = torch.cat(_a , dim=0 ) else: __a = self.vae.encode(_a ).latent_dist.sample(_a ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 0.1_8215 * init_latents __a = init_latents.repeat_interleave(_a , dim=0 ) __a = randn_tensor(init_latents.shape , generator=_a , device=_a , dtype=_a ) # get latents __a = self.scheduler.add_noise(_a , _a , _a ) __a = init_latents return latents def __UpperCAmelCase ( self , _a ): __a = self.coca_transform(_a ).unsqueeze(0 ) with torch.no_grad(), torch.cuda.amp.autocast(): __a = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) ) __a = self.coca_tokenizer.decode(generated[0].cpu().numpy() ) return generated.split('''<end_of_text>''' )[0].replace('''<start_of_text>''' , '''''' ).rstrip(''' .,''' ) def __UpperCAmelCase ( self , _a , _a ): __a = self.feature_extractor.preprocess(_a ) __a = torch.from_numpy(clip_image_input['''pixel_values'''][0] ).unsqueeze(0 ).to(self.device ).half() __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = image_embeddings_clip.repeat_interleave(_a , dim=0 ) return image_embeddings_clip @torch.enable_grad() def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , ): __a = latents.detach().requires_grad_() __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ): __a = self.scheduler.alphas_cumprod[timestep] __a = 1 - alpha_prod_t # compute predicted original sample from predicted noise also called # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf __a = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5 __a = torch.sqrt(_a ) __a = pred_original_sample * (fac) + latents * (1 - fac) elif isinstance(self.scheduler , _a ): __a = self.scheduler.sigmas[index] __a = latents - sigma * noise_pred else: raise ValueError(f'''scheduler type {type(self.scheduler )} not supported''' ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * sample __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = transforms.Resize(self.feature_extractor_size )(_a ) __a = self.normalize(_a ).to(latents.dtype ) __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = spherical_dist_loss(_a , _a ).mean() * clip_guidance_scale __a = -torch.autograd.grad(_a , _a )[0] if isinstance(self.scheduler , _a ): __a = latents.detach() + grads * (sigma**2) __a = noise_pred_original else: __a = noise_pred_original - torch.sqrt(_a ) * grads return noise_pred, latents @torch.no_grad() def __call__( self , _a , _a , _a = None , _a = None , _a = 512 , _a = 512 , _a = 0.6 , _a = 50 , _a = 7.5 , _a = 1 , _a = 0.0 , _a = 100 , _a = None , _a = "pil" , _a = True , _a = 0.8 , _a = 0.1 , _a = 0.1 , ): if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError(f'''You have passed {batch_size} batch_size, but only {len(_a )} generators.''' ) if height % 8 != 0 or width % 8 != 0: raise ValueError(f'''`height` and `width` have to be divisible by 8 but are {height} and {width}.''' ) if isinstance(_a , torch.Generator ) and batch_size > 1: __a = [generator] + [None] * (batch_size - 1) __a = [ ('''model''', self.coca_model is None), ('''tokenizer''', self.coca_tokenizer is None), ('''transform''', self.coca_transform is None), ] __a = [x[0] for x in coca_is_none if x[1]] __a = ''', '''.join(_a ) # generate prompts with coca model if prompt is None if content_prompt is None: if len(_a ): raise ValueError( f'''Content prompt is None and CoCa [{coca_is_none_str}] is None.''' f'''Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) if style_prompt is None: if len(_a ): raise ValueError( f'''Style prompt is None and CoCa [{coca_is_none_str}] is None.''' f''' Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) # get prompt text embeddings for content and style __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0] __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0] __a = slerp(_a , _a , _a ) # duplicate text embeddings for each generation per prompt __a = text_embeddings.repeat_interleave(_a , dim=0 ) # set timesteps __a = '''offset''' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() ) __a = {} if accepts_offset: __a = 1 self.scheduler.set_timesteps(_a , **_a ) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand self.scheduler.timesteps.to(self.device ) __a , __a = self.get_timesteps(_a , _a , self.device ) __a = timesteps[:1].repeat(_a ) # Preprocess image __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = slerp(_a , _a , _a ) if clip_guidance_scale > 0: __a = self.get_clip_image_embeddings(_a , _a ) __a = self.get_clip_image_embeddings(_a , _a ) __a = slerp( _a , _a , _a ) # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` # corresponds to doing no classifier free guidance. __a = guidance_scale > 1.0 # get unconditional embeddings for classifier free guidance if do_classifier_free_guidance: __a = content_text_input.input_ids.shape[-1] __a = self.tokenizer([''''''] , padding='''max_length''' , max_length=_a , return_tensors='''pt''' ) __a = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0] # duplicate unconditional embeddings for each generation per prompt __a = uncond_embeddings.repeat_interleave(_a , dim=0 ) # For classifier free guidance, we need to do two forward passes. # Here we concatenate the unconditional and text embeddings into a single batch # to avoid doing two forward passes __a = torch.cat([uncond_embeddings, text_embeddings] ) # get the initial random noise unless the user supplied it # Unlike in other pipelines, latents need to be generated in the target device # for 1-to-1 results reproducibility with the CompVis implementation. # However this currently doesn't work in `mps`. __a = (batch_size, self.unet.config.in_channels, height // 8, width // 8) __a = text_embeddings.dtype if latents is None: if self.device.type == "mps": # randn does not work reproducibly on mps __a = torch.randn(_a , generator=_a , device='''cpu''' , dtype=_a ).to( self.device ) else: __a = torch.randn(_a , generator=_a , device=self.device , dtype=_a ) else: if latents.shape != latents_shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {latents_shape}''' ) __a = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler __a = latents * self.scheduler.init_noise_sigma # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 # and should be between [0, 1] __a = '''eta''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) __a = {} if accepts_eta: __a = eta # check if the scheduler accepts generator __a = '''generator''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) if accepts_generator: __a = generator with self.progress_bar(total=_a ): for i, t in enumerate(_a ): # expand the latents if we are doing classifier free guidance __a = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample # perform classifier free guidance if do_classifier_free_guidance: __a , __a = noise_pred.chunk(2 ) __a = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) # perform clip guidance if clip_guidance_scale > 0: __a = ( text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings ) __a , __a = self.cond_fn( _a , _a , _a , _a , _a , _a , _a , ) # compute the previous noisy sample x_t -> x_t-1 __a = self.scheduler.step(_a , _a , _a , **_a ).prev_sample # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * latents __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image, None) return StableDiffusionPipelineOutput(images=_a , nsfw_content_detected=_a )
709
"""simple docstring""" import unittest from transformers import MraConfig, is_torch_available from transformers.testing_utils import require_torch, slow, torch_device from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask if is_torch_available(): import torch from transformers import ( MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, MraModel, ) from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=2 , _a=8 , _a=True , _a=True , _a=True , _a=True , _a=99 , _a=16 , _a=5 , _a=2 , _a=36 , _a="gelu" , _a=0.0 , _a=0.0 , _a=512 , _a=16 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ): __a = parent __a = batch_size __a = seq_length __a = is_training __a = use_input_mask __a = use_token_type_ids __a = use_labels __a = vocab_size __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = type_sequence_label_size __a = initializer_range __a = num_labels __a = num_choices __a = scope def __UpperCAmelCase ( self ): __a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) __a = None if self.use_input_mask: __a = random_attention_mask([self.batch_size, self.seq_length] ) __a = None if self.use_token_type_ids: __a = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size ) __a = None __a = None __a = None if self.use_labels: __a = ids_tensor([self.batch_size] , self.type_sequence_label_size ) __a = ids_tensor([self.batch_size, self.seq_length] , self.num_labels ) __a = ids_tensor([self.batch_size] , self.num_choices ) __a = self.get_config() return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels def __UpperCAmelCase ( self ): return MraConfig( 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=_a , initializer_range=self.initializer_range , ) def __UpperCAmelCase ( self ): __a = self.get_config() __a = 300 return config def __UpperCAmelCase ( self ): ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = self.prepare_config_and_inputs() __a = True __a = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] ) __a = 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraModel(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a ) __a = model(_a , token_type_ids=_a ) __a = model(_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , _a , _a , ): __a = True __a = MraModel(_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , encoder_attention_mask=_a , ) __a = model( _a , attention_mask=_a , token_type_ids=_a , encoder_hidden_states=_a , ) __a = model(_a , attention_mask=_a , token_type_ids=_a ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForMaskedLM(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = MraForQuestionAnswering(config=_a ) model.to(_a ) model.eval() __a = model( _a , attention_mask=_a , token_type_ids=_a , start_positions=_a , end_positions=_a , ) 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 __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForSequenceClassification(_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_labels __a = MraForTokenClassification(config=_a ) model.to(_a ) model.eval() __a = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a ): __a = self.num_choices __a = MraForMultipleChoice(config=_a ) model.to(_a ) model.eval() __a = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() __a = model( _a , attention_mask=_a , token_type_ids=_a , labels=_a , ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) ) def __UpperCAmelCase ( self ): __a = self.prepare_config_and_inputs() ( ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ( __a ) , ) = config_and_inputs __a = {'''input_ids''': input_ids, '''token_type_ids''': token_type_ids, '''attention_mask''': input_mask} return config, inputs_dict @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = ( ( MraModel, MraForMaskedLM, MraForMultipleChoice, MraForQuestionAnswering, MraForSequenceClassification, MraForTokenClassification, ) if is_torch_available() else () ) __UpperCAmelCase : str = False __UpperCAmelCase : Any = False __UpperCAmelCase : List[Any] = False __UpperCAmelCase : Optional[Any] = False __UpperCAmelCase : Dict = () def __UpperCAmelCase ( self ): __a = MraModelTester(self ) __a = ConfigTester(self , config_class=_a , hidden_size=37 ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() for type in ["absolute", "relative_key", "relative_key_query"]: __a = type self.model_tester.create_and_check_model(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_masked_lm(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_multiple_choice(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_question_answering(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_sequence_classification(*_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_token_classification(*_a ) @slow def __UpperCAmelCase ( self ): for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: __a = MraModel.from_pretrained(_a ) self.assertIsNotNone(_a ) @unittest.skip(reason='''MRA does not output attentions''' ) def __UpperCAmelCase ( self ): return @require_torch class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @slow def __UpperCAmelCase ( self ): __a = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = torch.Size((1, 256, 768) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[-0.0140, 0.0830, -0.0381], [0.1546, 0.1402, 0.0220], [0.1162, 0.0851, 0.0165]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' ) __a = torch.arange(256 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 256, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[9.2595, -3.6038, 11.8819], [9.3869, -3.2693, 11.0956], [11.8524, -3.4938, 13.1210]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) ) @slow def __UpperCAmelCase ( self ): __a = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' ) __a = torch.arange(4_096 ).unsqueeze(0 ) with torch.no_grad(): __a = model(_a )[0] __a = 50_265 __a = torch.Size((1, 4_096, vocab_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[[5.4789, -2.3564, 7.5064], [7.9067, -1.3369, 9.9668], [9.0712, -1.8106, 7.0380]]] ) self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1E-4 ) )
65
0
"""simple docstring""" import pytest from datasets import Dataset, DatasetDict, Features, NamedSplit, Value from datasets.io.text import TextDatasetReader from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.split == split if split else "train" @pytest.mark.parametrize('''path_type''' , [str, list] ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict: if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = text_path elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [text_path] __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for split in splits: __a = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str: __a = tmp_path / '''cache''' # CSV file loses col_1 string dtype information: default now is "int64" instead of "string" __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict: if split: __a = {split: text_path} else: __a = '''train''' __a = {'''train''': text_path, '''test''': text_path} __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() )
710
