code
stringlengths
82
54.1k
code_codestyle
int64
0
699
style_context
stringlengths
111
35.6k
style_context_codestyle
int64
0
699
label
int64
0
1
def lowerCamelCase__ ( lowercase = 100 ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = (n * (n + 1) // 2) ** 2 SCREAMING_SNAKE_CASE : Dict = n * (n + 1) * (2 * n + 1) // 6 return sum_cubes - sum_squares if __name__ == "__main__": print(F"""{solution() = }""")
62
from collections import Counter import numpy as np from sklearn import datasets from sklearn.model_selection import train_test_split snake_case = datasets.load_iris() snake_case = np.array(data["""data"""]) snake_case = np.array(data["""target"""]) snake_case = data["""target_names"""] snake_case , snake_case , snake_case , snake_case = train_test_split(X, y) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return np.linalg.norm(np.array(lowercase ) - np.array(lowercase ) ) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase=5 ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = zip(lowercase , lowercase ) # List of distances of all points from the point to be classified SCREAMING_SNAKE_CASE : Optional[int] = [] for data_point in data: SCREAMING_SNAKE_CASE : Tuple = euclidean_distance(data_point[0] , lowercase ) distances.append((distance, data_point[1]) ) # Choosing 'k' points with the least distances. SCREAMING_SNAKE_CASE : List[Any] = [i[1] for i in sorted(lowercase )[:k]] # Most commonly occurring class among them # is the class into which the point is classified SCREAMING_SNAKE_CASE : List[Any] = Counter(lowercase ).most_common(1 )[0][0] return classes[result] if __name__ == "__main__": print(classifier(X_train, y_train, classes, [4.4, 3.1, 1.3, 1.4]))
62
1
import os from shutil import copyfile from typing import Any, Dict, List, Optional, Tuple import sentencepiece as spm from ...tokenization_utils import AddedToken, BatchEncoding, PreTrainedTokenizer from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = """▁""" snake_case = {"""vocab_file""": """sentencepiece.bpe.model"""} snake_case = { """vocab_file""": { """facebook/nllb-200-distilled-600M""": ( """https://huggingface.co/facebook/nllb-200-distilled-600M/blob/main/sentencepiece.bpe.model""" ), } } snake_case = { """facebook/nllb-200-distilled-600M""": 1_024, } # fmt: off snake_case = ["""ace_Arab""", """ace_Latn""", """acm_Arab""", """acq_Arab""", """aeb_Arab""", """afr_Latn""", """ajp_Arab""", """aka_Latn""", """amh_Ethi""", """apc_Arab""", """arb_Arab""", """ars_Arab""", """ary_Arab""", """arz_Arab""", """asm_Beng""", """ast_Latn""", """awa_Deva""", """ayr_Latn""", """azb_Arab""", """azj_Latn""", """bak_Cyrl""", """bam_Latn""", """ban_Latn""", """bel_Cyrl""", """bem_Latn""", """ben_Beng""", """bho_Deva""", """bjn_Arab""", """bjn_Latn""", """bod_Tibt""", """bos_Latn""", """bug_Latn""", """bul_Cyrl""", """cat_Latn""", """ceb_Latn""", """ces_Latn""", """cjk_Latn""", """ckb_Arab""", """crh_Latn""", """cym_Latn""", """dan_Latn""", """deu_Latn""", """dik_Latn""", """dyu_Latn""", """dzo_Tibt""", """ell_Grek""", """eng_Latn""", """epo_Latn""", """est_Latn""", """eus_Latn""", """ewe_Latn""", """fao_Latn""", """pes_Arab""", """fij_Latn""", """fin_Latn""", """fon_Latn""", """fra_Latn""", """fur_Latn""", """fuv_Latn""", """gla_Latn""", """gle_Latn""", """glg_Latn""", """grn_Latn""", """guj_Gujr""", """hat_Latn""", """hau_Latn""", """heb_Hebr""", """hin_Deva""", """hne_Deva""", """hrv_Latn""", """hun_Latn""", """hye_Armn""", """ibo_Latn""", """ilo_Latn""", """ind_Latn""", """isl_Latn""", """ita_Latn""", """jav_Latn""", """jpn_Jpan""", """kab_Latn""", """kac_Latn""", """kam_Latn""", """kan_Knda""", """kas_Arab""", """kas_Deva""", """kat_Geor""", """knc_Arab""", """knc_Latn""", """kaz_Cyrl""", """kbp_Latn""", """kea_Latn""", """khm_Khmr""", """kik_Latn""", """kin_Latn""", """kir_Cyrl""", """kmb_Latn""", """kon_Latn""", """kor_Hang""", """kmr_Latn""", """lao_Laoo""", """lvs_Latn""", """lij_Latn""", """lim_Latn""", """lin_Latn""", """lit_Latn""", """lmo_Latn""", """ltg_Latn""", """ltz_Latn""", """lua_Latn""", """lug_Latn""", """luo_Latn""", """lus_Latn""", """mag_Deva""", """mai_Deva""", """mal_Mlym""", """mar_Deva""", """min_Latn""", """mkd_Cyrl""", """plt_Latn""", """mlt_Latn""", """mni_Beng""", """khk_Cyrl""", """mos_Latn""", """mri_Latn""", """zsm_Latn""", """mya_Mymr""", """nld_Latn""", """nno_Latn""", """nob_Latn""", """npi_Deva""", """nso_Latn""", """nus_Latn""", """nya_Latn""", """oci_Latn""", """gaz_Latn""", """ory_Orya""", """pag_Latn""", """pan_Guru""", """pap_Latn""", """pol_Latn""", """por_Latn""", """prs_Arab""", """pbt_Arab""", """quy_Latn""", """ron_Latn""", """run_Latn""", """rus_Cyrl""", """sag_Latn""", """san_Deva""", """sat_Beng""", """scn_Latn""", """shn_Mymr""", """sin_Sinh""", """slk_Latn""", """slv_Latn""", """smo_Latn""", """sna_Latn""", """snd_Arab""", """som_Latn""", """sot_Latn""", """spa_Latn""", """als_Latn""", """srd_Latn""", """srp_Cyrl""", """ssw_Latn""", """sun_Latn""", """swe_Latn""", """swh_Latn""", """szl_Latn""", """tam_Taml""", """tat_Cyrl""", """tel_Telu""", """tgk_Cyrl""", """tgl_Latn""", """tha_Thai""", """tir_Ethi""", """taq_Latn""", """taq_Tfng""", """tpi_Latn""", """tsn_Latn""", """tso_Latn""", """tuk_Latn""", """tum_Latn""", """tur_Latn""", """twi_Latn""", """tzm_Tfng""", """uig_Arab""", """ukr_Cyrl""", """umb_Latn""", """urd_Arab""", """uzn_Latn""", """vec_Latn""", """vie_Latn""", """war_Latn""", """wol_Latn""", """xho_Latn""", """ydd_Hebr""", """yor_Latn""", """yue_Hant""", """zho_Hans""", """zho_Hant""", """zul_Latn"""] class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = VOCAB_FILES_NAMES UpperCamelCase_ : int = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : List[str] = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Dict = ['''input_ids''', '''attention_mask'''] UpperCamelCase_ : List[int] = [] UpperCamelCase_ : List[int] = [] def __init__( self : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int="<s>" , UpperCAmelCase_ : Any="</s>" , UpperCAmelCase_ : Tuple="</s>" , UpperCAmelCase_ : Union[str, Any]="<s>" , UpperCAmelCase_ : str="<unk>" , UpperCAmelCase_ : Union[str, Any]="<pad>" , UpperCAmelCase_ : Dict="<mask>" , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Optional[int]=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Optional[Dict[str, Any]] = None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : Optional[Any] , ): # Mask token behave like a normal word, i.e. include the space before it SCREAMING_SNAKE_CASE : Union[str, Any] = AddedToken(UpperCAmelCase_ , lstrip=UpperCAmelCase_ , rstrip=UpperCAmelCase_ ) if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) else mask_token SCREAMING_SNAKE_CASE : Tuple = {} if sp_model_kwargs is None else sp_model_kwargs SCREAMING_SNAKE_CASE : str = legacy_behaviour super().__init__( bos_token=UpperCAmelCase_ , eos_token=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , src_lang=UpperCAmelCase_ , tgt_lang=UpperCAmelCase_ , additional_special_tokens=UpperCAmelCase_ , sp_model_kwargs=self.sp_model_kwargs , legacy_behaviour=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = spm.SentencePieceProcessor(**self.sp_model_kwargs ) self.sp_model.Load(str(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = vocab_file # Original fairseq vocab and spm vocab must be "aligned": # Vocab | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 # -------- | ------- | ------- | ------ | ------- | ---- | ---- | ---- | ---- | ---- | ---- # fairseq | '<s>' | '<pad>' | '</s>' | '<unk>' | 'an' | '▁n' | '▁m' | '▁t' | '▁k' | '▁a' # spm | '<unk>' | '<s>' | '</s>' | 'an' | '▁n' | '▁m' | '▁t' | '▁k' | '▁a' | '▁s' # Mimic fairseq token-to-id alignment for the first 4 token SCREAMING_SNAKE_CASE : Any = {"<s>": 0, "<pad>": 1, "</s>": 2, "<unk>": 3} # The first "real" token "," has position 4 in the original fairseq vocab and position 3 in the spm vocab SCREAMING_SNAKE_CASE : Union[str, Any] = 1 SCREAMING_SNAKE_CASE : List[Any] = len(self.sp_model ) SCREAMING_SNAKE_CASE : Union[str, Any] = { code: self.sp_model_size + i + self.fairseq_offset for i, code in enumerate(UpperCAmelCase_ ) } SCREAMING_SNAKE_CASE : Optional[Any] = {v: k for k, v in self.lang_code_to_id.items()} SCREAMING_SNAKE_CASE : str = len(self.sp_model ) + len(self.lang_code_to_id ) + self.fairseq_offset self.fairseq_tokens_to_ids.update(self.lang_code_to_id ) SCREAMING_SNAKE_CASE : Dict = {v: k for k, v in self.fairseq_tokens_to_ids.items()} SCREAMING_SNAKE_CASE : List[str] = list(self.lang_code_to_id.keys() ) if additional_special_tokens is not None: # Only add those special tokens if they are not already there. self._additional_special_tokens.extend( [t for t in additional_special_tokens if t not in self._additional_special_tokens] ) SCREAMING_SNAKE_CASE : int = src_lang if src_lang is not None else "eng_Latn" SCREAMING_SNAKE_CASE : Tuple = self.lang_code_to_id[self._src_lang] SCREAMING_SNAKE_CASE : Optional[int] = tgt_lang self.set_src_lang_special_tokens(self._src_lang ) def __getstate__( self : Any ): SCREAMING_SNAKE_CASE : Dict = self.__dict__.copy() SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : str = self.sp_model.serialized_model_proto() return state def __setstate__( self : Optional[Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Tuple = d # for backward compatibility if not hasattr(self , "sp_model_kwargs" ): SCREAMING_SNAKE_CASE : Any = {} SCREAMING_SNAKE_CASE : Optional[Any] = spm.SentencePieceProcessor(**self.sp_model_kwargs ) self.sp_model.LoadFromSerializedProto(self.sp_model_proto ) @property def _A ( self : Optional[int] ): return len(self.sp_model ) + len(self.lang_code_to_id ) + self.fairseq_offset + 1 # Plus 1 for the mask token @property def _A ( self : List[Any] ): return self._src_lang @src_lang.setter def _A ( self : Optional[int] , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Optional[int] = new_src_lang self.set_src_lang_special_tokens(self._src_lang ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None , UpperCAmelCase_ : bool = False ): if already_has_special_tokens: return super().get_special_tokens_mask( token_ids_a=UpperCAmelCase_ , token_ids_a=UpperCAmelCase_ , already_has_special_tokens=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [1] * len(self.prefix_tokens ) SCREAMING_SNAKE_CASE : List[str] = [1] * len(self.suffix_tokens ) if token_ids_a is None: return prefix_ones + ([0] * len(UpperCAmelCase_ )) + suffix_ones return prefix_ones + ([0] * len(UpperCAmelCase_ )) + ([0] * len(UpperCAmelCase_ )) + suffix_ones def _A ( self : Optional[int] , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): if token_ids_a is None: return self.prefix_tokens + token_ids_a + self.suffix_tokens # We don't expect to process pairs, but leave the pair logic for API consistency return self.prefix_tokens + token_ids_a + token_ids_a + self.suffix_tokens def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : Dict = [self.sep_token_id] SCREAMING_SNAKE_CASE : str = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0] def _A ( self : Dict , UpperCAmelCase_ : int , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] , UpperCAmelCase_ : Optional[str] , **UpperCAmelCase_ : List[str] ): if src_lang is None or tgt_lang is None: raise ValueError("Translation requires a `src_lang` and a `tgt_lang` for this model" ) SCREAMING_SNAKE_CASE : Optional[Any] = src_lang SCREAMING_SNAKE_CASE : Tuple = self(UpperCAmelCase_ , add_special_tokens=UpperCAmelCase_ , return_tensors=UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.convert_tokens_to_ids(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = tgt_lang_id return inputs def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = {self.convert_ids_to_tokens(UpperCAmelCase_ ): i for i in range(self.vocab_size )} vocab.update(self.added_tokens_encoder ) return vocab def _A ( self : int , UpperCAmelCase_ : str ): return self.sp_model.encode(UpperCAmelCase_ , out_type=UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : Optional[Any] ): if token in self.fairseq_tokens_to_ids: return self.fairseq_tokens_to_ids[token] SCREAMING_SNAKE_CASE : Any = self.sp_model.PieceToId(UpperCAmelCase_ ) # Need to return unknown token if the SP model returned 0 return spm_id + self.fairseq_offset if spm_id else self.unk_token_id def _A ( self : List[str] , UpperCAmelCase_ : Optional[Any] ): if index in self.fairseq_ids_to_tokens: return self.fairseq_ids_to_tokens[index] return self.sp_model.IdToPiece(index - self.fairseq_offset ) def _A ( self : Dict , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : Optional[int] = "".join(UpperCAmelCase_ ).replace(UpperCAmelCase_ , " " ).strip() return out_string def _A ( self : Any , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): if not os.path.isdir(UpperCAmelCase_ ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return SCREAMING_SNAKE_CASE : Any = os.path.join( UpperCAmelCase_ , (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"] ) if os.path.abspath(self.vocab_file ) != os.path.abspath(UpperCAmelCase_ ) and os.path.isfile(self.vocab_file ): copyfile(self.vocab_file , UpperCAmelCase_ ) elif not os.path.isfile(self.vocab_file ): with open(UpperCAmelCase_ , "wb" ) as fi: SCREAMING_SNAKE_CASE : Dict = self.sp_model.serialized_model_proto() fi.write(UpperCAmelCase_ ) return (out_vocab_file,) def _A ( self : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : str = "eng_Latn" , UpperCAmelCase_ : Optional[List[str]] = None , UpperCAmelCase_ : str = "fra_Latn" , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : str = src_lang SCREAMING_SNAKE_CASE : Optional[Any] = tgt_lang return super().prepare_seqaseq_batch(UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : List[str] ): return self.set_src_lang_special_tokens(self.src_lang ) def _A ( self : List[Any] ): return self.set_tgt_lang_special_tokens(self.tgt_lang ) def _A ( self : Dict , UpperCAmelCase_ : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = self.lang_code_to_id[src_lang] if self.legacy_behaviour: SCREAMING_SNAKE_CASE : Optional[Any] = [] SCREAMING_SNAKE_CASE : str = [self.eos_token_id, self.cur_lang_code] else: SCREAMING_SNAKE_CASE : Tuple = [self.cur_lang_code] SCREAMING_SNAKE_CASE : Tuple = [self.eos_token_id] def _A ( self : List[str] , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : List[Any] = self.lang_code_to_id[lang] if self.legacy_behaviour: SCREAMING_SNAKE_CASE : List[Any] = [] SCREAMING_SNAKE_CASE : List[str] = [self.eos_token_id, self.cur_lang_code] else: SCREAMING_SNAKE_CASE : str = [self.cur_lang_code] SCREAMING_SNAKE_CASE : Tuple = [self.eos_token_id]
62
from typing import Callable, Optional, Union from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """microsoft/xprophetnet-large-wiki100-cased""": ( """https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Tuple = '''xlm-prophetnet''' UpperCamelCase_ : Tuple = ['''past_key_values'''] UpperCamelCase_ : int = { '''num_attention_heads''': '''num_encoder_attention_heads''', } def __init__( self : Dict , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[Union[str, Callable]] = "gelu" , UpperCAmelCase_ : Optional[int] = 3_0522 , UpperCAmelCase_ : Optional[int] = 1024 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[int] = 512 , UpperCAmelCase_ : Optional[float] = 0.02 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 2 , UpperCAmelCase_ : Optional[int] = 32 , UpperCAmelCase_ : Optional[int] = 128 , UpperCAmelCase_ : Optional[bool] = False , UpperCAmelCase_ : Optional[float] = 0.0 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 1 , UpperCAmelCase_ : Optional[int] = 2 , **UpperCAmelCase_ : Dict , ): SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : str = hidden_size SCREAMING_SNAKE_CASE : Tuple = encoder_ffn_dim SCREAMING_SNAKE_CASE : List[Any] = num_encoder_layers SCREAMING_SNAKE_CASE : Any = num_encoder_attention_heads SCREAMING_SNAKE_CASE : Dict = decoder_ffn_dim SCREAMING_SNAKE_CASE : List[str] = num_decoder_layers SCREAMING_SNAKE_CASE : List[Any] = num_decoder_attention_heads SCREAMING_SNAKE_CASE : List[Any] = max_position_embeddings SCREAMING_SNAKE_CASE : Dict = init_std # Normal(0, this parameter) SCREAMING_SNAKE_CASE : Union[str, Any] = activation_function # parameters for xlmprophetnet SCREAMING_SNAKE_CASE : Dict = ngram SCREAMING_SNAKE_CASE : Any = num_buckets SCREAMING_SNAKE_CASE : str = relative_max_distance SCREAMING_SNAKE_CASE : str = disable_ngram_loss SCREAMING_SNAKE_CASE : Dict = eps # 3 Types of Dropout SCREAMING_SNAKE_CASE : Union[str, Any] = attention_dropout SCREAMING_SNAKE_CASE : Optional[int] = activation_dropout SCREAMING_SNAKE_CASE : List[Any] = dropout SCREAMING_SNAKE_CASE : int = use_cache super().__init__( pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , is_encoder_decoder=UpperCAmelCase_ , add_cross_attention=UpperCAmelCase_ , decoder_start_token_id=UpperCAmelCase_ , **UpperCAmelCase_ , ) @property def _A ( self : int ): return self.num_encoder_layers + self.num_decoder_layers @num_hidden_layers.setter def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): raise NotImplementedError( "This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and" " `num_decoder_layers`." )
62
1
from typing import Dict, Optional import numpy as np import datasets snake_case = """ IoU is the area of overlap between the predicted segmentation and the ground truth divided by the area of union between the predicted segmentation and the ground truth. For binary (two classes) or multi-class segmentation, the mean IoU of the image is calculated by taking the IoU of each class and averaging them. """ snake_case = """ Args: predictions (`List[ndarray]`): List of predicted segmentation maps, each of shape (height, width). Each segmentation map can be of a different size. references (`List[ndarray]`): List of ground truth segmentation maps, each of shape (height, width). Each segmentation map can be of a different size. num_labels (`int`): Number of classes (categories). ignore_index (`int`): Index that will be ignored during evaluation. nan_to_num (`int`, *optional*): If specified, NaN values will be replaced by the number defined by the user. label_map (`dict`, *optional*): If specified, dictionary mapping old label indices to new label indices. reduce_labels (`bool`, *optional*, defaults to `False`): Whether or not to reduce all label values of segmentation maps by 1. Usually used for datasets where 0 is used for background, and background itself is not included in all classes of a dataset (e.g. ADE20k). The background label will be replaced by 255. Returns: `Dict[str, float | ndarray]` comprising various elements: - *mean_iou* (`float`): Mean Intersection-over-Union (IoU averaged over all categories). - *mean_accuracy* (`float`): Mean accuracy (averaged over all categories). - *overall_accuracy* (`float`): Overall accuracy on all images. - *per_category_accuracy* (`ndarray` of shape `(num_labels,)`): Per category accuracy. - *per_category_iou* (`ndarray` of shape `(num_labels,)`): Per category IoU. Examples: >>> import numpy as np >>> mean_iou = datasets.load_metric(\"mean_iou\") >>> # suppose one has 3 different segmentation maps predicted >>> predicted_1 = np.array([[1, 2], [3, 4], [5, 255]]) >>> actual_1 = np.array([[0, 3], [5, 4], [6, 255]]) >>> predicted_2 = np.array([[2, 7], [9, 2], [3, 6]]) >>> actual_2 = np.array([[1, 7], [9, 2], [3, 6]]) >>> predicted_3 = np.array([[2, 2, 3], [8, 2, 4], [3, 255, 2]]) >>> actual_3 = np.array([[1, 2, 2], [8, 2, 1], [3, 255, 1]]) >>> predicted = [predicted_1, predicted_2, predicted_3] >>> ground_truth = [actual_1, actual_2, actual_3] >>> results = mean_iou.compute(predictions=predicted, references=ground_truth, num_labels=10, ignore_index=255, reduce_labels=False) >>> print(results) # doctest: +NORMALIZE_WHITESPACE {'mean_iou': 0.47750000000000004, 'mean_accuracy': 0.5916666666666666, 'overall_accuracy': 0.5263157894736842, 'per_category_iou': array([0. , 0. , 0.375, 0.4 , 0.5 , 0. , 0.5 , 1. , 1. , 1. ]), 'per_category_accuracy': array([0. , 0. , 0.75 , 0.66666667, 1. , 0. , 0.5 , 1. , 1. , 1. ])} """ snake_case = """\ @software{MMSegmentation_Contributors_OpenMMLab_Semantic_Segmentation_2020, author = {{MMSegmentation Contributors}}, license = {Apache-2.0}, month = {7}, title = {{OpenMMLab Semantic Segmentation Toolbox and Benchmark}}, url = {https://github.com/open-mmlab/mmsegmentation}, year = {2020} }""" def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase = None , lowercase = False , ): """simple docstring""" if label_map is not None: for old_id, new_id in label_map.items(): SCREAMING_SNAKE_CASE : str = new_id # turn into Numpy arrays SCREAMING_SNAKE_CASE : List[str] = np.array(lowercase ) SCREAMING_SNAKE_CASE : Tuple = np.array(lowercase ) if reduce_labels: SCREAMING_SNAKE_CASE : Dict = 255 SCREAMING_SNAKE_CASE : Optional[int] = label - 1 SCREAMING_SNAKE_CASE : List[Any] = 255 SCREAMING_SNAKE_CASE : Any = label != ignore_index SCREAMING_SNAKE_CASE : Tuple = np.not_equal(lowercase , lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = pred_label[mask] SCREAMING_SNAKE_CASE : Optional[Any] = np.array(lowercase )[mask] SCREAMING_SNAKE_CASE : Union[str, Any] = pred_label[pred_label == label] SCREAMING_SNAKE_CASE : int = np.histogram(lowercase , bins=lowercase , range=(0, num_labels - 1) )[0] SCREAMING_SNAKE_CASE : List[Any] = np.histogram(lowercase , bins=lowercase , range=(0, num_labels - 1) )[0] SCREAMING_SNAKE_CASE : Optional[Any] = np.histogram(lowercase , bins=lowercase , range=(0, num_labels - 1) )[0] SCREAMING_SNAKE_CASE : Dict = area_pred_label + area_label - area_intersect return area_intersect, area_union, area_pred_label, area_label def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase = None , lowercase = False , ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = np.zeros((num_labels,) , dtype=np.floataa ) SCREAMING_SNAKE_CASE : Any = np.zeros((num_labels,) , dtype=np.floataa ) SCREAMING_SNAKE_CASE : Dict = np.zeros((num_labels,) , dtype=np.floataa ) SCREAMING_SNAKE_CASE : Optional[Any] = np.zeros((num_labels,) , dtype=np.floataa ) for result, gt_seg_map in zip(lowercase , lowercase ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = intersect_and_union( lowercase , lowercase , lowercase , lowercase , lowercase , lowercase ) total_area_intersect += area_intersect total_area_union += area_union total_area_pred_label += area_pred_label total_area_label += area_label return total_area_intersect, total_area_union, total_area_pred_label, total_area_label def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase = None , lowercase = None , lowercase = False , ): """simple docstring""" SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = total_intersect_and_union( lowercase , lowercase , lowercase , lowercase , lowercase , lowercase ) # compute metrics SCREAMING_SNAKE_CASE : Union[str, Any] = {} SCREAMING_SNAKE_CASE : Union[str, Any] = total_area_intersect.sum() / total_area_label.sum() SCREAMING_SNAKE_CASE : Dict = total_area_intersect / total_area_union SCREAMING_SNAKE_CASE : List[Any] = total_area_intersect / total_area_label SCREAMING_SNAKE_CASE : Union[str, Any] = np.nanmean(lowercase ) SCREAMING_SNAKE_CASE : str = np.nanmean(lowercase ) SCREAMING_SNAKE_CASE : List[Any] = all_acc SCREAMING_SNAKE_CASE : Dict = iou SCREAMING_SNAKE_CASE : Optional[Any] = acc if nan_to_num is not None: SCREAMING_SNAKE_CASE : Optional[int] = {metric: np.nan_to_num(lowercase , nan=lowercase ) for metric, metric_value in metrics.items()} return metrics @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : List[str] ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( # 1st Seq - height dim, 2nd - width dim { "predictions": datasets.Sequence(datasets.Sequence(datasets.Value("uint16" ) ) ), "references": datasets.Sequence(datasets.Sequence(datasets.Value("uint16" ) ) ), } ) , reference_urls=[ "https://github.com/open-mmlab/mmsegmentation/blob/71c201b1813267d78764f306a297ca717827c4bf/mmseg/core/evaluation/metrics.py" ] , ) def _A ( self : Any , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : bool , UpperCAmelCase_ : Optional[int] = None , UpperCAmelCase_ : Optional[Dict[int, int]] = None , UpperCAmelCase_ : bool = False , ): SCREAMING_SNAKE_CASE : str = mean_iou( results=UpperCAmelCase_ , gt_seg_maps=UpperCAmelCase_ , num_labels=UpperCAmelCase_ , ignore_index=UpperCAmelCase_ , nan_to_num=UpperCAmelCase_ , label_map=UpperCAmelCase_ , reduce_labels=UpperCAmelCase_ , ) return iou_result
62
import gc import random import unittest import numpy as np import torch from diffusers import DDIMScheduler, KandinskyVaaPipeline, KandinskyVaaPriorPipeline, UNetaDConditionModel, VQModel from diffusers.utils import floats_tensor, load_numpy, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = KandinskyVaaPipeline UpperCamelCase_ : List[Any] = [ '''image_embeds''', '''negative_image_embeds''', ] UpperCamelCase_ : Tuple = ['''image_embeds''', '''negative_image_embeds'''] UpperCamelCase_ : Any = [ '''generator''', '''height''', '''width''', '''latents''', '''guidance_scale''', '''num_inference_steps''', '''return_dict''', '''guidance_scale''', '''num_images_per_prompt''', '''output_type''', '''return_dict''', ] UpperCamelCase_ : List[str] = False @property def _A ( self : List[Any] ): return 32 @property def _A ( self : List[Any] ): return 32 @property def _A ( self : Any ): return self.time_input_dim @property def _A ( self : Union[str, Any] ): return self.time_input_dim * 4 @property def _A ( self : Tuple ): return 100 @property def _A ( self : Optional[int] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = { "in_channels": 4, # Out channels is double in channels because predicts mean and variance "out_channels": 8, "addition_embed_type": "image", "down_block_types": ("ResnetDownsampleBlock2D", "SimpleCrossAttnDownBlock2D"), "up_block_types": ("SimpleCrossAttnUpBlock2D", "ResnetUpsampleBlock2D"), "mid_block_type": "UNetMidBlock2DSimpleCrossAttn", "block_out_channels": (self.block_out_channels_a, self.block_out_channels_a * 2), "layers_per_block": 1, "encoder_hid_dim": self.text_embedder_hidden_size, "encoder_hid_dim_type": "image_proj", "cross_attention_dim": self.cross_attention_dim, "attention_head_dim": 4, "resnet_time_scale_shift": "scale_shift", "class_embed_type": None, } SCREAMING_SNAKE_CASE : str = UNetaDConditionModel(**UpperCAmelCase_ ) return model @property def _A ( self : int ): return { "block_out_channels": [32, 64], "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"], "in_channels": 3, "latent_channels": 4, "layers_per_block": 1, "norm_num_groups": 8, "norm_type": "spatial", "num_vq_embeddings": 12, "out_channels": 3, "up_block_types": [ "AttnUpDecoderBlock2D", "UpDecoderBlock2D", ], "vq_embed_dim": 4, } @property def _A ( self : Any ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : List[str] = VQModel(**self.dummy_movq_kwargs ) return model def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.dummy_unet SCREAMING_SNAKE_CASE : str = self.dummy_movq SCREAMING_SNAKE_CASE : Optional[int] = DDIMScheduler( num_train_timesteps=1000 , beta_schedule="linear" , beta_start=0.00_085 , beta_end=0.012 , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , steps_offset=1 , prediction_type="epsilon" , thresholding=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "scheduler": scheduler, "movq": movq, } return components def _A ( self : List[str] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any]=0 ): SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(seed + 1 ) ).to( UpperCAmelCase_ ) if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : List[Any] = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = { "image_embeds": image_embeds, "negative_image_embeds": negative_image_embeds, "generator": generator, "height": 64, "width": 64, "guidance_scale": 4.0, "num_inference_steps": 2, "output_type": "np", } return inputs def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "cpu" SCREAMING_SNAKE_CASE : Tuple = self.get_dummy_components() SCREAMING_SNAKE_CASE : List[str] = self.pipeline_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(**self.get_dummy_inputs(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : str = output.images SCREAMING_SNAKE_CASE : Tuple = pipe( **self.get_dummy_inputs(UpperCAmelCase_ ) , return_dict=UpperCAmelCase_ , )[0] SCREAMING_SNAKE_CASE : str = image[0, -3:, -3:, -1] SCREAMING_SNAKE_CASE : Any = image_from_tuple[0, -3:, -3:, -1] assert image.shape == (1, 64, 64, 3) SCREAMING_SNAKE_CASE : List[Any] = np.array( [0.6_237_976, 1.0, 0.36_441_332, 1.0, 0.70_639_634, 0.29_877_186, 0.85_652_125, 0.5_216_843, 0.54_454_046] ) assert ( np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_slice.flatten()}''' assert ( np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_from_tuple_slice.flatten()}''' @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main" "/kandinskyv22/kandinskyv22_text2img_cat_fp16.npy" ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPriorPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-prior" , torch_dtype=torch.floataa ) pipe_prior.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-decoder" , torch_dtype=torch.floataa ) SCREAMING_SNAKE_CASE : Tuple = pipeline.to(UpperCAmelCase_ ) pipeline.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = "red cat, 4k photo" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = pipe_prior( UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=5 , negative_prompt="" , ).to_tuple() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = pipeline( image_embeds=UpperCAmelCase_ , negative_image_embeds=UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=100 , output_type="np" , ) SCREAMING_SNAKE_CASE : List[Any] = output.images[0] assert image.shape == (512, 512, 3) assert_mean_pixel_difference(UpperCAmelCase_ , UpperCAmelCase_ )
62
1
from collections import OrderedDict from typing import Mapping from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """roberta-base""": """https://huggingface.co/roberta-base/resolve/main/config.json""", """roberta-large""": """https://huggingface.co/roberta-large/resolve/main/config.json""", """roberta-large-mnli""": """https://huggingface.co/roberta-large-mnli/resolve/main/config.json""", """distilroberta-base""": """https://huggingface.co/distilroberta-base/resolve/main/config.json""", """roberta-base-openai-detector""": """https://huggingface.co/roberta-base-openai-detector/resolve/main/config.json""", """roberta-large-openai-detector""": """https://huggingface.co/roberta-large-openai-detector/resolve/main/config.json""", } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Union[str, Any] = '''roberta''' def __init__( self : str , UpperCAmelCase_ : List[Any]=5_0265 , UpperCAmelCase_ : Dict=768 , UpperCAmelCase_ : str=12 , UpperCAmelCase_ : str=12 , UpperCAmelCase_ : Any=3072 , UpperCAmelCase_ : Dict="gelu" , UpperCAmelCase_ : Union[str, Any]=0.1 , UpperCAmelCase_ : str=0.1 , UpperCAmelCase_ : int=512 , UpperCAmelCase_ : List[Any]=2 , UpperCAmelCase_ : Dict=0.02 , UpperCAmelCase_ : List[str]=1E-12 , UpperCAmelCase_ : Optional[int]=1 , UpperCAmelCase_ : Any=0 , UpperCAmelCase_ : str=2 , UpperCAmelCase_ : str="absolute" , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Dict=None , **UpperCAmelCase_ : Tuple , ): super().__init__(pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : Optional[Any] = hidden_size SCREAMING_SNAKE_CASE : Optional[Any] = num_hidden_layers SCREAMING_SNAKE_CASE : Optional[int] = num_attention_heads SCREAMING_SNAKE_CASE : Dict = hidden_act SCREAMING_SNAKE_CASE : List[Any] = intermediate_size SCREAMING_SNAKE_CASE : Tuple = hidden_dropout_prob SCREAMING_SNAKE_CASE : str = attention_probs_dropout_prob SCREAMING_SNAKE_CASE : int = max_position_embeddings SCREAMING_SNAKE_CASE : Union[str, Any] = type_vocab_size SCREAMING_SNAKE_CASE : Dict = initializer_range SCREAMING_SNAKE_CASE : int = layer_norm_eps SCREAMING_SNAKE_CASE : Tuple = position_embedding_type SCREAMING_SNAKE_CASE : Union[str, Any] = use_cache SCREAMING_SNAKE_CASE : List[str] = classifier_dropout class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @property def _A ( self : Optional[int] ): if self.task == "multiple-choice": SCREAMING_SNAKE_CASE : List[Any] = {0: "batch", 1: "choice", 2: "sequence"} else: SCREAMING_SNAKE_CASE : Tuple = {0: "batch", 1: "sequence"} return OrderedDict( [ ("input_ids", dynamic_axis), ("attention_mask", dynamic_axis), ] )
62