"""simple docstring""" from __future__ import annotations def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> list[str]: if partitions <= 0: raise ValueError('''partitions must be a positive number!''' ) if partitions > number_of_bytes: raise ValueError('''partitions can not > number_of_bytes!''' ) __a = number_of_bytes // partitions __a = [] for i in range(lowerCAmelCase__ ): __a = i * bytes_per_partition + 1 __a = ( number_of_bytes if i == partitions - 1 else (i + 1) * bytes_per_partition ) allocation_list.append(f'''{start_bytes}-{end_bytes}''' ) return allocation_list if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import os from typing import List, Optional, Union from ...image_processing_utils import BatchFeature from ...image_utils import ImageInput from ...processing_utils import ProcessorMixin from ...tokenization_utils_base import PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy from ...utils import TensorType from ..auto import AutoTokenizer class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Union[str, Any] = ['image_processor', 'tokenizer'] __UpperCAmelCase : int = 'BlipImageProcessor' __UpperCAmelCase : Dict = 'AutoTokenizer' def __init__( self , _a , _a , _a ): super().__init__(_a , _a ) # add QFormer tokenizer __a = qformer_tokenizer def __call__( self , _a = None , _a = None , _a = True , _a = False , _a = None , _a = None , _a = 0 , _a = None , _a = None , _a = False , _a = False , _a = False , _a = False , _a = False , _a = True , _a = None , **_a , ): if images is None and text is None: raise ValueError('''You have to specify at least images or text.''' ) __a = BatchFeature() if text is not None: __a = self.tokenizer( text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_token_type_ids=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , ) encoding.update(_a ) __a = self.qformer_tokenizer( text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_token_type_ids=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , ) __a = qformer_text_encoding.pop('''input_ids''' ) __a = qformer_text_encoding.pop('''attention_mask''' ) if images is not None: __a = self.image_processor(_a , return_tensors=_a ) encoding.update(_a ) return encoding def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.batch_decode(*_a , **_a ) def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.decode(*_a , **_a ) @property # Copied from transformers.models.blip.processing_blip.BlipProcessor.model_input_names def __UpperCAmelCase ( self ): __a = self.tokenizer.model_input_names __a = self.image_processor.model_input_names return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) ) def __UpperCAmelCase ( self , _a , **_a ): if os.path.isfile(_a ): raise ValueError(f'''Provided path ({save_directory}) should be a directory, not a file''' ) os.makedirs(_a , exist_ok=_a ) __a = os.path.join(_a , '''qformer_tokenizer''' ) self.qformer_tokenizer.save_pretrained(_a ) return super().save_pretrained(_a , **_a ) @classmethod def __UpperCAmelCase ( cls , _a , **_a ): __a = AutoTokenizer.from_pretrained(_a , subfolder='''qformer_tokenizer''' ) __a = cls._get_arguments_from_pretrained(_a , **_a ) args.append(_a ) return cls(*_a )
711
"""simple docstring""" def lowercase ( lowerCAmelCase__ : list , lowerCAmelCase__ : list , lowerCAmelCase__ : int ) -> list: __a = len(lowerCAmelCase__ ) __a = [[0] * n for i in range(lowerCAmelCase__ )] for i in range(lowerCAmelCase__ ): __a = y_points[i] for i in range(2 , lowerCAmelCase__ ): for j in range(lowerCAmelCase__ , lowerCAmelCase__ ): __a = ( (xa - x_points[j - i + 1]) * q[j][i - 1] - (xa - x_points[j]) * q[j - 1][i - 1] ) / (x_points[j] - x_points[j - i + 1]) return [q[n - 1][n - 1], q] if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import json import unittest import numpy as np from huggingface_hub import hf_hub_download from transformers.testing_utils import require_torch, require_vision 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 transformers import OneFormerImageProcessor from transformers.models.oneformer.image_processing_oneformer import binary_mask_to_rle from transformers.models.oneformer.modeling_oneformer import OneFormerForUniversalSegmentationOutput if is_vision_available(): from PIL import Image def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int]="shi-labs/oneformer_demo" ) -> Dict: with open(hf_hub_download(lowerCAmelCase__ , lowerCAmelCase__ , repo_type='''dataset''' ) , '''r''' ) as f: __a = json.load(lowerCAmelCase__ ) __a = {} __a = [] __a = [] for key, info in class_info.items(): __a = info['''name'''] class_names.append(info['''name'''] ) if info["isthing"]: thing_ids.append(int(lowerCAmelCase__ ) ) __a = thing_ids __a = class_names return metadata class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __init__( self , _a , _a=7 , _a=3 , _a=30 , _a=400 , _a=None , _a=True , _a=True , _a=[0.5, 0.5, 0.5] , _a=[0.5, 0.5, 0.5] , _a=10 , _a=False , _a=255 , _a="shi-labs/oneformer_demo" , _a="ade20k_panoptic.json" , _a=10 , ): __a = parent __a = batch_size __a = num_channels __a = min_resolution __a = max_resolution __a = do_resize __a = {'''shortest_edge''': 32, '''longest_edge''': 1_333} if size is None else size __a = do_normalize __a = image_mean __a = image_std __a = class_info_file __a = prepare_metadata(_a , _a ) __a = num_text __a = repo_path # for the post_process_functions __a = 2 __a = 10 __a = 10 __a = 3 __a = 4 __a = num_labels __a = do_reduce_labels __a = ignore_index def __UpperCAmelCase ( self ): return { "do_resize": self.do_resize, "size": self.size, "do_normalize": self.do_normalize, "image_mean": self.image_mean, "image_std": self.image_std, "num_labels": self.num_labels, "do_reduce_labels": self.do_reduce_labels, "ignore_index": self.ignore_index, "class_info_file": self.class_info_file, "metadata": self.metadata, "num_text": self.num_text, } def __UpperCAmelCase ( self , _a , _a=False ): if not batched: __a = image_inputs[0] if isinstance(_a , Image.Image ): __a , __a = image.size else: __a , __a = image.shape[1], image.shape[2] if w < h: __a = int(self.size['''shortest_edge'''] * h / w ) __a = self.size['''shortest_edge'''] elif w > h: __a = self.size['''shortest_edge'''] __a = int(self.size['''shortest_edge'''] * w / h ) else: __a = self.size['''shortest_edge'''] __a = self.size['''shortest_edge'''] else: __a = [] for image in image_inputs: __a , __a = self.get_expected_values([image] ) expected_values.append((expected_height, expected_width) ) __a = max(_a , key=lambda _a : item[0] )[0] __a = max(_a , key=lambda _a : item[1] )[1] return expected_height, expected_width def __UpperCAmelCase ( self ): return OneFormerForUniversalSegmentationOutput( # +1 for null class class_queries_logits=torch.randn((self.batch_size, self.num_queries, self.num_classes + 1) ) , masks_queries_logits=torch.randn((self.batch_size, self.num_queries, self.height, self.width) ) , ) @require_torch @require_vision class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : str = OneFormerImageProcessor if (is_vision_available() and is_torch_available()) else None # only for test_image_processing_common.test_image_proc_to_json_string __UpperCAmelCase : List[str] = image_processing_class def __UpperCAmelCase ( self ): __a = OneFormerImageProcessorTester(self ) @property def __UpperCAmelCase ( self ): return self.image_processing_tester.prepare_image_processor_dict() def __UpperCAmelCase ( self ): __a = self.image_processing_class(**self.image_processor_dict ) self.assertTrue(hasattr(_a , '''image_mean''' ) ) self.assertTrue(hasattr(_a , '''image_std''' ) ) self.assertTrue(hasattr(_a , '''do_normalize''' ) ) self.assertTrue(hasattr(_a , '''do_resize''' ) ) self.assertTrue(hasattr(_a , '''size''' ) ) self.assertTrue(hasattr(_a , '''ignore_index''' ) ) self.assertTrue(hasattr(_a , '''class_info_file''' ) ) self.assertTrue(hasattr(_a , '''num_text''' ) ) self.assertTrue(hasattr(_a , '''repo_path''' ) ) self.assertTrue(hasattr(_a , '''metadata''' ) ) self.assertTrue(hasattr(_a , '''do_reduce_labels''' ) ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): # Initialize image_processor __a = self.image_processing_class(**self.image_processor_dict ) # create random PIL images __a = prepare_image_inputs(self.image_processing_tester , equal_resolution=_a ) for image in image_inputs: self.assertIsInstance(_a , Image.Image ) # Test not batched input __a = image_processor(image_inputs[0] , ['''semantic'''] , return_tensors='''pt''' ).pixel_values __a , __a = self.image_processing_tester.get_expected_values(_a ) self.assertEqual( encoded_images.shape , (1, self.image_processing_tester.num_channels, expected_height, expected_width) , ) # Test batched __a , __a = self.image_processing_tester.get_expected_values(_a , batched=_a ) __a = image_processor( _a , ['''semantic'''] * len(_a ) , return_tensors='''pt''' ).pixel_values self.assertEqual( encoded_images.shape , ( self.image_processing_tester.batch_size, self.image_processing_tester.num_channels, expected_height, expected_width, ) , ) def __UpperCAmelCase ( self ): # Initialize image_processor __a = self.image_processing_class(**self.image_processor_dict ) # create random numpy tensors __a = prepare_image_inputs(self.image_processing_tester , equal_resolution=_a , numpify=_a ) for image in image_inputs: self.assertIsInstance(_a , np.ndarray ) # Test not batched input __a = image_processor(image_inputs[0] , ['''semantic'''] , return_tensors='''pt''' ).pixel_values __a , __a = self.image_processing_tester.get_expected_values(_a ) self.assertEqual( encoded_images.shape , (1, self.image_processing_tester.num_channels, expected_height, expected_width) , ) # Test batched __a , __a = self.image_processing_tester.get_expected_values(_a , batched=_a ) __a = image_processor( _a , ['''semantic'''] * len(_a ) , return_tensors='''pt''' ).pixel_values self.assertEqual( encoded_images.shape , ( self.image_processing_tester.batch_size, self.image_processing_tester.num_channels, expected_height, expected_width, ) , ) def __UpperCAmelCase ( self ): # Initialize image_processor __a = self.image_processing_class(**self.image_processor_dict ) # create random PyTorch tensors __a = prepare_image_inputs(self.image_processing_tester , equal_resolution=_a , torchify=_a ) for image in image_inputs: self.assertIsInstance(_a , torch.Tensor ) # Test not batched input __a = image_processor(image_inputs[0] , ['''semantic'''] , return_tensors='''pt''' ).pixel_values __a , __a = self.image_processing_tester.get_expected_values(_a ) self.assertEqual( encoded_images.shape , (1, self.image_processing_tester.num_channels, expected_height, expected_width) , ) # Test batched __a , __a = self.image_processing_tester.get_expected_values(_a , batched=_a ) __a = image_processor( _a , ['''semantic'''] * len(_a ) , return_tensors='''pt''' ).pixel_values self.assertEqual( encoded_images.shape , ( self.image_processing_tester.batch_size, self.image_processing_tester.num_channels, expected_height, expected_width, ) , ) def __UpperCAmelCase ( self , _a=False , _a=False , _a="np" ): __a = self.image_processing_class(**self.image_processor_dict ) # prepare image and target __a = self.image_processing_tester.num_labels __a = None __a = None __a = prepare_image_inputs(self.image_processing_tester , equal_resolution=_a ) if with_segmentation_maps: __a = num_labels if is_instance_map: __a = list(range(_a ) ) * 2 __a = dict(enumerate(_a ) ) __a = [ np.random.randint(0 , high * 2 , (img.size[1], img.size[0]) ).astype(np.uinta ) for img in image_inputs ] if segmentation_type == "pil": __a = [Image.fromarray(_a ) for annotation in annotations] __a = image_processor( _a , ['''semantic'''] * len(_a ) , _a , return_tensors='''pt''' , instance_id_to_semantic_id=_a , pad_and_return_pixel_mask=_a , ) return inputs def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): def common(_a=False , _a=None ): __a = self.comm_get_image_processor_inputs( with_segmentation_maps=_a , is_instance_map=_a , segmentation_type=_a ) __a = inputs['''mask_labels'''] __a = inputs['''class_labels'''] __a = inputs['''pixel_values'''] __a = inputs['''text_inputs'''] # check the batch_size for mask_label, class_label, text_input in zip(_a , _a , _a ): self.assertEqual(mask_label.shape[0] , class_label.shape[0] ) # this ensure padding has happened self.assertEqual(mask_label.shape[1:] , pixel_values.shape[2:] ) self.assertEqual(len(_a ) , self.image_processing_tester.num_text ) common() common(is_instance_map=_a ) common(is_instance_map=_a , segmentation_type='''pil''' ) common(is_instance_map=_a , segmentation_type='''pil''' ) def __UpperCAmelCase ( self ): __a = np.zeros((20, 50) ) __a = 1 __a = 1 __a = 1 __a = binary_mask_to_rle(_a ) self.assertEqual(len(_a ) , 4 ) self.assertEqual(rle[0] , 21 ) self.assertEqual(rle[1] , 45 ) def __UpperCAmelCase ( self ): __a = self.image_processing_class( num_labels=self.image_processing_tester.num_classes , max_seq_length=77 , task_seq_length=77 , class_info_file='''ade20k_panoptic.json''' , num_text=self.image_processing_tester.num_text , repo_path='''shi-labs/oneformer_demo''' , ) __a = self.image_processing_tester.get_fake_oneformer_outputs() __a = fature_extractor.post_process_semantic_segmentation(_a ) self.assertEqual(len(_a ) , self.image_processing_tester.batch_size ) self.assertEqual( segmentation[0].shape , ( self.image_processing_tester.height, self.image_processing_tester.width, ) , ) __a = [(1, 4) for i in range(self.image_processing_tester.batch_size )] __a = fature_extractor.post_process_semantic_segmentation(_a , target_sizes=_a ) self.assertEqual(segmentation[0].shape , target_sizes[0] ) def __UpperCAmelCase ( self ): __a = self.image_processing_class( num_labels=self.image_processing_tester.num_classes , max_seq_length=77 , task_seq_length=77 , class_info_file='''ade20k_panoptic.json''' , num_text=self.image_processing_tester.num_text , repo_path='''shi-labs/oneformer_demo''' , ) __a = self.image_processing_tester.get_fake_oneformer_outputs() __a = image_processor.post_process_instance_segmentation(_a , threshold=0 ) self.assertTrue(len(_a ) == self.image_processing_tester.batch_size ) for el in segmentation: self.assertTrue('''segmentation''' in el ) self.assertTrue('''segments_info''' in el ) self.assertEqual(type(el['''segments_info'''] ) , _a ) self.assertEqual( el['''segmentation'''].shape , (self.image_processing_tester.height, self.image_processing_tester.width) ) def __UpperCAmelCase ( self ): __a = self.image_processing_class( num_labels=self.image_processing_tester.num_classes , max_seq_length=77 , task_seq_length=77 , class_info_file='''ade20k_panoptic.json''' , num_text=self.image_processing_tester.num_text , repo_path='''shi-labs/oneformer_demo''' , ) __a = self.image_processing_tester.get_fake_oneformer_outputs() __a = image_processor.post_process_panoptic_segmentation(_a , threshold=0 ) self.assertTrue(len(_a ) == self.image_processing_tester.batch_size ) for el in segmentation: self.assertTrue('''segmentation''' in el ) self.assertTrue('''segments_info''' in el ) self.assertEqual(type(el['''segments_info'''] ) , _a ) self.assertEqual( el['''segmentation'''].shape , (self.image_processing_tester.height, self.image_processing_tester.width) )
712
"""simple docstring""" import json import os from functools import lru_cache from typing import List, Optional, Tuple import regex as re from ...tokenization_utils import AddedToken, PreTrainedTokenizer from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = {"vocab_file": "vocab.json", "merges_file": "merges.txt"} lowercase_ = { "vocab_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/vocab.json", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/vocab.json" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/vocab.json" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/vocab.json" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/vocab.json" ), }, "merges_file": { "allenai/longformer-base-4096": "https://huggingface.co/allenai/longformer-base-4096/resolve/main/merges.txt", "allenai/longformer-large-4096": ( "https://huggingface.co/allenai/longformer-large-4096/resolve/main/merges.txt" ), "allenai/longformer-large-4096-finetuned-triviaqa": ( "https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/merges.txt" ), "allenai/longformer-base-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/merges.txt" ), "allenai/longformer-large-4096-extra.pos.embd.only": ( "https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/merges.txt" ), }, } lowercase_ = { "allenai/longformer-base-4096": 4_0_9_6, "allenai/longformer-large-4096": 4_0_9_6, "allenai/longformer-large-4096-finetuned-triviaqa": 4_0_9_6, "allenai/longformer-base-4096-extra.pos.embd.only": 4_0_9_6, "allenai/longformer-large-4096-extra.pos.embd.only": 4_0_9_6, } @lru_cache() # Copied from transformers.models.roberta.tokenization_roberta.bytes_to_unicode def lowercase ( ) -> Union[str, Any]: __a = ( list(range(ord('''!''' ) , ord('''~''' ) + 1 ) ) + list(range(ord('''¡''' ) , ord('''¬''' ) + 1 ) ) + list(range(ord('''®''' ) , ord('''ÿ''' ) + 1 ) ) ) __a = bs[:] __a = 0 for b in range(2**8 ): if b not in bs: bs.append(lowerCAmelCase__ ) cs.append(2**8 + n ) n += 1 __a = [chr(lowerCAmelCase__ ) for n in cs] return dict(zip(lowerCAmelCase__ , lowerCAmelCase__ ) ) def lowercase ( lowerCAmelCase__ : Tuple ) -> Tuple: __a = set() __a = word[0] for char in word[1:]: pairs.add((prev_char, char) ) __a = char return pairs class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Optional[int] = VOCAB_FILES_NAMES __UpperCAmelCase : int = PRETRAINED_VOCAB_FILES_MAP __UpperCAmelCase : Dict = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES __UpperCAmelCase : Union[str, Any] = ['input_ids', 'attention_mask'] def __init__( self , _a , _a , _a="replace" , _a="<s>" , _a="</s>" , _a="</s>" , _a="<s>" , _a="<unk>" , _a="<pad>" , _a="<mask>" , _a=False , **_a , ): __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else bos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else eos_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else sep_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else cls_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else unk_token __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else pad_token # Mask token behave like a normal word, i.e. include the space before it __a = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else mask_token super().__init__( errors=_a , bos_token=_a , eos_token=_a , unk_token=_a , sep_token=_a , cls_token=_a , pad_token=_a , mask_token=_a , add_prefix_space=_a , **_a , ) with open(_a , encoding='''utf-8''' ) as vocab_handle: __a = json.load(_a ) __a = {v: k for k, v in self.encoder.items()} __a = errors # how to handle errors in decoding __a = bytes_to_unicode() __a = {v: k for k, v in self.byte_encoder.items()} with open(_a , encoding='''utf-8''' ) as merges_handle: __a = merges_handle.read().split('''\n''' )[1:-1] __a = [tuple(merge.split() ) for merge in bpe_merges] __a = dict(zip(_a , range(len(_a ) ) ) ) __a = {} __a = add_prefix_space # Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions __a = re.compile(R'''\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+''' ) @property def __UpperCAmelCase ( self ): return len(self.encoder ) def __UpperCAmelCase ( self ): return dict(self.encoder , **self.added_tokens_encoder ) def __UpperCAmelCase ( self , _a ): if token in self.cache: return self.cache[token] __a = tuple(_a ) __a = get_pairs(_a ) if not pairs: return token while True: __a = min(_a , key=lambda _a : self.bpe_ranks.get(_a , float('''inf''' ) ) ) if bigram not in self.bpe_ranks: break __a , __a = bigram __a = [] __a = 0 while i < len(_a ): try: __a = word.index(_a , _a ) except ValueError: new_word.extend(word[i:] ) break else: new_word.extend(word[i:j] ) __a = j if word[i] == first and i < len(_a ) - 1 and word[i + 1] == second: new_word.append(first + second ) i += 2 else: new_word.append(word[i] ) i += 1 __a = tuple(_a ) __a = new_word if len(_a ) == 1: break else: __a = get_pairs(_a ) __a = ''' '''.join(_a ) __a = word return word def __UpperCAmelCase ( self , _a ): __a = [] for token in re.findall(self.pat , _a ): __a = ''''''.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(_a ).split(''' ''' ) ) return bpe_tokens def __UpperCAmelCase ( self , _a ): return self.encoder.get(_a , self.encoder.get(self.unk_token ) ) def __UpperCAmelCase ( self , _a ): return self.decoder.get(_a ) def __UpperCAmelCase ( self , _a ): __a = ''''''.join(_a ) __a = bytearray([self.byte_decoder[c] for c in text] ).decode('''utf-8''' , errors=self.errors ) return text def __UpperCAmelCase ( self , _a , _a = None ): if not os.path.isdir(_a ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] ) __a = os.path.join( _a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] ) with open(_a , '''w''' , encoding='''utf-8''' ) as f: f.write(json.dumps(self.encoder , indent=2 , sort_keys=_a , ensure_ascii=_a ) + '''\n''' ) __a = 0 with open(_a , '''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 _a : 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!''' ) __a = token_index writer.write(''' '''.join(_a ) + '''\n''' ) index += 1 return vocab_file, merge_file def __UpperCAmelCase ( self , _a , _a = None ): if token_ids_a is None: return [self.cls_token_id] + token_ids_a + [self.sep_token_id] __a = [self.cls_token_id] __a = [self.sep_token_id] return cls + token_ids_a + sep + sep + token_ids_a + sep def __UpperCAmelCase ( self , _a , _a = None , _a = False ): if already_has_special_tokens: return super().get_special_tokens_mask( token_ids_a=_a , token_ids_a=_a , already_has_special_tokens=_a ) if token_ids_a is None: return [1] + ([0] * len(_a )) + [1] return [1] + ([0] * len(_a )) + [1, 1] + ([0] * len(_a )) + [1] def __UpperCAmelCase ( self , _a , _a = None ): __a = [self.sep_token_id] __a = [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 __UpperCAmelCase ( self , _a , _a=False , **_a ): __a = kwargs.pop('''add_prefix_space''' , self.add_prefix_space ) if (is_split_into_words or add_prefix_space) and (len(_a ) > 0 and not text[0].isspace()): __a = ''' ''' + text return (text, kwargs)
65
0
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import ( OptionalDependencyNotAvailable, _LazyModule, is_tf_available, is_torch_available, is_vision_available, ) lowercase_ = { "configuration_mobilevit": ["MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP", "MobileViTConfig", "MobileViTOnnxConfig"], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["MobileViTFeatureExtractor"] lowercase_ = ["MobileViTImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST", "MobileViTForImageClassification", "MobileViTForSemanticSegmentation", "MobileViTModel", "MobileViTPreTrainedModel", ] try: if not is_tf_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "TF_MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST", "TFMobileViTForImageClassification", "TFMobileViTForSemanticSegmentation", "TFMobileViTModel", "TFMobileViTPreTrainedModel", ] if TYPE_CHECKING: from .configuration_mobilevit import MOBILEVIT_PRETRAINED_CONFIG_ARCHIVE_MAP, MobileViTConfig, MobileViTOnnxConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .feature_extraction_mobilevit import MobileViTFeatureExtractor from .image_processing_mobilevit import MobileViTImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_mobilevit import ( MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST, MobileViTForImageClassification, MobileViTForSemanticSegmentation, MobileViTModel, MobileViTPreTrainedModel, ) try: if not is_tf_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_tf_mobilevit import ( TF_MOBILEVIT_PRETRAINED_MODEL_ARCHIVE_LIST, TFMobileViTForImageClassification, TFMobileViTForSemanticSegmentation, TFMobileViTModel, TFMobileViTPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
713
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging lowercase_ = logging.get_logger(__name__) lowercase_ = { "unc-nlp/lxmert-base-uncased": "https://huggingface.co/unc-nlp/lxmert-base-uncased/resolve/main/config.json", } class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : str = 'lxmert' __UpperCAmelCase : str = {} def __init__( self , _a=30_522 , _a=768 , _a=12 , _a=9_500 , _a=1_600 , _a=400 , _a=3_072 , _a="gelu" , _a=0.1 , _a=0.1 , _a=512 , _a=2 , _a=0.02 , _a=1E-12 , _a=9 , _a=5 , _a=5 , _a=2_048 , _a=4 , _a=6.67 , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , _a=True , **_a , ): __a = vocab_size __a = hidden_size __a = num_attention_heads __a = hidden_act __a = intermediate_size __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = max_position_embeddings __a = type_vocab_size __a = initializer_range __a = layer_norm_eps __a = num_qa_labels __a = num_object_labels __a = num_attr_labels __a = l_layers __a = x_layers __a = r_layers __a = visual_feat_dim __a = visual_pos_dim __a = visual_loss_normalizer __a = task_matched __a = task_mask_lm __a = task_obj_predict __a = task_qa __a = visual_obj_loss __a = visual_attr_loss __a = visual_feat_loss __a = {'''vision''': r_layers, '''cross_encoder''': x_layers, '''language''': l_layers} super().__init__(**_a )
65
0
"""simple docstring""" class __lowerCAmelCase : '''simple docstring''' def __init__( self ): __a = {} def __UpperCAmelCase ( self ): print(self.vertex ) for i in self.vertex: print(_a , ''' -> ''' , ''' -> '''.join([str(_a ) for j in self.vertex[i]] ) ) def __UpperCAmelCase ( self , _a , _a ): # check if vertex is already present, if from_vertex in self.vertex: self.vertex[from_vertex].append(_a ) else: # else make a new vertex __a = [to_vertex] def __UpperCAmelCase ( self ): # visited array for storing already visited nodes __a = [False] * len(self.vertex ) # call the recursive helper function for i in range(len(self.vertex ) ): if not visited[i]: self.dfs_recursive(_a , _a ) def __UpperCAmelCase ( self , _a , _a ): # mark start vertex as visited __a = True print(_a , end=''' ''' ) # Recur for all the vertices that are adjacent to this node for i in self.vertex: if not visited[i]: self.dfs_recursive(_a , _a ) if __name__ == "__main__": lowercase_ = Graph() g.add_edge(0, 1) g.add_edge(0, 2) g.add_edge(1, 2) g.add_edge(2, 0) g.add_edge(2, 3) g.add_edge(3, 3) g.print_graph() print("DFS:") g.dfs() # OUTPUT: # 0 -> 1 -> 2 # 1 -> 2 # 2 -> 0 -> 3 # 3 -> 3 # DFS: # 0 1 2 3
714
"""simple docstring""" import itertools import math def lowercase ( lowerCAmelCase__ : int ) -> bool: 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(lowerCAmelCase__ ) + 1 ) , 6 ): if number % i == 0 or number % (i + 2) == 0: return False return True def lowercase ( ) -> int: __a = 2 while True: if is_prime(lowerCAmelCase__ ): yield num num += 1 def lowercase ( lowerCAmelCase__ : int = 10001 ) -> int: return next(itertools.islice(prime_generator() , nth - 1 , lowerCAmelCase__ ) ) if __name__ == "__main__": print(F'''{solution() = }''')
65
0
"""simple docstring""" from ...configuration_utils import PretrainedConfig from ...utils import logging from ...utils.backbone_utils import BackboneConfigMixin, get_aligned_output_features_output_indices lowercase_ = logging.get_logger(__name__) lowercase_ = { "shi-labs/dinat-mini-in1k-224": "https://huggingface.co/shi-labs/dinat-mini-in1k-224/resolve/main/config.json", # See all Dinat models at https://huggingface.co/models?filter=dinat } class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE ): __UpperCAmelCase : Optional[int] = 'dinat' __UpperCAmelCase : str = { 'num_attention_heads': 'num_heads', 'num_hidden_layers': 'num_layers', } def __init__( self , _a=4 , _a=3 , _a=64 , _a=[3, 4, 6, 5] , _a=[2, 4, 8, 16] , _a=7 , _a=[[1, 8, 1], [1, 4, 1, 4], [1, 2, 1, 2, 1, 2], [1, 1, 1, 1, 1]] , _a=3.0 , _a=True , _a=0.0 , _a=0.0 , _a=0.1 , _a="gelu" , _a=0.02 , _a=1E-5 , _a=0.0 , _a=None , _a=None , **_a , ): super().__init__(**_a ) __a = patch_size __a = num_channels __a = embed_dim __a = depths __a = len(_a ) __a = num_heads __a = kernel_size __a = dilations __a = mlp_ratio __a = qkv_bias __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = drop_path_rate __a = hidden_act __a = layer_norm_eps __a = initializer_range # we set the hidden_size attribute in order to make Dinat work with VisionEncoderDecoderModel # this indicates the channel dimension after the last stage of the model __a = int(embed_dim * 2 ** (len(_a ) - 1) ) __a = layer_scale_init_value __a = ['''stem'''] + [f'''stage{idx}''' for idx in range(1 , len(_a ) + 1 )] __a , __a = get_aligned_output_features_output_indices( out_features=_a , out_indices=_a , stage_names=self.stage_names )
715
"""simple docstring""" from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_mask2former": [ "MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "Mask2FormerConfig", ], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["Mask2FormerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST", "Mask2FormerForUniversalSegmentation", "Mask2FormerModel", "Mask2FormerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_maskaformer import MASK2FORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, MaskaFormerConfig try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_maskaformer import MaskaFormerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_maskaformer import ( MASK2FORMER_PRETRAINED_MODEL_ARCHIVE_LIST, MaskaFormerForUniversalSegmentation, MaskaFormerModel, MaskaFormerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
65
0