import logging import os import quant_trainer import torch from torch.utils.data import DataLoader from transformers import Trainer, is_torch_tpu_available from transformers.trainer_utils import PredictionOutput snake_case = logging.getLogger(__name__) if is_torch_tpu_available(check_device=False): import torch_xla.core.xla_model as xm import torch_xla.debug.metrics as met class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , *UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Tuple=None , **UpperCAmelCase_ : Optional[Any] ): super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = eval_examples SCREAMING_SNAKE_CASE : List[Any] = post_process_function SCREAMING_SNAKE_CASE : Any = quant_trainer_args SCREAMING_SNAKE_CASE : Optional[Any] = 128 # default number of calibration samples def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple=None ): if calib_dataset is None and self.calib_dataset is None: raise ValueError("Trainer: calibration requires an calib_dataset." ) SCREAMING_SNAKE_CASE : str = calib_dataset if calib_dataset is not None else self.calib_dataset SCREAMING_SNAKE_CASE : str = self._remove_unused_columns(UpperCAmelCase_ , description="Calibration" ) return DataLoader( UpperCAmelCase_ , batch_size=self.args.eval_batch_size , collate_fn=self.data_collator , drop_last=self.args.dataloader_drop_last , num_workers=self.args.dataloader_num_workers , pin_memory=self.args.dataloader_pin_memory , shuffle=UpperCAmelCase_ , ) def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[int]=None ): SCREAMING_SNAKE_CASE : Any = self.train_dataset if calib_dataset is None else calib_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_calib_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args , calib=UpperCAmelCase_ ) model.eval() quant_trainer.enable_calibration(UpperCAmelCase_ ) logger.info("***** Running calibration *****" ) logger.info(f''' Num examples = {self.calib_num}''' ) logger.info(f''' Batch size = {calib_dataloader.batch_size}''' ) for step, inputs in enumerate(UpperCAmelCase_ ): # Prediction step SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = self.prediction_step(UpperCAmelCase_ , UpperCAmelCase_ , prediction_loss_only=UpperCAmelCase_ ) if (step + 1) * calib_dataloader.batch_size >= self.calib_num: break quant_trainer.finish_calibration(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : Optional[int] = model def _A ( self : List[Any] , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : str = "eval" ): SCREAMING_SNAKE_CASE : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset SCREAMING_SNAKE_CASE : Tuple = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.eval_examples if eval_examples is None else eval_examples # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Dict = self.compute_metrics SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : int = eval_loop( UpperCAmelCase_ , description="Evaluation" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : int = compute_metrics if self.post_process_function is not None and self.compute_metrics is not None: SCREAMING_SNAKE_CASE : List[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions ) SCREAMING_SNAKE_CASE : Any = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : Union[str, Any] = metrics.pop(UpperCAmelCase_ ) self.log(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[Any] = {} if self.args.tpu_metrics_debug or self.args.debug: # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.) xm.master_print(met.metrics_report() ) SCREAMING_SNAKE_CASE : List[str] = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCAmelCase_ ) return metrics def _A ( self : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : str = "test" ): SCREAMING_SNAKE_CASE : str = self.get_test_dataloader(UpperCAmelCase_ ) # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Tuple = self.compute_metrics SCREAMING_SNAKE_CASE : Union[str, Any] = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : str = eval_loop( UpperCAmelCase_ , description="Prediction" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : Union[str, Any] = compute_metrics if self.post_process_function is None or self.compute_metrics is None: return output SCREAMING_SNAKE_CASE : Optional[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions , "predict" ) SCREAMING_SNAKE_CASE : str = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : str = metrics.pop(UpperCAmelCase_ ) return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : int="./" ): SCREAMING_SNAKE_CASE : List[Any] = self.eval_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = next(iter(UpperCAmelCase_ ) ) # saving device - to make it consistent SCREAMING_SNAKE_CASE : int = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) # convert to tuple SCREAMING_SNAKE_CASE : Tuple = tuple(v.to(UpperCAmelCase_ ) for k, v in batch.items() ) logger.info("Converting model to be onnx compatible" ) from pytorch_quantization.nn import TensorQuantizer SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Any = self.model.to(UpperCAmelCase_ ) model.eval() model.float() SCREAMING_SNAKE_CASE : str = model.module if hasattr(UpperCAmelCase_ , "module" ) else model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : List[Any] = os.path.join(UpperCAmelCase_ , "model.onnx" ) logger.info(f'''exporting model to {output_model_file}''' ) SCREAMING_SNAKE_CASE : int = {0: "batch_size", 1: "seq_len"} torch.onnx.export( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , export_params=UpperCAmelCase_ , opset_version=13 , do_constant_folding=UpperCAmelCase_ , input_names=["input_ids", "attention_mask", "token_type_ids"] , output_names=["output_start_logits", "output_end_logits"] , dynamic_axes={ "input_ids": axes, "attention_mask": axes, "token_type_ids": axes, "output_start_logits": axes, "output_end_logits": axes, } , verbose=UpperCAmelCase_ , ) logger.info("onnx export finished" )
62
1
import sys import turtle def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return (pa[0] + pa[0]) / 2, (pa[1] + pa[1]) / 2 def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , ): """simple docstring""" my_pen.up() my_pen.goto(vertexa[0] , vertexa[1] ) my_pen.down() my_pen.goto(vertexa[0] , vertexa[1] ) my_pen.goto(vertexa[0] , vertexa[1] ) my_pen.goto(vertexa[0] , vertexa[1] ) if depth == 0: return triangle(lowercase , get_mid(lowercase , lowercase ) , get_mid(lowercase , lowercase ) , depth - 1 ) triangle(lowercase , get_mid(lowercase , lowercase ) , get_mid(lowercase , lowercase ) , depth - 1 ) triangle(lowercase , get_mid(lowercase , lowercase ) , get_mid(lowercase , lowercase ) , depth - 1 ) if __name__ == "__main__": if len(sys.argv) != 2: raise ValueError( """Correct format for using this script: """ """python fractals.py <int:depth_for_fractal>""" ) snake_case = turtle.Turtle() my_pen.ht() my_pen.speed(5) my_pen.pencolor("""red""") snake_case = [(-175, -125), (0, 175), (175, -125)] # vertices of triangle triangle(vertices[0], vertices[1], vertices[2], int(sys.argv[1]))
62
import os import unittest from transformers import LayoutLMTokenizer, LayoutLMTokenizerFast from transformers.models.layoutlm.tokenization_layoutlm import VOCAB_FILES_NAMES from transformers.testing_utils import require_tokenizers from ...test_tokenization_common import TokenizerTesterMixin @require_tokenizers class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = LayoutLMTokenizer UpperCamelCase_ : str = LayoutLMTokenizerFast UpperCamelCase_ : Any = True UpperCamelCase_ : Optional[Any] = True def _A ( self : Any ): super().setUp() SCREAMING_SNAKE_CASE : Optional[Any] = [ "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest", ] SCREAMING_SNAKE_CASE : Optional[Any] = 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 _A ( self : str , **UpperCAmelCase_ : Optional[int] ): return LayoutLMTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Any = "UNwant\u00E9d,running" SCREAMING_SNAKE_CASE : Union[str, Any] = "unwanted, running" return input_text, output_text def _A ( self : int ): SCREAMING_SNAKE_CASE : Optional[int] = self.tokenizer_class(self.vocab_file ) SCREAMING_SNAKE_CASE : Dict = tokenizer.tokenize("UNwant\u00E9d,running" ) self.assertListEqual(UpperCAmelCase_ , ["un", "##want", "##ed", ",", "runn", "##ing"] ) self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , [7, 4, 5, 10, 8, 9] ) def _A ( self : List[str] ): pass
62
1
import math import flax.linen as nn import jax.numpy as jnp def lowerCamelCase__ ( lowercase , lowercase , lowercase = 1 , lowercase = 1 , lowercase = 1.0E4 , lowercase = False , lowercase = 1.0 , ): """simple docstring""" assert timesteps.ndim == 1, "Timesteps should be a 1d-array" assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even''' SCREAMING_SNAKE_CASE : Union[str, Any] = float(embedding_dim // 2 ) SCREAMING_SNAKE_CASE : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift) SCREAMING_SNAKE_CASE : Optional[Any] = min_timescale * jnp.exp(jnp.arange(lowercase , dtype=jnp.floataa ) * -log_timescale_increment ) SCREAMING_SNAKE_CASE : Optional[int] = jnp.expand_dims(lowercase , 1 ) * jnp.expand_dims(lowercase , 0 ) # scale embeddings SCREAMING_SNAKE_CASE : Optional[int] = scale * emb if flip_sin_to_cos: SCREAMING_SNAKE_CASE : List[Any] = jnp.concatenate([jnp.cos(lowercase ), jnp.sin(lowercase )] , axis=1 ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = jnp.concatenate([jnp.sin(lowercase ), jnp.cos(lowercase )] , axis=1 ) SCREAMING_SNAKE_CASE : Tuple = jnp.reshape(lowercase , [jnp.shape(lowercase )[0], embedding_dim] ) return signal class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : jnp.dtype = jnp.floataa @nn.compact def __call__( self : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_1" )(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.silu(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_2" )(UpperCAmelCase_ ) return temb class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : bool = False UpperCamelCase_ : float = 1 @nn.compact def __call__( self : Optional[int] , UpperCAmelCase_ : int ): return get_sinusoidal_embeddings( UpperCAmelCase_ , embedding_dim=self.dim , flip_sin_to_cos=self.flip_sin_to_cos , freq_shift=self.freq_shift )
62
import argparse import json import os from collections import OrderedDict import numpy as np import tensorflow as tf import torch def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = os.path.join(args.tf_model_dir , "parameters.json" ) SCREAMING_SNAKE_CASE : Union[str, Any] = json.loads(open(lowercase ).read() ) if not params: raise ValueError( F'''It seems that the json file at {parameter_file} is empty. Make sure you have a correct json file.''' ) if not args.output.endswith(".pt" ): SCREAMING_SNAKE_CASE : Optional[int] = args.output + ".pt" SCREAMING_SNAKE_CASE : Any = OrderedDict() with tf.device("/CPU:0" ): SCREAMING_SNAKE_CASE : Union[str, Any] = tf.train.load_checkpoint(args.tf_model_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = reader.get_variable_to_shape_map() for key_name in shapes.keys(): SCREAMING_SNAKE_CASE : Any = reader.get_tensor(lowercase ).astype(np.floataa ) if key_name.endswith("/adam_m" ) or key_name.endswith("/adam_v" ): continue if key_name.startswith("pasts/" ): if key_name.startswith("pasts/mlp" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9] ) elif key_name.startswith("pasts/out" ): SCREAMING_SNAKE_CASE : Optional[int] = 8 SCREAMING_SNAKE_CASE : List[Any] = "model.sqout.%d.weight" % (player * 2) # enter to nn.Sequencial with Tanh, so 2 at a time SCREAMING_SNAKE_CASE : Optional[int] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.startswith("model/moe" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/switch_gating/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.router.classifier.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/softmlp/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.soft_bypass_mlp.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/wo/kernel" ) or key_name.endswith("/wi/kernel" ): SCREAMING_SNAKE_CASE : Optional[int] = key_name[-9:-7] for i in range(16 ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.feed_forward.mlp.experts.expert_%d.%s.weight" % (player, i, nlayer) SCREAMING_SNAKE_CASE : List[str] = ( vnp[i].transpose([1, 0] ).copy() ) # In Mesh-Tensorflow, it is one array, so it is divided SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name.startswith("model/mlp" ): SCREAMING_SNAKE_CASE : str = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/p1/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.mlp.wi.weight" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/p1/bias" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.wi.bias" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/kernel" ): SCREAMING_SNAKE_CASE : str = "model.blocks.%d.feed_forward.mlp.wo.weight" % player SCREAMING_SNAKE_CASE : List[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/bias" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.feed_forward.mlp.wo.bias" % player SCREAMING_SNAKE_CASE : str = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) elif key_name.startswith("model/ln" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.norm.bias" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : List[str] = "model.blocks.%d.feed_forward.norm.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/att" ): SCREAMING_SNAKE_CASE : Optional[int] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/qkv/kernel" ): SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # Compute same dimension as Mesh-tensorflow using einsum SCREAMING_SNAKE_CASE : List[str] = state[:, 0, :, :] SCREAMING_SNAKE_CASE : Tuple = state[:, 1, :, :] SCREAMING_SNAKE_CASE : List[Any] = state[:, 2, :, :] SCREAMING_SNAKE_CASE : Tuple = ( state_q.reshape([state_q.shape[0], state_q.shape[1] * state_q.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = ( state_k.reshape([state_k.shape[0], state_k.shape[1] * state_k.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = ( state_v.reshape([state_v.shape[0], state_v.shape[1] * state_v.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : int = "model.blocks.%d.self_attn.self_attn.q_proj.weight" % player SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = "model.blocks.%d.self_attn.self_attn.k_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = "model.blocks.%d.self_attn.self_attn.v_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/o/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.self_attn.self_attn.out_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = ( vnp.reshape([vnp.shape[0] * vnp.shape[1], vnp.shape[2]] ).transpose([1, 0] ).copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif key_name.startswith("model/an" ): SCREAMING_SNAKE_CASE : int = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.self_attn.norm.bias" % player SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[str] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.self_attn.norm.weight" % player SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif ( key_name.startswith("model/wte" ) or key_name.startswith("model/wpe" ) or key_name.startswith("model/ete" ) ): SCREAMING_SNAKE_CASE : str = {"wte": "embed_tokens", "wpe": "position_embeddings", "ete": "extra_position_embeddings"}[ key_name[-3:] ] SCREAMING_SNAKE_CASE : List[str] = "model.%s.weight" % nlayer SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) if key_name.startswith("model/wte" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "lm_head.weight" SCREAMING_SNAKE_CASE : List[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/wob" ): SCREAMING_SNAKE_CASE : List[Any] = "final_logits_bias" SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : List[str] = state.reshape((1, -1) ) SCREAMING_SNAKE_CASE : int = torch.tensor(lowercase ) elif key_name == "model/dense/kernel": SCREAMING_SNAKE_CASE : Optional[int] = "model.last_project.weight" SCREAMING_SNAKE_CASE : Dict = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name == "model/dense_1/bias": SCREAMING_SNAKE_CASE : str = "model.last_project.bias" SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) torch.save(lowercase , args.output ) if __name__ == "__main__": snake_case = argparse.ArgumentParser( description="""model converter.""", formatter_class=argparse.ArgumentDefaultsHelpFormatter ) parser.add_argument("""--tf_model_dir""", metavar="""PATH""", type=str, required=True, help="""import model""") parser.add_argument("""--output""", metavar="""PATH""", type=str, required=True, help="""output model""") snake_case = parser.parse_args() convert_tf_gptsan_to_pt(args)
62
1
import logging import os from dataclasses import dataclass from typing import List, Optional, Union import tqdm from filelock import FileLock from transformers import ( BartTokenizer, BartTokenizerFast, DataProcessor, PreTrainedTokenizer, RobertaTokenizer, RobertaTokenizerFast, XLMRobertaTokenizer, is_tf_available, is_torch_available, ) snake_case = logging.getLogger(__name__) @dataclass(frozen=lowerCAmelCase ) class SCREAMING_SNAKE_CASE : '''simple docstring''' UpperCamelCase_ : str UpperCamelCase_ : str UpperCamelCase_ : Optional[str] = None UpperCamelCase_ : Optional[str] = None UpperCamelCase_ : Optional[str] = None @dataclass(frozen=lowerCAmelCase ) class SCREAMING_SNAKE_CASE : '''simple docstring''' UpperCamelCase_ : List[int] UpperCamelCase_ : Optional[List[int]] = None UpperCamelCase_ : Optional[List[int]] = None UpperCamelCase_ : Optional[Union[int, float]] = None UpperCamelCase_ : Optional[int] = None if is_torch_available(): import torch from torch.utils.data import Dataset class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : List[InputFeatures] def __init__( self : int , UpperCAmelCase_ : str , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[int] = None , UpperCAmelCase_ : List[Any]=False , UpperCAmelCase_ : bool = False , ): SCREAMING_SNAKE_CASE : str = hans_processors[task]() SCREAMING_SNAKE_CASE : Any = os.path.join( UpperCAmelCase_ , "cached_{}_{}_{}_{}".format( "dev" if evaluate else "train" , tokenizer.__class__.__name__ , str(UpperCAmelCase_ ) , UpperCAmelCase_ , ) , ) SCREAMING_SNAKE_CASE : List[str] = processor.get_labels() if tokenizer.__class__ in ( RobertaTokenizer, RobertaTokenizerFast, XLMRobertaTokenizer, BartTokenizer, BartTokenizerFast, ): # HACK(label indices are swapped in RoBERTa pretrained model) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = label_list[2], label_list[1] SCREAMING_SNAKE_CASE : List[str] = label_list # Make sure only the first process in distributed training processes the dataset, # and the others will use the cache. SCREAMING_SNAKE_CASE : Optional[Any] = cached_features_file + ".lock" with FileLock(UpperCAmelCase_ ): if os.path.exists(UpperCAmelCase_ ) and not overwrite_cache: logger.info(f'''Loading features from cached file {cached_features_file}''' ) SCREAMING_SNAKE_CASE : List[Any] = torch.load(UpperCAmelCase_ ) else: logger.info(f'''Creating features from dataset file at {data_dir}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = ( processor.get_dev_examples(UpperCAmelCase_ ) if evaluate else processor.get_train_examples(UpperCAmelCase_ ) ) logger.info("Training examples: %s" , len(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = hans_convert_examples_to_features(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) logger.info("Saving features into cached file %s" , UpperCAmelCase_ ) torch.save(self.features , UpperCAmelCase_ ) def __len__( self : Optional[int] ): return len(self.features ) def __getitem__( self : int , UpperCAmelCase_ : str ): return self.features[i] def _A ( self : int ): return self.label_list if is_tf_available(): import tensorflow as tf class SCREAMING_SNAKE_CASE : '''simple docstring''' UpperCamelCase_ : List[InputFeatures] def __init__( self : Dict , UpperCAmelCase_ : str , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[int] = 128 , UpperCAmelCase_ : Optional[Any]=False , UpperCAmelCase_ : bool = False , ): SCREAMING_SNAKE_CASE : Optional[int] = hans_processors[task]() SCREAMING_SNAKE_CASE : int = processor.get_labels() if tokenizer.__class__ in ( RobertaTokenizer, RobertaTokenizerFast, XLMRobertaTokenizer, BartTokenizer, BartTokenizerFast, ): # HACK(label indices are swapped in RoBERTa pretrained model) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = label_list[2], label_list[1] SCREAMING_SNAKE_CASE : Tuple = label_list SCREAMING_SNAKE_CASE : List[str] = processor.get_dev_examples(UpperCAmelCase_ ) if evaluate else processor.get_train_examples(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = hans_convert_examples_to_features(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) def gen(): for ex_index, ex in tqdm.tqdm(enumerate(self.features ) , desc="convert examples to features" ): if ex_index % 1_0000 == 0: logger.info("Writing example %d of %d" % (ex_index, len(UpperCAmelCase_ )) ) yield ( { "example_id": 0, "input_ids": ex.input_ids, "attention_mask": ex.attention_mask, "token_type_ids": ex.token_type_ids, }, ex.label, ) SCREAMING_SNAKE_CASE : Any = tf.data.Dataset.from_generator( UpperCAmelCase_ , ( { "example_id": tf.intaa, "input_ids": tf.intaa, "attention_mask": tf.intaa, "token_type_ids": tf.intaa, }, tf.intaa, ) , ( { "example_id": tf.TensorShape([] ), "input_ids": tf.TensorShape([None, None] ), "attention_mask": tf.TensorShape([None, None] ), "token_type_ids": tf.TensorShape([None, None] ), }, tf.TensorShape([] ), ) , ) def _A ( self : List[str] ): return self.dataset def __len__( self : int ): return len(self.features ) def __getitem__( self : int , UpperCAmelCase_ : List[Any] ): return self.features[i] def _A ( self : Optional[int] ): return self.label_list class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[Any] , UpperCAmelCase_ : str ): return self._create_examples(self._read_tsv(os.path.join(UpperCAmelCase_ , "heuristics_train_set.txt" ) ) , "train" ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : str ): return self._create_examples(self._read_tsv(os.path.join(UpperCAmelCase_ , "heuristics_evaluation_set.txt" ) ) , "dev" ) def _A ( self : Optional[int] ): return ["contradiction", "entailment", "neutral"] def _A ( self : Tuple , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = [] for i, line in enumerate(UpperCAmelCase_ ): if i == 0: continue SCREAMING_SNAKE_CASE : Union[str, Any] = "%s-%s" % (set_type, line[0]) SCREAMING_SNAKE_CASE : int = line[5] SCREAMING_SNAKE_CASE : Any = line[6] SCREAMING_SNAKE_CASE : str = line[7][2:] if line[7].startswith("ex" ) else line[7] SCREAMING_SNAKE_CASE : Optional[int] = line[0] examples.append(InputExample(guid=UpperCAmelCase_ , text_a=UpperCAmelCase_ , text_b=UpperCAmelCase_ , label=UpperCAmelCase_ , pairID=UpperCAmelCase_ ) ) return examples def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = {label: i for i, label in enumerate(lowercase )} SCREAMING_SNAKE_CASE : List[Any] = [] for ex_index, example in tqdm.tqdm(enumerate(lowercase ) , desc="convert examples to features" ): if ex_index % 10000 == 0: logger.info("Writing example %d" % (ex_index) ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer( example.text_a , example.text_b , add_special_tokens=lowercase , max_length=lowercase , padding="max_length" , truncation=lowercase , return_overflowing_tokens=lowercase , ) SCREAMING_SNAKE_CASE : Optional[Any] = label_map[example.label] if example.label in label_map else 0 SCREAMING_SNAKE_CASE : List[str] = int(example.pairID ) features.append(InputFeatures(**lowercase , label=lowercase , pairID=lowercase ) ) for i, example in enumerate(examples[:5] ): logger.info("*** Example ***" ) logger.info(F'''guid: {example}''' ) logger.info(F'''features: {features[i]}''' ) return features snake_case = { """hans""": 3, } snake_case = { """hans""": HansProcessor, }
62
import json import os import shutil import tempfile import unittest import numpy as np from transformers import BertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer from transformers.testing_utils import require_tokenizers, require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor @require_tokenizers @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = tempfile.mkdtemp() # fmt: off SCREAMING_SNAKE_CASE : Any = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest"] # fmt: on SCREAMING_SNAKE_CASE : Dict = 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] ) ) SCREAMING_SNAKE_CASE : Optional[int] = { "do_resize": True, "size": {"height": 18, "width": 18}, "do_normalize": True, "image_mean": [0.5, 0.5, 0.5], "image_std": [0.5, 0.5, 0.5], } SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(self.tmpdirname , UpperCAmelCase_ ) with open(self.image_processor_file , "w" , encoding="utf-8" ) as fp: json.dump(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] , **UpperCAmelCase_ : List[str] ): return BertTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , **UpperCAmelCase_ : Any ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] ): shutil.rmtree(self.tmpdirname ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] SCREAMING_SNAKE_CASE : Union[str, Any] = [Image.fromarray(np.moveaxis(UpperCAmelCase_ , 0 , -1 ) ) for x in image_inputs] return image_inputs def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : int = VisionTextDualEncoderProcessor( tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer(bos_token="(BOS)" , eos_token="(EOS)" ) SCREAMING_SNAKE_CASE : Any = self.get_image_processor(do_normalize=UpperCAmelCase_ , padding_value=1.0 ) SCREAMING_SNAKE_CASE : str = VisionTextDualEncoderProcessor.from_pretrained( self.tmpdirname , bos_token="(BOS)" , eos_token="(EOS)" , do_normalize=UpperCAmelCase_ , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : int = image_processor(UpperCAmelCase_ , return_tensors="np" ) SCREAMING_SNAKE_CASE : Optional[int] = processor(images=UpperCAmelCase_ , 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 _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = "lower newer" SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = tokenizer(UpperCAmelCase_ ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : int = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer() SCREAMING_SNAKE_CASE : List[str] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = "lower newer" SCREAMING_SNAKE_CASE : Union[str, Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Optional[Any] = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , ["input_ids", "token_type_ids", "attention_mask", "pixel_values"] ) # test if it raises when no input is passed with self.assertRaises(UpperCAmelCase_ ): processor() def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] SCREAMING_SNAKE_CASE : Optional[Any] = processor.batch_decode(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer.batch_decode(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = "lower newer" SCREAMING_SNAKE_CASE : Any = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
62
1
import os def lowerCamelCase__ ( lowercase = "input.txt" ): """simple docstring""" with open(os.path.join(os.path.dirname(lowercase ) , lowercase ) ) as input_file: SCREAMING_SNAKE_CASE : List[str] = [ [int(lowercase ) for element in line.split("," )] for line in input_file.readlines() ] SCREAMING_SNAKE_CASE : Optional[Any] = len(lowercase ) SCREAMING_SNAKE_CASE : Dict = len(matrix[0] ) SCREAMING_SNAKE_CASE : Dict = [[-1 for _ in range(lowercase )] for _ in range(lowercase )] for i in range(lowercase ): SCREAMING_SNAKE_CASE : List[Any] = matrix[i][0] for j in range(1 , lowercase ): for i in range(lowercase ): SCREAMING_SNAKE_CASE : List[str] = minimal_path_sums[i][j - 1] + matrix[i][j] for i in range(1 , lowercase ): SCREAMING_SNAKE_CASE : Optional[int] = min( minimal_path_sums[i][j] , minimal_path_sums[i - 1][j] + matrix[i][j] ) for i in range(rows - 2 , -1 , -1 ): SCREAMING_SNAKE_CASE : List[str] = min( minimal_path_sums[i][j] , minimal_path_sums[i + 1][j] + matrix[i][j] ) return min(minimal_path_sums_row[-1] for minimal_path_sums_row in minimal_path_sums ) if __name__ == "__main__": print(F"""{solution() = }""")
62
import inspect import unittest from transformers import MobileNetVaConfig from transformers.testing_utils import require_torch, require_vision, slow, torch_device from transformers.utils import cached_property, is_torch_available, is_vision_available from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor from ...test_pipeline_mixin import PipelineTesterMixin if is_torch_available(): import torch from transformers import MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation, MobileNetVaModel from transformers.models.mobilenet_va.modeling_mobilenet_va import MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST if is_vision_available(): from PIL import Image from transformers import MobileNetVaImageProcessor class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.config_class(**self.inputs_dict ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "tf_padding" ) ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "depth_multiplier" ) ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : List[str]=13 , UpperCAmelCase_ : Any=3 , UpperCAmelCase_ : Union[str, Any]=32 , UpperCAmelCase_ : int=0.25 , UpperCAmelCase_ : Union[str, Any]=8 , UpperCAmelCase_ : Dict=8 , UpperCAmelCase_ : Optional[int]=6 , UpperCAmelCase_ : Optional[Any]=32 , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : str=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : str="relu6" , UpperCAmelCase_ : List[str]=1280 , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : Any=0.02 , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=10 , UpperCAmelCase_ : Optional[Any]=None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = parent SCREAMING_SNAKE_CASE : int = batch_size SCREAMING_SNAKE_CASE : Dict = num_channels SCREAMING_SNAKE_CASE : Dict = image_size SCREAMING_SNAKE_CASE : int = depth_multiplier SCREAMING_SNAKE_CASE : str = depth_divisible_by SCREAMING_SNAKE_CASE : Union[str, Any] = min_depth SCREAMING_SNAKE_CASE : int = expand_ratio SCREAMING_SNAKE_CASE : Tuple = tf_padding SCREAMING_SNAKE_CASE : List[str] = output_stride SCREAMING_SNAKE_CASE : Optional[int] = first_layer_is_expansion SCREAMING_SNAKE_CASE : Any = finegrained_output SCREAMING_SNAKE_CASE : Optional[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[int] = last_hidden_size if finegrained_output else int(last_hidden_size * depth_multiplier ) SCREAMING_SNAKE_CASE : Any = classifier_dropout_prob SCREAMING_SNAKE_CASE : Dict = use_labels SCREAMING_SNAKE_CASE : int = is_training SCREAMING_SNAKE_CASE : Dict = num_labels SCREAMING_SNAKE_CASE : Optional[Any] = initializer_range SCREAMING_SNAKE_CASE : Union[str, Any] = scope def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] ) SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : List[Any] = None if self.use_labels: SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size] , self.num_labels ) SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size, self.image_size, self.image_size] , self.num_labels ) SCREAMING_SNAKE_CASE : Tuple = self.get_config() return config, pixel_values, labels, pixel_labels def _A ( self : Optional[int] ): return MobileNetVaConfig( num_channels=self.num_channels , image_size=self.image_size , depth_multiplier=self.depth_multiplier , depth_divisible_by=self.depth_divisible_by , min_depth=self.min_depth , expand_ratio=self.expand_ratio , output_stride=self.output_stride , first_layer_is_expansion=self.first_layer_is_expansion , finegrained_output=self.finegrained_output , hidden_act=self.hidden_act , tf_padding=self.tf_padding , classifier_dropout_prob=self.classifier_dropout_prob , initializer_range=self.initializer_range , ) def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaModel(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ ) self.parent.assertEqual( result.last_hidden_state.shape , ( self.batch_size, self.last_hidden_size, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) self.parent.assertEqual( result.pooler_output.shape , (self.batch_size, self.last_hidden_size) , ) def _A ( self : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : str = self.num_labels SCREAMING_SNAKE_CASE : List[Any] = MobileNetVaForImageClassification(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Dict = MobileNetVaForSemanticSegmentation(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[int] = model(UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : List[Any] = self.prepare_config_and_inputs() SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = config_and_inputs SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": pixel_values} return config, inputs_dict @require_torch class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Any = ( (MobileNetVaModel, MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation) if is_torch_available() else () ) UpperCamelCase_ : List[Any] = ( { '''feature-extraction''': MobileNetVaModel, '''image-classification''': MobileNetVaForImageClassification, '''image-segmentation''': MobileNetVaForSemanticSegmentation, } if is_torch_available() else {} ) UpperCamelCase_ : Any = False UpperCamelCase_ : List[str] = False UpperCamelCase_ : int = False UpperCamelCase_ : str = False def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = MobileNetVaModelTester(self ) SCREAMING_SNAKE_CASE : Optional[Any] = MobileNetVaConfigTester(self , config_class=UpperCAmelCase_ , has_text_modality=UpperCAmelCase_ ) def _A ( self : Optional[Any] ): self.config_tester.run_common_tests() @unittest.skip(reason="MobileNetV2 does not use inputs_embeds" ) def _A ( self : List[Any] ): pass @unittest.skip(reason="MobileNetV2 does not support input and output embeddings" ) def _A ( self : Dict ): pass @unittest.skip(reason="MobileNetV2 does not output attentions" ) def _A ( self : Union[str, Any] ): pass def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : Tuple = model_class(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic SCREAMING_SNAKE_CASE : Dict = [*signature.parameters.keys()] SCREAMING_SNAKE_CASE : Union[str, Any] = ["pixel_values"] self.assertListEqual(arg_names[:1] , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*UpperCAmelCase_ ) def _A ( self : List[Any] ): def check_hidden_states_output(UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = model_class(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**self._prepare_for_class(UpperCAmelCase_ , UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : List[Any] = outputs.hidden_states SCREAMING_SNAKE_CASE : Any = 16 self.assertEqual(len(UpperCAmelCase_ ) , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : str = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) # check that output_hidden_states also work using config del inputs_dict["output_hidden_states"] SCREAMING_SNAKE_CASE : List[Any] = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Tuple = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_image_classification(*UpperCAmelCase_ ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_semantic_segmentation(*UpperCAmelCase_ ) @slow def _A ( self : Optional[Any] ): for model_name in MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: SCREAMING_SNAKE_CASE : int = MobileNetVaModel.from_pretrained(UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : int = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" ) return image @require_torch @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @cached_property def _A ( self : Optional[int] ): return ( MobileNetVaImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224" ) if is_vision_available() else None ) @slow def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaForImageClassification.from_pretrained("google/mobilenet_v2_1.0_224" ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.default_image_processor SCREAMING_SNAKE_CASE : Optional[int] = prepare_img() SCREAMING_SNAKE_CASE : Optional[int] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**UpperCAmelCase_ ) # verify the logits SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Size((1, 1001) ) self.assertEqual(outputs.logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = torch.tensor([0.2_445, -1.1_993, 0.1_905] ).to(UpperCAmelCase_ ) self.assertTrue(torch.allclose(outputs.logits[0, :3] , UpperCAmelCase_ , atol=1E-4 ) ) @slow def _A ( self : str ): SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaForSemanticSegmentation.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : int = model.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaImageProcessor.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : Optional[Any] = prepare_img() SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = outputs.logits # verify the logits SCREAMING_SNAKE_CASE : Dict = torch.Size((1, 21, 65, 65) ) self.assertEqual(logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.tensor( [ [[17.5_790, 17.7_581, 18.3_355], [18.3_257, 18.4_230, 18.8_973], [18.6_169, 18.8_650, 19.2_187]], [[-2.1_595, -2.0_977, -2.3_741], [-2.4_226, -2.3_028, -2.6_835], [-2.7_819, -2.5_991, -2.7_706]], [[4.2_058, 4.8_317, 4.7_638], [4.4_136, 5.0_361, 4.9_383], [4.5_028, 4.9_644, 4.8_734]], ] , device=UpperCAmelCase_ , ) self.assertTrue(torch.allclose(logits[0, :3, :3, :3] , UpperCAmelCase_ , atol=1E-4 ) )