"""simple docstring""" import inspect import unittest import numpy as np from tests.test_modeling_common import floats_tensor from transformers import MaskaFormerConfig, is_torch_available, is_vision_available from transformers.testing_utils import require_torch, require_torch_multi_gpu, require_vision, slow, torch_device from transformers.utils import cached_property from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin from ...test_pipeline_mixin import PipelineTesterMixin if is_torch_available(): import torch from transformers import MaskaFormerForUniversalSegmentation, MaskaFormerModel if is_vision_available(): from transformers import MaskaFormerImageProcessor if is_vision_available(): from PIL import Image class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=2 , _a=True , _a=False , _a=10 , _a=3 , _a=32 * 8 , _a=32 * 8 , _a=4 , _a=64 , ): __a = parent __a = batch_size __a = is_training __a = use_auxiliary_loss __a = num_queries __a = num_channels __a = min_size __a = max_size __a = num_labels __a = hidden_dim __a = hidden_dim def __UpperCAmelCase ( self ): __a = floats_tensor([self.batch_size, self.num_channels, self.min_size, self.max_size] ).to( _a ) __a = torch.ones([self.batch_size, self.min_size, self.max_size] , device=_a ) __a = ( torch.rand([self.batch_size, self.num_labels, self.min_size, self.max_size] , device=_a ) > 0.5 ).float() __a = (torch.rand((self.batch_size, self.num_labels) , device=_a ) > 0.5).long() __a = self.get_config() return config, pixel_values, pixel_mask, mask_labels, class_labels def __UpperCAmelCase ( self ): __a = MaskaFormerConfig( hidden_size=self.hidden_dim , ) __a = self.num_queries __a = self.num_labels __a = [1, 1, 1, 1] __a = self.num_channels __a = 64 __a = 128 __a = self.hidden_dim __a = self.hidden_dim __a = self.hidden_dim return config def __UpperCAmelCase ( self ): __a , __a , __a , __a , __a = self.prepare_config_and_inputs() __a = {'''pixel_values''': pixel_values, '''pixel_mask''': pixel_mask} return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = output.encoder_hidden_states __a = output.pixel_decoder_hidden_states __a = output.transformer_decoder_hidden_states self.parent.assertTrue(len(_a ) , len(config.backbone_config.depths ) ) self.parent.assertTrue(len(_a ) , len(config.backbone_config.depths ) ) self.parent.assertTrue(len(_a ) , config.decoder_layers ) def __UpperCAmelCase ( self , _a , _a , _a , _a=False ): with torch.no_grad(): __a = MaskaFormerModel(config=_a ) model.to(_a ) model.eval() __a = model(pixel_values=_a , pixel_mask=_a ) __a = model(_a , output_hidden_states=_a ) self.parent.assertEqual( output.transformer_decoder_last_hidden_state.shape , (self.batch_size, self.num_queries, self.hidden_dim) , ) # let's ensure the other two hidden state exists self.parent.assertTrue(output.pixel_decoder_last_hidden_state is not None ) self.parent.assertTrue(output.encoder_last_hidden_state is not None ) if output_hidden_states: self.check_output_hidden_state(_a , _a ) def __UpperCAmelCase ( self , _a , _a , _a , _a , _a ): __a = MaskaFormerForUniversalSegmentation(config=_a ) model.to(_a ) model.eval() def comm_check_on_output(_a ): # let's still check that all the required stuff is there self.parent.assertTrue(result.transformer_decoder_last_hidden_state is not None ) self.parent.assertTrue(result.pixel_decoder_last_hidden_state is not None ) self.parent.assertTrue(result.encoder_last_hidden_state is not None ) # okay, now we need to check the logits shape # due to the encoder compression, masks have a //4 spatial size self.parent.assertEqual( result.masks_queries_logits.shape , (self.batch_size, self.num_queries, self.min_size // 4, self.max_size // 4) , ) # + 1 for null class self.parent.assertEqual( result.class_queries_logits.shape , (self.batch_size, self.num_queries, self.num_labels + 1) ) with torch.no_grad(): __a = model(pixel_values=_a , pixel_mask=_a ) __a = model(_a ) comm_check_on_output(_a ) __a = model( pixel_values=_a , pixel_mask=_a , mask_labels=_a , class_labels=_a ) comm_check_on_output(_a ) self.parent.assertTrue(result.loss is not None ) self.parent.assertEqual(result.loss.shape , torch.Size([1] ) ) @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Dict = (MaskaFormerModel, MaskaFormerForUniversalSegmentation) if is_torch_available() else () __UpperCAmelCase : Optional[int] = {'feature-extraction': MaskaFormerModel} if is_torch_available() else {} __UpperCAmelCase : str = False __UpperCAmelCase : List[str] = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : Tuple = False def __UpperCAmelCase ( self ): __a = MaskaFormerModelTester(self ) __a = ConfigTester(self , config_class=_a , has_text_modality=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.create_and_check_maskaformer_model(_a , **_a , output_hidden_states=_a ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_maskaformer_instance_segmentation_head_model(*_a ) @unittest.skip(reason='''Mask2Former does not use inputs_embeds''' ) def __UpperCAmelCase ( self ): pass @unittest.skip(reason='''Mask2Former does not have a get_input_embeddings method''' ) def __UpperCAmelCase ( self ): pass @unittest.skip(reason='''Mask2Former is not a generative model''' ) def __UpperCAmelCase ( self ): pass @unittest.skip(reason='''Mask2Former does not use token embeddings''' ) def __UpperCAmelCase ( self ): pass @require_torch_multi_gpu @unittest.skip( reason='''Mask2Former has some layers using `add_module` which doesn\'t work well with `nn.DataParallel`''' ) def __UpperCAmelCase ( self ): pass @unittest.skip('''Will be fixed soon by reducing the size of the model used for common tests.''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ) __a = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic __a = [*signature.parameters.keys()] __a = ['''pixel_values'''] self.assertListEqual(arg_names[:1] , _a ) @slow def __UpperCAmelCase ( self ): for model_name in ["facebook/mask2former-swin-small-coco-instance"]: __a = MaskaFormerModel.from_pretrained(_a ) self.assertIsNotNone(_a ) def __UpperCAmelCase ( self ): __a = (self.model_tester.min_size,) * 2 __a = { '''pixel_values''': torch.randn((2, 3, *size) , device=_a ), '''mask_labels''': torch.randn((2, 10, *size) , device=_a ), '''class_labels''': torch.zeros(2 , 10 , device=_a ).long(), } __a = self.model_tester.get_config() __a = MaskaFormerForUniversalSegmentation(_a ).to(_a ) __a = model(**_a ) self.assertTrue(outputs.loss is not None ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.create_and_check_maskaformer_model(_a , **_a , output_hidden_states=_a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ).to(_a ) __a = model(**_a , output_attentions=_a ) self.assertTrue(outputs.attentions is not None ) def __UpperCAmelCase ( self ): if not self.model_tester.is_training: return __a = self.all_model_classes[1] __a , __a , __a , __a , __a = self.model_tester.prepare_config_and_inputs() __a = model_class(_a ) model.to(_a ) model.train() __a = model(_a , mask_labels=_a , class_labels=_a ).loss loss.backward() def __UpperCAmelCase ( self ): __a = self.all_model_classes[1] __a , __a , __a , __a , __a = self.model_tester.prepare_config_and_inputs() __a = True __a = True __a = model_class(_a ).to(_a ) model.train() __a = model(_a , mask_labels=_a , class_labels=_a ) __a = outputs.encoder_hidden_states[0] encoder_hidden_states.retain_grad() __a = outputs.pixel_decoder_hidden_states[0] pixel_decoder_hidden_states.retain_grad() __a = outputs.transformer_decoder_hidden_states[0] transformer_decoder_hidden_states.retain_grad() __a = outputs.attentions[0] attentions.retain_grad() outputs.loss.backward(retain_graph=_a ) self.assertIsNotNone(encoder_hidden_states.grad ) self.assertIsNotNone(pixel_decoder_hidden_states.grad ) self.assertIsNotNone(transformer_decoder_hidden_states.grad ) self.assertIsNotNone(attentions.grad ) lowercase_ = 1e-4 def lowercase ( ) -> Dict: __a = Image.open('''./tests/fixtures/tests_samples/COCO/000000039769.png''' ) return image @require_vision @slow class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' @cached_property def __UpperCAmelCase ( self ): return "facebook/mask2former-swin-small-coco-instance" @cached_property def __UpperCAmelCase ( self ): return MaskaFormerImageProcessor.from_pretrained(self.model_checkpoints ) if is_vision_available() else None def __UpperCAmelCase ( self ): __a = MaskaFormerModel.from_pretrained(self.model_checkpoints ).to(_a ) __a = self.default_image_processor __a = prepare_img() __a = image_processor(_a , return_tensors='''pt''' ).to(_a ) __a = inputs['''pixel_values'''].shape # check size is divisible by 32 self.assertTrue((inputs_shape[-1] % 32) == 0 and (inputs_shape[-2] % 32) == 0 ) # check size self.assertEqual(_a , (1, 3, 384, 384) ) with torch.no_grad(): __a = model(**_a ) __a = torch.tensor( [[-0.2790, -1.0717, -1.1668], [-0.5128, -0.3128, -0.4987], [-0.5832, 0.1971, -0.0197]] ).to(_a ) self.assertTrue( torch.allclose( outputs.encoder_last_hidden_state[0, 0, :3, :3] , _a , atol=_a ) ) __a = torch.tensor( [[0.8973, 1.1847, 1.1776], [1.1934, 1.5040, 1.5128], [1.1153, 1.4486, 1.4951]] ).to(_a ) self.assertTrue( torch.allclose( outputs.pixel_decoder_last_hidden_state[0, 0, :3, :3] , _a , atol=_a ) ) __a = torch.tensor( [[2.1152, 1.7000, -0.8603], [1.5808, 1.8004, -0.9353], [1.6043, 1.7495, -0.5999]] ).to(_a ) self.assertTrue( torch.allclose( outputs.transformer_decoder_last_hidden_state[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = MaskaFormerForUniversalSegmentation.from_pretrained(self.model_checkpoints ).to(_a ).eval() __a = self.default_image_processor __a = prepare_img() __a = image_processor(_a , return_tensors='''pt''' ).to(_a ) __a = inputs['''pixel_values'''].shape # check size is divisible by 32 self.assertTrue((inputs_shape[-1] % 32) == 0 and (inputs_shape[-2] % 32) == 0 ) # check size self.assertEqual(_a , (1, 3, 384, 384) ) with torch.no_grad(): __a = model(**_a ) # masks_queries_logits __a = outputs.masks_queries_logits self.assertEqual( masks_queries_logits.shape , (1, model.config.num_queries, inputs_shape[-2] // 4, inputs_shape[-1] // 4) ) __a = [ [-8.7839, -9.0056, -8.8121], [-7.4104, -7.0313, -6.5401], [-6.6105, -6.3427, -6.4675], ] __a = torch.tensor(_a ).to(_a ) self.assertTrue(torch.allclose(masks_queries_logits[0, 0, :3, :3] , _a , atol=_a ) ) # class_queries_logits __a = outputs.class_queries_logits self.assertEqual(class_queries_logits.shape , (1, model.config.num_queries, model.config.num_labels + 1) ) __a = torch.tensor( [ [1.8324, -8.0835, -4.1922], [0.8450, -9.0050, -3.6053], [0.3045, -7.7293, -3.0275], ] ).to(_a ) self.assertTrue(torch.allclose(outputs.class_queries_logits[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = MaskaFormerForUniversalSegmentation.from_pretrained(self.model_checkpoints ).to(_a ).eval() __a = self.default_image_processor __a = image_processor( [np.zeros((3, 800, 1_333) ), np.zeros((3, 800, 1_333) )] , segmentation_maps=[np.zeros((384, 384) ).astype(np.floataa ), np.zeros((384, 384) ).astype(np.floataa )] , return_tensors='''pt''' , ) __a = inputs['''pixel_values'''].to(_a ) __a = [el.to(_a ) for el in inputs['''mask_labels''']] __a = [el.to(_a ) for el in inputs['''class_labels''']] with torch.no_grad(): __a = model(**_a ) self.assertTrue(outputs.loss is not None )
716
"""simple docstring""" from sklearn.metrics import fa_score import datasets lowercase_ = "\nThe F1 score is the harmonic mean of the precision and recall. It can be computed with the equation:\nF1 = 2 * (precision * recall) / (precision + recall)\n" lowercase_ = "\nArgs:\n predictions (`list` of `int`): Predicted labels.\n references (`list` of `int`): Ground truth labels.\n labels (`list` of `int`): The set of labels to include when `average` is not set to `'binary'`, and the order of the labels if `average` is `None`. Labels present in the data can be excluded, for example to calculate a multiclass average ignoring a majority negative class. Labels not present in the data will result in 0 components in a macro average. For multilabel targets, labels are column indices. By default, all labels in `predictions` and `references` are used in sorted order. Defaults to None.\n pos_label (`int`): The class to be considered the positive class, in the case where `average` is set to `binary`. Defaults to 1.\n average (`string`): This parameter is required for multiclass/multilabel targets. If set to `None`, the scores for each class are returned. Otherwise, this determines the type of averaging performed on the data. Defaults to `'binary'`.\n\n - 'binary': Only report results for the class specified by `pos_label`. This is applicable only if the classes found in `predictions` and `references` are binary.\n - 'micro': Calculate metrics globally by counting the total true positives, false negatives and false positives.\n - 'macro': Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.\n - 'weighted': Calculate metrics for each label, and find their average weighted by support (the number of true instances for each label). This alters `'macro'` to account for label imbalance. This option can result in an F-score that is not between precision and recall.\n - 'samples': Calculate metrics for each instance, and find their average (only meaningful for multilabel classification).\n sample_weight (`list` of `float`): Sample weights Defaults to None.\n\nReturns:\n f1 (`float` or `array` of `float`): F1 score or list of f1 scores, depending on the value passed to `average`. Minimum possible value is 0. Maximum possible value is 1. Higher f1 scores are better.\n\nExamples:\n\n Example 1-A simple binary example\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0])\n >>> print(results)\n {'f1': 0.5}\n\n Example 2-The same simple binary example as in Example 1, but with `pos_label` set to `0`.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], pos_label=0)\n >>> print(round(results['f1'], 2))\n 0.67\n\n Example 3-The same simple binary example as in Example 1, but with `sample_weight` included.\n >>> f1_metric = datasets.load_metric(\"f1\")\n >>> results = f1_metric.compute(references=[0, 1, 0, 1, 0], predictions=[0, 0, 1, 1, 0], sample_weight=[0.9, 0.5, 3.9, 1.2, 0.3])\n >>> print(round(results['f1'], 2))\n 0.35\n\n Example 4-A multiclass example, with different values for the `average` input.\n >>> predictions = [0, 2, 1, 0, 0, 1]\n >>> references = [0, 1, 2, 0, 1, 2]\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"macro\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"micro\")\n >>> print(round(results['f1'], 2))\n 0.33\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=\"weighted\")\n >>> print(round(results['f1'], 2))\n 0.27\n >>> results = f1_metric.compute(predictions=predictions, references=references, average=None)\n >>> print(results)\n {'f1': array([0.8, 0. , 0. ])}\n" lowercase_ = "\n@article{scikit-learn,\n title={Scikit-learn: Machine Learning in {P}ython},\n author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.\n and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.\n and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and\n Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},\n journal={Journal of Machine Learning Research},\n volume={12},\n pages={2825--2830},\n year={2011}\n}\n" @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class __lowerCAmelCase ( datasets.Metric ): '''simple docstring''' def __UpperCAmelCase ( self ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { '''predictions''': datasets.Sequence(datasets.Value('''int32''' ) ), '''references''': datasets.Sequence(datasets.Value('''int32''' ) ), } if self.config_name == '''multilabel''' else { '''predictions''': datasets.Value('''int32''' ), '''references''': datasets.Value('''int32''' ), } ) , reference_urls=['''https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html'''] , ) def __UpperCAmelCase ( self , _a , _a , _a=None , _a=1 , _a="binary" , _a=None ): __a = fa_score( _a , _a , labels=_a , pos_label=_a , average=_a , sample_weight=_a ) return {"f1": float(_a ) if score.size == 1 else score}
65
0