62
1
import argparse import os import re import numpy as np import PIL import torch from timm import create_model from torch.optim.lr_scheduler import OneCycleLR from torch.utils.data import DataLoader, Dataset from torchvision.transforms import Compose, RandomResizedCrop, Resize, ToTensor from accelerate import Accelerator def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : str = fname.split(os.path.sep )[-1] return re.search(R"^(.*)_\d+\.jpg$" , lowercase ).groups()[0] class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : Dict , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Dict=None ): SCREAMING_SNAKE_CASE : Optional[int] = file_names SCREAMING_SNAKE_CASE : Tuple = image_transform SCREAMING_SNAKE_CASE : List[str] = label_to_id def __len__( self : Optional[int] ): return len(self.file_names ) def __getitem__( self : Any , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : Tuple = self.file_names[idx] SCREAMING_SNAKE_CASE : Tuple = PIL.Image.open(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = raw_image.convert("RGB" ) if self.image_transform is not None: SCREAMING_SNAKE_CASE : Tuple = self.image_transform(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = extract_label(UpperCAmelCase_ ) if self.label_to_id is not None: SCREAMING_SNAKE_CASE : Dict = self.label_to_id[label] return {"image": image, "label": label} def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" if args.with_tracking: SCREAMING_SNAKE_CASE : Dict = Accelerator( cpu=args.cpu , mixed_precision=args.mixed_precision , log_with="all" , project_dir=args.project_dir ) else: SCREAMING_SNAKE_CASE : Any = Accelerator(cpu=args.cpu , mixed_precision=args.mixed_precision ) # Sample hyper-parameters for learning rate, batch size, seed and a few other HPs SCREAMING_SNAKE_CASE : Dict = config["lr"] SCREAMING_SNAKE_CASE : List[str] = int(config["num_epochs"] ) SCREAMING_SNAKE_CASE : List[Any] = int(config["seed"] ) SCREAMING_SNAKE_CASE : Union[str, Any] = int(config["batch_size"] ) SCREAMING_SNAKE_CASE : List[str] = config["image_size"] if not isinstance(lowercase , (list, tuple) ): SCREAMING_SNAKE_CASE : Any = (image_size, image_size) # Parse out whether we are saving every epoch or after a certain number of batches if hasattr(args.checkpointing_steps , "isdigit" ): if args.checkpointing_steps == "epoch": SCREAMING_SNAKE_CASE : int = args.checkpointing_steps elif args.checkpointing_steps.isdigit(): SCREAMING_SNAKE_CASE : Dict = int(args.checkpointing_steps ) else: raise ValueError( F'''Argument `checkpointing_steps` must be either a number or `epoch`. `{args.checkpointing_steps}` passed.''' ) else: SCREAMING_SNAKE_CASE : Tuple = None # We need to initialize the trackers we use, and also store our configuration if args.with_tracking: SCREAMING_SNAKE_CASE : Optional[Any] = os.path.split(lowercase )[-1].split("." )[0] accelerator.init_trackers(lowercase , lowercase ) # Grab all the image filenames SCREAMING_SNAKE_CASE : Union[str, Any] = [os.path.join(args.data_dir , lowercase ) for fname in os.listdir(args.data_dir ) if fname.endswith(".jpg" )] # Build the label correspondences SCREAMING_SNAKE_CASE : List[str] = [extract_label(lowercase ) for fname in file_names] SCREAMING_SNAKE_CASE : Dict = list(set(lowercase ) ) id_to_label.sort() SCREAMING_SNAKE_CASE : str = {lbl: i for i, lbl in enumerate(lowercase )} # Set the seed before splitting the data. np.random.seed(lowercase ) torch.manual_seed(lowercase ) torch.cuda.manual_seed_all(lowercase ) # Split our filenames between train and validation SCREAMING_SNAKE_CASE : Optional[Any] = np.random.permutation(len(lowercase ) ) SCREAMING_SNAKE_CASE : List[Any] = int(0.8 * len(lowercase ) ) SCREAMING_SNAKE_CASE : int = random_perm[:cut] SCREAMING_SNAKE_CASE : int = random_perm[cut:] # For training we use a simple RandomResizedCrop SCREAMING_SNAKE_CASE : List[str] = Compose([RandomResizedCrop(lowercase , scale=(0.5, 1.0) ), ToTensor()] ) SCREAMING_SNAKE_CASE : int = PetsDataset( [file_names[i] for i in train_split] , image_transform=lowercase , label_to_id=lowercase ) # For evaluation, we use a deterministic Resize SCREAMING_SNAKE_CASE : List[Any] = Compose([Resize(lowercase ), ToTensor()] ) SCREAMING_SNAKE_CASE : Union[str, Any] = PetsDataset([file_names[i] for i in eval_split] , image_transform=lowercase , label_to_id=lowercase ) # Instantiate dataloaders. SCREAMING_SNAKE_CASE : Tuple = DataLoader(lowercase , shuffle=lowercase , batch_size=lowercase , num_workers=4 ) SCREAMING_SNAKE_CASE : Any = DataLoader(lowercase , shuffle=lowercase , batch_size=lowercase , num_workers=4 ) # Instantiate the model (we build the model here so that the seed also control new weights initialization) SCREAMING_SNAKE_CASE : Union[str, Any] = create_model("resnet50d" , pretrained=lowercase , num_classes=len(lowercase ) ) # 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). SCREAMING_SNAKE_CASE : int = model.to(accelerator.device ) # Freezing the base model for param in model.parameters(): SCREAMING_SNAKE_CASE : int = False for param in model.get_classifier().parameters(): SCREAMING_SNAKE_CASE : List[str] = True # We normalize the batches of images to be a bit faster. SCREAMING_SNAKE_CASE : str = torch.tensor(model.default_cfg["mean"] )[None, :, None, None].to(accelerator.device ) SCREAMING_SNAKE_CASE : str = torch.tensor(model.default_cfg["std"] )[None, :, None, None].to(accelerator.device ) # Instantiate optimizer SCREAMING_SNAKE_CASE : str = torch.optim.Adam(params=model.parameters() , lr=lr / 25 ) # Instantiate learning rate scheduler SCREAMING_SNAKE_CASE : str = OneCycleLR(optimizer=lowercase , max_lr=lowercase , epochs=lowercase , steps_per_epoch=len(lowercase ) ) # 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. SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : str = accelerator.prepare( lowercase , lowercase , lowercase , lowercase , lowercase ) # We need to keep track of how many total steps we have iterated over SCREAMING_SNAKE_CASE : Any = 0 # We also need to keep track of the starting epoch so files are named properly SCREAMING_SNAKE_CASE : Dict = 0 # Potentially load in the weights and states from a previous save if args.resume_from_checkpoint: if args.resume_from_checkpoint is not None or args.resume_from_checkpoint != "": accelerator.print(F'''Resumed from checkpoint: {args.resume_from_checkpoint}''' ) accelerator.load_state(args.resume_from_checkpoint ) SCREAMING_SNAKE_CASE : List[str] = os.path.basename(args.resume_from_checkpoint ) else: # Get the most recent checkpoint SCREAMING_SNAKE_CASE : Optional[Any] = [f.name for f in os.scandir(os.getcwd() ) if f.is_dir()] dirs.sort(key=os.path.getctime ) SCREAMING_SNAKE_CASE : Union[str, Any] = dirs[-1] # Sorts folders by date modified, most recent checkpoint is the last # Extract `epoch_{i}` or `step_{i}` SCREAMING_SNAKE_CASE : str = os.path.splitext(lowercase )[0] if "epoch" in training_difference: SCREAMING_SNAKE_CASE : List[Any] = int(training_difference.replace("epoch_" , "" ) ) + 1 SCREAMING_SNAKE_CASE : Dict = None else: SCREAMING_SNAKE_CASE : str = int(training_difference.replace("step_" , "" ) ) SCREAMING_SNAKE_CASE : Optional[Any] = resume_step // len(lowercase ) resume_step -= starting_epoch * len(lowercase ) # Now we train the model for epoch in range(lowercase , lowercase ): model.train() if args.with_tracking: SCREAMING_SNAKE_CASE : str = 0 if args.resume_from_checkpoint and epoch == starting_epoch and resume_step is not None: # We need to skip steps until we reach the resumed step SCREAMING_SNAKE_CASE : List[Any] = accelerator.skip_first_batches(lowercase , lowercase ) overall_step += resume_step else: # After the first iteration though, we need to go back to the original dataloader SCREAMING_SNAKE_CASE : Union[str, Any] = train_dataloader for batch in active_dataloader: # We could avoid this line since we set the accelerator with `device_placement=True`. SCREAMING_SNAKE_CASE : Optional[int] = {k: v.to(accelerator.device ) for k, v in batch.items()} SCREAMING_SNAKE_CASE : int = (batch["image"] - mean) / std SCREAMING_SNAKE_CASE : Dict = model(lowercase ) SCREAMING_SNAKE_CASE : List[Any] = torch.nn.functional.cross_entropy(lowercase , batch["label"] ) # We keep track of the loss at each epoch if args.with_tracking: total_loss += loss.detach().float() accelerator.backward(lowercase ) optimizer.step() lr_scheduler.step() optimizer.zero_grad() overall_step += 1 if isinstance(lowercase , lowercase ): SCREAMING_SNAKE_CASE : str = F'''step_{overall_step}''' if overall_step % checkpointing_steps == 0: if args.output_dir is not None: SCREAMING_SNAKE_CASE : List[str] = os.path.join(args.output_dir , lowercase ) accelerator.save_state(lowercase ) model.eval() SCREAMING_SNAKE_CASE : Optional[Any] = 0 SCREAMING_SNAKE_CASE : Optional[int] = 0 for step, batch in enumerate(lowercase ): # We could avoid this line since we set the accelerator with `device_placement=True`. SCREAMING_SNAKE_CASE : Dict = {k: v.to(accelerator.device ) for k, v in batch.items()} SCREAMING_SNAKE_CASE : Optional[int] = (batch["image"] - mean) / std with torch.no_grad(): SCREAMING_SNAKE_CASE : Tuple = model(lowercase ) SCREAMING_SNAKE_CASE : Tuple = outputs.argmax(dim=-1 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = accelerator.gather_for_metrics((predictions, batch["label"]) ) SCREAMING_SNAKE_CASE : Any = predictions == references num_elems += accurate_preds.shape[0] accurate += accurate_preds.long().sum() SCREAMING_SNAKE_CASE : Optional[Any] = accurate.item() / num_elems # Use accelerator.print to print only on the main process. accelerator.print(F'''epoch {epoch}: {100 * eval_metric:.2f}''' ) if args.with_tracking: accelerator.log( { "accuracy": 100 * eval_metric, "train_loss": total_loss.item() / len(lowercase ), "epoch": epoch, } , step=lowercase , ) if checkpointing_steps == "epoch": SCREAMING_SNAKE_CASE : List[Any] = F'''epoch_{epoch}''' if args.output_dir is not None: SCREAMING_SNAKE_CASE : Optional[int] = os.path.join(args.output_dir , lowercase ) accelerator.save_state(lowercase ) if args.with_tracking: accelerator.end_training() def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[Any] = argparse.ArgumentParser(description="Simple example of training script." ) parser.add_argument("--data_dir" , required=lowercase , help="The data folder on disk." ) parser.add_argument("--fp16" , action="store_true" , help="If passed, will use FP16 training." ) parser.add_argument( "--mixed_precision" , type=lowercase , default=lowercase , 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." ) parser.add_argument( "--checkpointing_steps" , type=lowercase , default=lowercase , help="Whether the various states should be saved at the end of every n steps, or 'epoch' for each epoch." , ) parser.add_argument( "--output_dir" , type=lowercase , default="." , help="Optional save directory where all checkpoint folders will be stored. Default is the current working directory." , ) parser.add_argument( "--resume_from_checkpoint" , type=lowercase , default=lowercase , help="If the training should continue from a checkpoint folder." , ) parser.add_argument( "--with_tracking" , action="store_true" , help="Whether to load in all available experiment trackers from the environment and use them for logging." , ) parser.add_argument( "--project_dir" , type=lowercase , default="logs" , help="Location on where to store experiment tracking logs` and relevent project information" , ) SCREAMING_SNAKE_CASE : List[Any] = parser.parse_args() SCREAMING_SNAKE_CASE : int = {"lr": 3E-2, "num_epochs": 3, "seed": 42, "batch_size": 64, "image_size": 224} training_function(lowercase , lowercase ) if __name__ == "__main__": main()
62
import math from typing import Dict, Iterable, List, Optional, Tuple, Union import numpy as np from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format from ...image_utils import ( IMAGENET_STANDARD_MEAN, IMAGENET_STANDARD_STD, ChannelDimension, ImageInput, PILImageResampling, get_image_size, is_torch_available, is_torch_tensor, make_list_of_images, to_numpy_array, valid_images, ) from ...utils import TensorType, is_vision_available, logging if is_torch_available(): import torch if is_vision_available(): import PIL snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" def constraint_to_multiple_of(lowercase , lowercase , lowercase=0 , lowercase=None ): SCREAMING_SNAKE_CASE : Any = round(val / multiple ) * multiple if max_val is not None and x > max_val: SCREAMING_SNAKE_CASE : Optional[int] = math.floor(val / multiple ) * multiple if x < min_val: SCREAMING_SNAKE_CASE : int = math.ceil(val / multiple ) * multiple return x SCREAMING_SNAKE_CASE : Tuple = (output_size, output_size) if isinstance(lowercase , lowercase ) else output_size SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = get_image_size(lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = output_size # determine new height and width SCREAMING_SNAKE_CASE : Tuple = output_height / input_height SCREAMING_SNAKE_CASE : List[Any] = output_width / input_width if keep_aspect_ratio: # scale as little as possible if abs(1 - scale_width ) < abs(1 - scale_height ): # fit width SCREAMING_SNAKE_CASE : List[str] = scale_width else: # fit height SCREAMING_SNAKE_CASE : Optional[Any] = scale_height SCREAMING_SNAKE_CASE : int = constraint_to_multiple_of(scale_height * input_height , multiple=lowercase ) SCREAMING_SNAKE_CASE : Dict = constraint_to_multiple_of(scale_width * input_width , multiple=lowercase ) return (new_height, new_width) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Any = ['''pixel_values'''] def __init__( self : Any , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Dict[str, int] = None , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BILINEAR , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Union[int, float] = 1 / 255 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , **UpperCAmelCase_ : str , ): super().__init__(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = size if size is not None else {"height": 384, "width": 384} SCREAMING_SNAKE_CASE : List[Any] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = do_resize SCREAMING_SNAKE_CASE : Optional[Any] = size SCREAMING_SNAKE_CASE : str = keep_aspect_ratio SCREAMING_SNAKE_CASE : int = ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample SCREAMING_SNAKE_CASE : List[str] = do_rescale SCREAMING_SNAKE_CASE : Tuple = rescale_factor SCREAMING_SNAKE_CASE : Union[str, Any] = do_normalize SCREAMING_SNAKE_CASE : Union[str, Any] = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN SCREAMING_SNAKE_CASE : int = image_std if image_std is not None else IMAGENET_STANDARD_STD def _A ( self : Tuple , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Dict[str, int] , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : int , ): SCREAMING_SNAKE_CASE : List[str] = get_size_dict(UpperCAmelCase_ ) if "height" not in size or "width" not in size: raise ValueError(f'''The size dictionary must contain the keys \'height\' and \'width\'. Got {size.keys()}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = get_resize_output_image_size( UpperCAmelCase_ , output_size=(size["height"], size["width"]) , keep_aspect_ratio=UpperCAmelCase_ , multiple=UpperCAmelCase_ , ) return resize(UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[int, float] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : Dict , ): return rescale(UpperCAmelCase_ , scale=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Dict , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : List[str] , ): return normalize(UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : ImageInput , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : PILImageResampling = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : float = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[str, TensorType]] = None , UpperCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **UpperCAmelCase_ : Optional[Any] , ): SCREAMING_SNAKE_CASE : Optional[int] = do_resize if do_resize is not None else self.do_resize SCREAMING_SNAKE_CASE : Dict = size if size is not None else self.size SCREAMING_SNAKE_CASE : Optional[int] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = keep_aspect_ratio if keep_aspect_ratio is not None else self.keep_aspect_ratio SCREAMING_SNAKE_CASE : Optional[Any] = ensure_multiple_of if ensure_multiple_of is not None else self.ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample if resample is not None else self.resample SCREAMING_SNAKE_CASE : Dict = do_rescale if do_rescale is not None else self.do_rescale SCREAMING_SNAKE_CASE : Optional[Any] = rescale_factor if rescale_factor is not None else self.rescale_factor SCREAMING_SNAKE_CASE : Optional[int] = do_normalize if do_normalize is not None else self.do_normalize SCREAMING_SNAKE_CASE : int = image_mean if image_mean is not None else self.image_mean SCREAMING_SNAKE_CASE : Tuple = image_std if image_std is not None else self.image_std SCREAMING_SNAKE_CASE : List[str] = make_list_of_images(UpperCAmelCase_ ) if not valid_images(UpperCAmelCase_ ): raise ValueError( "Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, " "torch.Tensor, tf.Tensor or jax.ndarray." ) if do_resize and size is None or resample is None: raise ValueError("Size and resample must be specified if do_resize is True." ) if do_rescale and rescale_factor is None: raise ValueError("Rescale factor must be specified if do_rescale is True." ) if do_normalize and (image_mean is None or image_std is None): raise ValueError("Image mean and std must be specified if do_normalize is True." ) # All transformations expect numpy arrays. SCREAMING_SNAKE_CASE : Any = [to_numpy_array(UpperCAmelCase_ ) for image in images] if do_resize: SCREAMING_SNAKE_CASE : Tuple = [self.resize(image=UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ ) for image in images] if do_rescale: SCREAMING_SNAKE_CASE : Tuple = [self.rescale(image=UpperCAmelCase_ , scale=UpperCAmelCase_ ) for image in images] if do_normalize: SCREAMING_SNAKE_CASE : Optional[Any] = [self.normalize(image=UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = [to_channel_dimension_format(UpperCAmelCase_ , UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": images} return BatchFeature(data=UpperCAmelCase_ , tensor_type=UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Tuple] = None ): SCREAMING_SNAKE_CASE : int = outputs.logits # Resize logits and compute semantic segmentation maps if target_sizes is not None: if len(UpperCAmelCase_ ) != len(UpperCAmelCase_ ): raise ValueError( "Make sure that you pass in as many target sizes as the batch dimension of the logits" ) if is_torch_tensor(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = target_sizes.numpy() SCREAMING_SNAKE_CASE : str = [] for idx in range(len(UpperCAmelCase_ ) ): SCREAMING_SNAKE_CASE : Dict = torch.nn.functional.interpolate( logits[idx].unsqueeze(dim=0 ) , size=target_sizes[idx] , mode="bilinear" , align_corners=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = resized_logits[0].argmax(dim=0 ) semantic_segmentation.append(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = logits.argmax(dim=1 ) SCREAMING_SNAKE_CASE : str = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )] return semantic_segmentation
62
1
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : List[Any] = int(lowercase ) # Initialize Result SCREAMING_SNAKE_CASE : Dict = [] # Traverse through all denomination for denomination in reversed(lowercase ): # Find denominations while int(lowercase ) >= int(lowercase ): total_value -= int(lowercase ) answer.append(lowercase ) # Append the "answers" array return answer # Driver Code if __name__ == "__main__": snake_case = [] snake_case = """0""" if ( input("""Do you want to enter your denominations ? (yY/n): """).strip().lower() == "y" ): snake_case = int(input("""Enter the number of denominations you want to add: """).strip()) for i in range(0, n): denominations.append(int(input(F"""Denomination {i}: """).strip())) snake_case = input("""Enter the change you want to make in Indian Currency: """).strip() else: # All denominations of Indian Currency if user does not enter snake_case = [1, 2, 5, 10, 20, 50, 100, 500, 2_000] snake_case = input("""Enter the change you want to make: """).strip() if int(value) == 0 or int(value) < 0: print("""The total value cannot be zero or negative.""") else: print(F"""Following is minimal change for {value}: """) snake_case = find_minimum_change(denominations, value) # Print result for i in range(len(answer)): print(answer[i], end=""" """)
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : List[str] , UpperCAmelCase_ : int = 6 ): SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None self.create_linked_list(UpperCAmelCase_ ) def _A ( self : List[Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[int] = Node() SCREAMING_SNAKE_CASE : str = current_node SCREAMING_SNAKE_CASE : Optional[int] = current_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node for _ in range(1 , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Tuple = Node() SCREAMING_SNAKE_CASE : Dict = current_node SCREAMING_SNAKE_CASE : Optional[Any] = previous_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node SCREAMING_SNAKE_CASE : Union[str, Any] = self.front SCREAMING_SNAKE_CASE : List[str] = previous_node def _A ( self : Union[str, Any] ): return ( self.front == self.rear and self.front is not None and self.front.data is None ) def _A ( self : Optional[int] ): self.check_can_perform_operation() return self.front.data if self.front else None def _A ( self : Optional[int] , UpperCAmelCase_ : Any ): if self.rear is None: return self.check_is_full() if not self.is_empty(): SCREAMING_SNAKE_CASE : List[str] = self.rear.next if self.rear: SCREAMING_SNAKE_CASE : Dict = data def _A ( self : List[str] ): self.check_can_perform_operation() if self.rear is None or self.front is None: return None if self.front == self.rear: SCREAMING_SNAKE_CASE : List[str] = self.front.data SCREAMING_SNAKE_CASE : Optional[int] = None return data SCREAMING_SNAKE_CASE : List[str] = self.front SCREAMING_SNAKE_CASE : List[str] = old_front.next SCREAMING_SNAKE_CASE : Optional[int] = old_front.data SCREAMING_SNAKE_CASE : List[str] = None return data def _A ( self : Any ): if self.is_empty(): raise Exception("Empty Queue" ) def _A ( self : Optional[Any] ): if self.rear and self.rear.next == self.front: raise Exception("Full Queue" ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any | None = None SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None if __name__ == "__main__": import doctest doctest.testmod()
62
1
import argparse import struct import unittest class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : int , UpperCAmelCase_ : bytes ): SCREAMING_SNAKE_CASE : Dict = data # Initialize hash values SCREAMING_SNAKE_CASE : List[str] = [ 0X6A_09_E6_67, 0XBB_67_AE_85, 0X3C_6E_F3_72, 0XA5_4F_F5_3A, 0X51_0E_52_7F, 0X9B_05_68_8C, 0X1F_83_D9_AB, 0X5B_E0_CD_19, ] # Initialize round constants SCREAMING_SNAKE_CASE : List[str] = [ 0X42_8A_2F_98, 0X71_37_44_91, 0XB5_C0_FB_CF, 0XE9_B5_DB_A5, 0X39_56_C2_5B, 0X59_F1_11_F1, 0X92_3F_82_A4, 0XAB_1C_5E_D5, 0XD8_07_AA_98, 0X12_83_5B_01, 0X24_31_85_BE, 0X55_0C_7D_C3, 0X72_BE_5D_74, 0X80_DE_B1_FE, 0X9B_DC_06_A7, 0XC1_9B_F1_74, 0XE4_9B_69_C1, 0XEF_BE_47_86, 0X0F_C1_9D_C6, 0X24_0C_A1_CC, 0X2D_E9_2C_6F, 0X4A_74_84_AA, 0X5C_B0_A9_DC, 0X76_F9_88_DA, 0X98_3E_51_52, 0XA8_31_C6_6D, 0XB0_03_27_C8, 0XBF_59_7F_C7, 0XC6_E0_0B_F3, 0XD5_A7_91_47, 0X06_CA_63_51, 0X14_29_29_67, 0X27_B7_0A_85, 0X2E_1B_21_38, 0X4D_2C_6D_FC, 0X53_38_0D_13, 0X65_0A_73_54, 0X76_6A_0A_BB, 0X81_C2_C9_2E, 0X92_72_2C_85, 0XA2_BF_E8_A1, 0XA8_1A_66_4B, 0XC2_4B_8B_70, 0XC7_6C_51_A3, 0XD1_92_E8_19, 0XD6_99_06_24, 0XF4_0E_35_85, 0X10_6A_A0_70, 0X19_A4_C1_16, 0X1E_37_6C_08, 0X27_48_77_4C, 0X34_B0_BC_B5, 0X39_1C_0C_B3, 0X4E_D8_AA_4A, 0X5B_9C_CA_4F, 0X68_2E_6F_F3, 0X74_8F_82_EE, 0X78_A5_63_6F, 0X84_C8_78_14, 0X8C_C7_02_08, 0X90_BE_FF_FA, 0XA4_50_6C_EB, 0XBE_F9_A3_F7, 0XC6_71_78_F2, ] SCREAMING_SNAKE_CASE : Tuple = self.preprocessing(self.data ) self.final_hash() @staticmethod def _A ( UpperCAmelCase_ : bytes ): SCREAMING_SNAKE_CASE : Optional[int] = b"\x80" + (b"\x00" * (63 - (len(UpperCAmelCase_ ) + 8) % 64)) SCREAMING_SNAKE_CASE : List[Any] = struct.pack(">Q" , (len(UpperCAmelCase_ ) * 8) ) return data + padding + big_endian_integer def _A ( self : Optional[Any] ): # Convert into blocks of 64 bytes SCREAMING_SNAKE_CASE : int = [ self.preprocessed_data[x : x + 64] for x in range(0 , len(self.preprocessed_data ) , 64 ) ] for block in self.blocks: # Convert the given block into a list of 4 byte integers SCREAMING_SNAKE_CASE : Union[str, Any] = list(struct.unpack(">16L" , UpperCAmelCase_ ) ) # add 48 0-ed integers words += [0] * 48 SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = self.hashes for index in range(0 , 64 ): if index > 15: # modify the zero-ed indexes at the end of the array SCREAMING_SNAKE_CASE : Optional[int] = ( self.ror(words[index - 15] , 7 ) ^ self.ror(words[index - 15] , 18 ) ^ (words[index - 15] >> 3) ) SCREAMING_SNAKE_CASE : int = ( self.ror(words[index - 2] , 17 ) ^ self.ror(words[index - 2] , 19 ) ^ (words[index - 2] >> 10) ) SCREAMING_SNAKE_CASE : List[str] = ( words[index - 16] + sa + words[index - 7] + sa ) % 0X1_00_00_00_00 # Compression SCREAMING_SNAKE_CASE : Any = self.ror(UpperCAmelCase_ , 6 ) ^ self.ror(UpperCAmelCase_ , 11 ) ^ self.ror(UpperCAmelCase_ , 25 ) SCREAMING_SNAKE_CASE : Any = (e & f) ^ ((~e & 0XFF_FF_FF_FF) & g) SCREAMING_SNAKE_CASE : Dict = ( h + sa + ch + self.round_constants[index] + words[index] ) % 0X1_00_00_00_00 SCREAMING_SNAKE_CASE : List[Any] = self.ror(UpperCAmelCase_ , 2 ) ^ self.ror(UpperCAmelCase_ , 13 ) ^ self.ror(UpperCAmelCase_ , 22 ) SCREAMING_SNAKE_CASE : Optional[Any] = (a & b) ^ (a & c) ^ (b & c) SCREAMING_SNAKE_CASE : Optional[int] = (sa + maj) % 0X1_00_00_00_00 SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = ( g, f, e, ((d + tempa) % 0X1_00_00_00_00), c, b, a, ((tempa + tempa) % 0X1_00_00_00_00), ) SCREAMING_SNAKE_CASE : List[str] = [a, b, c, d, e, f, g, h] # Modify final values SCREAMING_SNAKE_CASE : Union[str, Any] = [ ((element + mutated_hash_values[index]) % 0X1_00_00_00_00) for index, element in enumerate(self.hashes ) ] SCREAMING_SNAKE_CASE : Any = "".join([hex(UpperCAmelCase_ )[2:].zfill(8 ) for value in self.hashes] ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : int ): return 0XFF_FF_FF_FF & (value << (32 - rotations)) | (value >> rotations) class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Optional[Any] ): import hashlib SCREAMING_SNAKE_CASE : int = bytes("Test String" , "utf-8" ) self.assertEqual(SHAaaa(UpperCAmelCase_ ).hash , hashlib.shaaaa(UpperCAmelCase_ ).hexdigest() ) def lowerCamelCase__ ( ): """simple docstring""" import doctest doctest.testmod() SCREAMING_SNAKE_CASE : int = argparse.ArgumentParser() parser.add_argument( "-s" , "--string" , dest="input_string" , default="Hello World!! Welcome to Cryptography" , help="Hash the string" , ) parser.add_argument( "-f" , "--file" , dest="input_file" , help="Hash contents of a file" ) SCREAMING_SNAKE_CASE : Tuple = parser.parse_args() SCREAMING_SNAKE_CASE : str = args.input_string # hash input should be a bytestring if args.input_file: with open(args.input_file , "rb" ) as f: SCREAMING_SNAKE_CASE : int = f.read() else: SCREAMING_SNAKE_CASE : Optional[Any] = bytes(lowercase , "utf-8" ) print(SHAaaa(lowercase ).hash ) if __name__ == "__main__": main()
62
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return 1 if input_a == input_a else 0 def lowerCamelCase__ ( ): """simple docstring""" assert xnor_gate(0 , 0 ) == 1 assert xnor_gate(0 , 1 ) == 0 assert xnor_gate(1 , 0 ) == 0 assert xnor_gate(1 , 1 ) == 1 if __name__ == "__main__": print(xnor_gate(0, 0)) print(xnor_gate(0, 1)) print(xnor_gate(1, 0)) print(xnor_gate(1, 1))
62
1
import os from collections.abc import Iterator def lowerCamelCase__ ( lowercase = "." ): """simple docstring""" for dir_path, dir_names, filenames in os.walk(lowercase ): SCREAMING_SNAKE_CASE : Tuple = [d for d in dir_names if d != "scripts" and d[0] not in "._"] for filename in filenames: if filename == "__init__.py": continue if os.path.splitext(lowercase )[1] in (".py", ".ipynb"): yield os.path.join(lowercase , lowercase ).lstrip("./" ) def lowerCamelCase__ ( lowercase ): """simple docstring""" return F'''{i * ' '}*''' if i else "\n##" def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = old_path.split(os.sep ) for i, new_part in enumerate(new_path.split(os.sep ) ): if (i + 1 > len(lowercase ) or old_parts[i] != new_part) and new_part: print(F'''{md_prefix(lowercase )} {new_part.replace('_' , ' ' ).title()}''' ) return new_path def lowerCamelCase__ ( lowercase = "." ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = "" for filepath in sorted(good_file_paths(lowercase ) ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = os.path.split(lowercase ) if filepath != old_path: SCREAMING_SNAKE_CASE : Dict = print_path(lowercase , lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = (filepath.count(os.sep ) + 1) if filepath else 0 SCREAMING_SNAKE_CASE : int = F'''{filepath}/{filename}'''.replace(" " , "%20" ) SCREAMING_SNAKE_CASE : Tuple = os.path.splitext(filename.replace("_" , " " ).title() )[0] print(F'''{md_prefix(lowercase )} [{filename}]({url})''' ) if __name__ == "__main__": print_directory_md(""".""")
62
import math import flax.linen as nn import jax.numpy as jnp def lowerCamelCase__ ( lowercase , lowercase , lowercase = 1 , lowercase = 1 , lowercase = 1.0E4 , lowercase = False , lowercase = 1.0 , ): """simple docstring""" assert timesteps.ndim == 1, "Timesteps should be a 1d-array" assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even''' SCREAMING_SNAKE_CASE : Union[str, Any] = float(embedding_dim // 2 ) SCREAMING_SNAKE_CASE : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift) SCREAMING_SNAKE_CASE : Optional[Any] = min_timescale * jnp.exp(jnp.arange(lowercase , dtype=jnp.floataa ) * -log_timescale_increment ) SCREAMING_SNAKE_CASE : Optional[int] = jnp.expand_dims(lowercase , 1 ) * jnp.expand_dims(lowercase , 0 ) # scale embeddings SCREAMING_SNAKE_CASE : Optional[int] = scale * emb if flip_sin_to_cos: SCREAMING_SNAKE_CASE : List[Any] = jnp.concatenate([jnp.cos(lowercase ), jnp.sin(lowercase )] , axis=1 ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = jnp.concatenate([jnp.sin(lowercase ), jnp.cos(lowercase )] , axis=1 ) SCREAMING_SNAKE_CASE : Tuple = jnp.reshape(lowercase , [jnp.shape(lowercase )[0], embedding_dim] ) return signal class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : jnp.dtype = jnp.floataa @nn.compact def __call__( self : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_1" )(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.silu(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_2" )(UpperCAmelCase_ ) return temb class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : bool = False UpperCamelCase_ : float = 1 @nn.compact def __call__( self : Optional[int] , UpperCAmelCase_ : int ): return get_sinusoidal_embeddings( UpperCAmelCase_ , embedding_dim=self.dim , flip_sin_to_cos=self.flip_sin_to_cos , freq_shift=self.freq_shift )
62
1
import argparse import json from pathlib import Path import requests import torch from huggingface_hub import hf_hub_download from PIL import Image from transformers import ( MobileViTConfig, MobileViTForImageClassification, MobileViTForSemanticSegmentation, MobileViTImageProcessor, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = MobileViTConfig() # size of the architecture if "mobilevit_s" in mobilevit_name: SCREAMING_SNAKE_CASE : List[Any] = [144, 192, 240] SCREAMING_SNAKE_CASE : Tuple = [16, 32, 64, 96, 128, 160, 640] elif "mobilevit_xs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [96, 120, 144] SCREAMING_SNAKE_CASE : Dict = [16, 32, 48, 64, 80, 96, 384] elif "mobilevit_xxs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [64, 80, 96] SCREAMING_SNAKE_CASE : List[str] = [16, 16, 24, 48, 64, 80, 320] SCREAMING_SNAKE_CASE : int = 0.05 SCREAMING_SNAKE_CASE : int = 2.0 if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : str = 512 SCREAMING_SNAKE_CASE : List[str] = 16 SCREAMING_SNAKE_CASE : Union[str, Any] = 21 SCREAMING_SNAKE_CASE : Dict = "pascal-voc-id2label.json" else: SCREAMING_SNAKE_CASE : Optional[Any] = 1000 SCREAMING_SNAKE_CASE : Optional[Any] = "imagenet-1k-id2label.json" SCREAMING_SNAKE_CASE : Any = "huggingface/label-files" SCREAMING_SNAKE_CASE : Tuple = json.load(open(hf_hub_download(lowercase , lowercase , repo_type="dataset" ) , "r" ) ) SCREAMING_SNAKE_CASE : List[str] = {int(lowercase ): v for k, v in idalabel.items()} SCREAMING_SNAKE_CASE : Optional[Any] = idalabel SCREAMING_SNAKE_CASE : str = {v: k for k, v in idalabel.items()} return config def lowerCamelCase__ ( lowercase , lowercase=False ): """simple docstring""" for i in range(1 , 6 ): if F'''layer_{i}.''' in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(F'''layer_{i}.''' , F'''encoder.layer.{i - 1}.''' ) if "conv_1." in name: SCREAMING_SNAKE_CASE : Dict = name.replace("conv_1." , "conv_stem." ) if ".block." in name: SCREAMING_SNAKE_CASE : List[str] = name.replace(".block." , "." ) if "exp_1x1" in name: SCREAMING_SNAKE_CASE : str = name.replace("exp_1x1" , "expand_1x1" ) if "red_1x1" in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("red_1x1" , "reduce_1x1" ) if ".local_rep.conv_3x3." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".local_rep.conv_3x3." , ".conv_kxk." ) if ".local_rep.conv_1x1." in name: SCREAMING_SNAKE_CASE : int = name.replace(".local_rep.conv_1x1." , ".conv_1x1." ) if ".norm." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".norm." , ".normalization." ) if ".conv." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".conv." , ".convolution." ) if ".conv_proj." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".conv_proj." , ".conv_projection." ) for i in range(0 , 2 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.layer.{j}.''' ) for i in range(2 , 6 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.''' ) if "expand_1x1" in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace("expand_1x1" , "downsampling_layer.expand_1x1" ) if "conv_3x3" in name: SCREAMING_SNAKE_CASE : str = name.replace("conv_3x3" , "downsampling_layer.conv_3x3" ) if "reduce_1x1" in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("reduce_1x1" , "downsampling_layer.reduce_1x1" ) for i in range(2 , 5 ): if F'''.global_rep.{i}.weight''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.global_rep.{i}.weight''' , ".layernorm.weight" ) if F'''.global_rep.{i}.bias''' in name: SCREAMING_SNAKE_CASE : str = name.replace(F'''.global_rep.{i}.bias''' , ".layernorm.bias" ) if ".global_rep." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".global_rep." , ".transformer." ) if ".pre_norm_mha.0." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_mha.0." , ".layernorm_before." ) if ".pre_norm_mha.1.out_proj." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_mha.1.out_proj." , ".attention.output.dense." ) if ".pre_norm_ffn.0." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(".pre_norm_ffn.0." , ".layernorm_after." ) if ".pre_norm_ffn.1." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_ffn.1." , ".intermediate.dense." ) if ".pre_norm_ffn.4." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_ffn.4." , ".output.dense." ) if ".transformer." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".transformer." , ".transformer.layer." ) if ".aspp_layer." in name: SCREAMING_SNAKE_CASE : int = name.replace(".aspp_layer." , "." ) if ".aspp_pool." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".aspp_pool." , "." ) if "seg_head." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("seg_head." , "segmentation_head." ) if "segmentation_head.classifier.classifier." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace("segmentation_head.classifier.classifier." , "segmentation_head.classifier." ) if "classifier.fc." in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("classifier.fc." , "classifier." ) elif (not base_model) and ("segmentation_head." not in name): SCREAMING_SNAKE_CASE : List[Any] = "mobilevit." + name return name def lowerCamelCase__ ( lowercase , lowercase , lowercase=False ): """simple docstring""" if base_model: SCREAMING_SNAKE_CASE : Optional[int] = "" else: SCREAMING_SNAKE_CASE : Any = "mobilevit." for key in orig_state_dict.copy().keys(): SCREAMING_SNAKE_CASE : Union[str, Any] = orig_state_dict.pop(lowercase ) if key[:8] == "encoder.": SCREAMING_SNAKE_CASE : int = key[8:] if "qkv" in key: SCREAMING_SNAKE_CASE : Optional[int] = key.split("." ) SCREAMING_SNAKE_CASE : Any = int(key_split[0][6:] ) - 1 SCREAMING_SNAKE_CASE : List[Any] = int(key_split[3] ) SCREAMING_SNAKE_CASE : List[Any] = model.get_submodule(F'''{model_prefix}encoder.layer.{layer_num}''' ) SCREAMING_SNAKE_CASE : int = layer.transformer.layer[transformer_num].attention.attention.all_head_size SCREAMING_SNAKE_CASE : Union[str, Any] = ( F'''{model_prefix}encoder.layer.{layer_num}.transformer.layer.{transformer_num}.attention.attention.''' ) if "weight" in key: SCREAMING_SNAKE_CASE : Optional[int] = val[:dim, :] SCREAMING_SNAKE_CASE : Tuple = val[dim : dim * 2, :] SCREAMING_SNAKE_CASE : Dict = val[-dim:, :] else: SCREAMING_SNAKE_CASE : str = val[:dim] SCREAMING_SNAKE_CASE : Union[str, Any] = val[dim : dim * 2] SCREAMING_SNAKE_CASE : Union[str, Any] = val[-dim:] else: SCREAMING_SNAKE_CASE : List[Any] = val return orig_state_dict def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = "http://images.cocodataset.org/val2017/000000039769.jpg" SCREAMING_SNAKE_CASE : Union[str, Any] = Image.open(requests.get(lowercase , stream=lowercase ).raw ) return im @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase=False ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = get_mobilevit_config(lowercase ) # load original state_dict SCREAMING_SNAKE_CASE : Union[str, Any] = torch.load(lowercase , map_location="cpu" ) # load 🤗 model if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : List[str] = MobileViTForSemanticSegmentation(lowercase ).eval() else: SCREAMING_SNAKE_CASE : str = MobileViTForImageClassification(lowercase ).eval() SCREAMING_SNAKE_CASE : Any = convert_state_dict(lowercase , lowercase ) model.load_state_dict(lowercase ) # Check outputs on an image, prepared by MobileViTImageProcessor SCREAMING_SNAKE_CASE : List[Any] = MobileViTImageProcessor(crop_size=config.image_size , size=config.image_size + 32 ) SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=prepare_img() , return_tensors="pt" ) SCREAMING_SNAKE_CASE : List[Any] = model(**lowercase ) SCREAMING_SNAKE_CASE : str = outputs.logits if mobilevit_name.startswith("deeplabv3_" ): assert logits.shape == (1, 21, 32, 32) if mobilevit_name == "deeplabv3_mobilevit_s": SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor( [ [[6.2065, 6.1292, 6.2070], [6.1079, 6.1254, 6.1747], [6.0042, 6.1071, 6.1034]], [[-6.9253, -6.8653, -7.0398], [-7.3218, -7.3983, -7.3670], [-7.1961, -7.2482, -7.1569]], [[-4.4723, -4.4348, -4.3769], [-5.3629, -5.4632, -5.4598], [-5.1587, -5.3402, -5.5059]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor( [ [[5.4449, 5.5733, 5.6314], [5.1815, 5.3930, 5.5963], [5.1656, 5.4333, 5.4853]], [[-9.4423, -9.7766, -9.6714], [-9.1581, -9.5720, -9.5519], [-9.1006, -9.6458, -9.5703]], [[-7.7721, -7.3716, -7.1583], [-8.4599, -8.0624, -7.7944], [-8.4172, -7.8366, -7.5025]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xxs": SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [ [[6.9811, 6.9743, 7.3123], [7.1777, 7.1931, 7.3938], [7.5633, 7.8050, 7.8901]], [[-10.5536, -10.2332, -10.2924], [-10.2336, -9.8624, -9.5964], [-10.8840, -10.8158, -10.6659]], [[-3.4938, -3.0631, -2.8620], [-3.4205, -2.8135, -2.6875], [-3.4179, -2.7945, -2.8750]], ] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3, :3, :3] , lowercase , atol=1E-4 ) else: assert logits.shape == (1, 1000) if mobilevit_name == "mobilevit_s": SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([-0.9866, 0.2392, -1.1241] ) elif mobilevit_name == "mobilevit_xs": SCREAMING_SNAKE_CASE : Dict = torch.tensor([-2.4761, -0.9399, -1.9587] ) elif mobilevit_name == "mobilevit_xxs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor([-1.9364, -1.2327, -0.4653] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3] , lowercase , atol=1E-4 ) Path(lowercase ).mkdir(exist_ok=lowercase ) print(F'''Saving model {mobilevit_name} to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowercase ) print(F'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowercase ) if push_to_hub: SCREAMING_SNAKE_CASE : List[str] = { "mobilevit_s": "mobilevit-small", "mobilevit_xs": "mobilevit-x-small", "mobilevit_xxs": "mobilevit-xx-small", "deeplabv3_mobilevit_s": "deeplabv3-mobilevit-small", "deeplabv3_mobilevit_xs": "deeplabv3-mobilevit-x-small", "deeplabv3_mobilevit_xxs": "deeplabv3-mobilevit-xx-small", } print("Pushing to the hub..." ) SCREAMING_SNAKE_CASE : int = model_mapping[mobilevit_name] image_processor.push_to_hub(lowercase , organization="apple" ) model.push_to_hub(lowercase , organization="apple" ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument( """--mobilevit_name""", default="""mobilevit_s""", type=str, help=( """Name of the MobileViT model you'd like to convert. Should be one of 'mobilevit_s', 'mobilevit_xs',""" """ 'mobilevit_xxs', 'deeplabv3_mobilevit_s', 'deeplabv3_mobilevit_xs', 'deeplabv3_mobilevit_xxs'.""" ), ) parser.add_argument( """--checkpoint_path""", required=True, type=str, help="""Path to the original state dict (.pt file).""" ) parser.add_argument( """--pytorch_dump_folder_path""", required=True, type=str, help="""Path to the output PyTorch model directory.""" ) parser.add_argument( """--push_to_hub""", action="""store_true""", help="""Whether or not to push the converted model to the 🤗 hub.""" ) snake_case = parser.parse_args() convert_movilevit_checkpoint( args.mobilevit_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub )
62
# Copyright 2023 The HuggingFace Inc. team. All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import numpy as np import torch from ..models.clipseg import CLIPSegForImageSegmentation from ..utils import is_vision_available, requires_backends from .base import PipelineTool if is_vision_available(): from PIL import Image class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = ( '''This is a tool that creates a segmentation mask of an image according to a label. It cannot create an image.''' '''It takes two arguments named `image` which should be the original image, and `label` which should be a text ''' '''describing the elements what should be identified in the segmentation mask. The tool returns the mask.''' ) UpperCamelCase_ : Union[str, Any] = '''CIDAS/clipseg-rd64-refined''' UpperCamelCase_ : Any = '''image_segmenter''' UpperCamelCase_ : int = CLIPSegForImageSegmentation UpperCamelCase_ : Optional[Any] = ['''image''', '''text'''] UpperCamelCase_ : int = ['''image'''] def __init__( self : str , *UpperCAmelCase_ : Union[str, Any] , **UpperCAmelCase_ : Union[str, Any] ): requires_backends(self , ["vision"] ) super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : "Image" , UpperCAmelCase_ : str ): return self.pre_processor(text=[label] , images=[image] , padding=UpperCAmelCase_ , return_tensors="pt" ) def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): with torch.no_grad(): SCREAMING_SNAKE_CASE : str = self.model(**UpperCAmelCase_ ).logits return logits def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = outputs.cpu().detach().numpy() SCREAMING_SNAKE_CASE : Optional[int] = 0 SCREAMING_SNAKE_CASE : str = 1 return Image.fromarray((array * 255).astype(np.uinta ) )
62
1
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_fnet import FNetTokenizer else: snake_case = None snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """spiece.model""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """google/fnet-base""": """https://huggingface.co/google/fnet-base/resolve/main/spiece.model""", """google/fnet-large""": """https://huggingface.co/google/fnet-large/resolve/main/spiece.model""", }, """tokenizer_file""": { """google/fnet-base""": """https://huggingface.co/google/fnet-base/resolve/main/tokenizer.json""", """google/fnet-large""": """https://huggingface.co/google/fnet-large/resolve/main/tokenizer.json""", }, } snake_case = { """google/fnet-base""": 512, """google/fnet-large""": 512, } snake_case = """▁""" class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[int] = VOCAB_FILES_NAMES UpperCamelCase_ : str = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Dict = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : str = ['''input_ids''', '''token_type_ids'''] UpperCamelCase_ : Optional[int] = FNetTokenizer def __init__( self : Tuple , UpperCAmelCase_ : Dict=None , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : int=False , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : Any="<unk>" , UpperCAmelCase_ : Optional[Any]="[SEP]" , UpperCAmelCase_ : str="<pad>" , UpperCAmelCase_ : Union[str, Any]="[CLS]" , UpperCAmelCase_ : int="[MASK]" , **UpperCAmelCase_ : Dict , ): # 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. SCREAMING_SNAKE_CASE : str = ( AddedToken(UpperCAmelCase_ , lstrip=UpperCAmelCase_ , rstrip=UpperCAmelCase_ , normalized=UpperCAmelCase_ ) if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) else mask_token ) super().__init__( UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , do_lower_case=UpperCAmelCase_ , remove_space=UpperCAmelCase_ , keep_accents=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : str = do_lower_case SCREAMING_SNAKE_CASE : Optional[int] = remove_space SCREAMING_SNAKE_CASE : Tuple = keep_accents SCREAMING_SNAKE_CASE : List[str] = vocab_file SCREAMING_SNAKE_CASE : Union[str, Any] = False if not self.vocab_file else True def _A ( self : int , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : Union[str, Any] = [self.sep_token_id] SCREAMING_SNAKE_CASE : List[str] = [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 _A ( self : Any , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : Dict = [self.sep_token_id] SCREAMING_SNAKE_CASE : Union[str, Any] = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1] def _A ( self : str , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): if not os.path.isdir(UpperCAmelCase_ ): logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' ) return SCREAMING_SNAKE_CASE : Any = os.path.join( UpperCAmelCase_ , (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"] ) if os.path.abspath(self.vocab_file ) != os.path.abspath(UpperCAmelCase_ ): copyfile(self.vocab_file , UpperCAmelCase_ ) return (out_vocab_file,)
62
import json import os import unittest from transformers.models.blenderbot_small.tokenization_blenderbot_small import ( VOCAB_FILES_NAMES, BlenderbotSmallTokenizer, ) from ...test_tokenization_common import TokenizerTesterMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = BlenderbotSmallTokenizer UpperCamelCase_ : int = False def _A ( self : Union[str, Any] ): super().setUp() SCREAMING_SNAKE_CASE : List[Any] = ["__start__", "adapt", "act", "ap@@", "te", "__end__", "__unk__"] SCREAMING_SNAKE_CASE : Optional[Any] = dict(zip(UpperCAmelCase_ , range(len(UpperCAmelCase_ ) ) ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = ["#version: 0.2", "a p", "t e</w>", "ap t</w>", "a d", "ad apt</w>", "a c", "ac t</w>", ""] SCREAMING_SNAKE_CASE : int = {"unk_token": "__unk__", "bos_token": "__start__", "eos_token": "__end__"} SCREAMING_SNAKE_CASE : Union[str, Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["vocab_file"] ) SCREAMING_SNAKE_CASE : str = 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(UpperCAmelCase_ ) + "\n" ) with open(self.merges_file , "w" , encoding="utf-8" ) as fp: fp.write("\n".join(UpperCAmelCase_ ) ) def _A ( self : List[Any] , **UpperCAmelCase_ : str ): kwargs.update(self.special_tokens_map ) return BlenderbotSmallTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] , UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : int = "adapt act apte" return input_text, output_text def _A ( self : str ): SCREAMING_SNAKE_CASE : int = BlenderbotSmallTokenizer(self.vocab_file , self.merges_file , **self.special_tokens_map ) SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : List[str] = ["adapt", "act", "ap@@", "te"] SCREAMING_SNAKE_CASE : Optional[Any] = tokenizer.tokenize(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = [tokenizer.bos_token] + tokens + [tokenizer.eos_token] SCREAMING_SNAKE_CASE : Tuple = [0, 1, 2, 3, 4, 5] self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , UpperCAmelCase_ ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Union[str, Any] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) assert tok("sam" ).input_ids == [1384] SCREAMING_SNAKE_CASE : str = "I am a small frog." SCREAMING_SNAKE_CASE : List[Any] = tok([src_text] , padding=UpperCAmelCase_ , truncation=UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : int = tok.batch_decode(UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ , clean_up_tokenization_spaces=UpperCAmelCase_ )[0] assert src_text != decoded # I wish it did! assert decoded == "i am a small frog ." def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : List[str] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) SCREAMING_SNAKE_CASE : Tuple = "I am a small frog ." SCREAMING_SNAKE_CASE : Optional[int] = "." SCREAMING_SNAKE_CASE : Dict = tok(UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : Optional[Any] = tok(UpperCAmelCase_ )["input_ids"] assert encoded[-1] == encoded_dot[0]
62
1
import unittest from transformers.testing_utils import require_bsa from transformers.utils import is_bsa_available from ...test_feature_extraction_common import FeatureExtractionSavingTestMixin if is_bsa_available(): from transformers import MarkupLMFeatureExtractor class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def __init__( self : int , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Dict = parent def _A ( self : Dict ): return {} def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = "<HTML>\n\n <HEAD>\n <TITLE>sample document</TITLE>\n </HEAD>\n\n <BODY BGCOLOR=\"FFFFFF\">\n <HR>\n <a href=\"http://google.com\">Goog</a>\n <H1>This is one header</H1>\n <H2>This is a another Header</H2>\n <P>Travel from\n <P>\n <B>SFO to JFK</B>\n <BR>\n <B><I>on May 2, 2015 at 2:00 pm. For details go to confirm.com </I></B>\n <HR>\n <div style=\"color:#0000FF\">\n <h3>Traveler <b> name </b> is\n <p> John Doe </p>\n </div>" SCREAMING_SNAKE_CASE : Union[str, Any] = "\n <!DOCTYPE html>\n <html>\n <body>\n\n <h1>My First Heading</h1>\n <p>My first paragraph.</p>\n\n </body>\n </html>\n " return [html_string_a, html_string_a] @require_bsa class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Any = MarkupLMFeatureExtractor if is_bsa_available() else None def _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = MarkupLMFeatureExtractionTester(self ) @property def _A ( self : Union[str, Any] ): return self.feature_extract_tester.prepare_feat_extract_dict() def _A ( self : Optional[Any] ): # Initialize feature_extractor SCREAMING_SNAKE_CASE : Optional[int] = self.feature_extraction_class() # Test not batched input SCREAMING_SNAKE_CASE : Dict = get_html_strings()[0] SCREAMING_SNAKE_CASE : Dict = feature_extractor(UpperCAmelCase_ ) # fmt: off SCREAMING_SNAKE_CASE : Dict = [["sample document", "Goog", "This is one header", "This is a another Header", "Travel from", "SFO to JFK", "on May 2, 2015 at 2:00 pm. For details go to confirm.com", "Traveler", "name", "is", "John Doe"]] SCREAMING_SNAKE_CASE : Any = [["/html/head/title", "/html/body/a", "/html/body/h1", "/html/body/h2", "/html/body/p", "/html/body/p/p/b[1]", "/html/body/p/p/b[2]/i", "/html/body/p/p/div/h3", "/html/body/p/p/div/h3/b", "/html/body/p/p/div/h3", "/html/body/p/p/div/h3/p"]] # fmt: on self.assertEqual(encoding.nodes , UpperCAmelCase_ ) self.assertEqual(encoding.xpaths , UpperCAmelCase_ ) # Test batched SCREAMING_SNAKE_CASE : Tuple = get_html_strings() SCREAMING_SNAKE_CASE : Dict = feature_extractor(UpperCAmelCase_ ) # fmt: off SCREAMING_SNAKE_CASE : List[Any] = expected_nodes + [["My First Heading", "My first paragraph."]] SCREAMING_SNAKE_CASE : int = expected_xpaths + [["/html/body/h1", "/html/body/p"]] self.assertEqual(len(encoding.nodes ) , 2 ) self.assertEqual(len(encoding.xpaths ) , 2 ) self.assertEqual(encoding.nodes , UpperCAmelCase_ ) self.assertEqual(encoding.xpaths , UpperCAmelCase_ )
62
import argparse from collections import OrderedDict from pathlib import Path import torch from transformers import ( VisualBertConfig, VisualBertForMultipleChoice, VisualBertForPreTraining, VisualBertForQuestionAnswering, VisualBertForVisualReasoning, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) snake_case = [ ("""bert.bert""", """visual_bert"""), ("""bert.cls""", """cls"""), ("""bert.classifier""", """cls"""), ("""token_type_embeddings_visual""", """visual_token_type_embeddings"""), ("""position_embeddings_visual""", """visual_position_embeddings"""), ("""projection""", """visual_projection"""), ] snake_case = [ """nlvr2_coco_pre_trained.th""", """nlvr2_fine_tuned.th""", """nlvr2_pre_trained.th""", """vcr_coco_pre_train.th""", """vcr_fine_tune.th""", """vcr_pre_train.th""", """vqa_coco_pre_trained.th""", """vqa_fine_tuned.th""", """vqa_pre_trained.th""", ] def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = torch.load(lowercase , map_location="cpu" ) return sd def lowerCamelCase__ ( lowercase , lowercase , lowercase=rename_keys_prefix ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = OrderedDict() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.arange(config.max_position_embeddings ).expand((1, -1) ) # detector_d = OrderedDict() for key in d: if "detector" in key: # detector_d[key.replace('detector.','')] = d[key] continue SCREAMING_SNAKE_CASE : Union[str, Any] = key for name_pair in rename_keys_prefix: SCREAMING_SNAKE_CASE : Any = new_key.replace(name_pair[0] , name_pair[1] ) SCREAMING_SNAKE_CASE : Dict = d[key] if key == "bert.cls.predictions.decoder.weight": # Old bert code didn't have `decoder.bias`, but was added separately SCREAMING_SNAKE_CASE : Union[str, Any] = new_d["cls.predictions.bias"] return new_d @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( checkpoint_path.split("/" )[-1] in ACCEPTABLE_CHECKPOINTS ), F'''The checkpoint provided must be in {ACCEPTABLE_CHECKPOINTS}.''' # Get Config if "pre" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = "pretraining" if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = {"visual_embedding_dim": 512} elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = {"visual_embedding_dim": 2048} elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Any = {"visual_embedding_dim": 2048} elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 1024} else: raise NotImplementedError(F'''No implementation found for `{checkpoint_path}`.''' ) else: if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 512} SCREAMING_SNAKE_CASE : Tuple = "multichoice" elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[str] = {"visual_embedding_dim": 2048} SCREAMING_SNAKE_CASE : str = "vqa_advanced" elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Tuple = {"visual_embedding_dim": 2048, "num_labels": 3129} SCREAMING_SNAKE_CASE : Optional[Any] = "vqa" elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = { "visual_embedding_dim": 1024, "num_labels": 2, } SCREAMING_SNAKE_CASE : Tuple = "nlvr" SCREAMING_SNAKE_CASE : List[Any] = VisualBertConfig(**lowercase ) # Load State Dict SCREAMING_SNAKE_CASE : List[str] = load_state_dict(lowercase ) SCREAMING_SNAKE_CASE : Any = get_new_dict(lowercase , lowercase ) if model_type == "pretraining": SCREAMING_SNAKE_CASE : int = VisualBertForPreTraining(lowercase ) elif model_type == "vqa": SCREAMING_SNAKE_CASE : Tuple = VisualBertForQuestionAnswering(lowercase ) elif model_type == "nlvr": SCREAMING_SNAKE_CASE : List[Any] = VisualBertForVisualReasoning(lowercase ) elif model_type == "multichoice": SCREAMING_SNAKE_CASE : Optional[Any] = VisualBertForMultipleChoice(lowercase ) model.load_state_dict(lowercase ) # Save Checkpoints Path(lowercase ).mkdir(exist_ok=lowercase ) model.save_pretrained(lowercase ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument("""orig_checkpoint_path""", type=str, help="""A path to .th on local filesystem.""") parser.add_argument("""pytorch_dump_folder_path""", type=str, help="""Path to the output PyTorch model.""") snake_case = parser.parse_args() convert_visual_bert_checkpoint(args.orig_checkpoint_path, args.pytorch_dump_folder_path)
62
1
import enum import shutil import sys snake_case , snake_case = shutil.get_terminal_size() snake_case = {"""UP""": """A""", """DOWN""": """B""", """RIGHT""": """C""", """LEFT""": """D"""} class SCREAMING_SNAKE_CASE ( enum.Enum ): '''simple docstring''' UpperCamelCase_ : Union[str, Any] = 0 UpperCamelCase_ : Any = 1 def lowerCamelCase__ ( lowercase , lowercase="" ): """simple docstring""" sys.stdout.write(str(lowercase ) + end ) sys.stdout.flush() def lowerCamelCase__ ( lowercase , lowercase , lowercase="" ): """simple docstring""" forceWrite(F'''\u001b[{color}m{content}\u001b[0m''' , lowercase ) def lowerCamelCase__ ( ): """simple docstring""" forceWrite("\r" ) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" forceWrite(F'''\033[{num_lines}{CURSOR_TO_CHAR[direction.upper()]}''' ) def lowerCamelCase__ ( ): """simple docstring""" forceWrite(" " * TERMINAL_WIDTH ) reset_cursor() def lowerCamelCase__ ( ): """simple docstring""" reset_cursor() forceWrite("-" * TERMINAL_WIDTH )
62
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available snake_case = { """configuration_jukebox""": [ """JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP""", """JukeboxConfig""", """JukeboxPriorConfig""", """JukeboxVQVAEConfig""", ], """tokenization_jukebox""": ["""JukeboxTokenizer"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST""", """JukeboxModel""", """JukeboxPreTrainedModel""", """JukeboxVQVAE""", """JukeboxPrior""", ] if TYPE_CHECKING: from .configuration_jukebox import ( JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP, JukeboxConfig, JukeboxPriorConfig, JukeboxVQVAEConfig, ) from .tokenization_jukebox import JukeboxTokenizer try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_jukebox import ( JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST, JukeboxModel, JukeboxPreTrainedModel, JukeboxPrior, JukeboxVQVAE, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
1
from __future__ import annotations from math import pi def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if (inductance, frequency, reactance).count(0 ) != 1: raise ValueError("One and only one argument must be 0" ) if inductance < 0: raise ValueError("Inductance cannot be negative" ) if frequency < 0: raise ValueError("Frequency cannot be negative" ) if reactance < 0: raise ValueError("Inductive reactance cannot be negative" ) if inductance == 0: return {"inductance": reactance / (2 * pi * frequency)} elif frequency == 0: return {"frequency": reactance / (2 * pi * inductance)} elif reactance == 0: return {"reactance": 2 * pi * frequency * inductance} else: raise ValueError("Exactly one argument must be 0" ) if __name__ == "__main__": import doctest doctest.testmod()
62
from typing import Dict, List from nltk.translate import gleu_score import datasets from datasets import MetricInfo snake_case = """\ @misc{wu2016googles, title={Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation}, author={Yonghui Wu and Mike Schuster and Zhifeng Chen and Quoc V. Le and Mohammad Norouzi and Wolfgang Macherey and Maxim Krikun and Yuan Cao and Qin Gao and Klaus Macherey and Jeff Klingner and Apurva Shah and Melvin Johnson and Xiaobing Liu and Łukasz Kaiser and Stephan Gouws and Yoshikiyo Kato and Taku Kudo and Hideto Kazawa and Keith Stevens and George Kurian and Nishant Patil and Wei Wang and Cliff Young and Jason Smith and Jason Riesa and Alex Rudnick and Oriol Vinyals and Greg Corrado and Macduff Hughes and Jeffrey Dean}, year={2016}, eprint={1609.08144}, archivePrefix={arXiv}, primaryClass={cs.CL} } """ snake_case = """\ The BLEU score has some undesirable properties when used for single sentences, as it was designed to be a corpus measure. We therefore use a slightly different score for our RL experiments which we call the 'GLEU score'. For the GLEU score, we record all sub-sequences of 1, 2, 3 or 4 tokens in output and target sequence (n-grams). We then compute a recall, which is the ratio of the number of matching n-grams to the number of total n-grams in the target (ground truth) sequence, and a precision, which is the ratio of the number of matching n-grams to the number of total n-grams in the generated output sequence. Then GLEU score is simply the minimum of recall and precision. This GLEU score's range is always between 0 (no matches) and 1 (all match) and it is symmetrical when switching output and target. According to our experiments, GLEU score correlates quite well with the BLEU metric on a corpus level but does not have its drawbacks for our per sentence reward objective. """ snake_case = """\ Computes corpus-level Google BLEU (GLEU) score of translated segments against one or more references. Instead of averaging the sentence level GLEU scores (i.e. macro-average precision), Wu et al. (2016) sum up the matching tokens and the max of hypothesis and reference tokens for each sentence, then compute using the aggregate values. Args: predictions (list of str): list of translations to score. Each translation should be tokenized into a list of tokens. references (list of list of str): list of lists of references for each translation. Each reference should be tokenized into a list of tokens. min_len (int): The minimum order of n-gram this function should extract. Defaults to 1. max_len (int): The maximum order of n-gram this function should extract. Defaults to 4. Returns: 'google_bleu': google_bleu score Examples: Example 1: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.44 Example 2: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.61 Example 3: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references, min_len=2) >>> print(round(results[\"google_bleu\"], 2)) 0.53 Example 4: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses,references=list_of_references, min_len=2, max_len=6) >>> print(round(results[\"google_bleu\"], 2)) 0.4 """ @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : Optional[Any] ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { "predictions": datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ), "references": datasets.Sequence( datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ) , id="references" ), } ) , ) def _A ( self : List[Any] , UpperCAmelCase_ : List[List[List[str]]] , UpperCAmelCase_ : List[List[str]] , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : int = 4 , ): return { "google_bleu": gleu_score.corpus_gleu( list_of_references=UpperCAmelCase_ , hypotheses=UpperCAmelCase_ , min_len=UpperCAmelCase_ , max_len=UpperCAmelCase_ ) }
62
1