"""simple docstring""" import pytest from datasets.parallel import ParallelBackendConfig, parallel_backend from datasets.utils.py_utils import map_nested from .utils import require_dill_gt_0_3_2, require_joblibspark, require_not_windows def lowercase ( lowerCAmelCase__ : Tuple ) -> Optional[int]: # picklable for multiprocessing return i + 1 @require_dill_gt_0_3_2 @require_joblibspark @require_not_windows def lowercase ( ) -> Any: with parallel_backend('''spark''' ): assert ParallelBackendConfig.backend_name == "spark" __a = [1, 2, 3] with pytest.raises(lowerCAmelCase__ ): with parallel_backend('''unsupported backend''' ): map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=2 ) with pytest.raises(lowerCAmelCase__ ): with parallel_backend('''unsupported backend''' ): map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=-1 ) @require_dill_gt_0_3_2 @require_joblibspark @require_not_windows @pytest.mark.parametrize('''num_proc''' , [2, -1] ) def lowercase ( lowerCAmelCase__ : Optional[Any] ) -> List[Any]: __a = [1, 2] __a = {'''a''': 1, '''b''': 2} __a = {'''a''': [1, 2], '''b''': [3, 4]} __a = {'''a''': {'''1''': 1}, '''b''': 2} __a = {'''a''': 1, '''b''': 2, '''c''': 3, '''d''': 4} __a = [2, 3] __a = {'''a''': 2, '''b''': 3} __a = {'''a''': [2, 3], '''b''': [4, 5]} __a = {'''a''': {'''1''': 2}, '''b''': 3} __a = {'''a''': 2, '''b''': 3, '''c''': 4, '''d''': 5} with parallel_backend('''spark''' ): assert map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=lowerCAmelCase__ ) == expected_map_nested_sa assert map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=lowerCAmelCase__ ) == expected_map_nested_sa assert map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=lowerCAmelCase__ ) == expected_map_nested_sa assert map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=lowerCAmelCase__ ) == expected_map_nested_sa assert map_nested(lowerCAmelCase__ , lowerCAmelCase__ , num_proc=lowerCAmelCase__ ) == expected_map_nested_sa
717
"""simple docstring""" from ..utils import DummyObject, requires_backends class __lowerCAmelCase ( metaclass=__SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = ['onnx'] def __init__( self , *_a , **_a ): requires_backends(self , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] ) @classmethod def __UpperCAmelCase ( cls , *_a , **_a ): requires_backends(cls , ['''onnx'''] )
65
0
"""simple docstring""" from .constants import ( MODEL_NAME, OPTIMIZER_NAME, RNG_STATE_NAME, SAFE_WEIGHTS_INDEX_NAME, SAFE_WEIGHTS_NAME, SCALER_NAME, SCHEDULER_NAME, TORCH_LAUNCH_PARAMS, WEIGHTS_INDEX_NAME, WEIGHTS_NAME, ) from .dataclasses import ( BnbQuantizationConfig, ComputeEnvironment, CustomDtype, DeepSpeedPlugin, DistributedDataParallelKwargs, DistributedType, DynamoBackend, FPaRecipeKwargs, FullyShardedDataParallelPlugin, GradientAccumulationPlugin, GradScalerKwargs, InitProcessGroupKwargs, KwargsHandler, LoggerType, MegatronLMPlugin, PrecisionType, ProjectConfiguration, RNGType, SageMakerDistributedType, TensorInformation, TorchDynamoPlugin, ) from .environment import get_int_from_env, parse_choice_from_env, parse_flag_from_env from .imports import ( get_ccl_version, is_abit_bnb_available, is_abit_bnb_available, is_aim_available, is_bfaa_available, is_bnb_available, is_botoa_available, is_ccl_available, is_comet_ml_available, is_datasets_available, is_deepspeed_available, is_fpa_available, is_ipex_available, is_megatron_lm_available, is_mlflow_available, is_mps_available, is_npu_available, is_rich_available, is_safetensors_available, is_sagemaker_available, is_tensorboard_available, is_tpu_available, is_transformers_available, is_wandb_available, is_xpu_available, ) from .modeling import ( check_device_map, check_tied_parameters_in_config, check_tied_parameters_on_same_device, compute_module_sizes, convert_file_size_to_int, dtype_byte_size, find_tied_parameters, get_balanced_memory, get_max_layer_size, get_max_memory, get_mixed_precision_context_manager, id_tensor_storage, infer_auto_device_map, load_checkpoint_in_model, load_offloaded_weights, load_state_dict, named_module_tensors, retie_parameters, set_module_tensor_to_device, shard_checkpoint, ) from .offload import ( OffloadedWeightsLoader, PrefixedDataset, extract_submodules_state_dict, load_offloaded_weight, offload_state_dict, offload_weight, save_offload_index, ) from .operations import ( broadcast, broadcast_object_list, concatenate, convert_outputs_to_fpaa, convert_to_fpaa, find_batch_size, find_device, gather, gather_object, get_data_structure, honor_type, initialize_tensors, is_namedtuple, is_tensor_information, is_torch_tensor, listify, pad_across_processes, recursively_apply, reduce, send_to_device, slice_tensors, ) from .versions import compare_versions, is_torch_version if is_deepspeed_available(): from .deepspeed import ( DeepSpeedEngineWrapper, DeepSpeedOptimizerWrapper, DeepSpeedSchedulerWrapper, DummyOptim, DummyScheduler, HfDeepSpeedConfig, ) from .bnb import has_abit_bnb_layers, load_and_quantize_model from .fsdp_utils import load_fsdp_model, load_fsdp_optimizer, save_fsdp_model, save_fsdp_optimizer from .launch import ( PrepareForLaunch, _filter_args, prepare_deepspeed_cmd_env, prepare_multi_gpu_env, prepare_sagemager_args_inputs, prepare_simple_launcher_cmd_env, prepare_tpu, ) from .megatron_lm import ( AbstractTrainStep, BertTrainStep, GPTTrainStep, MegatronEngine, MegatronLMDummyDataLoader, MegatronLMDummyScheduler, MegatronLMOptimizerWrapper, MegatronLMSchedulerWrapper, TaTrainStep, avg_losses_across_data_parallel_group, gather_across_data_parallel_groups, ) from .megatron_lm import initialize as megatron_lm_initialize from .megatron_lm import prepare_data_loader as megatron_lm_prepare_data_loader from .megatron_lm import prepare_model as megatron_lm_prepare_model from .megatron_lm import prepare_optimizer as megatron_lm_prepare_optimizer from .megatron_lm import prepare_scheduler as megatron_lm_prepare_scheduler from .memory import find_executable_batch_size, release_memory from .other import ( extract_model_from_parallel, get_pretty_name, is_port_in_use, merge_dicts, patch_environment, save, wait_for_everyone, write_basic_config, ) from .random import set_seed, synchronize_rng_state, synchronize_rng_states from .torch_xla import install_xla from .tqdm import tqdm from .transformer_engine import convert_model, has_transformer_engine_layers
718
"""simple docstring""" 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 lowercase_ = 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 __lowerCAmelCase : '''simple docstring''' def __init__( self , _a , _a=16 , _a=13 , _a=7 , _a=14 , _a=10 , _a=19 , _a=5 , _a=4 , _a=True , _a=16 , _a=2 , _a=4 , _a=4 , _a="gelu" , _a=0.1 , _a=0.1 , _a=[1, 2, 3, 4, 5] , _a=25 , _a=5 , ): __a = d_model __a = parent __a = batch_size __a = prediction_length __a = context_length __a = cardinality __a = num_time_features __a = lags_sequence __a = embedding_dimension __a = is_training __a = hidden_size __a = num_hidden_layers __a = num_attention_heads __a = intermediate_size __a = hidden_act __a = hidden_dropout_prob __a = attention_probs_dropout_prob __a = context_length __a = prediction_length + label_length __a = label_length __a = moving_average __a = autocorrelation_factor def __UpperCAmelCase ( self ): 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 __UpperCAmelCase ( self , _a ): __a = config.context_length + max(config.lags_sequence ) __a = ids_tensor([self.batch_size, 1] , config.cardinality[0] ) __a = floats_tensor([self.batch_size, _past_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, _past_length] ) __a = floats_tensor([self.batch_size, _past_length] ) > 0.5 # decoder inputs __a = floats_tensor([self.batch_size, config.prediction_length, config.num_time_features] ) __a = floats_tensor([self.batch_size, config.prediction_length] ) __a = { '''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 __UpperCAmelCase ( self ): __a = self.get_config() __a = self.prepare_autoformer_inputs_dict(_a ) return config, inputs_dict def __UpperCAmelCase ( self ): __a , __a = self.prepare_config_and_inputs() return config, inputs_dict def __UpperCAmelCase ( self , _a , _a ): __a = AutoformerModel(config=_a ).to(_a ).eval() __a = model(**_a ) __a = outputs.encoder_last_hidden_state __a = outputs.last_hidden_state with tempfile.TemporaryDirectory() as tmpdirname: __a = model.get_encoder() encoder.save_pretrained(_a ) __a = AutoformerEncoder.from_pretrained(_a ).to(_a ) __a , __a , __a , __a , __a = model.create_network_inputs(**_a ) __a , __a = model.decomposition_layer(transformer_inputs[:, : config.context_length, ...] ) __a = torch.cat( (transformer_inputs[:, : config.context_length, ...], feature[:, : config.context_length, ...]) , dim=-1 , ) __a = encoder(inputs_embeds=_a )[0] self.parent.assertTrue((encoder_last_hidden_state_a - encoder_last_hidden_state).abs().max().item() < 1E-3 ) __a = ( torch.mean(transformer_inputs[:, : config.context_length, ...] , dim=1 ) .unsqueeze(1 ) .repeat(1 , config.prediction_length , 1 ) ) __a = torch.zeros( [transformer_inputs.shape[0], config.prediction_length, transformer_inputs.shape[2]] , device=enc_input.device , ) __a = torch.cat( ( torch.cat((seasonal_input[:, -config.label_length :, ...], zeros) , dim=1 ), feature[:, config.context_length - config.label_length :, ...], ) , dim=-1 , ) __a = 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: __a = model.get_decoder() decoder.save_pretrained(_a ) __a = AutoformerDecoder.from_pretrained(_a ).to(_a ) __a = decoder( trend=_a , inputs_embeds=_a , encoder_hidden_states=_a , )[0] self.parent.assertTrue((last_hidden_state_a - last_hidden_state).abs().max().item() < 1E-3 ) @require_torch class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Any = (AutoformerModel, AutoformerForPrediction) if is_torch_available() else () __UpperCAmelCase : int = (AutoformerForPrediction,) if is_torch_available() else () __UpperCAmelCase : Any = {'feature-extraction': AutoformerModel} if is_torch_available() else {} __UpperCAmelCase : int = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False __UpperCAmelCase : Any = False __UpperCAmelCase : Optional[int] = False __UpperCAmelCase : int = False def __UpperCAmelCase ( self ): __a = AutoformerModelTester(self ) __a = ConfigTester(self , config_class=_a , has_text_modality=_a ) def __UpperCAmelCase ( self ): self.config_tester.run_common_tests() def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs() for model_class in self.all_model_classes: __a = model_class(_a ) with tempfile.TemporaryDirectory() as tmpdirname: model.save_pretrained(_a ) __a , __a = model_class.from_pretrained(_a , output_loading_info=_a ) self.assertEqual(info['''missing_keys'''] , [] ) def __UpperCAmelCase ( self ): __a = self.model_tester.prepare_config_and_inputs_for_common() self.model_tester.check_encoder_decoder_model_standalone(*_a ) @unittest.skip(reason='''Model has no tokens embeddings''' ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = inspect.signature(getattr(_a , '''forward''' ) ) # The main input is the name of the argument after `self` __a = list(model_signature.parameters.keys() )[1] self.assertEqual(AutoformerModel.main_input_name , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: __a = model_class(_a ) __a = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic __a = [*signature.parameters.keys()] __a = [ '''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(_a )] , _a ) def __UpperCAmelCase ( self ): __a , __a = self.model_tester.prepare_config_and_inputs_for_common() __a = True __a = getattr(self.model_tester , '''seq_length''' , _a ) __a = getattr(self.model_tester , '''decoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''encoder_seq_length''' , _a ) __a = getattr(self.model_tester , '''d_model''' , _a ) __a = getattr(self.model_tester , '''num_attention_heads''' , _a ) __a = d_model // num_attention_heads for model_class in self.all_model_classes: __a = True __a = False __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) # check that output_attentions also work using config del inputs_dict["output_attentions"] __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) __a = outputs.encoder_attentions self.assertEqual(len(_a ) , self.model_tester.num_hidden_layers ) self.assertListEqual( list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, encoder_seq_length, dim] , ) __a = len(_a ) __a = 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(_a , _a ) # decoder attentions __a = outputs.decoder_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = outputs.cross_attentions self.assertIsInstance(_a , (list, tuple) ) self.assertEqual(len(_a ) , 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 __a = True __a = True __a = model_class(_a ) model.to(_a ) model.eval() with torch.no_grad(): __a = model(**self._prepare_for_class(_a , _a ) ) self.assertEqual(out_len + 2 , len(_a ) ) __a = outputs.encoder_attentions if config.is_encoder_decoder else outputs.attentions self.assertEqual(len(_a ) , 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 __UpperCAmelCase ( self ): super().test_retain_grad_hidden_states_attentions() def lowercase ( lowerCAmelCase__ : Optional[int]="train-batch.pt" ) -> List[str]: __a = hf_hub_download(repo_id='''hf-internal-testing/tourism-monthly-batch''' , filename=lowerCAmelCase__ , repo_type='''dataset''' ) __a = torch.load(lowerCAmelCase__ , map_location=lowerCAmelCase__ ) return batch @require_torch @slow class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' def __UpperCAmelCase ( self ): __a = AutoformerModel.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch() with torch.no_grad(): __a = 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] __a = torch.Size( (64, model.config.prediction_length + model.config.label_length, model.config.feature_size) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[0.3593, -1.3398, 0.6330], [0.2279, 1.5396, -0.1792], [0.0450, 1.3225, -0.2335]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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 __a = torch.Size((64, model.config.context_length, model.config.d_model) ) self.assertEqual(output.shape , _a ) __a = torch.tensor( [[-0.0734, -0.9036, 0.8358], [4.7186, 2.4113, 1.9581], [1.7953, 2.3558, 1.2970]] , device=_a ) self.assertTrue(torch.allclose(output[0, :3, :3] , _a , atol=_a ) ) def __UpperCAmelCase ( self ): __a = AutoformerForPrediction.from_pretrained('''huggingface/autoformer-tourism-monthly''' ).to(_a ) __a = prepare_batch('''val-batch.pt''' ) with torch.no_grad(): __a = 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'''] , ) __a = torch.Size((64, model.config.num_parallel_samples, model.config.prediction_length) ) self.assertEqual(outputs.sequences.shape , _a ) __a = torch.tensor([3130.6763, 4056.5293, 7053.0786] , device=_a ) __a = outputs.sequences.mean(dim=1 ) self.assertTrue(torch.allclose(mean_prediction[0, -3:] , _a , rtol=1E-1 ) )
65
0