import gc import random import unittest import numpy as np import torch from diffusers import DDIMScheduler, KandinskyVaaPipeline, KandinskyVaaPriorPipeline, UNetaDConditionModel, VQModel from diffusers.utils import floats_tensor, load_numpy, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = KandinskyVaaPipeline UpperCamelCase_ : List[Any] = [ '''image_embeds''', '''negative_image_embeds''', ] UpperCamelCase_ : Tuple = ['''image_embeds''', '''negative_image_embeds'''] UpperCamelCase_ : Any = [ '''generator''', '''height''', '''width''', '''latents''', '''guidance_scale''', '''num_inference_steps''', '''return_dict''', '''guidance_scale''', '''num_images_per_prompt''', '''output_type''', '''return_dict''', ] UpperCamelCase_ : List[str] = False @property def _A ( self : List[Any] ): return 32 @property def _A ( self : List[Any] ): return 32 @property def _A ( self : Any ): return self.time_input_dim @property def _A ( self : Union[str, Any] ): return self.time_input_dim * 4 @property def _A ( self : Tuple ): return 100 @property def _A ( self : Optional[int] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = { "in_channels": 4, # Out channels is double in channels because predicts mean and variance "out_channels": 8, "addition_embed_type": "image", "down_block_types": ("ResnetDownsampleBlock2D", "SimpleCrossAttnDownBlock2D"), "up_block_types": ("SimpleCrossAttnUpBlock2D", "ResnetUpsampleBlock2D"), "mid_block_type": "UNetMidBlock2DSimpleCrossAttn", "block_out_channels": (self.block_out_channels_a, self.block_out_channels_a * 2), "layers_per_block": 1, "encoder_hid_dim": self.text_embedder_hidden_size, "encoder_hid_dim_type": "image_proj", "cross_attention_dim": self.cross_attention_dim, "attention_head_dim": 4, "resnet_time_scale_shift": "scale_shift", "class_embed_type": None, } SCREAMING_SNAKE_CASE : str = UNetaDConditionModel(**UpperCAmelCase_ ) return model @property def _A ( self : int ): return { "block_out_channels": [32, 64], "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"], "in_channels": 3, "latent_channels": 4, "layers_per_block": 1, "norm_num_groups": 8, "norm_type": "spatial", "num_vq_embeddings": 12, "out_channels": 3, "up_block_types": [ "AttnUpDecoderBlock2D", "UpDecoderBlock2D", ], "vq_embed_dim": 4, } @property def _A ( self : Any ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : List[str] = VQModel(**self.dummy_movq_kwargs ) return model def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.dummy_unet SCREAMING_SNAKE_CASE : str = self.dummy_movq SCREAMING_SNAKE_CASE : Optional[int] = DDIMScheduler( num_train_timesteps=1000 , beta_schedule="linear" , beta_start=0.00_085 , beta_end=0.012 , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , steps_offset=1 , prediction_type="epsilon" , thresholding=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "scheduler": scheduler, "movq": movq, } return components def _A ( self : List[str] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any]=0 ): SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(seed + 1 ) ).to( UpperCAmelCase_ ) if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : List[Any] = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = { "image_embeds": image_embeds, "negative_image_embeds": negative_image_embeds, "generator": generator, "height": 64, "width": 64, "guidance_scale": 4.0, "num_inference_steps": 2, "output_type": "np", } return inputs def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "cpu" SCREAMING_SNAKE_CASE : Tuple = self.get_dummy_components() SCREAMING_SNAKE_CASE : List[str] = self.pipeline_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(**self.get_dummy_inputs(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : str = output.images SCREAMING_SNAKE_CASE : Tuple = pipe( **self.get_dummy_inputs(UpperCAmelCase_ ) , return_dict=UpperCAmelCase_ , )[0] SCREAMING_SNAKE_CASE : str = image[0, -3:, -3:, -1] SCREAMING_SNAKE_CASE : Any = image_from_tuple[0, -3:, -3:, -1] assert image.shape == (1, 64, 64, 3) SCREAMING_SNAKE_CASE : List[Any] = np.array( [0.6_237_976, 1.0, 0.36_441_332, 1.0, 0.70_639_634, 0.29_877_186, 0.85_652_125, 0.5_216_843, 0.54_454_046] ) assert ( np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_slice.flatten()}''' assert ( np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_from_tuple_slice.flatten()}''' @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main" "/kandinskyv22/kandinskyv22_text2img_cat_fp16.npy" ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPriorPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-prior" , torch_dtype=torch.floataa ) pipe_prior.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-decoder" , torch_dtype=torch.floataa ) SCREAMING_SNAKE_CASE : Tuple = pipeline.to(UpperCAmelCase_ ) pipeline.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = "red cat, 4k photo" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = pipe_prior( UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=5 , negative_prompt="" , ).to_tuple() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = pipeline( image_embeds=UpperCAmelCase_ , negative_image_embeds=UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=100 , output_type="np" , ) SCREAMING_SNAKE_CASE : List[Any] = output.images[0] assert image.shape == (512, 512, 3) assert_mean_pixel_difference(UpperCAmelCase_ , UpperCAmelCase_ )
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : int , UpperCAmelCase_ : float = 0 ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = row, column SCREAMING_SNAKE_CASE : Optional[Any] = [[default_value for c in range(UpperCAmelCase_ )] for r in range(UpperCAmelCase_ )] def __str__( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = f'''Matrix consist of {self.row} rows and {self.column} columns\n''' # Make string identifier SCREAMING_SNAKE_CASE : Dict = 0 for row_vector in self.array: for obj in row_vector: SCREAMING_SNAKE_CASE : Optional[Any] = max(UpperCAmelCase_ , len(str(UpperCAmelCase_ ) ) ) SCREAMING_SNAKE_CASE : Optional[int] = f'''%{max_element_length}s''' # Make string and return def single_line(UpperCAmelCase_ : list[float] ) -> str: nonlocal string_format_identifier SCREAMING_SNAKE_CASE : Optional[int] = "[" line += ", ".join(string_format_identifier % (obj,) for obj in row_vector ) line += "]" return line s += "\n".join(single_line(UpperCAmelCase_ ) for row_vector in self.array ) return s def __repr__( self : Dict ): return str(self ) def _A ( self : Optional[int] , UpperCAmelCase_ : tuple[int, int] ): if not (isinstance(UpperCAmelCase_ , (list, tuple) ) and len(UpperCAmelCase_ ) == 2): return False elif not (0 <= loc[0] < self.row and 0 <= loc[1] < self.column): return False else: return True def __getitem__( self : Any , UpperCAmelCase_ : tuple[int, int] ): assert self.validate_indicies(UpperCAmelCase_ ) return self.array[loc[0]][loc[1]] def __setitem__( self : int , UpperCAmelCase_ : tuple[int, int] , UpperCAmelCase_ : float ): assert self.validate_indicies(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = value def __add__( self : Optional[int] , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == another.row and self.column == another.column # Add SCREAMING_SNAKE_CASE : str = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : Any = self[r, c] + another[r, c] return result def __neg__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = -self[r, c] return result def __sub__( self : Optional[Any] , UpperCAmelCase_ : Matrix ): return self + (-another) def __mul__( self : Dict , UpperCAmelCase_ : int | float | Matrix ): if isinstance(UpperCAmelCase_ , (int, float) ): # Scalar multiplication SCREAMING_SNAKE_CASE : Any = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = self[r, c] * another return result elif isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): # Matrix multiplication assert self.column == another.row SCREAMING_SNAKE_CASE : Any = Matrix(self.row , another.column ) for r in range(self.row ): for c in range(another.column ): for i in range(self.column ): result[r, c] += self[r, i] * another[i, c] return result else: SCREAMING_SNAKE_CASE : List[str] = f'''Unsupported type given for another ({type(UpperCAmelCase_ )})''' raise TypeError(UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : List[str] = Matrix(self.column , self.row ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : List[str] = self[r, c] return result def _A ( self : Union[str, Any] , UpperCAmelCase_ : Matrix , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) and isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == self.column == u.row == v.row # u, v should be column vector assert u.column == v.column == 1 # u, v should be column vector # Calculate SCREAMING_SNAKE_CASE : Tuple = v.transpose() SCREAMING_SNAKE_CASE : int = (v_t * self * u)[0, 0] + 1 if numerator_factor == 0: return None # It's not invertable return self - ((self * u) * (v_t * self) * (1.0 / numerator_factor)) # Testing if __name__ == "__main__": def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 3 , 0 ) for i in range(3 ): SCREAMING_SNAKE_CASE : str = 1 print(F'''a^(-1) is {ainv}''' ) # u, v SCREAMING_SNAKE_CASE : Optional[int] = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = 1, 2, -3 SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = 4, -2, 5 print(F'''u is {u}''' ) print(F'''v is {v}''' ) print(F'''uv^T is {u * v.transpose()}''' ) # Sherman Morrison print(F'''(a + uv^T)^(-1) is {ainv.sherman_morrison(lowercase , lowercase )}''' ) def lowerCamelCase__ ( ): """simple docstring""" import doctest doctest.testmod() testa()
62
1
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tokenizers_available, is_torch_available snake_case = { """configuration_biogpt""": ["""BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP""", """BioGptConfig"""], """tokenization_biogpt""": ["""BioGptTokenizer"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST""", """BioGptForCausalLM""", """BioGptForTokenClassification""", """BioGptForSequenceClassification""", """BioGptModel""", """BioGptPreTrainedModel""", ] if TYPE_CHECKING: from .configuration_biogpt import BIOGPT_PRETRAINED_CONFIG_ARCHIVE_MAP, BioGptConfig from .tokenization_biogpt import BioGptTokenizer try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_biogpt import ( BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST, BioGptForCausalLM, BioGptForSequenceClassification, BioGptForTokenClassification, BioGptModel, BioGptPreTrainedModel, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
import json from typing import List, Optional, Tuple from tokenizers import normalizers from tokenizers.pre_tokenizers import BertPreTokenizer, PreTokenizer from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging from .tokenization_roformer import RoFormerTokenizer from .tokenization_utils import JiebaPreTokenizer snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """vocab.txt""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """junnyu/roformer_chinese_small""": """https://huggingface.co/junnyu/roformer_chinese_small/resolve/main/vocab.txt""", """junnyu/roformer_chinese_base""": """https://huggingface.co/junnyu/roformer_chinese_base/resolve/main/vocab.txt""", """junnyu/roformer_chinese_char_small""": ( """https://huggingface.co/junnyu/roformer_chinese_char_small/resolve/main/vocab.txt""" ), """junnyu/roformer_chinese_char_base""": ( """https://huggingface.co/junnyu/roformer_chinese_char_base/resolve/main/vocab.txt""" ), """junnyu/roformer_small_discriminator""": ( """https://huggingface.co/junnyu/roformer_small_discriminator/resolve/main/vocab.txt""" ), """junnyu/roformer_small_generator""": ( """https://huggingface.co/junnyu/roformer_small_generator/resolve/main/vocab.txt""" ), } } snake_case = { """junnyu/roformer_chinese_small""": 1_536, """junnyu/roformer_chinese_base""": 1_536, """junnyu/roformer_chinese_char_small""": 512, """junnyu/roformer_chinese_char_base""": 512, """junnyu/roformer_small_discriminator""": 128, """junnyu/roformer_small_generator""": 128, } snake_case = { """junnyu/roformer_chinese_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_base""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_base""": {"""do_lower_case""": True}, """junnyu/roformer_small_discriminator""": {"""do_lower_case""": True}, """junnyu/roformer_small_generator""": {"""do_lower_case""": True}, } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = VOCAB_FILES_NAMES UpperCamelCase_ : int = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Tuple = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : List[Any] = PRETRAINED_INIT_CONFIGURATION UpperCamelCase_ : Any = RoFormerTokenizer def __init__( self : Tuple , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : List[str]="[UNK]" , UpperCAmelCase_ : Any="[SEP]" , UpperCAmelCase_ : Any="[PAD]" , UpperCAmelCase_ : List[str]="[CLS]" , UpperCAmelCase_ : str="[MASK]" , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : Optional[Any]=None , **UpperCAmelCase_ : List[str] , ): super().__init__( UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , do_lower_case=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , tokenize_chinese_chars=UpperCAmelCase_ , strip_accents=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = json.loads(self.backend_tokenizer.normalizer.__getstate__() ) if ( pre_tok_state.get("lowercase" , UpperCAmelCase_ ) != do_lower_case or pre_tok_state.get("strip_accents" , UpperCAmelCase_ ) != strip_accents ): SCREAMING_SNAKE_CASE : Optional[Any] = getattr(UpperCAmelCase_ , pre_tok_state.pop("type" ) ) SCREAMING_SNAKE_CASE : Any = do_lower_case SCREAMING_SNAKE_CASE : List[str] = strip_accents SCREAMING_SNAKE_CASE : Tuple = pre_tok_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = do_lower_case def __getstate__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = self.__dict__.copy() SCREAMING_SNAKE_CASE : Optional[Any] = BertPreTokenizer() return state def __setstate__( self : Tuple , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = d SCREAMING_SNAKE_CASE : Dict = self.__dict__["_tokenizer"].get_vocab() SCREAMING_SNAKE_CASE : Any = PreTokenizer.custom(JiebaPreTokenizer(UpperCAmelCase_ ) ) def _A ( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Union[str, Any]=None ): SCREAMING_SNAKE_CASE : List[Any] = [self.cls_token_id] + token_ids_a + [self.sep_token_id] if token_ids_a: output += token_ids_a + [self.sep_token_id] return output def _A ( self : Tuple , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : List[Any] = [self.sep_token_id] SCREAMING_SNAKE_CASE : Optional[int] = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1] def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): SCREAMING_SNAKE_CASE : Optional[int] = self._tokenizer.model.save(UpperCAmelCase_ , name=UpperCAmelCase_ ) return tuple(UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Any=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : Union[str, Any] = BertPreTokenizer() return super().save_pretrained(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ )
62
1
snake_case = {"""a""": ["""c""", """b"""], """b""": ["""d""", """e"""], """c""": [], """d""": [], """e""": []} snake_case = ["""a""", """b""", """c""", """d""", """e"""] def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = start # add current to visited visited.append(lowercase ) SCREAMING_SNAKE_CASE : Any = edges[current] for neighbor in neighbors: # if neighbor not in visited, visit if neighbor not in visited: SCREAMING_SNAKE_CASE : Union[str, Any] = topological_sort(lowercase , lowercase , lowercase ) # if all neighbors visited add current to sort sort.append(lowercase ) # if all vertices haven't been visited select a new one to visit if len(lowercase ) != len(lowercase ): for vertice in vertices: if vertice not in visited: SCREAMING_SNAKE_CASE : List[Any] = topological_sort(lowercase , lowercase , lowercase ) # return sort return sort if __name__ == "__main__": snake_case = topological_sort("""a""", [], []) print(sort)
62
def lowerCamelCase__ ( lowercase ): """simple docstring""" if not isinstance(lowercase , lowercase ): raise TypeError("only integers accepted as input" ) else: SCREAMING_SNAKE_CASE : Optional[int] = str(abs(lowercase ) ) SCREAMING_SNAKE_CASE : str = [list(lowercase ) for char in range(len(lowercase ) )] for index in range(len(lowercase ) ): num_transpositions[index].pop(lowercase ) return max( int("".join(list(lowercase ) ) ) for transposition in num_transpositions ) if __name__ == "__main__": __import__("""doctest""").testmod()
62
1
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 SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @slow def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = AutoModelForSeqaSeqLM.from_pretrained("google/mt5-small" , return_dict=UpperCAmelCase_ ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = AutoTokenizer.from_pretrained("google/mt5-small" ) SCREAMING_SNAKE_CASE : str = tokenizer("Hello there" , return_tensors="pt" ).input_ids SCREAMING_SNAKE_CASE : Any = tokenizer("Hi I am" , return_tensors="pt" ).input_ids SCREAMING_SNAKE_CASE : str = model(input_ids.to(UpperCAmelCase_ ) , labels=labels.to(UpperCAmelCase_ ) ).loss SCREAMING_SNAKE_CASE : List[str] = -(labels.shape[-1] * loss.item()) SCREAMING_SNAKE_CASE : Any = -84.9_127 self.assertTrue(abs(mtf_score - EXPECTED_SCORE ) < 1E-4 )
62
# This model implementation is heavily inspired by https://github.com/haofanwang/ControlNet-for-Diffusers/ import gc import random import tempfile import unittest import numpy as np import torch from PIL import Image from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, ControlNetModel, DDIMScheduler, StableDiffusionControlNetImgaImgPipeline, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_controlnet import MultiControlNetModel from diffusers.utils import floats_tensor, load_image, load_numpy, randn_tensor, slow, torch_device from diffusers.utils.import_utils import is_xformers_available from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..pipeline_params import ( IMAGE_TO_IMAGE_IMAGE_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_PARAMS, ) from ..test_pipelines_common import ( PipelineKarrasSchedulerTesterMixin, PipelineLatentTesterMixin, PipelineTesterMixin, ) enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : List[Any] = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[Any] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Tuple = IMAGE_TO_IMAGE_IMAGE_PARAMS.union({'''control_image'''} ) UpperCamelCase_ : Any = IMAGE_TO_IMAGE_IMAGE_PARAMS def _A ( self : List[str] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : str = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : str , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Any = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Optional[Any] = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = 2 SCREAMING_SNAKE_CASE : Union[str, Any] = randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ) SCREAMING_SNAKE_CASE : Tuple = floats_tensor(control_image.shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : str = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : List[str] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : int ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : Union[str, Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : str = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[int] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Dict = frozenset([] ) # TO_DO: add image_params once refactored VaeImageProcessor.preprocess def _A ( self : Optional[Any] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) def init_weights(UpperCAmelCase_ : List[Any] ): if isinstance(UpperCAmelCase_ , torch.nn.Convad ): torch.nn.init.normal(m.weight ) m.bias.data.fill_(1.0 ) SCREAMING_SNAKE_CASE : List[str] = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Dict = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Any = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : Tuple = MultiControlNetModel([controlneta, controlneta] ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Dict = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = 2 SCREAMING_SNAKE_CASE : Tuple = [ randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), ] SCREAMING_SNAKE_CASE : Optional[int] = floats_tensor(control_image[0].shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : Optional[Any] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : Any = self.get_dummy_components() SCREAMING_SNAKE_CASE : str = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = 10.0 SCREAMING_SNAKE_CASE : Any = 4 SCREAMING_SNAKE_CASE : Optional[int] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : Optional[int] = pipe(**UpperCAmelCase_ )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : Any = scale SCREAMING_SNAKE_CASE : List[str] = pipe(**UpperCAmelCase_ , control_guidance_start=0.1 , control_guidance_end=0.2 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : List[Any] = pipe(**UpperCAmelCase_ , control_guidance_start=[0.1, 0.3] , control_guidance_end=[0.2, 0.7] )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = steps SCREAMING_SNAKE_CASE : Dict = scale SCREAMING_SNAKE_CASE : Dict = pipe(**UpperCAmelCase_ , control_guidance_start=0.4 , control_guidance_end=[0.5, 0.8] )[0] # make sure that all outputs are different assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 def _A ( self : Union[str, Any] ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : List[Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = self.get_dummy_components() SCREAMING_SNAKE_CASE : Dict = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) with tempfile.TemporaryDirectory() as tmpdir: try: # save_pretrained is not implemented for Multi-ControlNet pipe.save_pretrained(UpperCAmelCase_ ) except NotImplementedError: pass @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Optional[Any] ): super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny" ) SCREAMING_SNAKE_CASE : Union[str, Any] = StableDiffusionControlNetImgaImgPipeline.from_pretrained( "runwayml/stable-diffusion-v1-5" , safety_checker=UpperCAmelCase_ , controlnet=UpperCAmelCase_ ) pipe.enable_model_cpu_offload() pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.Generator(device="cpu" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : str = "evil space-punk bird" SCREAMING_SNAKE_CASE : Optional[Any] = load_image( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/bird_canny.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : Optional[int] = load_image( "https://huggingface.co/lllyasviel/sd-controlnet-canny/resolve/main/images/bird.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : str = pipe( UpperCAmelCase_ , UpperCAmelCase_ , control_image=UpperCAmelCase_ , generator=UpperCAmelCase_ , output_type="np" , num_inference_steps=50 , strength=0.6 , ) SCREAMING_SNAKE_CASE : int = output.images[0] assert image.shape == (512, 512, 3) SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/img2img.npy" ) assert np.abs(expected_image - image ).max() < 9E-2
62
1
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 SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Union[str, Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : List[str]=13 , UpperCAmelCase_ : List[str]=7 , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : List[Any]=True , UpperCAmelCase_ : Dict=99 , UpperCAmelCase_ : Optional[int]=16 , UpperCAmelCase_ : str=36 , UpperCAmelCase_ : Optional[Any]=6 , UpperCAmelCase_ : Union[str, Any]=6 , UpperCAmelCase_ : List[Any]=6 , UpperCAmelCase_ : Optional[int]=37 , UpperCAmelCase_ : Union[str, Any]="gelu" , UpperCAmelCase_ : Optional[Any]=0.1 , UpperCAmelCase_ : Optional[Any]=0.1 , UpperCAmelCase_ : Tuple=512 , UpperCAmelCase_ : Union[str, Any]=16 , UpperCAmelCase_ : List[str]=2 , UpperCAmelCase_ : Optional[int]=0.02 , UpperCAmelCase_ : List[str]=3 , UpperCAmelCase_ : Union[str, Any]=4 , UpperCAmelCase_ : Tuple=None , ): SCREAMING_SNAKE_CASE : str = parent SCREAMING_SNAKE_CASE : str = batch_size SCREAMING_SNAKE_CASE : int = seq_length SCREAMING_SNAKE_CASE : Union[str, Any] = is_training SCREAMING_SNAKE_CASE : Tuple = use_input_mask SCREAMING_SNAKE_CASE : Any = use_token_type_ids SCREAMING_SNAKE_CASE : str = use_labels SCREAMING_SNAKE_CASE : Union[str, Any] = vocab_size SCREAMING_SNAKE_CASE : List[str] = embedding_size SCREAMING_SNAKE_CASE : Union[str, Any] = hidden_size SCREAMING_SNAKE_CASE : Union[str, Any] = num_hidden_layers SCREAMING_SNAKE_CASE : Optional[Any] = num_hidden_groups SCREAMING_SNAKE_CASE : List[str] = num_attention_heads SCREAMING_SNAKE_CASE : Optional[Any] = intermediate_size SCREAMING_SNAKE_CASE : Tuple = hidden_act SCREAMING_SNAKE_CASE : Optional[Any] = hidden_dropout_prob SCREAMING_SNAKE_CASE : Union[str, Any] = attention_probs_dropout_prob SCREAMING_SNAKE_CASE : List[Any] = max_position_embeddings SCREAMING_SNAKE_CASE : int = type_vocab_size SCREAMING_SNAKE_CASE : Tuple = type_sequence_label_size SCREAMING_SNAKE_CASE : Dict = initializer_range SCREAMING_SNAKE_CASE : Union[str, Any] = num_labels SCREAMING_SNAKE_CASE : Dict = num_choices SCREAMING_SNAKE_CASE : List[str] = scope def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Any = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size ) SCREAMING_SNAKE_CASE : Union[str, Any] = None if self.use_input_mask: SCREAMING_SNAKE_CASE : Optional[int] = random_attention_mask([self.batch_size, self.seq_length] ) SCREAMING_SNAKE_CASE : int = None if self.use_token_type_ids: SCREAMING_SNAKE_CASE : Union[str, Any] = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size ) SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : Union[str, Any] = None if self.use_labels: SCREAMING_SNAKE_CASE : Dict = ids_tensor([self.batch_size] , self.type_sequence_label_size ) SCREAMING_SNAKE_CASE : Union[str, Any] = ids_tensor([self.batch_size, self.seq_length] , self.num_labels ) SCREAMING_SNAKE_CASE : int = ids_tensor([self.batch_size] , self.num_choices ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_config() return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels def _A ( self : Any ): 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 _A ( self : Union[str, Any] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : List[str] = AlbertModel(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Dict = model(UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = model(UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ ) self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) ) self.parent.assertEqual(result.pooler_output.shape , (self.batch_size, self.hidden_size) ) def _A ( self : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : str , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Union[str, Any] = AlbertForPreTraining(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[Any] = model( UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , labels=UpperCAmelCase_ , sentence_order_label=UpperCAmelCase_ , ) 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 _A ( self : Dict , UpperCAmelCase_ : str , UpperCAmelCase_ : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Any , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : List[Any] = AlbertForMaskedLM(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : str = model(UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) ) def _A ( self : Any , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : int , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : List[str] = AlbertForQuestionAnswering(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Union[str, Any] = model( UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , start_positions=UpperCAmelCase_ , end_positions=UpperCAmelCase_ , ) self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) ) self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Any = AlbertForSequenceClassification(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : str = model(UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Any , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Union[str, Any] = AlbertForTokenClassification(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[Any] = model(UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) ) def _A ( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = self.num_choices SCREAMING_SNAKE_CASE : Optional[Any] = AlbertForMultipleChoice(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : int = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() SCREAMING_SNAKE_CASE : List[Any] = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() SCREAMING_SNAKE_CASE : Dict = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous() SCREAMING_SNAKE_CASE : int = model( UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , labels=UpperCAmelCase_ , ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) ) def _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = self.prepare_config_and_inputs() ( ( SCREAMING_SNAKE_CASE ) , ( SCREAMING_SNAKE_CASE ) , ( SCREAMING_SNAKE_CASE ) , ( SCREAMING_SNAKE_CASE ) , ( SCREAMING_SNAKE_CASE ) , ( SCREAMING_SNAKE_CASE ) , ( SCREAMING_SNAKE_CASE ) , ) : Tuple = config_and_inputs SCREAMING_SNAKE_CASE : Dict = {"input_ids": input_ids, "token_type_ids": token_type_ids, "attention_mask": input_mask} return config, inputs_dict @require_torch class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : str = ( ( AlbertModel, AlbertForPreTraining, AlbertForMaskedLM, AlbertForMultipleChoice, AlbertForSequenceClassification, AlbertForTokenClassification, AlbertForQuestionAnswering, ) if is_torch_available() else () ) UpperCamelCase_ : List[Any] = ( { '''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 _A ( self : int , UpperCAmelCase_ : Dict , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[int]=False ): SCREAMING_SNAKE_CASE : Dict = super()._prepare_for_class(UpperCAmelCase_ , UpperCAmelCase_ , return_labels=UpperCAmelCase_ ) if return_labels: if model_class in get_values(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Any = torch.zeros( (self.model_tester.batch_size, self.model_tester.seq_length) , dtype=torch.long , device=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = torch.zeros( self.model_tester.batch_size , dtype=torch.long , device=UpperCAmelCase_ ) return inputs_dict def _A ( self : str ): SCREAMING_SNAKE_CASE : Optional[Any] = AlbertModelTester(self ) SCREAMING_SNAKE_CASE : str = ConfigTester(self , config_class=UpperCAmelCase_ , hidden_size=37 ) def _A ( self : Any ): self.config_tester.run_common_tests() def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : Dict = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*UpperCAmelCase_ ) def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : int = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_pretraining(*UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_masked_lm(*UpperCAmelCase_ ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Tuple = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_multiple_choice(*UpperCAmelCase_ ) def _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_question_answering(*UpperCAmelCase_ ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_sequence_classification(*UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : str = self.model_tester.prepare_config_and_inputs() for type in ["absolute", "relative_key", "relative_key_query"]: SCREAMING_SNAKE_CASE : int = type self.model_tester.create_and_check_model(*UpperCAmelCase_ ) @slow def _A ( self : Optional[int] ): for model_name in ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: SCREAMING_SNAKE_CASE : Dict = AlbertModel.from_pretrained(UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) @require_torch class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @slow def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = AlbertModel.from_pretrained("albert-base-v2" ) SCREAMING_SNAKE_CASE : Tuple = torch.tensor([[0, 345, 232, 328, 740, 140, 1695, 69, 6078, 1588, 2]] ) SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]] ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Tuple = model(UpperCAmelCase_ , attention_mask=UpperCAmelCase_ )[0] SCREAMING_SNAKE_CASE : str = torch.Size((1, 11, 768) ) self.assertEqual(output.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [[[-0.6_513, 1.5_035, -0.2_766], [-0.6_515, 1.5_046, -0.2_780], [-0.6_512, 1.5_049, -0.2_784]]] ) self.assertTrue(torch.allclose(output[:, 1:4, 1:4] , UpperCAmelCase_ , atol=1E-4 ) )
62
import argparse import json from pathlib import Path import requests import torch from huggingface_hub import hf_hub_download from PIL import Image from transformers import ( MobileViTConfig, MobileViTForImageClassification, MobileViTForSemanticSegmentation, MobileViTImageProcessor, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = MobileViTConfig() # size of the architecture if "mobilevit_s" in mobilevit_name: SCREAMING_SNAKE_CASE : List[Any] = [144, 192, 240] SCREAMING_SNAKE_CASE : Tuple = [16, 32, 64, 96, 128, 160, 640] elif "mobilevit_xs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [96, 120, 144] SCREAMING_SNAKE_CASE : Dict = [16, 32, 48, 64, 80, 96, 384] elif "mobilevit_xxs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [64, 80, 96] SCREAMING_SNAKE_CASE : List[str] = [16, 16, 24, 48, 64, 80, 320] SCREAMING_SNAKE_CASE : int = 0.05 SCREAMING_SNAKE_CASE : int = 2.0 if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : str = 512 SCREAMING_SNAKE_CASE : List[str] = 16 SCREAMING_SNAKE_CASE : Union[str, Any] = 21 SCREAMING_SNAKE_CASE : Dict = "pascal-voc-id2label.json" else: SCREAMING_SNAKE_CASE : Optional[Any] = 1000 SCREAMING_SNAKE_CASE : Optional[Any] = "imagenet-1k-id2label.json" SCREAMING_SNAKE_CASE : Any = "huggingface/label-files" SCREAMING_SNAKE_CASE : Tuple = json.load(open(hf_hub_download(lowercase , lowercase , repo_type="dataset" ) , "r" ) ) SCREAMING_SNAKE_CASE : List[str] = {int(lowercase ): v for k, v in idalabel.items()} SCREAMING_SNAKE_CASE : Optional[Any] = idalabel SCREAMING_SNAKE_CASE : str = {v: k for k, v in idalabel.items()} return config def lowerCamelCase__ ( lowercase , lowercase=False ): """simple docstring""" for i in range(1 , 6 ): if F'''layer_{i}.''' in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(F'''layer_{i}.''' , F'''encoder.layer.{i - 1}.''' ) if "conv_1." in name: SCREAMING_SNAKE_CASE : Dict = name.replace("conv_1." , "conv_stem." ) if ".block." in name: SCREAMING_SNAKE_CASE : List[str] = name.replace(".block." , "." ) if "exp_1x1" in name: SCREAMING_SNAKE_CASE : str = name.replace("exp_1x1" , "expand_1x1" ) if "red_1x1" in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("red_1x1" , "reduce_1x1" ) if ".local_rep.conv_3x3." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".local_rep.conv_3x3." , ".conv_kxk." ) if ".local_rep.conv_1x1." in name: SCREAMING_SNAKE_CASE : int = name.replace(".local_rep.conv_1x1." , ".conv_1x1." ) if ".norm." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".norm." , ".normalization." ) if ".conv." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".conv." , ".convolution." ) if ".conv_proj." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".conv_proj." , ".conv_projection." ) for i in range(0 , 2 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.layer.{j}.''' ) for i in range(2 , 6 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.''' ) if "expand_1x1" in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace("expand_1x1" , "downsampling_layer.expand_1x1" ) if "conv_3x3" in name: SCREAMING_SNAKE_CASE : str = name.replace("conv_3x3" , "downsampling_layer.conv_3x3" ) if "reduce_1x1" in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("reduce_1x1" , "downsampling_layer.reduce_1x1" ) for i in range(2 , 5 ): if F'''.global_rep.{i}.weight''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.global_rep.{i}.weight''' , ".layernorm.weight" ) if F'''.global_rep.{i}.bias''' in name: SCREAMING_SNAKE_CASE : str = name.replace(F'''.global_rep.{i}.bias''' , ".layernorm.bias" ) if ".global_rep." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".global_rep." , ".transformer." ) if ".pre_norm_mha.0." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_mha.0." , ".layernorm_before." ) if ".pre_norm_mha.1.out_proj." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_mha.1.out_proj." , ".attention.output.dense." ) if ".pre_norm_ffn.0." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(".pre_norm_ffn.0." , ".layernorm_after." ) if ".pre_norm_ffn.1." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_ffn.1." , ".intermediate.dense." ) if ".pre_norm_ffn.4." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_ffn.4." , ".output.dense." ) if ".transformer." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".transformer." , ".transformer.layer." ) if ".aspp_layer." in name: SCREAMING_SNAKE_CASE : int = name.replace(".aspp_layer." , "." ) if ".aspp_pool." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".aspp_pool." , "." ) if "seg_head." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("seg_head." , "segmentation_head." ) if "segmentation_head.classifier.classifier." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace("segmentation_head.classifier.classifier." , "segmentation_head.classifier." ) if "classifier.fc." in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("classifier.fc." , "classifier." ) elif (not base_model) and ("segmentation_head." not in name): SCREAMING_SNAKE_CASE : List[Any] = "mobilevit." + name return name def lowerCamelCase__ ( lowercase , lowercase , lowercase=False ): """simple docstring""" if base_model: SCREAMING_SNAKE_CASE : Optional[int] = "" else: SCREAMING_SNAKE_CASE : Any = "mobilevit." for key in orig_state_dict.copy().keys(): SCREAMING_SNAKE_CASE : Union[str, Any] = orig_state_dict.pop(lowercase ) if key[:8] == "encoder.": SCREAMING_SNAKE_CASE : int = key[8:] if "qkv" in key: SCREAMING_SNAKE_CASE : Optional[int] = key.split("." ) SCREAMING_SNAKE_CASE : Any = int(key_split[0][6:] ) - 1 SCREAMING_SNAKE_CASE : List[Any] = int(key_split[3] ) SCREAMING_SNAKE_CASE : List[Any] = model.get_submodule(F'''{model_prefix}encoder.layer.{layer_num}''' ) SCREAMING_SNAKE_CASE : int = layer.transformer.layer[transformer_num].attention.attention.all_head_size SCREAMING_SNAKE_CASE : Union[str, Any] = ( F'''{model_prefix}encoder.layer.{layer_num}.transformer.layer.{transformer_num}.attention.attention.''' ) if "weight" in key: SCREAMING_SNAKE_CASE : Optional[int] = val[:dim, :] SCREAMING_SNAKE_CASE : Tuple = val[dim : dim * 2, :] SCREAMING_SNAKE_CASE : Dict = val[-dim:, :] else: SCREAMING_SNAKE_CASE : str = val[:dim] SCREAMING_SNAKE_CASE : Union[str, Any] = val[dim : dim * 2] SCREAMING_SNAKE_CASE : Union[str, Any] = val[-dim:] else: SCREAMING_SNAKE_CASE : List[Any] = val return orig_state_dict def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = "http://images.cocodataset.org/val2017/000000039769.jpg" SCREAMING_SNAKE_CASE : Union[str, Any] = Image.open(requests.get(lowercase , stream=lowercase ).raw ) return im @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase=False ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = get_mobilevit_config(lowercase ) # load original state_dict SCREAMING_SNAKE_CASE : Union[str, Any] = torch.load(lowercase , map_location="cpu" ) # load 🤗 model if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : List[str] = MobileViTForSemanticSegmentation(lowercase ).eval() else: SCREAMING_SNAKE_CASE : str = MobileViTForImageClassification(lowercase ).eval() SCREAMING_SNAKE_CASE : Any = convert_state_dict(lowercase , lowercase ) model.load_state_dict(lowercase ) # Check outputs on an image, prepared by MobileViTImageProcessor SCREAMING_SNAKE_CASE : List[Any] = MobileViTImageProcessor(crop_size=config.image_size , size=config.image_size + 32 ) SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=prepare_img() , return_tensors="pt" ) SCREAMING_SNAKE_CASE : List[Any] = model(**lowercase ) SCREAMING_SNAKE_CASE : str = outputs.logits if mobilevit_name.startswith("deeplabv3_" ): assert logits.shape == (1, 21, 32, 32) if mobilevit_name == "deeplabv3_mobilevit_s": SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor( [ [[6.2065, 6.1292, 6.2070], [6.1079, 6.1254, 6.1747], [6.0042, 6.1071, 6.1034]], [[-6.9253, -6.8653, -7.0398], [-7.3218, -7.3983, -7.3670], [-7.1961, -7.2482, -7.1569]], [[-4.4723, -4.4348, -4.3769], [-5.3629, -5.4632, -5.4598], [-5.1587, -5.3402, -5.5059]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor( [ [[5.4449, 5.5733, 5.6314], [5.1815, 5.3930, 5.5963], [5.1656, 5.4333, 5.4853]], [[-9.4423, -9.7766, -9.6714], [-9.1581, -9.5720, -9.5519], [-9.1006, -9.6458, -9.5703]], [[-7.7721, -7.3716, -7.1583], [-8.4599, -8.0624, -7.7944], [-8.4172, -7.8366, -7.5025]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xxs": SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [ [[6.9811, 6.9743, 7.3123], [7.1777, 7.1931, 7.3938], [7.5633, 7.8050, 7.8901]], [[-10.5536, -10.2332, -10.2924], [-10.2336, -9.8624, -9.5964], [-10.8840, -10.8158, -10.6659]], [[-3.4938, -3.0631, -2.8620], [-3.4205, -2.8135, -2.6875], [-3.4179, -2.7945, -2.8750]], ] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3, :3, :3] , lowercase , atol=1E-4 ) else: assert logits.shape == (1, 1000) if mobilevit_name == "mobilevit_s": SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([-0.9866, 0.2392, -1.1241] ) elif mobilevit_name == "mobilevit_xs": SCREAMING_SNAKE_CASE : Dict = torch.tensor([-2.4761, -0.9399, -1.9587] ) elif mobilevit_name == "mobilevit_xxs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor([-1.9364, -1.2327, -0.4653] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3] , lowercase , atol=1E-4 ) Path(lowercase ).mkdir(exist_ok=lowercase ) print(F'''Saving model {mobilevit_name} to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowercase ) print(F'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowercase ) if push_to_hub: SCREAMING_SNAKE_CASE : List[str] = { "mobilevit_s": "mobilevit-small", "mobilevit_xs": "mobilevit-x-small", "mobilevit_xxs": "mobilevit-xx-small", "deeplabv3_mobilevit_s": "deeplabv3-mobilevit-small", "deeplabv3_mobilevit_xs": "deeplabv3-mobilevit-x-small", "deeplabv3_mobilevit_xxs": "deeplabv3-mobilevit-xx-small", } print("Pushing to the hub..." ) SCREAMING_SNAKE_CASE : int = model_mapping[mobilevit_name] image_processor.push_to_hub(lowercase , organization="apple" ) model.push_to_hub(lowercase , organization="apple" ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument( """--mobilevit_name""", default="""mobilevit_s""", type=str, help=( """Name of the MobileViT model you'd like to convert. Should be one of 'mobilevit_s', 'mobilevit_xs',""" """ 'mobilevit_xxs', 'deeplabv3_mobilevit_s', 'deeplabv3_mobilevit_xs', 'deeplabv3_mobilevit_xxs'.""" ), ) parser.add_argument( """--checkpoint_path""", required=True, type=str, help="""Path to the original state dict (.pt file).""" ) parser.add_argument( """--pytorch_dump_folder_path""", required=True, type=str, help="""Path to the output PyTorch model directory.""" ) parser.add_argument( """--push_to_hub""", action="""store_true""", help="""Whether or not to push the converted model to the 🤗 hub.""" ) snake_case = parser.parse_args() convert_movilevit_checkpoint( args.mobilevit_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub )
62
1
import logging import os import sys import warnings from dataclasses import dataclass, field from random import randint from typing import Optional import datasets import evaluate import numpy as np from datasets import DatasetDict, load_dataset import transformers from transformers import ( AutoConfig, AutoFeatureExtractor, AutoModelForAudioClassification, HfArgumentParser, Trainer, TrainingArguments, set_seed, ) from transformers.trainer_utils import get_last_checkpoint from transformers.utils import check_min_version, send_example_telemetry from transformers.utils.versions import require_version snake_case = logging.getLogger(__name__) # Will error if the minimal version of Transformers is not installed. Remove at your own risks. check_min_version("""4.31.0""") require_version("""datasets>=1.14.0""", """To fix: pip install -r examples/pytorch/audio-classification/requirements.txt""") def lowerCamelCase__ ( lowercase , lowercase , lowercase = 16000 ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = int(round(sample_rate * max_length ) ) if len(lowercase ) <= sample_length: return wav SCREAMING_SNAKE_CASE : Any = randint(0 , len(lowercase ) - sample_length - 1 ) return wav[random_offset : random_offset + sample_length] @dataclass class SCREAMING_SNAKE_CASE : '''simple docstring''' UpperCamelCase_ : Optional[str] = field(default=lowerCAmelCase , metadata={'''help''': '''Name of a dataset from the datasets package'''} ) UpperCamelCase_ : Optional[str] = field( default=lowerCAmelCase , metadata={'''help''': '''The configuration name of the dataset to use (via the datasets library).'''} ) UpperCamelCase_ : Optional[str] = field( default=lowerCAmelCase , metadata={'''help''': '''A file containing the training audio paths and labels.'''} ) UpperCamelCase_ : Optional[str] = field( default=lowerCAmelCase , metadata={'''help''': '''A file containing the validation audio paths and labels.'''} ) UpperCamelCase_ : str = field( default='''train''' , metadata={ '''help''': '''The name of the training data set split to use (via the datasets library). Defaults to \'train\'''' } , ) UpperCamelCase_ : str = field( default='''validation''' , metadata={ '''help''': ( '''The name of the training data set split to use (via the datasets library). Defaults to \'validation\'''' ) } , ) UpperCamelCase_ : str = field( default='''audio''' , metadata={'''help''': '''The name of the dataset column containing the audio data. Defaults to \'audio\''''} , ) UpperCamelCase_ : str = field( default='''label''' , metadata={'''help''': '''The name of the dataset column containing the labels. Defaults to \'label\''''} ) UpperCamelCase_ : Optional[int] = field( default=lowerCAmelCase , metadata={ '''help''': ( '''For debugging purposes or quicker training, truncate the number of training examples to this ''' '''value if set.''' ) } , ) UpperCamelCase_ : Optional[int] = field( default=lowerCAmelCase , metadata={ '''help''': ( '''For debugging purposes or quicker training, truncate the number of evaluation examples to this ''' '''value if set.''' ) } , ) UpperCamelCase_ : float = field( default=2_0 , metadata={'''help''': '''Audio clips will be randomly cut to this length during training if the value is set.'''} , ) @dataclass class SCREAMING_SNAKE_CASE : '''simple docstring''' UpperCamelCase_ : str = field( default='''facebook/wav2vec2-base''' , metadata={'''help''': '''Path to pretrained model or model identifier from huggingface.co/models'''} , ) UpperCamelCase_ : Optional[str] = field( default=lowerCAmelCase , metadata={'''help''': '''Pretrained config name or path if not the same as model_name'''} ) UpperCamelCase_ : Optional[str] = field( default=lowerCAmelCase , metadata={'''help''': '''Where do you want to store the pretrained models downloaded from the Hub'''} ) UpperCamelCase_ : str = field( default='''main''' , metadata={'''help''': '''The specific model version to use (can be a branch name, tag name or commit id).'''} , ) UpperCamelCase_ : Optional[str] = field( default=lowerCAmelCase , metadata={'''help''': '''Name or path of preprocessor config.'''} ) UpperCamelCase_ : bool = field( default=lowerCAmelCase , metadata={'''help''': '''Whether to freeze the feature encoder layers of the model.'''} ) UpperCamelCase_ : bool = field( default=lowerCAmelCase , metadata={'''help''': '''Whether to generate an attention mask in the feature extractor.'''} ) UpperCamelCase_ : bool = field( default=lowerCAmelCase , metadata={ '''help''': ( '''Will use the token generated when running `huggingface-cli login` (necessary to use this script ''' '''with private models).''' ) } , ) UpperCamelCase_ : Optional[bool] = field( default=lowerCAmelCase , metadata={'''help''': '''Whether to freeze the feature extractor layers of the model.'''} ) UpperCamelCase_ : bool = field( default=lowerCAmelCase , metadata={'''help''': '''Will enable to load a pretrained model whose head dimensions are different.'''} , ) def _A ( self : int ): if not self.freeze_feature_extractor and self.freeze_feature_encoder: warnings.warn( "The argument `--freeze_feature_extractor` is deprecated and " "will be removed in a future version. Use `--freeze_feature_encoder`" "instead. Setting `freeze_feature_encoder==True`." , UpperCAmelCase_ , ) if self.freeze_feature_extractor and not self.freeze_feature_encoder: raise ValueError( "The argument `--freeze_feature_extractor` is deprecated and " "should not be used in combination with `--freeze_feature_encoder`." "Only make use of `--freeze_feature_encoder`." ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : str = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments) ) if len(sys.argv ) == 2 and sys.argv[1].endswith(".json" ): # If we pass only one argument to the script and it's the path to a json file, # let's parse it to get our arguments. SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) ) else: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = parser.parse_args_into_dataclasses() # Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The # information sent is the one passed as arguments along with your Python/PyTorch versions. send_example_telemetry("run_audio_classification" , lowercase , lowercase ) # Setup logging logging.basicConfig( format="%(asctime)s - %(levelname)s - %(name)s - %(message)s" , datefmt="%m/%d/%Y %H:%M:%S" , handlers=[logging.StreamHandler(sys.stdout )] , ) if training_args.should_log: # The default of training_args.log_level is passive, so we set log level at info here to have that default. transformers.utils.logging.set_verbosity_info() SCREAMING_SNAKE_CASE : str = training_args.get_process_log_level() logger.setLevel(lowercase ) transformers.utils.logging.set_verbosity(lowercase ) transformers.utils.logging.enable_default_handler() transformers.utils.logging.enable_explicit_format() # Log on each process the small summary: logger.warning( F'''Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu} ''' + F'''distributed training: {bool(training_args.local_rank != -1 )}, 16-bits training: {training_args.fpaa}''' ) logger.info(F'''Training/evaluation parameters {training_args}''' ) # Set seed before initializing model. set_seed(training_args.seed ) # Detecting last checkpoint. SCREAMING_SNAKE_CASE : List[Any] = None if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir: SCREAMING_SNAKE_CASE : str = get_last_checkpoint(training_args.output_dir ) if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0: raise ValueError( F'''Output directory ({training_args.output_dir}) already exists and is not empty. ''' "Use --overwrite_output_dir to train from scratch." ) elif last_checkpoint is not None and training_args.resume_from_checkpoint is None: logger.info( F'''Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change ''' "the `--output_dir` or add `--overwrite_output_dir` to train from scratch." ) # Initialize our dataset and prepare it for the audio classification task. SCREAMING_SNAKE_CASE : str = DatasetDict() SCREAMING_SNAKE_CASE : str = load_dataset( data_args.dataset_name , data_args.dataset_config_name , split=data_args.train_split_name , use_auth_token=True if model_args.use_auth_token else None , ) SCREAMING_SNAKE_CASE : List[str] = load_dataset( data_args.dataset_name , data_args.dataset_config_name , split=data_args.eval_split_name , use_auth_token=True if model_args.use_auth_token else None , ) if data_args.audio_column_name not in raw_datasets["train"].column_names: raise ValueError( F'''--audio_column_name {data_args.audio_column_name} not found in dataset \'{data_args.dataset_name}\'. ''' "Make sure to set `--audio_column_name` to the correct audio column - one of " F'''{', '.join(raw_datasets['train'].column_names )}.''' ) if data_args.label_column_name not in raw_datasets["train"].column_names: raise ValueError( F'''--label_column_name {data_args.label_column_name} not found in dataset \'{data_args.dataset_name}\'. ''' "Make sure to set `--label_column_name` to the correct text column - one of " F'''{', '.join(raw_datasets['train'].column_names )}.''' ) # Setting `return_attention_mask=True` is the way to get a correctly masked mean-pooling over # transformer outputs in the classifier, but it doesn't always lead to better accuracy SCREAMING_SNAKE_CASE : Any = AutoFeatureExtractor.from_pretrained( model_args.feature_extractor_name or model_args.model_name_or_path , return_attention_mask=model_args.attention_mask , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , ) # `datasets` takes care of automatically loading and resampling the audio, # so we just need to set the correct target sampling rate. SCREAMING_SNAKE_CASE : List[str] = raw_datasets.cast_column( data_args.audio_column_name , datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate ) ) SCREAMING_SNAKE_CASE : int = feature_extractor.model_input_names[0] def train_transforms(lowercase ): SCREAMING_SNAKE_CASE : Optional[int] = [] for audio in batch[data_args.audio_column_name]: SCREAMING_SNAKE_CASE : Optional[Any] = random_subsample( audio["array"] , max_length=data_args.max_length_seconds , sample_rate=feature_extractor.sampling_rate ) subsampled_wavs.append(lowercase ) SCREAMING_SNAKE_CASE : Any = feature_extractor(lowercase , sampling_rate=feature_extractor.sampling_rate ) SCREAMING_SNAKE_CASE : Tuple = {model_input_name: inputs.get(lowercase )} SCREAMING_SNAKE_CASE : Dict = list(batch[data_args.label_column_name] ) return output_batch def val_transforms(lowercase ): SCREAMING_SNAKE_CASE : Optional[Any] = [audio["array"] for audio in batch[data_args.audio_column_name]] SCREAMING_SNAKE_CASE : Any = feature_extractor(lowercase , sampling_rate=feature_extractor.sampling_rate ) SCREAMING_SNAKE_CASE : str = {model_input_name: inputs.get(lowercase )} SCREAMING_SNAKE_CASE : List[str] = list(batch[data_args.label_column_name] ) return output_batch # Prepare label mappings. # We'll include these in the model's config to get human readable labels in the Inference API. SCREAMING_SNAKE_CASE : Optional[Any] = raw_datasets["train"].features[data_args.label_column_name].names SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = {}, {} for i, label in enumerate(lowercase ): SCREAMING_SNAKE_CASE : int = str(lowercase ) SCREAMING_SNAKE_CASE : List[str] = label # Load the accuracy metric from the datasets package SCREAMING_SNAKE_CASE : str = evaluate.load("accuracy" ) # Define our compute_metrics function. It takes an `EvalPrediction` object (a namedtuple with # `predictions` and `label_ids` fields) and has to return a dictionary string to float. def compute_metrics(lowercase ): SCREAMING_SNAKE_CASE : Any = np.argmax(eval_pred.predictions , axis=1 ) return metric.compute(predictions=lowercase , references=eval_pred.label_ids ) SCREAMING_SNAKE_CASE : List[str] = AutoConfig.from_pretrained( model_args.config_name or model_args.model_name_or_path , num_labels=len(lowercase ) , labelaid=lowercase , idalabel=lowercase , finetuning_task="audio-classification" , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , ) SCREAMING_SNAKE_CASE : Union[str, Any] = AutoModelForAudioClassification.from_pretrained( model_args.model_name_or_path , from_tf=bool(".ckpt" in model_args.model_name_or_path ) , config=lowercase , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , ignore_mismatched_sizes=model_args.ignore_mismatched_sizes , ) # freeze the convolutional waveform encoder if model_args.freeze_feature_encoder: model.freeze_feature_encoder() if training_args.do_train: if data_args.max_train_samples is not None: SCREAMING_SNAKE_CASE : Tuple = ( raw_datasets["train"].shuffle(seed=training_args.seed ).select(range(data_args.max_train_samples ) ) ) # Set the training transforms raw_datasets["train"].set_transform(lowercase , output_all_columns=lowercase ) if training_args.do_eval: if data_args.max_eval_samples is not None: SCREAMING_SNAKE_CASE : Tuple = ( raw_datasets["eval"].shuffle(seed=training_args.seed ).select(range(data_args.max_eval_samples ) ) ) # Set the validation transforms raw_datasets["eval"].set_transform(lowercase , output_all_columns=lowercase ) # Initialize our trainer SCREAMING_SNAKE_CASE : Union[str, Any] = Trainer( model=lowercase , args=lowercase , train_dataset=raw_datasets["train"] if training_args.do_train else None , eval_dataset=raw_datasets["eval"] if training_args.do_eval else None , compute_metrics=lowercase , tokenizer=lowercase , ) # Training if training_args.do_train: SCREAMING_SNAKE_CASE : List[Any] = None if training_args.resume_from_checkpoint is not None: SCREAMING_SNAKE_CASE : Union[str, Any] = training_args.resume_from_checkpoint elif last_checkpoint is not None: SCREAMING_SNAKE_CASE : Optional[Any] = last_checkpoint SCREAMING_SNAKE_CASE : Optional[int] = trainer.train(resume_from_checkpoint=lowercase ) trainer.save_model() trainer.log_metrics("train" , train_result.metrics ) trainer.save_metrics("train" , train_result.metrics ) trainer.save_state() # Evaluation if training_args.do_eval: SCREAMING_SNAKE_CASE : int = trainer.evaluate() trainer.log_metrics("eval" , lowercase ) trainer.save_metrics("eval" , lowercase ) # Write model card and (optionally) push to hub SCREAMING_SNAKE_CASE : List[Any] = { "finetuned_from": model_args.model_name_or_path, "tasks": "audio-classification", "dataset": data_args.dataset_name, "tags": ["audio-classification"], } if training_args.push_to_hub: trainer.push_to_hub(**lowercase ) else: trainer.create_model_card(**lowercase ) if __name__ == "__main__": main()
62
from collections import Counter import numpy as np from sklearn import datasets from sklearn.model_selection import train_test_split snake_case = datasets.load_iris() snake_case = np.array(data["""data"""]) snake_case = np.array(data["""target"""]) snake_case = data["""target_names"""] snake_case , snake_case , snake_case , snake_case = train_test_split(X, y) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return np.linalg.norm(np.array(lowercase ) - np.array(lowercase ) ) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase=5 ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = zip(lowercase , lowercase ) # List of distances of all points from the point to be classified SCREAMING_SNAKE_CASE : Optional[int] = [] for data_point in data: SCREAMING_SNAKE_CASE : Tuple = euclidean_distance(data_point[0] , lowercase ) distances.append((distance, data_point[1]) ) # Choosing 'k' points with the least distances. SCREAMING_SNAKE_CASE : List[Any] = [i[1] for i in sorted(lowercase )[:k]] # Most commonly occurring class among them # is the class into which the point is classified SCREAMING_SNAKE_CASE : List[Any] = Counter(lowercase ).most_common(1 )[0][0] return classes[result] if __name__ == "__main__": print(classifier(X_train, y_train, classes, [4.4, 3.1, 1.3, 1.4]))
62
1
import copy from typing import Dict, Optional from ...configuration_utils import PretrainedConfig from ...utils import logging from ..auto import CONFIG_MAPPING from ..detr import DetrConfig from ..swin import SwinConfig snake_case = { """facebook/maskformer-swin-base-ade""": ( """https://huggingface.co/facebook/maskformer-swin-base-ade/blob/main/config.json""" ) # See all MaskFormer models at https://huggingface.co/models?filter=maskformer } snake_case = logging.get_logger(__name__) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : str = '''maskformer''' UpperCamelCase_ : List[str] = {'''hidden_size''': '''mask_feature_size'''} UpperCamelCase_ : List[Any] = ['''resnet''', '''swin'''] UpperCamelCase_ : Tuple = ['''detr'''] def __init__( self : Any , UpperCAmelCase_ : int = 256 , UpperCAmelCase_ : int = 256 , UpperCAmelCase_ : float = 0.1 , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : Optional[Dict] = None , UpperCAmelCase_ : Optional[Dict] = None , UpperCAmelCase_ : float = 0.02 , UpperCAmelCase_ : float = 1.0 , UpperCAmelCase_ : float = 1.0 , UpperCAmelCase_ : float = 1.0 , UpperCAmelCase_ : float = 20.0 , UpperCAmelCase_ : Optional[bool] = None , **UpperCAmelCase_ : List[str] , ): if backbone_config is None: # fall back to https://huggingface.co/microsoft/swin-base-patch4-window12-384-in22k SCREAMING_SNAKE_CASE : List[str] = SwinConfig( image_size=384 , in_channels=3 , patch_size=4 , embed_dim=128 , depths=[2, 2, 18, 2] , num_heads=[4, 8, 16, 32] , window_size=12 , drop_path_rate=0.3 , out_features=["stage1", "stage2", "stage3", "stage4"] , ) if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[str] = backbone_config.pop("model_type" ) SCREAMING_SNAKE_CASE : int = CONFIG_MAPPING[backbone_model_type] SCREAMING_SNAKE_CASE : List[str] = config_class.from_dict(UpperCAmelCase_ ) # verify that the backbone is supported if backbone_config.model_type not in self.backbones_supported: logger.warning_once( f'''Backbone {backbone_config.model_type} is not a supported model and may not be compatible with MaskFormer. ''' f'''Supported model types: {','.join(self.backbones_supported )}''' ) if decoder_config is None: # fall back to https://huggingface.co/facebook/detr-resnet-50 SCREAMING_SNAKE_CASE : Union[str, Any] = DetrConfig() else: # verify that the decoder is supported SCREAMING_SNAKE_CASE : int = ( decoder_config.pop("model_type" ) if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) else decoder_config.model_type ) if decoder_type not in self.decoders_supported: raise ValueError( f'''Transformer Decoder {decoder_type} not supported, please use one of''' f''' {','.join(self.decoders_supported )}''' ) if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Optional[Any] = CONFIG_MAPPING[decoder_type] SCREAMING_SNAKE_CASE : Dict = config_class.from_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = backbone_config SCREAMING_SNAKE_CASE : List[Any] = decoder_config # main feature dimension for the model SCREAMING_SNAKE_CASE : str = fpn_feature_size SCREAMING_SNAKE_CASE : Optional[int] = mask_feature_size # initializer SCREAMING_SNAKE_CASE : str = init_std SCREAMING_SNAKE_CASE : str = init_xavier_std # Hungarian matcher && loss SCREAMING_SNAKE_CASE : Union[str, Any] = cross_entropy_weight SCREAMING_SNAKE_CASE : List[str] = dice_weight SCREAMING_SNAKE_CASE : Dict = mask_weight SCREAMING_SNAKE_CASE : Tuple = use_auxiliary_loss SCREAMING_SNAKE_CASE : Optional[Any] = no_object_weight SCREAMING_SNAKE_CASE : Tuple = output_auxiliary_logits SCREAMING_SNAKE_CASE : int = self.decoder_config.encoder_attention_heads SCREAMING_SNAKE_CASE : int = self.decoder_config.num_hidden_layers super().__init__(**UpperCAmelCase_ ) @classmethod def _A ( cls : str , UpperCAmelCase_ : PretrainedConfig , UpperCAmelCase_ : PretrainedConfig , **UpperCAmelCase_ : int ): return cls( backbone_config=UpperCAmelCase_ , decoder_config=UpperCAmelCase_ , **UpperCAmelCase_ , ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = copy.deepcopy(self.__dict__ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.backbone_config.to_dict() SCREAMING_SNAKE_CASE : Tuple = self.decoder_config.to_dict() SCREAMING_SNAKE_CASE : Dict = self.__class__.model_type return output
62
from typing import Callable, Optional, Union from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """microsoft/xprophetnet-large-wiki100-cased""": ( """https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Tuple = '''xlm-prophetnet''' UpperCamelCase_ : Tuple = ['''past_key_values'''] UpperCamelCase_ : int = { '''num_attention_heads''': '''num_encoder_attention_heads''', } def __init__( self : Dict , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[Union[str, Callable]] = "gelu" , UpperCAmelCase_ : Optional[int] = 3_0522 , UpperCAmelCase_ : Optional[int] = 1024 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[int] = 512 , UpperCAmelCase_ : Optional[float] = 0.02 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 2 , UpperCAmelCase_ : Optional[int] = 32 , UpperCAmelCase_ : Optional[int] = 128 , UpperCAmelCase_ : Optional[bool] = False , UpperCAmelCase_ : Optional[float] = 0.0 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 1 , UpperCAmelCase_ : Optional[int] = 2 , **UpperCAmelCase_ : Dict , ): SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : str = hidden_size SCREAMING_SNAKE_CASE : Tuple = encoder_ffn_dim SCREAMING_SNAKE_CASE : List[Any] = num_encoder_layers SCREAMING_SNAKE_CASE : Any = num_encoder_attention_heads SCREAMING_SNAKE_CASE : Dict = decoder_ffn_dim SCREAMING_SNAKE_CASE : List[str] = num_decoder_layers SCREAMING_SNAKE_CASE : List[Any] = num_decoder_attention_heads SCREAMING_SNAKE_CASE : List[Any] = max_position_embeddings SCREAMING_SNAKE_CASE : Dict = init_std # Normal(0, this parameter) SCREAMING_SNAKE_CASE : Union[str, Any] = activation_function # parameters for xlmprophetnet SCREAMING_SNAKE_CASE : Dict = ngram SCREAMING_SNAKE_CASE : Any = num_buckets SCREAMING_SNAKE_CASE : str = relative_max_distance SCREAMING_SNAKE_CASE : str = disable_ngram_loss SCREAMING_SNAKE_CASE : Dict = eps # 3 Types of Dropout SCREAMING_SNAKE_CASE : Union[str, Any] = attention_dropout SCREAMING_SNAKE_CASE : Optional[int] = activation_dropout SCREAMING_SNAKE_CASE : List[Any] = dropout SCREAMING_SNAKE_CASE : int = use_cache super().__init__( pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , is_encoder_decoder=UpperCAmelCase_ , add_cross_attention=UpperCAmelCase_ , decoder_start_token_id=UpperCAmelCase_ , **UpperCAmelCase_ , ) @property def _A ( self : int ): return self.num_encoder_layers + self.num_decoder_layers @num_hidden_layers.setter def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): raise NotImplementedError( "This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and" " `num_decoder_layers`." )
62
1
import os import warnings from typing import List, Optional from ...tokenization_utils_base import BatchEncoding from ...utils import logging from .configuration_rag import RagConfig snake_case = logging.get_logger(__name__) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = question_encoder SCREAMING_SNAKE_CASE : str = generator SCREAMING_SNAKE_CASE : Optional[Any] = self.question_encoder def _A ( self : Union[str, Any] , UpperCAmelCase_ : Any ): if os.path.isfile(UpperCAmelCase_ ): raise ValueError(f'''Provided path ({save_directory}) should be a directory, not a file''' ) os.makedirs(UpperCAmelCase_ , exist_ok=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(UpperCAmelCase_ , "question_encoder_tokenizer" ) SCREAMING_SNAKE_CASE : Any = os.path.join(UpperCAmelCase_ , "generator_tokenizer" ) self.question_encoder.save_pretrained(UpperCAmelCase_ ) self.generator.save_pretrained(UpperCAmelCase_ ) @classmethod def _A ( cls : List[Any] , UpperCAmelCase_ : List[Any] , **UpperCAmelCase_ : List[str] ): # dynamically import AutoTokenizer from ..auto.tokenization_auto import AutoTokenizer SCREAMING_SNAKE_CASE : Optional[int] = kwargs.pop("config" , UpperCAmelCase_ ) if config is None: SCREAMING_SNAKE_CASE : List[Any] = RagConfig.from_pretrained(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = AutoTokenizer.from_pretrained( UpperCAmelCase_ , config=config.question_encoder , subfolder="question_encoder_tokenizer" ) SCREAMING_SNAKE_CASE : str = AutoTokenizer.from_pretrained( UpperCAmelCase_ , config=config.generator , subfolder="generator_tokenizer" ) return cls(question_encoder=UpperCAmelCase_ , generator=UpperCAmelCase_ ) def __call__( self : Optional[Any] , *UpperCAmelCase_ : int , **UpperCAmelCase_ : List[str] ): return self.current_tokenizer(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[int] , *UpperCAmelCase_ : Union[str, Any] , **UpperCAmelCase_ : Tuple ): return self.generator.batch_decode(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , *UpperCAmelCase_ : Optional[int] , **UpperCAmelCase_ : int ): return self.generator.decode(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.question_encoder def _A ( self : Dict ): SCREAMING_SNAKE_CASE : List[Any] = self.generator def _A ( self : Optional[int] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[List[str]] = None , UpperCAmelCase_ : Optional[int] = None , UpperCAmelCase_ : Optional[int] = None , UpperCAmelCase_ : str = "longest" , UpperCAmelCase_ : str = None , UpperCAmelCase_ : bool = True , **UpperCAmelCase_ : Optional[Any] , ): warnings.warn( "`prepare_seq2seq_batch` is deprecated and will be removed in version 5 of 🤗 Transformers. Use the " "regular `__call__` method to prepare your inputs and the tokenizer under the `with_target_tokenizer` " "context manager to prepare your targets. See the documentation of your specific tokenizer for more " "details" , UpperCAmelCase_ , ) if max_length is None: SCREAMING_SNAKE_CASE : Union[str, Any] = self.current_tokenizer.model_max_length SCREAMING_SNAKE_CASE : List[str] = self( UpperCAmelCase_ , add_special_tokens=UpperCAmelCase_ , return_tensors=UpperCAmelCase_ , max_length=UpperCAmelCase_ , padding=UpperCAmelCase_ , truncation=UpperCAmelCase_ , **UpperCAmelCase_ , ) if tgt_texts is None: return model_inputs # Process tgt_texts if max_target_length is None: SCREAMING_SNAKE_CASE : Optional[int] = self.current_tokenizer.model_max_length SCREAMING_SNAKE_CASE : str = self( text_target=UpperCAmelCase_ , add_special_tokens=UpperCAmelCase_ , return_tensors=UpperCAmelCase_ , padding=UpperCAmelCase_ , max_length=UpperCAmelCase_ , truncation=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : str = labels["input_ids"] return model_inputs
62
import gc import random import unittest import numpy as np import torch from diffusers import DDIMScheduler, KandinskyVaaPipeline, KandinskyVaaPriorPipeline, UNetaDConditionModel, VQModel from diffusers.utils import floats_tensor, load_numpy, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = KandinskyVaaPipeline UpperCamelCase_ : List[Any] = [ '''image_embeds''', '''negative_image_embeds''', ] UpperCamelCase_ : Tuple = ['''image_embeds''', '''negative_image_embeds'''] UpperCamelCase_ : Any = [ '''generator''', '''height''', '''width''', '''latents''', '''guidance_scale''', '''num_inference_steps''', '''return_dict''', '''guidance_scale''', '''num_images_per_prompt''', '''output_type''', '''return_dict''', ] UpperCamelCase_ : List[str] = False @property def _A ( self : List[Any] ): return 32 @property def _A ( self : List[Any] ): return 32 @property def _A ( self : Any ): return self.time_input_dim @property def _A ( self : Union[str, Any] ): return self.time_input_dim * 4 @property def _A ( self : Tuple ): return 100 @property def _A ( self : Optional[int] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = { "in_channels": 4, # Out channels is double in channels because predicts mean and variance "out_channels": 8, "addition_embed_type": "image", "down_block_types": ("ResnetDownsampleBlock2D", "SimpleCrossAttnDownBlock2D"), "up_block_types": ("SimpleCrossAttnUpBlock2D", "ResnetUpsampleBlock2D"), "mid_block_type": "UNetMidBlock2DSimpleCrossAttn", "block_out_channels": (self.block_out_channels_a, self.block_out_channels_a * 2), "layers_per_block": 1, "encoder_hid_dim": self.text_embedder_hidden_size, "encoder_hid_dim_type": "image_proj", "cross_attention_dim": self.cross_attention_dim, "attention_head_dim": 4, "resnet_time_scale_shift": "scale_shift", "class_embed_type": None, } SCREAMING_SNAKE_CASE : str = UNetaDConditionModel(**UpperCAmelCase_ ) return model @property def _A ( self : int ): return { "block_out_channels": [32, 64], "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"], "in_channels": 3, "latent_channels": 4, "layers_per_block": 1, "norm_num_groups": 8, "norm_type": "spatial", "num_vq_embeddings": 12, "out_channels": 3, "up_block_types": [ "AttnUpDecoderBlock2D", "UpDecoderBlock2D", ], "vq_embed_dim": 4, } @property def _A ( self : Any ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : List[str] = VQModel(**self.dummy_movq_kwargs ) return model def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.dummy_unet SCREAMING_SNAKE_CASE : str = self.dummy_movq SCREAMING_SNAKE_CASE : Optional[int] = DDIMScheduler( num_train_timesteps=1000 , beta_schedule="linear" , beta_start=0.00_085 , beta_end=0.012 , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , steps_offset=1 , prediction_type="epsilon" , thresholding=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "scheduler": scheduler, "movq": movq, } return components def _A ( self : List[str] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any]=0 ): SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(seed + 1 ) ).to( UpperCAmelCase_ ) if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : List[Any] = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = { "image_embeds": image_embeds, "negative_image_embeds": negative_image_embeds, "generator": generator, "height": 64, "width": 64, "guidance_scale": 4.0, "num_inference_steps": 2, "output_type": "np", } return inputs def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "cpu" SCREAMING_SNAKE_CASE : Tuple = self.get_dummy_components() SCREAMING_SNAKE_CASE : List[str] = self.pipeline_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(**self.get_dummy_inputs(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : str = output.images SCREAMING_SNAKE_CASE : Tuple = pipe( **self.get_dummy_inputs(UpperCAmelCase_ ) , return_dict=UpperCAmelCase_ , )[0] SCREAMING_SNAKE_CASE : str = image[0, -3:, -3:, -1] SCREAMING_SNAKE_CASE : Any = image_from_tuple[0, -3:, -3:, -1] assert image.shape == (1, 64, 64, 3) SCREAMING_SNAKE_CASE : List[Any] = np.array( [0.6_237_976, 1.0, 0.36_441_332, 1.0, 0.70_639_634, 0.29_877_186, 0.85_652_125, 0.5_216_843, 0.54_454_046] ) assert ( np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_slice.flatten()}''' assert ( np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_from_tuple_slice.flatten()}''' @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main" "/kandinskyv22/kandinskyv22_text2img_cat_fp16.npy" ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPriorPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-prior" , torch_dtype=torch.floataa ) pipe_prior.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-decoder" , torch_dtype=torch.floataa ) SCREAMING_SNAKE_CASE : Tuple = pipeline.to(UpperCAmelCase_ ) pipeline.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = "red cat, 4k photo" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = pipe_prior( UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=5 , negative_prompt="" , ).to_tuple() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = pipeline( image_embeds=UpperCAmelCase_ , negative_image_embeds=UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=100 , output_type="np" , ) SCREAMING_SNAKE_CASE : List[Any] = output.images[0] assert image.shape == (512, 512, 3) assert_mean_pixel_difference(UpperCAmelCase_ , UpperCAmelCase_ )
62
1
import baseaa def lowerCamelCase__ ( lowercase ): """simple docstring""" return baseaa.aaaencode(string.encode("utf-8" ) ) def lowerCamelCase__ ( lowercase ): """simple docstring""" return baseaa.aaadecode(lowercase ).decode("utf-8" ) if __name__ == "__main__": import doctest doctest.testmod()
62