"""simple docstring""" import warnings from contextlib import contextmanager from ...processing_utils import ProcessorMixin from .feature_extraction_wavaveca import WavaVecaFeatureExtractor from .tokenization_wavaveca import WavaVecaCTCTokenizer class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : List[str] = 'Wav2Vec2FeatureExtractor' __UpperCAmelCase : List[str] = 'AutoTokenizer' def __init__( self , _a , _a ): super().__init__(_a , _a ) __a = self.feature_extractor __a = False @classmethod def __UpperCAmelCase ( cls , _a , **_a ): try: return super().from_pretrained(_a , **_a ) except OSError: warnings.warn( f'''Loading a tokenizer inside {cls.__name__} from a config that does not''' ''' include a `tokenizer_class` attribute is deprecated and will be ''' '''removed in v5. Please add `\'tokenizer_class\': \'Wav2Vec2CTCTokenizer\'`''' ''' attribute to either your `config.json` or `tokenizer_config.json` ''' '''file to suppress this warning: ''' , _a , ) __a = WavaVecaFeatureExtractor.from_pretrained(_a , **_a ) __a = WavaVecaCTCTokenizer.from_pretrained(_a , **_a ) return cls(feature_extractor=_a , tokenizer=_a ) def __call__( self , *_a , **_a ): # For backward compatibility if self._in_target_context_manager: return self.current_processor(*_a , **_a ) if "raw_speech" in kwargs: warnings.warn('''Using `raw_speech` as a keyword argument is deprecated. Use `audio` instead.''' ) __a = kwargs.pop('''raw_speech''' ) else: __a = kwargs.pop('''audio''' , _a ) __a = kwargs.pop('''sampling_rate''' , _a ) __a = kwargs.pop('''text''' , _a ) if len(_a ) > 0: __a = args[0] __a = args[1:] if audio is None and text is None: raise ValueError('''You need to specify either an `audio` or `text` input to process.''' ) if audio is not None: __a = self.feature_extractor(_a , *_a , sampling_rate=_a , **_a ) if text is not None: __a = self.tokenizer(_a , **_a ) if text is None: return inputs elif audio is None: return encodings else: __a = encodings['''input_ids'''] return inputs def __UpperCAmelCase ( self , *_a , **_a ): # For backward compatibility if self._in_target_context_manager: return self.current_processor.pad(*_a , **_a ) __a = kwargs.pop('''input_features''' , _a ) __a = kwargs.pop('''labels''' , _a ) if len(_a ) > 0: __a = args[0] __a = args[1:] if input_features is not None: __a = self.feature_extractor.pad(_a , *_a , **_a ) if labels is not None: __a = self.tokenizer.pad(_a , **_a ) if labels is None: return input_features elif input_features is None: return labels else: __a = labels['''input_ids'''] return input_features def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.batch_decode(*_a , **_a ) def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.decode(*_a , **_a ) @contextmanager def __UpperCAmelCase ( self ): warnings.warn( '''`as_target_processor` is deprecated and will be removed in v5 of Transformers. You can process your ''' '''labels by using the argument `text` of the regular `__call__` method (either in the same call as ''' '''your audio inputs, or in a separate call.''' ) __a = True __a = self.tokenizer yield __a = self.feature_extractor __a = False
719
"""simple docstring""" import inspect from typing import Optional, Union import numpy as np import PIL import torch from torch.nn import functional as F from torchvision import transforms from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DiffusionPipeline, DPMSolverMultistepScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, randn_tensor, ) def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[str] ) -> Any: if isinstance(lowerCAmelCase__ , torch.Tensor ): return image elif isinstance(lowerCAmelCase__ , PIL.Image.Image ): __a = [image] if isinstance(image[0] , PIL.Image.Image ): __a = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['''lanczos'''] ) )[None, :] for i in image] __a = np.concatenate(lowerCAmelCase__ , axis=0 ) __a = np.array(lowerCAmelCase__ ).astype(np.floataa ) / 2_55.0 __a = image.transpose(0 , 3 , 1 , 2 ) __a = 2.0 * image - 1.0 __a = torch.from_numpy(lowerCAmelCase__ ) elif isinstance(image[0] , torch.Tensor ): __a = torch.cat(lowerCAmelCase__ , dim=0 ) return image def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : str , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Optional[int]=0.99_95 ) -> int: if not isinstance(lowerCAmelCase__ , np.ndarray ): __a = True __a = va.device __a = va.cpu().numpy() __a = va.cpu().numpy() __a = np.sum(va * va / (np.linalg.norm(lowerCAmelCase__ ) * np.linalg.norm(lowerCAmelCase__ )) ) if np.abs(lowerCAmelCase__ ) > DOT_THRESHOLD: __a = (1 - t) * va + t * va else: __a = np.arccos(lowerCAmelCase__ ) __a = np.sin(lowerCAmelCase__ ) __a = theta_a * t __a = np.sin(lowerCAmelCase__ ) __a = np.sin(theta_a - theta_t ) / sin_theta_a __a = sin_theta_t / sin_theta_a __a = sa * va + sa * va if inputs_are_torch: __a = torch.from_numpy(lowerCAmelCase__ ).to(lowerCAmelCase__ ) return va def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : List[Any] ) -> int: __a = F.normalize(lowerCAmelCase__ , dim=-1 ) __a = F.normalize(lowerCAmelCase__ , dim=-1 ) return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Any ) -> List[str]: for param in model.parameters(): __a = value class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' def __init__( self , _a , _a , _a , _a , _a , _a , _a , _a=None , _a=None , _a=None , ): super().__init__() self.register_modules( vae=_a , text_encoder=_a , clip_model=_a , tokenizer=_a , unet=_a , scheduler=_a , feature_extractor=_a , coca_model=_a , coca_tokenizer=_a , coca_transform=_a , ) __a = ( feature_extractor.size if isinstance(feature_extractor.size , _a ) else feature_extractor.size['''shortest_edge'''] ) __a = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std ) set_requires_grad(self.text_encoder , _a ) set_requires_grad(self.clip_model , _a ) def __UpperCAmelCase ( self , _a = "auto" ): if slice_size == "auto": # half the attention head size is usually a good trade-off between # speed and memory __a = self.unet.config.attention_head_dim // 2 self.unet.set_attention_slice(_a ) def __UpperCAmelCase ( self ): self.enable_attention_slicing(_a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.vae , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self ): set_requires_grad(self.unet , _a ) def __UpperCAmelCase ( self , _a , _a , _a ): # get the original timestep using init_timestep __a = min(int(num_inference_steps * strength ) , _a ) __a = max(num_inference_steps - init_timestep , 0 ) __a = self.scheduler.timesteps[t_start:] return timesteps, num_inference_steps - t_start def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a=None ): if not isinstance(_a , torch.Tensor ): raise ValueError(f'''`image` has to be of type `torch.Tensor` but is {type(_a )}''' ) __a = image.to(device=_a , dtype=_a ) if isinstance(_a , _a ): __a = [ self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_a ) ] __a = torch.cat(_a , dim=0 ) else: __a = self.vae.encode(_a ).latent_dist.sample(_a ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 0.1_8215 * init_latents __a = init_latents.repeat_interleave(_a , dim=0 ) __a = randn_tensor(init_latents.shape , generator=_a , device=_a , dtype=_a ) # get latents __a = self.scheduler.add_noise(_a , _a , _a ) __a = init_latents return latents def __UpperCAmelCase ( self , _a ): __a = self.coca_transform(_a ).unsqueeze(0 ) with torch.no_grad(), torch.cuda.amp.autocast(): __a = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) ) __a = self.coca_tokenizer.decode(generated[0].cpu().numpy() ) return generated.split('''<end_of_text>''' )[0].replace('''<start_of_text>''' , '''''' ).rstrip(''' .,''' ) def __UpperCAmelCase ( self , _a , _a ): __a = self.feature_extractor.preprocess(_a ) __a = torch.from_numpy(clip_image_input['''pixel_values'''][0] ).unsqueeze(0 ).to(self.device ).half() __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = image_embeddings_clip.repeat_interleave(_a , dim=0 ) return image_embeddings_clip @torch.enable_grad() def __UpperCAmelCase ( self , _a , _a , _a , _a , _a , _a , _a , ): __a = latents.detach().requires_grad_() __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ): __a = self.scheduler.alphas_cumprod[timestep] __a = 1 - alpha_prod_t # compute predicted original sample from predicted noise also called # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf __a = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5 __a = torch.sqrt(_a ) __a = pred_original_sample * (fac) + latents * (1 - fac) elif isinstance(self.scheduler , _a ): __a = self.scheduler.sigmas[index] __a = latents - sigma * noise_pred else: raise ValueError(f'''scheduler type {type(self.scheduler )} not supported''' ) # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * sample __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = transforms.Resize(self.feature_extractor_size )(_a ) __a = self.normalize(_a ).to(latents.dtype ) __a = self.clip_model.get_image_features(_a ) __a = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_a ) __a = spherical_dist_loss(_a , _a ).mean() * clip_guidance_scale __a = -torch.autograd.grad(_a , _a )[0] if isinstance(self.scheduler , _a ): __a = latents.detach() + grads * (sigma**2) __a = noise_pred_original else: __a = noise_pred_original - torch.sqrt(_a ) * grads return noise_pred, latents @torch.no_grad() def __call__( self , _a , _a , _a = None , _a = None , _a = 512 , _a = 512 , _a = 0.6 , _a = 50 , _a = 7.5 , _a = 1 , _a = 0.0 , _a = 100 , _a = None , _a = "pil" , _a = True , _a = 0.8 , _a = 0.1 , _a = 0.1 , ): if isinstance(_a , _a ) and len(_a ) != batch_size: raise ValueError(f'''You have passed {batch_size} batch_size, but only {len(_a )} generators.''' ) if height % 8 != 0 or width % 8 != 0: raise ValueError(f'''`height` and `width` have to be divisible by 8 but are {height} and {width}.''' ) if isinstance(_a , torch.Generator ) and batch_size > 1: __a = [generator] + [None] * (batch_size - 1) __a = [ ('''model''', self.coca_model is None), ('''tokenizer''', self.coca_tokenizer is None), ('''transform''', self.coca_transform is None), ] __a = [x[0] for x in coca_is_none if x[1]] __a = ''', '''.join(_a ) # generate prompts with coca model if prompt is None if content_prompt is None: if len(_a ): raise ValueError( f'''Content prompt is None and CoCa [{coca_is_none_str}] is None.''' f'''Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) if style_prompt is None: if len(_a ): raise ValueError( f'''Style prompt is None and CoCa [{coca_is_none_str}] is None.''' f''' Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.''' ) __a = self.get_image_description(_a ) # get prompt text embeddings for content and style __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0] __a = self.tokenizer( _a , padding='''max_length''' , max_length=self.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , ) __a = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0] __a = slerp(_a , _a , _a ) # duplicate text embeddings for each generation per prompt __a = text_embeddings.repeat_interleave(_a , dim=0 ) # set timesteps __a = '''offset''' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() ) __a = {} if accepts_offset: __a = 1 self.scheduler.set_timesteps(_a , **_a ) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand self.scheduler.timesteps.to(self.device ) __a , __a = self.get_timesteps(_a , _a , self.device ) __a = timesteps[:1].repeat(_a ) # Preprocess image __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = preprocess(_a , _a , _a ) __a = self.prepare_latents( _a , _a , _a , text_embeddings.dtype , self.device , _a ) __a = slerp(_a , _a , _a ) if clip_guidance_scale > 0: __a = self.get_clip_image_embeddings(_a , _a ) __a = self.get_clip_image_embeddings(_a , _a ) __a = slerp( _a , _a , _a ) # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` # corresponds to doing no classifier free guidance. __a = guidance_scale > 1.0 # get unconditional embeddings for classifier free guidance if do_classifier_free_guidance: __a = content_text_input.input_ids.shape[-1] __a = self.tokenizer([''''''] , padding='''max_length''' , max_length=_a , return_tensors='''pt''' ) __a = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0] # duplicate unconditional embeddings for each generation per prompt __a = uncond_embeddings.repeat_interleave(_a , dim=0 ) # For classifier free guidance, we need to do two forward passes. # Here we concatenate the unconditional and text embeddings into a single batch # to avoid doing two forward passes __a = torch.cat([uncond_embeddings, text_embeddings] ) # get the initial random noise unless the user supplied it # Unlike in other pipelines, latents need to be generated in the target device # for 1-to-1 results reproducibility with the CompVis implementation. # However this currently doesn't work in `mps`. __a = (batch_size, self.unet.config.in_channels, height // 8, width // 8) __a = text_embeddings.dtype if latents is None: if self.device.type == "mps": # randn does not work reproducibly on mps __a = torch.randn(_a , generator=_a , device='''cpu''' , dtype=_a ).to( self.device ) else: __a = torch.randn(_a , generator=_a , device=self.device , dtype=_a ) else: if latents.shape != latents_shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {latents_shape}''' ) __a = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler __a = latents * self.scheduler.init_noise_sigma # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 # and should be between [0, 1] __a = '''eta''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) __a = {} if accepts_eta: __a = eta # check if the scheduler accepts generator __a = '''generator''' in set(inspect.signature(self.scheduler.step ).parameters.keys() ) if accepts_generator: __a = generator with self.progress_bar(total=_a ): for i, t in enumerate(_a ): # expand the latents if we are doing classifier free guidance __a = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents __a = self.scheduler.scale_model_input(_a , _a ) # predict the noise residual __a = self.unet(_a , _a , encoder_hidden_states=_a ).sample # perform classifier free guidance if do_classifier_free_guidance: __a , __a = noise_pred.chunk(2 ) __a = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) # perform clip guidance if clip_guidance_scale > 0: __a = ( text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings ) __a , __a = self.cond_fn( _a , _a , _a , _a , _a , _a , _a , ) # compute the previous noisy sample x_t -> x_t-1 __a = self.scheduler.step(_a , _a , _a , **_a ).prev_sample # Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor __a = 1 / 0.1_8215 * latents __a = self.vae.decode(_a ).sample __a = (image / 2 + 0.5).clamp(0 , 1 ) __a = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": __a = self.numpy_to_pil(_a ) if not return_dict: return (image, None) return StableDiffusionPipelineOutput(images=_a , nsfw_content_detected=_a )
65
0
"""simple docstring""" import pytest from datasets import inspect_metric, list_metrics, load_metric @pytest.fixture def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' , set() ) @pytest.fixture def lowercase ( lowerCAmelCase__ : Any ) -> Any: class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = metric_id class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Any = [MetricMock(__SCREAMING_SNAKE_CASE ) for metric_id in ['accuracy', 'mse', 'precision', 'codeparrot/apps_metric']] def __UpperCAmelCase ( self ): return self._metrics monkeypatch.setattr('''datasets.inspect.huggingface_hub''' , HfhMock() ) @pytest.mark.parametrize( '''func, args''' , [(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Tuple ) -> Optional[int]: if "tmp_path" in args: __a = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args ) with pytest.warns(lowerCAmelCase__ , match='''https://huggingface.co/docs/evaluate''' ): func(*lowerCAmelCase__ )
720