import logging import os import quant_trainer import torch from torch.utils.data import DataLoader from transformers import Trainer, is_torch_tpu_available from transformers.trainer_utils import PredictionOutput snake_case = logging.getLogger(__name__) if is_torch_tpu_available(check_device=False): import torch_xla.core.xla_model as xm import torch_xla.debug.metrics as met class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , *UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Tuple=None , **UpperCAmelCase_ : Optional[Any] ): super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = eval_examples SCREAMING_SNAKE_CASE : List[Any] = post_process_function SCREAMING_SNAKE_CASE : Any = quant_trainer_args SCREAMING_SNAKE_CASE : Optional[Any] = 128 # default number of calibration samples def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple=None ): if calib_dataset is None and self.calib_dataset is None: raise ValueError("Trainer: calibration requires an calib_dataset." ) SCREAMING_SNAKE_CASE : str = calib_dataset if calib_dataset is not None else self.calib_dataset SCREAMING_SNAKE_CASE : str = self._remove_unused_columns(UpperCAmelCase_ , description="Calibration" ) return DataLoader( UpperCAmelCase_ , batch_size=self.args.eval_batch_size , collate_fn=self.data_collator , drop_last=self.args.dataloader_drop_last , num_workers=self.args.dataloader_num_workers , pin_memory=self.args.dataloader_pin_memory , shuffle=UpperCAmelCase_ , ) def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[int]=None ): SCREAMING_SNAKE_CASE : Any = self.train_dataset if calib_dataset is None else calib_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_calib_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args , calib=UpperCAmelCase_ ) model.eval() quant_trainer.enable_calibration(UpperCAmelCase_ ) logger.info("***** Running calibration *****" ) logger.info(f''' Num examples = {self.calib_num}''' ) logger.info(f''' Batch size = {calib_dataloader.batch_size}''' ) for step, inputs in enumerate(UpperCAmelCase_ ): # Prediction step SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = self.prediction_step(UpperCAmelCase_ , UpperCAmelCase_ , prediction_loss_only=UpperCAmelCase_ ) if (step + 1) * calib_dataloader.batch_size >= self.calib_num: break quant_trainer.finish_calibration(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : Optional[int] = model def _A ( self : List[Any] , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : str = "eval" ): SCREAMING_SNAKE_CASE : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset SCREAMING_SNAKE_CASE : Tuple = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.eval_examples if eval_examples is None else eval_examples # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Dict = self.compute_metrics SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : int = eval_loop( UpperCAmelCase_ , description="Evaluation" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : int = compute_metrics if self.post_process_function is not None and self.compute_metrics is not None: SCREAMING_SNAKE_CASE : List[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions ) SCREAMING_SNAKE_CASE : Any = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : Union[str, Any] = metrics.pop(UpperCAmelCase_ ) self.log(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[Any] = {} if self.args.tpu_metrics_debug or self.args.debug: # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.) xm.master_print(met.metrics_report() ) SCREAMING_SNAKE_CASE : List[str] = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCAmelCase_ ) return metrics def _A ( self : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : str = "test" ): SCREAMING_SNAKE_CASE : str = self.get_test_dataloader(UpperCAmelCase_ ) # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Tuple = self.compute_metrics SCREAMING_SNAKE_CASE : Union[str, Any] = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : str = eval_loop( UpperCAmelCase_ , description="Prediction" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : Union[str, Any] = compute_metrics if self.post_process_function is None or self.compute_metrics is None: return output SCREAMING_SNAKE_CASE : Optional[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions , "predict" ) SCREAMING_SNAKE_CASE : str = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : str = metrics.pop(UpperCAmelCase_ ) return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : int="./" ): SCREAMING_SNAKE_CASE : List[Any] = self.eval_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = next(iter(UpperCAmelCase_ ) ) # saving device - to make it consistent SCREAMING_SNAKE_CASE : int = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) # convert to tuple SCREAMING_SNAKE_CASE : Tuple = tuple(v.to(UpperCAmelCase_ ) for k, v in batch.items() ) logger.info("Converting model to be onnx compatible" ) from pytorch_quantization.nn import TensorQuantizer SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Any = self.model.to(UpperCAmelCase_ ) model.eval() model.float() SCREAMING_SNAKE_CASE : str = model.module if hasattr(UpperCAmelCase_ , "module" ) else model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : List[Any] = os.path.join(UpperCAmelCase_ , "model.onnx" ) logger.info(f'''exporting model to {output_model_file}''' ) SCREAMING_SNAKE_CASE : int = {0: "batch_size", 1: "seq_len"} torch.onnx.export( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , export_params=UpperCAmelCase_ , opset_version=13 , do_constant_folding=UpperCAmelCase_ , input_names=["input_ids", "attention_mask", "token_type_ids"] , output_names=["output_start_logits", "output_end_logits"] , dynamic_axes={ "input_ids": axes, "attention_mask": axes, "token_type_ids": axes, "output_start_logits": axes, "output_end_logits": axes, } , verbose=UpperCAmelCase_ , ) logger.info("onnx export finished" )
62
1
import argparse from tax import checkpoints from transformers import AutoConfig, FlaxAutoModelForSeqaSeqLM def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = AutoConfig.from_pretrained(lowercase ) SCREAMING_SNAKE_CASE : str = FlaxAutoModelForSeqaSeqLM.from_config(config=lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = checkpoints.load_tax_checkpoint(lowercase ) SCREAMING_SNAKE_CASE : Union[str, Any] = "wi_0" in tax_model["target"]["encoder"]["layers_0"]["mlp"] if config.model_type == "t5": SCREAMING_SNAKE_CASE : Optional[Any] = "SelfAttention" if config.model_type == "longt5" and config.encoder_attention_type == "local": SCREAMING_SNAKE_CASE : str = "LocalSelfAttention" elif config.model_type == "longt5" and config.encoder_attention_type == "transient-global": SCREAMING_SNAKE_CASE : Tuple = "TransientGlobalSelfAttention" else: raise ValueError( "Given config is expected to have `model_type='t5'`, or `model_type='longt5` with `encoder_attention_type`" " attribute with a value from ['local', 'transient-global]." ) # Encoder for layer_index in range(config.num_layers ): SCREAMING_SNAKE_CASE : Optional[int] = F'''layers_{str(lowercase )}''' # Self-Attention SCREAMING_SNAKE_CASE : Tuple = tax_model["target"]["encoder"][layer_name]["attention"]["key"]["kernel"] SCREAMING_SNAKE_CASE : List[Any] = tax_model["target"]["encoder"][layer_name]["attention"]["out"]["kernel"] SCREAMING_SNAKE_CASE : Optional[Any] = tax_model["target"]["encoder"][layer_name]["attention"]["query"]["kernel"] SCREAMING_SNAKE_CASE : int = tax_model["target"]["encoder"][layer_name]["attention"]["value"]["kernel"] # Global input layer norm if config.model_type == "longt5" and config.encoder_attention_type == "transient-global": SCREAMING_SNAKE_CASE : Tuple = tax_model["target"]["encoder"][layer_name]["attention"]["T5LayerNorm_0"]["scale"] # Layer Normalization SCREAMING_SNAKE_CASE : List[Any] = tax_model["target"]["encoder"][layer_name]["pre_attention_layer_norm"]["scale"] if split_mlp_wi: SCREAMING_SNAKE_CASE : Optional[int] = tax_model["target"]["encoder"][layer_name]["mlp"]["wi_0"]["kernel"] SCREAMING_SNAKE_CASE : Any = tax_model["target"]["encoder"][layer_name]["mlp"]["wi_1"]["kernel"] else: SCREAMING_SNAKE_CASE : Tuple = tax_model["target"]["encoder"][layer_name]["mlp"]["wi"]["kernel"] SCREAMING_SNAKE_CASE : int = tax_model["target"]["encoder"][layer_name]["mlp"]["wo"]["kernel"] # Layer Normalization SCREAMING_SNAKE_CASE : int = tax_model["target"]["encoder"][layer_name]["pre_mlp_layer_norm"]["scale"] # Assigning SCREAMING_SNAKE_CASE : Any = flax_model.params["encoder"]["block"][str(lowercase )]["layer"] SCREAMING_SNAKE_CASE : List[Any] = tax_attention_key SCREAMING_SNAKE_CASE : Optional[int] = tax_attention_out SCREAMING_SNAKE_CASE : List[str] = tax_attention_query SCREAMING_SNAKE_CASE : int = tax_attention_value SCREAMING_SNAKE_CASE : str = tax_attention_layer_norm # Global input layer norm if config.model_type == "longt5" and config.encoder_attention_type == "transient-global": SCREAMING_SNAKE_CASE : str = tax_global_layer_norm if split_mlp_wi: SCREAMING_SNAKE_CASE : Optional[Any] = tax_mlp_wi_a SCREAMING_SNAKE_CASE : Tuple = tax_mlp_wi_a else: SCREAMING_SNAKE_CASE : Tuple = tax_mlp_wi SCREAMING_SNAKE_CASE : Union[str, Any] = tax_mlp_wo SCREAMING_SNAKE_CASE : Any = tax_mlp_layer_norm SCREAMING_SNAKE_CASE : Optional[int] = flax_model_encoder_layer_block # Only for layer 0: SCREAMING_SNAKE_CASE : Optional[Any] = tax_model["target"]["encoder"]["relpos_bias"]["rel_embedding"].T SCREAMING_SNAKE_CASE : Optional[int] = tax_encoder_rel_embedding # Side/global relative position_bias + layer norm if config.model_type == "longt5" and config.encoder_attention_type == "transient-global": SCREAMING_SNAKE_CASE : List[Any] = tax_model["target"]["encoder"]["side_relpos_bias"]["rel_embedding"].T SCREAMING_SNAKE_CASE : str = tax_encoder_global_rel_embedding # Assigning SCREAMING_SNAKE_CASE : List[Any] = tax_model["target"]["encoder"]["encoder_norm"]["scale"] SCREAMING_SNAKE_CASE : Optional[Any] = tax_encoder_norm # Decoder for layer_index in range(config.num_layers ): SCREAMING_SNAKE_CASE : str = F'''layers_{str(lowercase )}''' # Self-Attention SCREAMING_SNAKE_CASE : List[str] = tax_model["target"]["decoder"][layer_name]["self_attention"]["key"]["kernel"] SCREAMING_SNAKE_CASE : Union[str, Any] = tax_model["target"]["decoder"][layer_name]["self_attention"]["out"]["kernel"] SCREAMING_SNAKE_CASE : Union[str, Any] = tax_model["target"]["decoder"][layer_name]["self_attention"]["query"]["kernel"] SCREAMING_SNAKE_CASE : List[Any] = tax_model["target"]["decoder"][layer_name]["self_attention"]["value"]["kernel"] # Layer Normalization SCREAMING_SNAKE_CASE : str = tax_model["target"]["decoder"][layer_name]["pre_self_attention_layer_norm"][ "scale" ] # Encoder-Decoder-Attention SCREAMING_SNAKE_CASE : str = tax_model["target"]["decoder"][layer_name]["encoder_decoder_attention"] SCREAMING_SNAKE_CASE : Dict = tax_enc_dec_attention_module["key"]["kernel"] SCREAMING_SNAKE_CASE : List[str] = tax_enc_dec_attention_module["out"]["kernel"] SCREAMING_SNAKE_CASE : Optional[int] = tax_enc_dec_attention_module["query"]["kernel"] SCREAMING_SNAKE_CASE : List[Any] = tax_enc_dec_attention_module["value"]["kernel"] # Layer Normalization SCREAMING_SNAKE_CASE : Optional[Any] = tax_model["target"]["decoder"][layer_name]["pre_cross_attention_layer_norm"]["scale"] # MLP if split_mlp_wi: SCREAMING_SNAKE_CASE : Dict = tax_model["target"]["decoder"][layer_name]["mlp"]["wi_0"]["kernel"] SCREAMING_SNAKE_CASE : Dict = tax_model["target"]["decoder"][layer_name]["mlp"]["wi_1"]["kernel"] else: SCREAMING_SNAKE_CASE : Dict = tax_model["target"]["decoder"][layer_name]["mlp"]["wi"]["kernel"] SCREAMING_SNAKE_CASE : Optional[int] = tax_model["target"]["decoder"][layer_name]["mlp"]["wo"]["kernel"] # Layer Normalization SCREAMING_SNAKE_CASE : Optional[int] = tax_model["target"]["decoder"][layer_name]["pre_mlp_layer_norm"]["scale"] # Assigning SCREAMING_SNAKE_CASE : List[str] = flax_model.params["decoder"]["block"][str(lowercase )]["layer"] SCREAMING_SNAKE_CASE : Union[str, Any] = tax_attention_key SCREAMING_SNAKE_CASE : Optional[Any] = tax_attention_out SCREAMING_SNAKE_CASE : List[Any] = tax_attention_query SCREAMING_SNAKE_CASE : Union[str, Any] = tax_attention_value SCREAMING_SNAKE_CASE : Any = tax_pre_attention_layer_norm SCREAMING_SNAKE_CASE : Any = tax_enc_dec_attention_key SCREAMING_SNAKE_CASE : Union[str, Any] = tax_enc_dec_attention_out SCREAMING_SNAKE_CASE : Any = tax_enc_dec_attention_query SCREAMING_SNAKE_CASE : Dict = tax_enc_dec_attention_value SCREAMING_SNAKE_CASE : Union[str, Any] = tax_cross_layer_norm if split_mlp_wi: SCREAMING_SNAKE_CASE : Optional[int] = tax_mlp_wi_a SCREAMING_SNAKE_CASE : int = tax_mlp_wi_a else: SCREAMING_SNAKE_CASE : Tuple = tax_mlp_wi SCREAMING_SNAKE_CASE : Optional[int] = tax_mlp_wo SCREAMING_SNAKE_CASE : str = txa_mlp_layer_norm SCREAMING_SNAKE_CASE : List[str] = flax_model_decoder_layer_block # Decoder Normalization SCREAMING_SNAKE_CASE : int = tax_model["target"]["decoder"]["decoder_norm"]["scale"] SCREAMING_SNAKE_CASE : List[str] = txa_decoder_norm # Only for layer 0: SCREAMING_SNAKE_CASE : List[str] = tax_model["target"]["decoder"]["relpos_bias"]["rel_embedding"].T SCREAMING_SNAKE_CASE : List[Any] = tax_decoder_rel_embedding # Token Embeddings SCREAMING_SNAKE_CASE : List[Any] = tax_model["target"]["token_embedder"]["embedding"] SCREAMING_SNAKE_CASE : Tuple = txa_token_embeddings # LM Head (only in v1.1 and LongT5 checkpoints) if "logits_dense" in tax_model["target"]["decoder"]: SCREAMING_SNAKE_CASE : int = tax_model["target"]["decoder"]["logits_dense"]["kernel"] flax_model.save_pretrained(lowercase ) print("T5X Model was sucessfully converted!" ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument( """--t5x_checkpoint_path""", default=None, type=str, required=True, help="""Path the T5X checkpoint.""" ) parser.add_argument("""--config_name""", default=None, type=str, required=True, help="""Config name of LongT5/T5 model.""") parser.add_argument( """--flax_dump_folder_path""", default=None, type=str, required=True, help="""Path to the output FLAX model.""" ) snake_case = parser.parse_args() convert_tax_checkpoint_to_flax(args.tax_checkpoint_path, args.config_name, args.flax_dump_folder_path)
62
import os import unittest from transformers import LayoutLMTokenizer, LayoutLMTokenizerFast from transformers.models.layoutlm.tokenization_layoutlm import VOCAB_FILES_NAMES from transformers.testing_utils import require_tokenizers from ...test_tokenization_common import TokenizerTesterMixin @require_tokenizers class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = LayoutLMTokenizer UpperCamelCase_ : str = LayoutLMTokenizerFast UpperCamelCase_ : Any = True UpperCamelCase_ : Optional[Any] = True def _A ( self : Any ): super().setUp() SCREAMING_SNAKE_CASE : Optional[Any] = [ "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest", ] SCREAMING_SNAKE_CASE : Optional[Any] = 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 _A ( self : str , **UpperCAmelCase_ : Optional[int] ): return LayoutLMTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Any = "UNwant\u00E9d,running" SCREAMING_SNAKE_CASE : Union[str, Any] = "unwanted, running" return input_text, output_text def _A ( self : int ): SCREAMING_SNAKE_CASE : Optional[int] = self.tokenizer_class(self.vocab_file ) SCREAMING_SNAKE_CASE : Dict = tokenizer.tokenize("UNwant\u00E9d,running" ) self.assertListEqual(UpperCAmelCase_ , ["un", "##want", "##ed", ",", "runn", "##ing"] ) self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , [7, 4, 5, 10, 8, 9] ) def _A ( self : List[str] ): pass
62
1
import pyarrow.parquet as pq import pytest from datasets import Audio, Dataset, DatasetDict, Features, NamedSplit, Sequence, Value, config from datasets.features.image import Image from datasets.io.parquet import ParquetDatasetReader, ParquetDatasetWriter, get_writer_batch_size from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) assert dataset.num_rows == 4 assert dataset.num_columns == 3 assert dataset.column_names == ["col_1", "col_2", "col_3"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize("keep_in_memory" , [False, True] ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = tmp_path / "cache" SCREAMING_SNAKE_CASE : str = {"col_1": "string", "col_2": "int64", "col_3": "float64"} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): SCREAMING_SNAKE_CASE : int = ParquetDatasetReader(lowercase , cache_dir=lowercase , keep_in_memory=lowercase ).read() _check_parquet_dataset(lowercase , lowercase ) @pytest.mark.parametrize( "features" , [ None, {"col_1": "string", "col_2": "int64", "col_3": "float64"}, {"col_1": "string", "col_2": "string", "col_3": "string"}, {"col_1": "int32", "col_2": "int32", "col_3": "int32"}, {"col_1": "float32", "col_2": "float32", "col_3": "float32"}, ] , ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = tmp_path / "cache" SCREAMING_SNAKE_CASE : int = {"col_1": "string", "col_2": "int64", "col_3": "float64"} SCREAMING_SNAKE_CASE : str = features.copy() if features else default_expected_features SCREAMING_SNAKE_CASE : str = ( Features({feature: Value(lowercase ) for feature, dtype in features.items()} ) if features is not None else None ) SCREAMING_SNAKE_CASE : List[Any] = ParquetDatasetReader(lowercase , features=lowercase , cache_dir=lowercase ).read() _check_parquet_dataset(lowercase , lowercase ) @pytest.mark.parametrize("split" , [None, NamedSplit("train" ), "train", "test"] ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = tmp_path / "cache" SCREAMING_SNAKE_CASE : Any = {"col_1": "string", "col_2": "int64", "col_3": "float64"} SCREAMING_SNAKE_CASE : str = ParquetDatasetReader(lowercase , cache_dir=lowercase , split=lowercase ).read() _check_parquet_dataset(lowercase , lowercase ) assert dataset.split == split if split else "train" @pytest.mark.parametrize("path_type" , [str, list] ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if issubclass(lowercase , lowercase ): SCREAMING_SNAKE_CASE : Union[str, Any] = parquet_path elif issubclass(lowercase , lowercase ): SCREAMING_SNAKE_CASE : Any = [parquet_path] SCREAMING_SNAKE_CASE : str = tmp_path / "cache" SCREAMING_SNAKE_CASE : str = {"col_1": "string", "col_2": "int64", "col_3": "float64"} SCREAMING_SNAKE_CASE : str = ParquetDatasetReader(lowercase , cache_dir=lowercase ).read() _check_parquet_dataset(lowercase , lowercase ) def lowerCamelCase__ ( lowercase , lowercase , lowercase=("train",) ): """simple docstring""" assert isinstance(lowercase , lowercase ) for split in splits: SCREAMING_SNAKE_CASE : Union[str, Any] = dataset_dict[split] assert dataset.num_rows == 4 assert dataset.num_columns == 3 assert dataset.column_names == ["col_1", "col_2", "col_3"] for feature, expected_dtype in expected_features.items(): assert dataset.features[feature].dtype == expected_dtype @pytest.mark.parametrize("keep_in_memory" , [False, True] ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = tmp_path / "cache" SCREAMING_SNAKE_CASE : str = {"col_1": "string", "col_2": "int64", "col_3": "float64"} with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase(): SCREAMING_SNAKE_CASE : List[Any] = ParquetDatasetReader( {"train": parquet_path} , cache_dir=lowercase , keep_in_memory=lowercase ).read() _check_parquet_datasetdict(lowercase , lowercase ) @pytest.mark.parametrize( "features" , [ None, {"col_1": "string", "col_2": "int64", "col_3": "float64"}, {"col_1": "string", "col_2": "string", "col_3": "string"}, {"col_1": "int32", "col_2": "int32", "col_3": "int32"}, {"col_1": "float32", "col_2": "float32", "col_3": "float32"}, ] , ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = tmp_path / "cache" SCREAMING_SNAKE_CASE : int = {"col_1": "string", "col_2": "int64", "col_3": "float64"} SCREAMING_SNAKE_CASE : Optional[int] = features.copy() if features else default_expected_features SCREAMING_SNAKE_CASE : Dict = ( Features({feature: Value(lowercase ) for feature, dtype in features.items()} ) if features is not None else None ) SCREAMING_SNAKE_CASE : Tuple = ParquetDatasetReader({"train": parquet_path} , features=lowercase , cache_dir=lowercase ).read() _check_parquet_datasetdict(lowercase , lowercase ) @pytest.mark.parametrize("split" , [None, NamedSplit("train" ), "train", "test"] ) def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if split: SCREAMING_SNAKE_CASE : Dict = {split: parquet_path} else: SCREAMING_SNAKE_CASE : Optional[Any] = "train" SCREAMING_SNAKE_CASE : Dict = {"train": parquet_path, "test": parquet_path} SCREAMING_SNAKE_CASE : str = tmp_path / "cache" SCREAMING_SNAKE_CASE : Union[str, Any] = {"col_1": "string", "col_2": "int64", "col_3": "float64"} SCREAMING_SNAKE_CASE : int = ParquetDatasetReader(lowercase , cache_dir=lowercase ).read() _check_parquet_datasetdict(lowercase , lowercase , splits=list(path.keys() ) ) assert all(dataset[split].split == split for split in path.keys() ) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = ParquetDatasetWriter(lowercase , tmp_path / "foo.parquet" ) assert writer.write() > 0 SCREAMING_SNAKE_CASE : Optional[int] = pq.ParquetFile(tmp_path / "foo.parquet" ) SCREAMING_SNAKE_CASE : Union[str, Any] = pf.read() assert dataset.data.table == output_table def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = str(shared_datadir / "test_image_rgb.jpg" ) SCREAMING_SNAKE_CASE : Union[str, Any] = {"image": [image_path]} SCREAMING_SNAKE_CASE : Any = Features({"image": Image()} ) SCREAMING_SNAKE_CASE : List[Any] = Dataset.from_dict(lowercase , features=lowercase ) SCREAMING_SNAKE_CASE : Tuple = ParquetDatasetWriter(lowercase , tmp_path / "foo.parquet" ) assert writer.write() > 0 SCREAMING_SNAKE_CASE : Dict = Dataset.from_parquet(str(tmp_path / "foo.parquet" ) ) assert dataset.features == reloaded_dataset.features SCREAMING_SNAKE_CASE : Dict = ParquetDatasetReader(str(tmp_path / "foo.parquet" ) , streaming=lowercase ).read() assert dataset.features == reloaded_iterable_dataset.features @pytest.mark.parametrize( "feature, expected" , [ (Features({"foo": Value("int32" )} ), None), (Features({"image": Image(), "foo": Value("int32" )} ), config.PARQUET_ROW_GROUP_SIZE_FOR_IMAGE_DATASETS), (Features({"nested": Sequence(Audio() )} ), config.PARQUET_ROW_GROUP_SIZE_FOR_AUDIO_DATASETS), ] , ) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert get_writer_batch_size(lowercase ) == expected
62
import argparse import json import os from collections import OrderedDict import numpy as np import tensorflow as tf import torch def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = os.path.join(args.tf_model_dir , "parameters.json" ) SCREAMING_SNAKE_CASE : Union[str, Any] = json.loads(open(lowercase ).read() ) if not params: raise ValueError( F'''It seems that the json file at {parameter_file} is empty. Make sure you have a correct json file.''' ) if not args.output.endswith(".pt" ): SCREAMING_SNAKE_CASE : Optional[int] = args.output + ".pt" SCREAMING_SNAKE_CASE : Any = OrderedDict() with tf.device("/CPU:0" ): SCREAMING_SNAKE_CASE : Union[str, Any] = tf.train.load_checkpoint(args.tf_model_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = reader.get_variable_to_shape_map() for key_name in shapes.keys(): SCREAMING_SNAKE_CASE : Any = reader.get_tensor(lowercase ).astype(np.floataa ) if key_name.endswith("/adam_m" ) or key_name.endswith("/adam_v" ): continue if key_name.startswith("pasts/" ): if key_name.startswith("pasts/mlp" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9] ) elif key_name.startswith("pasts/out" ): SCREAMING_SNAKE_CASE : Optional[int] = 8 SCREAMING_SNAKE_CASE : List[Any] = "model.sqout.%d.weight" % (player * 2) # enter to nn.Sequencial with Tanh, so 2 at a time SCREAMING_SNAKE_CASE : Optional[int] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.startswith("model/moe" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/switch_gating/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.router.classifier.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/softmlp/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.soft_bypass_mlp.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/wo/kernel" ) or key_name.endswith("/wi/kernel" ): SCREAMING_SNAKE_CASE : Optional[int] = key_name[-9:-7] for i in range(16 ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.feed_forward.mlp.experts.expert_%d.%s.weight" % (player, i, nlayer) SCREAMING_SNAKE_CASE : List[str] = ( vnp[i].transpose([1, 0] ).copy() ) # In Mesh-Tensorflow, it is one array, so it is divided SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name.startswith("model/mlp" ): SCREAMING_SNAKE_CASE : str = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/p1/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.mlp.wi.weight" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/p1/bias" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.wi.bias" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/kernel" ): SCREAMING_SNAKE_CASE : str = "model.blocks.%d.feed_forward.mlp.wo.weight" % player SCREAMING_SNAKE_CASE : List[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/bias" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.feed_forward.mlp.wo.bias" % player SCREAMING_SNAKE_CASE : str = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) elif key_name.startswith("model/ln" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.norm.bias" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : List[str] = "model.blocks.%d.feed_forward.norm.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/att" ): SCREAMING_SNAKE_CASE : Optional[int] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/qkv/kernel" ): SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # Compute same dimension as Mesh-tensorflow using einsum SCREAMING_SNAKE_CASE : List[str] = state[:, 0, :, :] SCREAMING_SNAKE_CASE : Tuple = state[:, 1, :, :] SCREAMING_SNAKE_CASE : List[Any] = state[:, 2, :, :] SCREAMING_SNAKE_CASE : Tuple = ( state_q.reshape([state_q.shape[0], state_q.shape[1] * state_q.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = ( state_k.reshape([state_k.shape[0], state_k.shape[1] * state_k.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = ( state_v.reshape([state_v.shape[0], state_v.shape[1] * state_v.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : int = "model.blocks.%d.self_attn.self_attn.q_proj.weight" % player SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = "model.blocks.%d.self_attn.self_attn.k_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = "model.blocks.%d.self_attn.self_attn.v_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/o/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.self_attn.self_attn.out_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = ( vnp.reshape([vnp.shape[0] * vnp.shape[1], vnp.shape[2]] ).transpose([1, 0] ).copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif key_name.startswith("model/an" ): SCREAMING_SNAKE_CASE : int = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.self_attn.norm.bias" % player SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[str] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.self_attn.norm.weight" % player SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif ( key_name.startswith("model/wte" ) or key_name.startswith("model/wpe" ) or key_name.startswith("model/ete" ) ): SCREAMING_SNAKE_CASE : str = {"wte": "embed_tokens", "wpe": "position_embeddings", "ete": "extra_position_embeddings"}[ key_name[-3:] ] SCREAMING_SNAKE_CASE : List[str] = "model.%s.weight" % nlayer SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) if key_name.startswith("model/wte" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "lm_head.weight" SCREAMING_SNAKE_CASE : List[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/wob" ): SCREAMING_SNAKE_CASE : List[Any] = "final_logits_bias" SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : List[str] = state.reshape((1, -1) ) SCREAMING_SNAKE_CASE : int = torch.tensor(lowercase ) elif key_name == "model/dense/kernel": SCREAMING_SNAKE_CASE : Optional[int] = "model.last_project.weight" SCREAMING_SNAKE_CASE : Dict = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name == "model/dense_1/bias": SCREAMING_SNAKE_CASE : str = "model.last_project.bias" SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) torch.save(lowercase , args.output ) if __name__ == "__main__": snake_case = argparse.ArgumentParser( description="""model converter.""", formatter_class=argparse.ArgumentDefaultsHelpFormatter ) parser.add_argument("""--tf_model_dir""", metavar="""PATH""", type=str, required=True, help="""import model""") parser.add_argument("""--output""", metavar="""PATH""", type=str, required=True, help="""output model""") snake_case = parser.parse_args() convert_tf_gptsan_to_pt(args)
62
1
from collections import OrderedDict from typing import Any, Mapping, Optional from ... import PreTrainedTokenizer from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfig, OnnxConfigWithPast, OnnxSeqaSeqConfigWithPast from ...onnx.utils import compute_effective_axis_dimension from ...utils import TensorType, is_torch_available, logging snake_case = logging.get_logger(__name__) snake_case = { """Helsinki-NLP/opus-mt-en-de""": """https://huggingface.co/Helsinki-NLP/opus-mt-en-de/resolve/main/config.json""", # See all Marian models at https://huggingface.co/models?filter=marian } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : List[Any] = '''marian''' UpperCamelCase_ : Tuple = ['''past_key_values'''] UpperCamelCase_ : Union[str, Any] = {'''num_attention_heads''': '''encoder_attention_heads''', '''hidden_size''': '''d_model'''} def __init__( self : Optional[int] , UpperCAmelCase_ : str=5_8101 , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Union[str, Any]=1024 , UpperCAmelCase_ : Optional[int]=12 , UpperCAmelCase_ : Optional[Any]=4096 , UpperCAmelCase_ : List[Any]=16 , UpperCAmelCase_ : str=12 , UpperCAmelCase_ : Optional[int]=4096 , UpperCAmelCase_ : Tuple=16 , UpperCAmelCase_ : int=0.0 , UpperCAmelCase_ : List[Any]=0.0 , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : Dict="gelu" , UpperCAmelCase_ : Optional[Any]=1024 , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : List[Any]=0.0 , UpperCAmelCase_ : List[Any]=0.0 , UpperCAmelCase_ : Optional[Any]=0.02 , UpperCAmelCase_ : Optional[Any]=5_8100 , UpperCAmelCase_ : Optional[Any]=False , UpperCAmelCase_ : Dict=5_8100 , UpperCAmelCase_ : Tuple=0 , UpperCAmelCase_ : Optional[int]=0 , UpperCAmelCase_ : List[Any]=True , **UpperCAmelCase_ : Optional[Any] , ): SCREAMING_SNAKE_CASE : int = vocab_size SCREAMING_SNAKE_CASE : Optional[int] = decoder_vocab_size or vocab_size SCREAMING_SNAKE_CASE : Dict = max_position_embeddings SCREAMING_SNAKE_CASE : Optional[Any] = d_model SCREAMING_SNAKE_CASE : List[str] = encoder_ffn_dim SCREAMING_SNAKE_CASE : Dict = encoder_layers SCREAMING_SNAKE_CASE : int = encoder_attention_heads SCREAMING_SNAKE_CASE : Optional[int] = decoder_ffn_dim SCREAMING_SNAKE_CASE : Union[str, Any] = decoder_layers SCREAMING_SNAKE_CASE : List[Any] = decoder_attention_heads SCREAMING_SNAKE_CASE : Optional[Any] = dropout SCREAMING_SNAKE_CASE : Optional[Any] = attention_dropout SCREAMING_SNAKE_CASE : Optional[int] = activation_dropout SCREAMING_SNAKE_CASE : Union[str, Any] = activation_function SCREAMING_SNAKE_CASE : List[str] = init_std SCREAMING_SNAKE_CASE : List[str] = encoder_layerdrop SCREAMING_SNAKE_CASE : Tuple = decoder_layerdrop SCREAMING_SNAKE_CASE : Optional[Any] = use_cache SCREAMING_SNAKE_CASE : List[str] = encoder_layers SCREAMING_SNAKE_CASE : Tuple = scale_embedding # scale factor will be sqrt(d_model) if True SCREAMING_SNAKE_CASE : List[Any] = share_encoder_decoder_embeddings super().__init__( pad_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , is_encoder_decoder=UpperCAmelCase_ , decoder_start_token_id=UpperCAmelCase_ , forced_eos_token_id=UpperCAmelCase_ , **UpperCAmelCase_ , ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @property # Copied from transformers.models.bart.configuration_bart.BartOnnxConfig.inputs def _A ( self : int ): if self.task in ["default", "seq2seq-lm"]: SCREAMING_SNAKE_CASE : Union[str, Any] = OrderedDict( [ ("input_ids", {0: "batch", 1: "encoder_sequence"}), ("attention_mask", {0: "batch", 1: "encoder_sequence"}), ] ) if self.use_past: SCREAMING_SNAKE_CASE : Tuple = {0: "batch"} SCREAMING_SNAKE_CASE : int = {0: "batch", 1: "past_decoder_sequence + sequence"} else: SCREAMING_SNAKE_CASE : Optional[int] = {0: "batch", 1: "decoder_sequence"} SCREAMING_SNAKE_CASE : int = {0: "batch", 1: "decoder_sequence"} if self.use_past: self.fill_with_past_key_values_(UpperCAmelCase_ , direction="inputs" ) elif self.task == "causal-lm": # TODO: figure this case out. SCREAMING_SNAKE_CASE : Optional[int] = OrderedDict( [ ("input_ids", {0: "batch", 1: "encoder_sequence"}), ("attention_mask", {0: "batch", 1: "encoder_sequence"}), ] ) if self.use_past: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : str = self.num_layers for i in range(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Optional[Any] = {0: "batch", 2: "past_sequence + sequence"} SCREAMING_SNAKE_CASE : Union[str, Any] = {0: "batch", 2: "past_sequence + sequence"} else: SCREAMING_SNAKE_CASE : Tuple = OrderedDict( [ ("input_ids", {0: "batch", 1: "encoder_sequence"}), ("attention_mask", {0: "batch", 1: "encoder_sequence"}), ("decoder_input_ids", {0: "batch", 1: "decoder_sequence"}), ("decoder_attention_mask", {0: "batch", 1: "decoder_sequence"}), ] ) return common_inputs @property # Copied from transformers.models.bart.configuration_bart.BartOnnxConfig.outputs def _A ( self : str ): if self.task in ["default", "seq2seq-lm"]: SCREAMING_SNAKE_CASE : str = super().outputs else: SCREAMING_SNAKE_CASE : Optional[Any] = super(UpperCAmelCase_ , self ).outputs if self.use_past: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = self.num_layers for i in range(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Optional[int] = {0: "batch", 2: "past_sequence + sequence"} SCREAMING_SNAKE_CASE : Union[str, Any] = {0: "batch", 2: "past_sequence + sequence"} return common_outputs def _A ( self : List[str] , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : Optional[TensorType] = None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = self._generate_dummy_inputs_for_encoder_and_decoder( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) # Generate decoder inputs SCREAMING_SNAKE_CASE : List[str] = seq_length if not self.use_past else 1 SCREAMING_SNAKE_CASE : List[str] = self._generate_dummy_inputs_for_encoder_and_decoder( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = {f'''decoder_{name}''': tensor for name, tensor in decoder_inputs.items()} SCREAMING_SNAKE_CASE : str = dict(**UpperCAmelCase_ , **UpperCAmelCase_ ) if self.use_past: if not is_torch_available(): raise ValueError("Cannot generate dummy past_keys inputs without PyTorch installed." ) else: import torch SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = common_inputs["input_ids"].shape SCREAMING_SNAKE_CASE : Union[str, Any] = common_inputs["decoder_input_ids"].shape[1] SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = self.num_attention_heads SCREAMING_SNAKE_CASE : Any = ( batch, num_encoder_attention_heads, encoder_seq_length, self._config.hidden_size // num_encoder_attention_heads, ) SCREAMING_SNAKE_CASE : List[str] = decoder_seq_length + 3 SCREAMING_SNAKE_CASE : Optional[Any] = ( batch, num_decoder_attention_heads, decoder_past_length, self._config.hidden_size // num_decoder_attention_heads, ) SCREAMING_SNAKE_CASE : Optional[int] = torch.cat( [common_inputs["decoder_attention_mask"], torch.ones(UpperCAmelCase_ , UpperCAmelCase_ )] , dim=1 ) SCREAMING_SNAKE_CASE : Optional[Any] = [] # If the number of encoder and decoder layers are present in the model configuration, both are considered SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = self.num_layers SCREAMING_SNAKE_CASE : Dict = min(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = max(UpperCAmelCase_ , UpperCAmelCase_ ) - min_num_layers SCREAMING_SNAKE_CASE : Optional[int] = "encoder" if num_encoder_layers > num_decoder_layers else "decoder" for _ in range(UpperCAmelCase_ ): common_inputs["past_key_values"].append( ( torch.zeros(UpperCAmelCase_ ), torch.zeros(UpperCAmelCase_ ), torch.zeros(UpperCAmelCase_ ), torch.zeros(UpperCAmelCase_ ), ) ) # TODO: test this. SCREAMING_SNAKE_CASE : Optional[int] = encoder_shape if remaining_side_name == "encoder" else decoder_shape for _ in range(UpperCAmelCase_ , UpperCAmelCase_ ): common_inputs["past_key_values"].append((torch.zeros(UpperCAmelCase_ ), torch.zeros(UpperCAmelCase_ )) ) return common_inputs def _A ( self : int , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : Optional[TensorType] = None , ): SCREAMING_SNAKE_CASE : List[str] = self._generate_dummy_inputs_for_encoder_and_decoder( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) if self.use_past: if not is_torch_available(): raise ValueError("Cannot generate dummy past_keys inputs without PyTorch installed." ) else: import torch SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = common_inputs["input_ids"].shape # Not using the same length for past_key_values SCREAMING_SNAKE_CASE : Union[str, Any] = seqlen + 2 SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.num_layers SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = self.num_attention_heads SCREAMING_SNAKE_CASE : Union[str, Any] = ( batch, num_encoder_attention_heads, past_key_values_length, self._config.hidden_size // num_encoder_attention_heads, ) SCREAMING_SNAKE_CASE : List[str] = common_inputs["attention_mask"].dtype SCREAMING_SNAKE_CASE : int = torch.cat( [common_inputs["attention_mask"], torch.ones(UpperCAmelCase_ , UpperCAmelCase_ , dtype=UpperCAmelCase_ )] , dim=1 ) SCREAMING_SNAKE_CASE : Dict = [ (torch.zeros(UpperCAmelCase_ ), torch.zeros(UpperCAmelCase_ )) for _ in range(UpperCAmelCase_ ) ] return common_inputs def _A ( self : str , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : Optional[TensorType] = None , ): # Copied from OnnxConfig.generate_dummy_inputs # Did not use super(OnnxConfigWithPast, self).generate_dummy_inputs for code clarity. # If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX SCREAMING_SNAKE_CASE : Tuple = compute_effective_axis_dimension( UpperCAmelCase_ , fixed_dimension=OnnxConfig.default_fixed_batch , num_token_to_add=0 ) # If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX SCREAMING_SNAKE_CASE : Dict = tokenizer.num_special_tokens_to_add(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = compute_effective_axis_dimension( UpperCAmelCase_ , fixed_dimension=OnnxConfig.default_fixed_sequence , num_token_to_add=UpperCAmelCase_ ) # Generate dummy inputs according to compute batch and sequence SCREAMING_SNAKE_CASE : Dict = [" ".join([tokenizer.unk_token] ) * seq_length] * batch_size SCREAMING_SNAKE_CASE : Tuple = dict(tokenizer(UpperCAmelCase_ , return_tensors=UpperCAmelCase_ ) ) return common_inputs def _A ( self : Optional[Any] , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : Optional[TensorType] = None , ): if self.task in ["default", "seq2seq-lm"]: SCREAMING_SNAKE_CASE : Optional[Any] = self._generate_dummy_inputs_for_default_and_seqaseq_lm( UpperCAmelCase_ , batch_size=UpperCAmelCase_ , seq_length=UpperCAmelCase_ , is_pair=UpperCAmelCase_ , framework=UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = self._generate_dummy_inputs_for_causal_lm( UpperCAmelCase_ , batch_size=UpperCAmelCase_ , seq_length=UpperCAmelCase_ , is_pair=UpperCAmelCase_ , framework=UpperCAmelCase_ ) return common_inputs def _A ( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int ): if self.task in ["default", "seq2seq-lm"]: SCREAMING_SNAKE_CASE : Optional[int] = super()._flatten_past_key_values_(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Any = super(UpperCAmelCase_ , self )._flatten_past_key_values_( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) @property def _A ( self : List[Any] ): return 1E-4