"""simple docstring""" # tests directory-specific settings - this file is run automatically # by pytest before any tests are run import sys import warnings from os.path import abspath, dirname, join # allow having multiple repository checkouts and not needing to remember to rerun # 'pip install -e .[dev]' when switching between checkouts and running tests. lowercase_ = abspath(join(dirname(dirname(dirname(__file__))), "src")) sys.path.insert(1, git_repo_path) # silence FutureWarning warnings in tests since often we can't act on them until # they become normal warnings - i.e. the tests still need to test the current functionality warnings.simplefilter(action="ignore", category=FutureWarning) def lowercase ( lowerCAmelCase__ : List[Any] ) -> str: from transformers.testing_utils import pytest_addoption_shared pytest_addoption_shared(lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : int ) -> Union[str, Any]: from transformers.testing_utils import pytest_terminal_summary_main __a = terminalreporter.config.getoption('''--make-reports''' ) if make_reports: pytest_terminal_summary_main(lowerCAmelCase__ , id=lowerCAmelCase__ )
65
0
"""simple docstring""" from __future__ import annotations from collections import deque from collections.abc import Iterator from dataclasses import dataclass @dataclass class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : int __UpperCAmelCase : int class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = [[] for _ in range(_a )] __a = size def __getitem__( self , _a ): return iter(self._graph[vertex] ) @property def __UpperCAmelCase ( self ): return self._size def __UpperCAmelCase ( self , _a , _a , _a ): if weight not in (0, 1): raise ValueError('''Edge weight must be either 0 or 1.''' ) if to_vertex < 0 or to_vertex >= self.size: raise ValueError('''Vertex indexes must be in [0; size).''' ) self._graph[from_vertex].append(Edge(_a , _a ) ) def __UpperCAmelCase ( self , _a , _a ): __a = deque([start_vertex] ) __a = [None] * self.size __a = 0 while queue: __a = queue.popleft() __a = distances[current_vertex] if current_distance is None: continue for edge in self[current_vertex]: __a = current_distance + edge.weight __a = distances[edge.destination_vertex] if ( isinstance(_a , _a ) and new_distance >= dest_vertex_distance ): continue __a = new_distance if edge.weight == 0: queue.appendleft(edge.destination_vertex ) else: queue.append(edge.destination_vertex ) if distances[finish_vertex] is None: raise ValueError('''No path from start_vertex to finish_vertex.''' ) return distances[finish_vertex] if __name__ == "__main__": import doctest doctest.testmod()
721
"""simple docstring""" 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. lowercase_ = {"LayoutLMv2Config", "LayoutLMv3Config"} @is_pipeline_test class __lowerCAmelCase ( unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : Optional[Any] = MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING __UpperCAmelCase : Any = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING if model_mapping is not None: __UpperCAmelCase : Any = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP} if tf_model_mapping is not None: __UpperCAmelCase : Dict = { config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP } def __UpperCAmelCase ( self , _a , _a , _a ): __a = ZeroShotClassificationPipeline( model=_a , tokenizer=_a , candidate_labels=['''polics''', '''health'''] ) return classifier, ["Who are you voting for in 2020?", "My stomach hurts."] def __UpperCAmelCase ( self , _a , _a ): __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # No kwarg __a = classifier('''Who are you voting for in 2020?''' , ['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics'''] ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels='''politics, public health''' ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier('''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health'''] ) self.assertEqual( _a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} ) self.assertAlmostEqual(sum(nested_simplify(outputs['''scores'''] ) ) , 1.0 ) __a = classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''This text is about {}''' ) self.assertEqual(_a , {'''sequence''': ANY(_a ), '''labels''': [ANY(_a )], '''scores''': [ANY(_a )]} ) # https://github.com/huggingface/transformers/issues/13846 __a = classifier(['''I am happy'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(1 ) ] , ) __a = classifier(['''I am happy''', '''I am sad'''] , ['''positive''', '''negative'''] ) self.assertEqual( _a , [ {'''sequence''': ANY(_a ), '''labels''': [ANY(_a ), ANY(_a )], '''scores''': [ANY(_a ), ANY(_a )]} for i in range(2 ) ] , ) with self.assertRaises(_a ): classifier('''''' , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier(_a , candidate_labels='''politics''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels='''''' ) with self.assertRaises(_a ): classifier('''Who are you voting for in 2020?''' , candidate_labels=_a ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template='''Not formatting template''' , ) with self.assertRaises(_a ): classifier( '''Who are you voting for in 2020?''' , candidate_labels='''politics''' , hypothesis_template=_a , ) self.run_entailment_id(_a ) def __UpperCAmelCase ( self , _a ): __a = zero_shot_classifier.model.config __a = config.labelaid __a = zero_shot_classifier.entailment_id __a = {'''LABEL_0''': 0, '''LABEL_1''': 1, '''LABEL_2''': 2} self.assertEqual(zero_shot_classifier.entailment_id , -1 ) __a = {'''entailment''': 0, '''neutral''': 1, '''contradiction''': 2} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 0, '''NON-ENTAIL''': 1} self.assertEqual(zero_shot_classifier.entailment_id , 0 ) __a = {'''ENTAIL''': 2, '''NEUTRAL''': 1, '''CONTR''': 0} self.assertEqual(zero_shot_classifier.entailment_id , 2 ) __a = original_labelaid self.assertEqual(_a , zero_shot_classifier.entailment_id ) @require_torch def __UpperCAmelCase ( self ): __a = 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 __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''pt''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @require_tf def __UpperCAmelCase ( self ): __a = pipeline( '''zero-shot-classification''' , model='''sshleifer/tiny-distilbert-base-cased-distilled-squad''' , framework='''tf''' , ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''science''', '''public health''', '''politics'''], '''scores''': [0.333, 0.333, 0.333], } , ) @slow @require_torch def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''pt''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , ) @slow @require_tf def __UpperCAmelCase ( self ): __a = pipeline('''zero-shot-classification''' , model='''roberta-large-mnli''' , framework='''tf''' ) __a = zero_shot_classifier( '''Who are you voting for in 2020?''' , candidate_labels=['''politics''', '''public health''', '''science'''] ) self.assertEqual( nested_simplify(_a ) , { '''sequence''': '''Who are you voting for in 2020?''', '''labels''': ['''politics''', '''public health''', '''science'''], '''scores''': [0.976, 0.015, 0.009], } , ) __a = 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=_a , ) self.assertEqual( nested_simplify(_a ) , { '''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''': [0.817, 0.713, 0.018, 0.018], } , )
65
0
"""simple docstring""" import warnings from ...processing_utils import ProcessorMixin from ...tokenization_utils_base import BatchEncoding class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' __UpperCAmelCase : Dict = ['image_processor', 'tokenizer'] __UpperCAmelCase : Optional[int] = 'ChineseCLIPImageProcessor' __UpperCAmelCase : Union[str, Any] = ('BertTokenizer', 'BertTokenizerFast') def __init__( self , _a=None , _a=None , **_a ): __a = None if "feature_extractor" in kwargs: warnings.warn( '''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`''' ''' instead.''' , _a , ) __a = kwargs.pop('''feature_extractor''' ) __a = image_processor if image_processor is not None else feature_extractor if image_processor is None: raise ValueError('''You need to specify an `image_processor`.''' ) if tokenizer is None: raise ValueError('''You need to specify a `tokenizer`.''' ) super().__init__(_a , _a ) __a = self.image_processor def __call__( self , _a=None , _a=None , _a=None , **_a ): 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: __a = self.tokenizer(_a , return_tensors=_a , **_a ) if images is not None: __a = self.image_processor(_a , return_tensors=_a , **_a ) if text is not None and images is not None: __a = image_features.pixel_values return encoding elif text is not None: return encoding else: return BatchEncoding(data=dict(**_a ) , tensor_type=_a ) def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.batch_decode(*_a , **_a ) def __UpperCAmelCase ( self , *_a , **_a ): return self.tokenizer.decode(*_a , **_a ) @property def __UpperCAmelCase ( self ): __a = self.tokenizer.model_input_names __a = self.image_processor.model_input_names return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) ) @property def __UpperCAmelCase ( self ): warnings.warn( '''`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead.''' , _a , ) return self.image_processor_class
700
"""simple docstring""" import pytest from datasets import inspect_metric, list_metrics, load_metric @pytest.fixture def lowercase ( lowerCAmelCase__ : Optional[int] ) -> int: monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' , set() ) @pytest.fixture def lowercase ( lowerCAmelCase__ : Any ) -> Any: class __lowerCAmelCase : '''simple docstring''' def __init__( self , _a ): __a = metric_id class __lowerCAmelCase : '''simple docstring''' __UpperCAmelCase : Any = [MetricMock(__SCREAMING_SNAKE_CASE ) for metric_id in ['accuracy', 'mse', 'precision', 'codeparrot/apps_metric']] def __UpperCAmelCase ( self ): return self._metrics monkeypatch.setattr('''datasets.inspect.huggingface_hub''' , HfhMock() ) @pytest.mark.parametrize( '''func, args''' , [(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Tuple ) -> Optional[int]: if "tmp_path" in args: __a = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args ) with pytest.warns(lowerCAmelCase__ , match='''https://huggingface.co/docs/evaluate''' ): func(*lowerCAmelCase__ )
65
0
"""simple docstring""" from __future__ import annotations import math def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int , lowerCAmelCase__ : bool , lowerCAmelCase__ : list[int] , lowerCAmelCase__ : float ) -> int: if depth < 0: raise ValueError('''Depth cannot be less than 0''' ) if not scores: raise ValueError('''Scores cannot be empty''' ) if depth == height: return scores[node_index] return ( max( minimax(depth + 1 , node_index * 2 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) , minimax(depth + 1 , node_index * 2 + 1 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) , ) if is_max else min( minimax(depth + 1 , node_index * 2 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) , minimax(depth + 1 , node_index * 2 + 1 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ) , ) ) def lowercase ( ) -> None: __a = [90, 23, 6, 33, 21, 65, 123, 34423] __a = math.log(len(lowerCAmelCase__ ) , 2 ) print(f'''Optimal value : {minimax(0 , 0 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ )}''' ) if __name__ == "__main__": import doctest doctest.testmod() main()
701
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int ) -> int: if not isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) or number < 0: raise ValueError('''Input must be a non-negative integer''' ) __a = 0 while number: # This way we arrive at next set bit (next 1) instead of looping # through each bit and checking for 1s hence the # loop won't run 32 times it will only run the number of `1` times number &= number - 1 count += 1 return count if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import argparse import logging from collections import namedtuple import torch from model_bertabs import BertAbsSummarizer from models.model_builder import AbsSummarizer # The authors' implementation from transformers import BertTokenizer logging.basicConfig(level=logging.INFO) lowercase_ : Optional[int] = logging.getLogger(__name__) lowercase_ : Union[str, Any] = "Hello world! cécé herlolip" lowercase_ : int = namedtuple( "BertAbsConfig", [ "temp_dir", "large", "use_bert_emb", "finetune_bert", "encoder", "share_emb", "max_pos", "enc_layers", "enc_hidden_size", "enc_heads", "enc_ff_size", "enc_dropout", "dec_layers", "dec_hidden_size", "dec_heads", "dec_ff_size", "dec_dropout", ], ) def lowercase ( lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : int ) -> Optional[Any]: __a = BertAbsConfig( temp_dir='''.''' , finetune_bert=lowerCAmelCase__ , large=lowerCAmelCase__ , share_emb=lowerCAmelCase__ , use_bert_emb=lowerCAmelCase__ , encoder='''bert''' , max_pos=512 , enc_layers=6 , enc_hidden_size=512 , enc_heads=8 , enc_ff_size=512 , enc_dropout=0.2 , dec_layers=6 , dec_hidden_size=768 , dec_heads=8 , dec_ff_size=2048 , dec_dropout=0.2 , ) __a = torch.load(lowerCAmelCase__ , lambda lowerCAmelCase__ , lowerCAmelCase__ : storage ) __a = AbsSummarizer(lowerCAmelCase__ , torch.device('''cpu''' ) , lowerCAmelCase__ ) original.eval() __a = BertAbsSummarizer(lowerCAmelCase__ , torch.device('''cpu''' ) ) new_model.eval() # ------------------- # Convert the weights # ------------------- logging.info('''convert the model''' ) new_model.bert.load_state_dict(original.bert.state_dict() ) new_model.decoder.load_state_dict(original.decoder.state_dict() ) new_model.generator.load_state_dict(original.generator.state_dict() ) # ---------------------------------- # Make sure the outpus are identical # ---------------------------------- logging.info('''Make sure that the models\' outputs are identical''' ) __a = BertTokenizer.from_pretrained('''bert-base-uncased''' ) # prepare the model inputs __a = tokenizer.encode('''This is sample éàalj\'-.''' ) encoder_input_ids.extend([tokenizer.pad_token_id] * (512 - len(lowerCAmelCase__ )) ) __a = torch.tensor(lowerCAmelCase__ ).unsqueeze(0 ) __a = tokenizer.encode('''This is sample 3 éàalj\'-.''' ) decoder_input_ids.extend([tokenizer.pad_token_id] * (512 - len(lowerCAmelCase__ )) ) __a = torch.tensor(lowerCAmelCase__ ).unsqueeze(0 ) # failsafe to make sure the weights reset does not affect the # loaded weights. assert torch.max(torch.abs(original.generator[0].weight - new_model.generator[0].weight ) ) == 0 # forward pass __a = encoder_input_ids __a = decoder_input_ids __a = __a = None __a = None __a = __a = None __a = __a = None __a = None # The original model does not apply the geneator layer immediatly but rather in # the beam search (where it combines softmax + linear layer). Since we already # apply the softmax in our generation process we only apply the linear layer here. # We make sure that the outputs of the full stack are identical __a = original(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ )[0] __a = original.generator(lowerCAmelCase__ ) __a = new_model( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ )[0] __a = new_model.generator(lowerCAmelCase__ ) __a = torch.max(torch.abs(output_converted_model - output_original_model ) ).item() print('''Maximum absolute difference beween weights: {:.2f}'''.format(lowerCAmelCase__ ) ) __a = torch.max(torch.abs(output_converted_generator - output_original_generator ) ).item() print('''Maximum absolute difference beween weights: {:.2f}'''.format(lowerCAmelCase__ ) ) __a = torch.allclose(lowerCAmelCase__ , lowerCAmelCase__ , atol=1e-3 ) if are_identical: logging.info('''all weights are equal up to 1e-3''' ) else: raise ValueError('''the weights are different. The new model is likely different from the original one.''' ) # The model has been saved with torch.save(model) and this is bound to the exact # directory structure. We save the state_dict instead. logging.info('''saving the model\'s state dictionary''' ) torch.save( new_model.state_dict() , '''./bertabs-finetuned-cnndm-extractive-abstractive-summarization/pytorch_model.bin''' ) if __name__ == "__main__": lowercase_ : Union[str, Any] = argparse.ArgumentParser() parser.add_argument( "--bertabs_checkpoint_path", default=None, type=str, required=True, help="Path the official PyTorch dump.", ) parser.add_argument( "--pytorch_dump_folder_path", default=None, type=str, required=True, help="Path to the output PyTorch model.", ) lowercase_ : Optional[int] = parser.parse_args() convert_bertabs_checkpoints( args.bertabs_checkpoint_path, args.pytorch_dump_folder_path, )
702
"""simple docstring""" import pytest from datasets import Dataset, DatasetDict, Features, NamedSplit, Value from datasets.io.text import TextDatasetReader from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) assert dataset.split == split if split else "train" @pytest.mark.parametrize('''path_type''' , [str, list] ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict: if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = text_path elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ): __a = [text_path] __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ ) def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]: assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) for split in splits: __a = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 1 assert dataset.column_names == ["text"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize('''keep_in_memory''' , [False, True] ) def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]: __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): __a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize( '''features''' , [ None, {'''text''': '''string'''}, {'''text''': '''int32'''}, {'''text''': '''float32'''}, ] , ) def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str: __a = tmp_path / '''cache''' # CSV file loses col_1 string dtype information: default now is "int64" instead of "string" __a = {'''text''': '''string'''} __a = features.copy() if features else default_expected_features __a = ( Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None ) __a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ ) @pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] ) def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict: if split: __a = {split: text_path} else: __a = '''train''' __a = {'''train''': text_path, '''test''': text_path} __a = tmp_path / '''cache''' __a = {'''text''': '''string'''} __a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read() _check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() )
65
0
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available lowercase_ = { "configuration_bridgetower": [ "BRIDGETOWER_PRETRAINED_CONFIG_ARCHIVE_MAP", "BridgeTowerConfig", "BridgeTowerTextConfig", "BridgeTowerVisionConfig", ], "processing_bridgetower": ["BridgeTowerProcessor"], } try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = ["BridgeTowerImageProcessor"] try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: lowercase_ = [ "BRIDGETOWER_PRETRAINED_MODEL_ARCHIVE_LIST", "BridgeTowerForContrastiveLearning", "BridgeTowerForImageAndTextRetrieval", "BridgeTowerForMaskedLM", "BridgeTowerModel", "BridgeTowerPreTrainedModel", ] if TYPE_CHECKING: from .configuration_bridgetower import ( BRIDGETOWER_PRETRAINED_CONFIG_ARCHIVE_MAP, BridgeTowerConfig, BridgeTowerTextConfig, BridgeTowerVisionConfig, ) from .processing_bridgetower import BridgeTowerProcessor try: if not is_vision_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .image_processing_bridgetower import BridgeTowerImageProcessor try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_bridgetower import ( BRIDGETOWER_PRETRAINED_MODEL_ARCHIVE_LIST, BridgeTowerForContrastiveLearning, BridgeTowerForImageAndTextRetrieval, BridgeTowerForMaskedLM, BridgeTowerModel, BridgeTowerPreTrainedModel, ) else: import sys lowercase_ = _LazyModule(__name__, globals()["__file__"], _import_structure)
703
"""simple docstring""" def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : int ) -> bool: return numa ^ numa < 0 if __name__ == "__main__": import doctest doctest.testmod()
65
0
"""simple docstring""" import copy from typing import TYPE_CHECKING, Any, Mapping, Optional, OrderedDict from packaging import version from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfig from ...utils import logging from ..auto.configuration_auto import AutoConfig if TYPE_CHECKING: from ... import PreTrainedTokenizerBase, TensorType lowercase_ = logging.get_logger(__name__) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' snake_case : str = 'vision-encoder-decoder' snake_case : Union[str, Any] = True def __init__( self , **_a ): super().__init__(**_a ) if "encoder" not in kwargs or "decoder" not in kwargs: raise ValueError( f'''A configuraton of type {self.model_type} cannot be instantiated because ''' f'''not both `encoder` and `decoder` sub-configurations are passed, but only {kwargs}''' ) __a = kwargs.pop('''encoder''' ) __a = encoder_config.pop('''model_type''' ) __a = kwargs.pop('''decoder''' ) __a = decoder_config.pop('''model_type''' ) __a = AutoConfig.for_model(_a , **_a ) __a = AutoConfig.for_model(_a , **_a ) __a = True @classmethod def __UpperCAmelCase ( cls , _a , _a , **_a ): logger.info('''Setting `config.is_decoder=True` and `config.add_cross_attention=True` for decoder_config''' ) __a = True __a = True return cls(encoder=encoder_config.to_dict() , decoder=decoder_config.to_dict() , **_a ) def __UpperCAmelCase ( self ): __a = copy.deepcopy(self.__dict__ ) __a = self.encoder.to_dict() __a = self.decoder.to_dict() __a = self.__class__.model_type return output class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' snake_case : int = version.parse('1.11' ) @property def __UpperCAmelCase ( self ): return OrderedDict( [ ('''pixel_values''', {0: '''batch''', 1: '''num_channels''', 2: '''height''', 3: '''width'''}), ] ) @property def __UpperCAmelCase ( self ): return 1E-4 @property def __UpperCAmelCase ( self ): return OrderedDict({'''last_hidden_state''': {0: '''batch''', 1: '''encoder_sequence'''}} ) class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' @property def __UpperCAmelCase ( self ): __a = OrderedDict() __a = {0: '''batch''', 1: '''past_decoder_sequence + sequence'''} __a = {0: '''batch''', 1: '''past_decoder_sequence + sequence'''} __a = {0: '''batch''', 1: '''encoder_sequence'''} return common_inputs def __UpperCAmelCase ( self , _a , _a = -1 , _a = -1 , _a = False , _a = None , ): import torch __a = OrderedDict() __a = super().generate_dummy_inputs( _a , batch_size=_a , seq_length=_a , is_pair=_a , framework=_a ) __a , __a = dummy_input['''input_ids'''].shape __a = (batch, encoder_sequence, self._config.encoder_hidden_size) __a = dummy_input.pop('''input_ids''' ) __a = dummy_input.pop('''attention_mask''' ) __a = torch.zeros(_a ) return common_inputs class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ): '''simple docstring''' @property def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self , _a ): return VisionEncoderDecoderEncoderOnnxConfig(_a ) def __UpperCAmelCase ( self , _a , _a , _a = "default" ): __a = encoder_config.hidden_size return VisionEncoderDecoderDecoderOnnxConfig(_a , _a )
704
"""simple docstring""" import unittest from transformers import SPIECE_UNDERLINE from transformers.models.speechta import SpeechTaTokenizer from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow from transformers.tokenization_utils import AddedToken from ...test_tokenization_common import TokenizerTesterMixin lowercase_ = get_tests_dir("fixtures/test_sentencepiece_bpe_char.model") @require_sentencepiece @require_tokenizers class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE , unittest.TestCase ): '''simple docstring''' __UpperCAmelCase : int = SpeechTaTokenizer __UpperCAmelCase : Tuple = False __UpperCAmelCase : Dict = True def __UpperCAmelCase ( self ): super().setUp() # We have a SentencePiece fixture for testing __a = SpeechTaTokenizer(_a ) __a = AddedToken('''<mask>''' , lstrip=_a , rstrip=_a ) __a = mask_token tokenizer.add_special_tokens({'''mask_token''': mask_token} ) tokenizer.add_tokens(['''<ctc_blank>'''] ) tokenizer.save_pretrained(self.tmpdirname ) def __UpperCAmelCase ( self , _a ): __a = '''this is a test''' __a = '''this is a test''' return input_text, output_text def __UpperCAmelCase ( self , _a , _a=False , _a=20 , _a=5 ): __a , __a = self.get_input_output_texts(_a ) __a = tokenizer.encode(_a , add_special_tokens=_a ) __a = tokenizer.decode(_a , clean_up_tokenization_spaces=_a ) return text, ids def __UpperCAmelCase ( self ): __a = '''<pad>''' __a = 1 self.assertEqual(self.get_tokenizer()._convert_token_to_id(_a ) , _a ) self.assertEqual(self.get_tokenizer()._convert_id_to_token(_a ) , _a ) def __UpperCAmelCase ( self ): __a = list(self.get_tokenizer().get_vocab().keys() ) self.assertEqual(vocab_keys[0] , '''<s>''' ) self.assertEqual(vocab_keys[1] , '''<pad>''' ) self.assertEqual(vocab_keys[-4] , '''œ''' ) self.assertEqual(vocab_keys[-2] , '''<mask>''' ) self.assertEqual(vocab_keys[-1] , '''<ctc_blank>''' ) self.assertEqual(len(_a ) , 81 ) def __UpperCAmelCase ( self ): self.assertEqual(self.get_tokenizer().vocab_size , 79 ) def __UpperCAmelCase ( self ): __a = self.get_tokenizers(do_lower_case=_a ) for tokenizer in tokenizers: with self.subTest(f'''{tokenizer.__class__.__name__}''' ): __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) # We usually have added tokens from the start in tests because our vocab fixtures are # smaller than the original vocabs - let's not assert this # self.assertEqual(vocab_size, all_size) __a = ['''aaaaa bbbbbb''', '''cccccccccdddddddd'''] __a = tokenizer.add_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size + len(_a ) ) __a = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 4 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) __a = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''} __a = tokenizer.add_special_tokens(_a ) __a = tokenizer.vocab_size __a = len(_a ) self.assertNotEqual(_a , 0 ) self.assertEqual(_a , _a ) self.assertEqual(_a , len(_a ) ) self.assertEqual(_a , all_size_a + len(_a ) ) __a = tokenizer.encode( '''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=_a ) self.assertGreaterEqual(len(_a ) , 6 ) self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[0] , tokens[1] ) self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 ) self.assertGreater(tokens[-3] , tokens[-4] ) self.assertEqual(tokens[0] , tokenizer.eos_token_id ) self.assertEqual(tokens[-3] , tokenizer.pad_token_id ) def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): pass def __UpperCAmelCase ( self ): __a = self.get_tokenizer() __a = tokenizer.tokenize('''This is a test''' ) # fmt: off self.assertListEqual(_a , [SPIECE_UNDERLINE, '''T''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''a''', SPIECE_UNDERLINE, '''t''', '''e''', '''s''', '''t'''] ) # fmt: on self.assertListEqual( tokenizer.convert_tokens_to_ids(_a ) , [4, 32, 11, 10, 12, 4, 10, 12, 4, 7, 4, 6, 5, 12, 6] , ) __a = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''92000''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) __a = tokenizer.convert_tokens_to_ids(_a ) # fmt: off self.assertListEqual(_a , [4, 30, 4, 20, 7, 12, 4, 25, 8, 13, 9, 4, 10, 9, 4, 3, 23, 4, 7, 9, 14, 4, 6, 11, 10, 12, 4, 10, 12, 4, 19, 7, 15, 12, 73, 26] ) # fmt: on __a = tokenizer.convert_ids_to_tokens(_a ) self.assertListEqual( _a , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''<unk>''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] ) @slow def __UpperCAmelCase ( self ): # Use custom sequence because this tokenizer does not handle numbers. __a = [ '''Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides ''' '''general-purpose architectures (BERT, GPT, RoBERTa, XLM, DistilBert, XLNet...) for Natural ''' '''Language Understanding (NLU) and Natural Language Generation (NLG) with over thirty-two pretrained ''' '''models in one hundred plus languages and deep interoperability between Jax, PyTorch and TensorFlow.''', '''BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly ''' '''conditioning on both left and right context in all layers.''', '''The quick brown fox jumps over the lazy dog.''', ] # fmt: off __a = { '''input_ids''': [ [4, 32, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 64, 19, 8, 13, 18, 5, 13, 15, 22, 4, 28, 9, 8, 20, 9, 4, 7, 12, 4, 24, 22, 6, 8, 13, 17, 11, 39, 6, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 7, 9, 14, 4, 24, 22, 6, 8, 13, 17, 11, 39, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 39, 25, 5, 13, 6, 63, 4, 24, 13, 8, 27, 10, 14, 5, 12, 4, 21, 5, 9, 5, 13, 7, 15, 39, 24, 16, 13, 24, 8, 12, 5, 4, 7, 13, 17, 11, 10, 6, 5, 17, 6, 16, 13, 5, 12, 4, 64, 40, 47, 54, 32, 23, 4, 53, 49, 32, 23, 4, 54, 8, 40, 47, 54, 32, 7, 23, 4, 69, 52, 43, 23, 4, 51, 10, 12, 6, 10, 15, 40, 5, 13, 6, 23, 4, 69, 52, 48, 5, 6, 26, 26, 26, 63, 4, 19, 8, 13, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 61, 9, 14, 5, 13, 12, 6, 7, 9, 14, 10, 9, 21, 4, 64, 48, 52, 61, 63, 4, 7, 9, 14, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 53, 5, 9, 5, 13, 7, 6, 10, 8, 9, 4, 64, 48, 52, 53, 63, 4, 20, 10, 6, 11, 4, 8, 27, 5, 13, 4, 6, 11, 10, 13, 6, 22, 39, 6, 20, 8, 4, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 4, 18, 8, 14, 5, 15, 12, 4, 10, 9, 4, 8, 9, 5, 4, 11, 16, 9, 14, 13, 5, 14, 4, 24, 15, 16, 12, 4, 15, 7, 9, 21, 16, 7, 21, 5, 12, 4, 7, 9, 14, 4, 14, 5, 5, 24, 4, 10, 9, 6, 5, 13, 8, 24, 5, 13, 7, 25, 10, 15, 10, 6, 22, 4, 25, 5, 6, 20, 5, 5, 9, 4, 58, 7, 37, 23, 4, 49, 22, 32, 8, 13, 17, 11, 4, 7, 9, 14, 4, 32, 5, 9, 12, 8, 13, 55, 15, 8, 20, 26, 2], [4, 40, 47, 54, 32, 4, 10, 12, 4, 14, 5, 12, 10, 21, 9, 5, 14, 4, 6, 8, 4, 24, 13, 5, 39, 6, 13, 7, 10, 9, 4, 14, 5, 5, 24, 4, 25, 10, 14, 10, 13, 5, 17, 6, 10, 8, 9, 7, 15, 4, 13, 5, 24, 13, 5, 12, 5, 9, 6, 7, 6, 10, 8, 9, 12, 4, 19, 13, 8, 18, 4, 16, 9, 15, 7, 25, 5, 15, 5, 14, 4, 6, 5, 37, 6, 4, 25, 22, 4, 46, 8, 10, 9, 6, 15, 22, 4, 17, 8, 9, 14, 10, 6, 10, 8, 9, 10, 9, 21, 4, 8, 9, 4, 25, 8, 6, 11, 4, 15, 5, 19, 6, 4, 7, 9, 14, 4, 13, 10, 21, 11, 6, 4, 17, 8, 9, 6, 5, 37, 6, 4, 10, 9, 4, 7, 15, 15, 4, 15, 7, 22, 5, 13, 12, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [4, 32, 11, 5, 4, 45, 16, 10, 17, 28, 4, 25, 13, 8, 20, 9, 4, 19, 8, 37, 4, 46, 16, 18, 24, 12, 4, 8, 27, 5, 13, 4, 6, 11, 5, 4, 15, 7, 57, 22, 4, 14, 8, 21, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], ], '''attention_mask''': [ [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], ] } # fmt: on self.tokenizer_integration_test_util( expected_encoding=_a , model_name='''microsoft/speecht5_asr''' , revision='''c5ef64c71905caeccde0e4462ef3f9077224c524''' , sequences=_a , )
65
0