62
import json import os import shutil import tempfile import unittest import numpy as np from transformers import BertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer from transformers.testing_utils import require_tokenizers, require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor @require_tokenizers @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = tempfile.mkdtemp() # fmt: off SCREAMING_SNAKE_CASE : Any = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest"] # fmt: on SCREAMING_SNAKE_CASE : Dict = 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] ) ) SCREAMING_SNAKE_CASE : Optional[int] = { "do_resize": True, "size": {"height": 18, "width": 18}, "do_normalize": True, "image_mean": [0.5, 0.5, 0.5], "image_std": [0.5, 0.5, 0.5], } SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(self.tmpdirname , UpperCAmelCase_ ) with open(self.image_processor_file , "w" , encoding="utf-8" ) as fp: json.dump(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] , **UpperCAmelCase_ : List[str] ): return BertTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , **UpperCAmelCase_ : Any ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] ): shutil.rmtree(self.tmpdirname ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] SCREAMING_SNAKE_CASE : Union[str, Any] = [Image.fromarray(np.moveaxis(UpperCAmelCase_ , 0 , -1 ) ) for x in image_inputs] return image_inputs def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : int = VisionTextDualEncoderProcessor( tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer(bos_token="(BOS)" , eos_token="(EOS)" ) SCREAMING_SNAKE_CASE : Any = self.get_image_processor(do_normalize=UpperCAmelCase_ , padding_value=1.0 ) SCREAMING_SNAKE_CASE : str = VisionTextDualEncoderProcessor.from_pretrained( self.tmpdirname , bos_token="(BOS)" , eos_token="(EOS)" , do_normalize=UpperCAmelCase_ , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : int = image_processor(UpperCAmelCase_ , return_tensors="np" ) SCREAMING_SNAKE_CASE : Optional[int] = processor(images=UpperCAmelCase_ , 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 _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = "lower newer" SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = tokenizer(UpperCAmelCase_ ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : int = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer() SCREAMING_SNAKE_CASE : List[str] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = "lower newer" SCREAMING_SNAKE_CASE : Union[str, Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Optional[Any] = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , ["input_ids", "token_type_ids", "attention_mask", "pixel_values"] ) # test if it raises when no input is passed with self.assertRaises(UpperCAmelCase_ ): processor() def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] SCREAMING_SNAKE_CASE : Optional[Any] = processor.batch_decode(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer.batch_decode(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = "lower newer" SCREAMING_SNAKE_CASE : Any = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
62
1
from __future__ import annotations snake_case = [ [-1, 0], # left [0, -1], # down [1, 0], # right [0, 1], # up ] def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase , ): """simple docstring""" SCREAMING_SNAKE_CASE : int = [ [0 for col in range(len(grid[0] ) )] for row in range(len(lowercase ) ) ] # the reference grid SCREAMING_SNAKE_CASE : Any = 1 SCREAMING_SNAKE_CASE : int = [ [0 for col in range(len(grid[0] ) )] for row in range(len(lowercase ) ) ] # the action grid SCREAMING_SNAKE_CASE : Tuple = init[0] SCREAMING_SNAKE_CASE : str = init[1] SCREAMING_SNAKE_CASE : Union[str, Any] = 0 SCREAMING_SNAKE_CASE : Tuple = g + heuristic[x][y] # cost from starting cell to destination cell SCREAMING_SNAKE_CASE : str = [[f, g, x, y]] SCREAMING_SNAKE_CASE : Any = False # flag that is set when search is complete SCREAMING_SNAKE_CASE : List[Any] = False # flag set if we can't find expand while not found and not resign: if len(lowercase ) == 0: raise ValueError("Algorithm is unable to find solution" ) else: # to choose the least costliest action so as to move closer to the goal cell.sort() cell.reverse() SCREAMING_SNAKE_CASE : Dict = cell.pop() SCREAMING_SNAKE_CASE : str = next_cell[2] SCREAMING_SNAKE_CASE : Dict = next_cell[3] SCREAMING_SNAKE_CASE : Tuple = next_cell[1] if x == goal[0] and y == goal[1]: SCREAMING_SNAKE_CASE : Optional[Any] = True else: for i in range(len(lowercase ) ): # to try out different valid actions SCREAMING_SNAKE_CASE : Optional[Any] = x + DIRECTIONS[i][0] SCREAMING_SNAKE_CASE : str = y + DIRECTIONS[i][1] if xa >= 0 and xa < len(lowercase ) and ya >= 0 and ya < len(grid[0] ): if closed[xa][ya] == 0 and grid[xa][ya] == 0: SCREAMING_SNAKE_CASE : List[Any] = g + cost SCREAMING_SNAKE_CASE : Dict = ga + heuristic[xa][ya] cell.append([fa, ga, xa, ya] ) SCREAMING_SNAKE_CASE : Optional[int] = 1 SCREAMING_SNAKE_CASE : Tuple = i SCREAMING_SNAKE_CASE : Any = [] SCREAMING_SNAKE_CASE : Optional[Any] = goal[0] SCREAMING_SNAKE_CASE : Any = goal[1] invpath.append([x, y] ) # we get the reverse path from here while x != init[0] or y != init[1]: SCREAMING_SNAKE_CASE : List[str] = x - DIRECTIONS[action[x][y]][0] SCREAMING_SNAKE_CASE : Optional[int] = y - DIRECTIONS[action[x][y]][1] SCREAMING_SNAKE_CASE : int = xa SCREAMING_SNAKE_CASE : Optional[Any] = ya invpath.append([x, y] ) SCREAMING_SNAKE_CASE : Any = [] for i in range(len(lowercase ) ): path.append(invpath[len(lowercase ) - 1 - i] ) return path, action if __name__ == "__main__": snake_case = [ [0, 1, 0, 0, 0, 0], [0, 1, 0, 0, 0, 0], # 0 are free path whereas 1's are obstacles [0, 1, 0, 0, 0, 0], [0, 1, 0, 0, 1, 0], [0, 0, 0, 0, 1, 0], ] snake_case = [0, 0] # all coordinates are given in format [y,x] snake_case = [len(grid) - 1, len(grid[0]) - 1] snake_case = 1 # the cost map which pushes the path closer to the goal snake_case = [[0 for row in range(len(grid[0]))] for col in range(len(grid))] for i in range(len(grid)): for j in range(len(grid[0])): snake_case = abs(i - goal[0]) + abs(j - goal[1]) if grid[i][j] == 1: # added extra penalty in the heuristic map snake_case = 99 snake_case , snake_case = search(grid, init, goal, cost, heuristic) print("""ACTION MAP""") for i in range(len(action)): print(action[i]) for i in range(len(path)): print(path[i])
62
import inspect import unittest from transformers import MobileNetVaConfig from transformers.testing_utils import require_torch, require_vision, slow, torch_device from transformers.utils import cached_property, is_torch_available, is_vision_available from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor from ...test_pipeline_mixin import PipelineTesterMixin if is_torch_available(): import torch from transformers import MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation, MobileNetVaModel from transformers.models.mobilenet_va.modeling_mobilenet_va import MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST if is_vision_available(): from PIL import Image from transformers import MobileNetVaImageProcessor class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.config_class(**self.inputs_dict ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "tf_padding" ) ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "depth_multiplier" ) ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : List[str]=13 , UpperCAmelCase_ : Any=3 , UpperCAmelCase_ : Union[str, Any]=32 , UpperCAmelCase_ : int=0.25 , UpperCAmelCase_ : Union[str, Any]=8 , UpperCAmelCase_ : Dict=8 , UpperCAmelCase_ : Optional[int]=6 , UpperCAmelCase_ : Optional[Any]=32 , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : str=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : str="relu6" , UpperCAmelCase_ : List[str]=1280 , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : Any=0.02 , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=10 , UpperCAmelCase_ : Optional[Any]=None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = parent SCREAMING_SNAKE_CASE : int = batch_size SCREAMING_SNAKE_CASE : Dict = num_channels SCREAMING_SNAKE_CASE : Dict = image_size SCREAMING_SNAKE_CASE : int = depth_multiplier SCREAMING_SNAKE_CASE : str = depth_divisible_by SCREAMING_SNAKE_CASE : Union[str, Any] = min_depth SCREAMING_SNAKE_CASE : int = expand_ratio SCREAMING_SNAKE_CASE : Tuple = tf_padding SCREAMING_SNAKE_CASE : List[str] = output_stride SCREAMING_SNAKE_CASE : Optional[int] = first_layer_is_expansion SCREAMING_SNAKE_CASE : Any = finegrained_output SCREAMING_SNAKE_CASE : Optional[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[int] = last_hidden_size if finegrained_output else int(last_hidden_size * depth_multiplier ) SCREAMING_SNAKE_CASE : Any = classifier_dropout_prob SCREAMING_SNAKE_CASE : Dict = use_labels SCREAMING_SNAKE_CASE : int = is_training SCREAMING_SNAKE_CASE : Dict = num_labels SCREAMING_SNAKE_CASE : Optional[Any] = initializer_range SCREAMING_SNAKE_CASE : Union[str, Any] = scope def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] ) SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : List[Any] = None if self.use_labels: SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size] , self.num_labels ) SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size, self.image_size, self.image_size] , self.num_labels ) SCREAMING_SNAKE_CASE : Tuple = self.get_config() return config, pixel_values, labels, pixel_labels def _A ( self : Optional[int] ): return MobileNetVaConfig( num_channels=self.num_channels , image_size=self.image_size , depth_multiplier=self.depth_multiplier , depth_divisible_by=self.depth_divisible_by , min_depth=self.min_depth , expand_ratio=self.expand_ratio , output_stride=self.output_stride , first_layer_is_expansion=self.first_layer_is_expansion , finegrained_output=self.finegrained_output , hidden_act=self.hidden_act , tf_padding=self.tf_padding , classifier_dropout_prob=self.classifier_dropout_prob , initializer_range=self.initializer_range , ) def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaModel(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ ) self.parent.assertEqual( result.last_hidden_state.shape , ( self.batch_size, self.last_hidden_size, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) self.parent.assertEqual( result.pooler_output.shape , (self.batch_size, self.last_hidden_size) , ) def _A ( self : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : str = self.num_labels SCREAMING_SNAKE_CASE : List[Any] = MobileNetVaForImageClassification(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Dict = MobileNetVaForSemanticSegmentation(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[int] = model(UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : List[Any] = self.prepare_config_and_inputs() SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = config_and_inputs SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": pixel_values} return config, inputs_dict @require_torch class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Any = ( (MobileNetVaModel, MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation) if is_torch_available() else () ) UpperCamelCase_ : List[Any] = ( { '''feature-extraction''': MobileNetVaModel, '''image-classification''': MobileNetVaForImageClassification, '''image-segmentation''': MobileNetVaForSemanticSegmentation, } if is_torch_available() else {} ) UpperCamelCase_ : Any = False UpperCamelCase_ : List[str] = False UpperCamelCase_ : int = False UpperCamelCase_ : str = False def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = MobileNetVaModelTester(self ) SCREAMING_SNAKE_CASE : Optional[Any] = MobileNetVaConfigTester(self , config_class=UpperCAmelCase_ , has_text_modality=UpperCAmelCase_ ) def _A ( self : Optional[Any] ): self.config_tester.run_common_tests() @unittest.skip(reason="MobileNetV2 does not use inputs_embeds" ) def _A ( self : List[Any] ): pass @unittest.skip(reason="MobileNetV2 does not support input and output embeddings" ) def _A ( self : Dict ): pass @unittest.skip(reason="MobileNetV2 does not output attentions" ) def _A ( self : Union[str, Any] ): pass def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : Tuple = model_class(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic SCREAMING_SNAKE_CASE : Dict = [*signature.parameters.keys()] SCREAMING_SNAKE_CASE : Union[str, Any] = ["pixel_values"] self.assertListEqual(arg_names[:1] , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*UpperCAmelCase_ ) def _A ( self : List[Any] ): def check_hidden_states_output(UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = model_class(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**self._prepare_for_class(UpperCAmelCase_ , UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : List[Any] = outputs.hidden_states SCREAMING_SNAKE_CASE : Any = 16 self.assertEqual(len(UpperCAmelCase_ ) , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : str = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) # check that output_hidden_states also work using config del inputs_dict["output_hidden_states"] SCREAMING_SNAKE_CASE : List[Any] = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Tuple = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_image_classification(*UpperCAmelCase_ ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_semantic_segmentation(*UpperCAmelCase_ ) @slow def _A ( self : Optional[Any] ): for model_name in MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: SCREAMING_SNAKE_CASE : int = MobileNetVaModel.from_pretrained(UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : int = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" ) return image @require_torch @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @cached_property def _A ( self : Optional[int] ): return ( MobileNetVaImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224" ) if is_vision_available() else None ) @slow def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaForImageClassification.from_pretrained("google/mobilenet_v2_1.0_224" ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.default_image_processor SCREAMING_SNAKE_CASE : Optional[int] = prepare_img() SCREAMING_SNAKE_CASE : Optional[int] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**UpperCAmelCase_ ) # verify the logits SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Size((1, 1001) ) self.assertEqual(outputs.logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = torch.tensor([0.2_445, -1.1_993, 0.1_905] ).to(UpperCAmelCase_ ) self.assertTrue(torch.allclose(outputs.logits[0, :3] , UpperCAmelCase_ , atol=1E-4 ) ) @slow def _A ( self : str ): SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaForSemanticSegmentation.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : int = model.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaImageProcessor.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : Optional[Any] = prepare_img() SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = outputs.logits # verify the logits SCREAMING_SNAKE_CASE : Dict = torch.Size((1, 21, 65, 65) ) self.assertEqual(logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.tensor( [ [[17.5_790, 17.7_581, 18.3_355], [18.3_257, 18.4_230, 18.8_973], [18.6_169, 18.8_650, 19.2_187]], [[-2.1_595, -2.0_977, -2.3_741], [-2.4_226, -2.3_028, -2.6_835], [-2.7_819, -2.5_991, -2.7_706]], [[4.2_058, 4.8_317, 4.7_638], [4.4_136, 5.0_361, 4.9_383], [4.5_028, 4.9_644, 4.8_734]], ] , device=UpperCAmelCase_ , ) self.assertTrue(torch.allclose(logits[0, :3, :3, :3] , UpperCAmelCase_ , atol=1E-4 ) )
62
1
from dataclasses import dataclass from typing import Optional, Tuple import torch from torch import nn from transformers import RobertaPreTrainedModel, XLMRobertaConfig, XLMRobertaModel from transformers.utils import ModelOutput @dataclass class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[torch.FloatTensor] = None UpperCamelCase_ : torch.FloatTensor = None UpperCamelCase_ : Optional[Tuple[torch.FloatTensor]] = None UpperCamelCase_ : Optional[Tuple[torch.FloatTensor]] = None class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : Optional[Any] , UpperCAmelCase_ : Optional[Any]=1 , UpperCAmelCase_ : str=0 , UpperCAmelCase_ : str=2 , UpperCAmelCase_ : Optional[int]=512 , UpperCAmelCase_ : str="cls" , UpperCAmelCase_ : int=False , UpperCAmelCase_ : Union[str, Any]=True , **UpperCAmelCase_ : List[Any] , ): super().__init__(pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = project_dim SCREAMING_SNAKE_CASE : Optional[Any] = pooler_fn SCREAMING_SNAKE_CASE : Optional[Any] = learn_encoder SCREAMING_SNAKE_CASE : Optional[int] = use_attention_mask class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Dict = [r'''pooler''', r'''logit_scale'''] UpperCamelCase_ : Tuple = [r'''position_ids''', r'''predictions.decoder.bias'''] UpperCamelCase_ : Tuple = '''roberta''' UpperCamelCase_ : Optional[Any] = RobertaSeriesConfig def __init__( self : Union[str, Any] , UpperCAmelCase_ : Optional[Any] ): super().__init__(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = XLMRobertaModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.Linear(config.hidden_size , config.project_dim ) SCREAMING_SNAKE_CASE : List[Any] = getattr(UpperCAmelCase_ , "has_pre_transformation" , UpperCAmelCase_ ) if self.has_pre_transformation: SCREAMING_SNAKE_CASE : Union[str, Any] = nn.Linear(config.hidden_size , config.project_dim ) SCREAMING_SNAKE_CASE : List[str] = nn.LayerNorm(config.hidden_size , eps=config.layer_norm_eps ) self.post_init() def _A ( self : Optional[Any] , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[torch.Tensor] = None , UpperCAmelCase_ : Optional[bool] = None , UpperCAmelCase_ : Optional[bool] = None , UpperCAmelCase_ : Optional[bool] = None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = return_dict if return_dict is not None else self.config.use_return_dict SCREAMING_SNAKE_CASE : Optional[Any] = self.base_model( input_ids=UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , token_type_ids=UpperCAmelCase_ , position_ids=UpperCAmelCase_ , head_mask=UpperCAmelCase_ , inputs_embeds=UpperCAmelCase_ , encoder_hidden_states=UpperCAmelCase_ , encoder_attention_mask=UpperCAmelCase_ , output_attentions=UpperCAmelCase_ , output_hidden_states=True if self.has_pre_transformation else output_hidden_states , return_dict=UpperCAmelCase_ , ) if self.has_pre_transformation: SCREAMING_SNAKE_CASE : str = outputs["hidden_states"][-2] SCREAMING_SNAKE_CASE : str = self.pre_LN(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = self.transformation_pre(UpperCAmelCase_ ) return TransformationModelOutput( projection_state=UpperCAmelCase_ , last_hidden_state=outputs.last_hidden_state , hidden_states=outputs.hidden_states , attentions=outputs.attentions , ) else: SCREAMING_SNAKE_CASE : str = self.transformation(outputs.last_hidden_state ) return TransformationModelOutput( projection_state=UpperCAmelCase_ , last_hidden_state=outputs.last_hidden_state , hidden_states=outputs.hidden_states , attentions=outputs.attentions , )
62
import math from typing import Dict, Iterable, List, Optional, Tuple, Union import numpy as np from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format from ...image_utils import ( IMAGENET_STANDARD_MEAN, IMAGENET_STANDARD_STD, ChannelDimension, ImageInput, PILImageResampling, get_image_size, is_torch_available, is_torch_tensor, make_list_of_images, to_numpy_array, valid_images, ) from ...utils import TensorType, is_vision_available, logging if is_torch_available(): import torch if is_vision_available(): import PIL snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" def constraint_to_multiple_of(lowercase , lowercase , lowercase=0 , lowercase=None ): SCREAMING_SNAKE_CASE : Any = round(val / multiple ) * multiple if max_val is not None and x > max_val: SCREAMING_SNAKE_CASE : Optional[int] = math.floor(val / multiple ) * multiple if x < min_val: SCREAMING_SNAKE_CASE : int = math.ceil(val / multiple ) * multiple return x SCREAMING_SNAKE_CASE : Tuple = (output_size, output_size) if isinstance(lowercase , lowercase ) else output_size SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = get_image_size(lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = output_size # determine new height and width SCREAMING_SNAKE_CASE : Tuple = output_height / input_height SCREAMING_SNAKE_CASE : List[Any] = output_width / input_width if keep_aspect_ratio: # scale as little as possible if abs(1 - scale_width ) < abs(1 - scale_height ): # fit width SCREAMING_SNAKE_CASE : List[str] = scale_width else: # fit height SCREAMING_SNAKE_CASE : Optional[Any] = scale_height SCREAMING_SNAKE_CASE : int = constraint_to_multiple_of(scale_height * input_height , multiple=lowercase ) SCREAMING_SNAKE_CASE : Dict = constraint_to_multiple_of(scale_width * input_width , multiple=lowercase ) return (new_height, new_width) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Any = ['''pixel_values'''] def __init__( self : Any , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Dict[str, int] = None , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BILINEAR , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Union[int, float] = 1 / 255 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , **UpperCAmelCase_ : str , ): super().__init__(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = size if size is not None else {"height": 384, "width": 384} SCREAMING_SNAKE_CASE : List[Any] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = do_resize SCREAMING_SNAKE_CASE : Optional[Any] = size SCREAMING_SNAKE_CASE : str = keep_aspect_ratio SCREAMING_SNAKE_CASE : int = ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample SCREAMING_SNAKE_CASE : List[str] = do_rescale SCREAMING_SNAKE_CASE : Tuple = rescale_factor SCREAMING_SNAKE_CASE : Union[str, Any] = do_normalize SCREAMING_SNAKE_CASE : Union[str, Any] = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN SCREAMING_SNAKE_CASE : int = image_std if image_std is not None else IMAGENET_STANDARD_STD def _A ( self : Tuple , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Dict[str, int] , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : int , ): SCREAMING_SNAKE_CASE : List[str] = get_size_dict(UpperCAmelCase_ ) if "height" not in size or "width" not in size: raise ValueError(f'''The size dictionary must contain the keys \'height\' and \'width\'. Got {size.keys()}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = get_resize_output_image_size( UpperCAmelCase_ , output_size=(size["height"], size["width"]) , keep_aspect_ratio=UpperCAmelCase_ , multiple=UpperCAmelCase_ , ) return resize(UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[int, float] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : Dict , ): return rescale(UpperCAmelCase_ , scale=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Dict , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : List[str] , ): return normalize(UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : ImageInput , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : PILImageResampling = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : float = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[str, TensorType]] = None , UpperCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **UpperCAmelCase_ : Optional[Any] , ): SCREAMING_SNAKE_CASE : Optional[int] = do_resize if do_resize is not None else self.do_resize SCREAMING_SNAKE_CASE : Dict = size if size is not None else self.size SCREAMING_SNAKE_CASE : Optional[int] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = keep_aspect_ratio if keep_aspect_ratio is not None else self.keep_aspect_ratio SCREAMING_SNAKE_CASE : Optional[Any] = ensure_multiple_of if ensure_multiple_of is not None else self.ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample if resample is not None else self.resample SCREAMING_SNAKE_CASE : Dict = do_rescale if do_rescale is not None else self.do_rescale SCREAMING_SNAKE_CASE : Optional[Any] = rescale_factor if rescale_factor is not None else self.rescale_factor SCREAMING_SNAKE_CASE : Optional[int] = do_normalize if do_normalize is not None else self.do_normalize SCREAMING_SNAKE_CASE : int = image_mean if image_mean is not None else self.image_mean SCREAMING_SNAKE_CASE : Tuple = image_std if image_std is not None else self.image_std SCREAMING_SNAKE_CASE : List[str] = make_list_of_images(UpperCAmelCase_ ) if not valid_images(UpperCAmelCase_ ): raise ValueError( "Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, " "torch.Tensor, tf.Tensor or jax.ndarray." ) if do_resize and size is None or resample is None: raise ValueError("Size and resample must be specified if do_resize is True." ) if do_rescale and rescale_factor is None: raise ValueError("Rescale factor must be specified if do_rescale is True." ) if do_normalize and (image_mean is None or image_std is None): raise ValueError("Image mean and std must be specified if do_normalize is True." ) # All transformations expect numpy arrays. SCREAMING_SNAKE_CASE : Any = [to_numpy_array(UpperCAmelCase_ ) for image in images] if do_resize: SCREAMING_SNAKE_CASE : Tuple = [self.resize(image=UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ ) for image in images] if do_rescale: SCREAMING_SNAKE_CASE : Tuple = [self.rescale(image=UpperCAmelCase_ , scale=UpperCAmelCase_ ) for image in images] if do_normalize: SCREAMING_SNAKE_CASE : Optional[Any] = [self.normalize(image=UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = [to_channel_dimension_format(UpperCAmelCase_ , UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": images} return BatchFeature(data=UpperCAmelCase_ , tensor_type=UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Tuple] = None ): SCREAMING_SNAKE_CASE : int = outputs.logits # Resize logits and compute semantic segmentation maps if target_sizes is not None: if len(UpperCAmelCase_ ) != len(UpperCAmelCase_ ): raise ValueError( "Make sure that you pass in as many target sizes as the batch dimension of the logits" ) if is_torch_tensor(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = target_sizes.numpy() SCREAMING_SNAKE_CASE : str = [] for idx in range(len(UpperCAmelCase_ ) ): SCREAMING_SNAKE_CASE : Dict = torch.nn.functional.interpolate( logits[idx].unsqueeze(dim=0 ) , size=target_sizes[idx] , mode="bilinear" , align_corners=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = resized_logits[0].argmax(dim=0 ) semantic_segmentation.append(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = logits.argmax(dim=1 ) SCREAMING_SNAKE_CASE : str = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )] return semantic_segmentation
62
1
def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = [0] * len(lowercase ) SCREAMING_SNAKE_CASE : Any = [] SCREAMING_SNAKE_CASE : Union[str, Any] = [] SCREAMING_SNAKE_CASE : Dict = 0 for values in graph.values(): for i in values: indegree[i] += 1 for i in range(len(lowercase ) ): if indegree[i] == 0: queue.append(lowercase ) while queue: SCREAMING_SNAKE_CASE : Dict = queue.pop(0 ) cnt += 1 topo.append(lowercase ) for x in graph[vertex]: indegree[x] -= 1 if indegree[x] == 0: queue.append(lowercase ) if cnt != len(lowercase ): print("Cycle exists" ) else: print(lowercase ) # Adjacency List of Graph snake_case = {0: [1, 2], 1: [3], 2: [3], 3: [4, 5], 4: [], 5: []} topological_sort(graph)
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : List[str] , UpperCAmelCase_ : int = 6 ): SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None self.create_linked_list(UpperCAmelCase_ ) def _A ( self : List[Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[int] = Node() SCREAMING_SNAKE_CASE : str = current_node SCREAMING_SNAKE_CASE : Optional[int] = current_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node for _ in range(1 , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Tuple = Node() SCREAMING_SNAKE_CASE : Dict = current_node SCREAMING_SNAKE_CASE : Optional[Any] = previous_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node SCREAMING_SNAKE_CASE : Union[str, Any] = self.front SCREAMING_SNAKE_CASE : List[str] = previous_node def _A ( self : Union[str, Any] ): return ( self.front == self.rear and self.front is not None and self.front.data is None ) def _A ( self : Optional[int] ): self.check_can_perform_operation() return self.front.data if self.front else None def _A ( self : Optional[int] , UpperCAmelCase_ : Any ): if self.rear is None: return self.check_is_full() if not self.is_empty(): SCREAMING_SNAKE_CASE : List[str] = self.rear.next if self.rear: SCREAMING_SNAKE_CASE : Dict = data def _A ( self : List[str] ): self.check_can_perform_operation() if self.rear is None or self.front is None: return None if self.front == self.rear: SCREAMING_SNAKE_CASE : List[str] = self.front.data SCREAMING_SNAKE_CASE : Optional[int] = None return data SCREAMING_SNAKE_CASE : List[str] = self.front SCREAMING_SNAKE_CASE : List[str] = old_front.next SCREAMING_SNAKE_CASE : Optional[int] = old_front.data SCREAMING_SNAKE_CASE : List[str] = None return data def _A ( self : Any ): if self.is_empty(): raise Exception("Empty Queue" ) def _A ( self : Optional[Any] ): if self.rear and self.rear.next == self.front: raise Exception("Full Queue" ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any | None = None SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None if __name__ == "__main__": import doctest doctest.testmod()
62
1
def lowerCamelCase__ ( lowercase ): """simple docstring""" if any(not isinstance(lowercase , lowercase ) or x < 0 for x in sequence ): raise TypeError("Sequence must be list of non-negative integers" ) for _ in range(len(lowercase ) ): for i, (rod_upper, rod_lower) in enumerate(zip(lowercase , sequence[1:] ) ): if rod_upper > rod_lower: sequence[i] -= rod_upper - rod_lower sequence[i + 1] += rod_upper - rod_lower return sequence if __name__ == "__main__": assert bead_sort([5, 4, 3, 2, 1]) == [1, 2, 3, 4, 5] assert bead_sort([7, 9, 4, 3, 5]) == [3, 4, 5, 7, 9]
62
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return 1 if input_a == input_a else 0 def lowerCamelCase__ ( ): """simple docstring""" assert xnor_gate(0 , 0 ) == 1 assert xnor_gate(0 , 1 ) == 0 assert xnor_gate(1 , 0 ) == 0 assert xnor_gate(1 , 1 ) == 1 if __name__ == "__main__": print(xnor_gate(0, 0)) print(xnor_gate(0, 1)) print(xnor_gate(1, 0)) print(xnor_gate(1, 1))
62
1
import argparse import logging import os import datasets import tensorflow as tf from transformers import AutoTokenizer snake_case = logging.getLogger(__name__) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = argparse.ArgumentParser( description="Prepare TFRecord shards from pre-tokenized samples of the wikitext dataset." ) parser.add_argument( "--dataset_name" , type=lowercase , default="wikitext" , help="Name of the training. Explore datasets at: hf.co/datasets." , ) parser.add_argument( "--dataset_config" , type=lowercase , default="wikitext-103-raw-v1" , help="Configuration name of the dataset." ) parser.add_argument( "--tokenizer_name_or_path" , type=lowercase , default="sayakpaul/unigram-tokenizer-wikitext" , help="Tokenizer identifier. Can be a local filepath or a Hub identifier." , ) parser.add_argument( "--shard_size" , type=lowercase , default=1000 , help="Number of entries to go in a single shard." , ) parser.add_argument("--split" , type=lowercase , default="train" , choices=["train", "test", "validation"] ) parser.add_argument( "--limit" , default=lowercase , type=lowercase , help="Limit the number of shards (used for debugging)." , ) parser.add_argument( "--max_length" , type=lowercase , default=512 , help="Maximum sequence length. For training on TPUs, it helps to have a maximum" " sequence length that is a multiple of 8." , ) parser.add_argument( "--output_dir" , default="tf-tpu" , type=lowercase , help="Output directory where the TFRecord shards will be saved. If the" " path is appended with `gs://` ('gs://tf-tpu', for example) then the TFRecord" " shards will be directly saved to a Google Cloud Storage bucket." , ) SCREAMING_SNAKE_CASE : Optional[int] = parser.parse_args() return args def lowerCamelCase__ ( lowercase ): """simple docstring""" def fn(lowercase ): return tokenizer(examples["text"] ) return fn def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = [] for i in range(len(tokenized_data["input_ids"] ) ): SCREAMING_SNAKE_CASE : Any = { "input_ids": tf.train.Feature(intaa_list=tf.train.IntaaList(value=tokenized_data["input_ids"][i] ) ), "attention_mask": tf.train.Feature( intaa_list=tf.train.IntaaList(value=tokenized_data["attention_mask"][i] ) ), } SCREAMING_SNAKE_CASE : str = tf.train.Features(feature=lowercase ) SCREAMING_SNAKE_CASE : List[Any] = tf.train.Example(features=lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = example.SerializeToString() records.append(lowercase ) return records def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = datasets.load_dataset(args.dataset_name , args.dataset_config , split=args.split ) if args.limit is not None: SCREAMING_SNAKE_CASE : Union[str, Any] = min(len(lowercase ) , args.limit ) SCREAMING_SNAKE_CASE : List[str] = dataset.select(range(lowercase ) ) print(F'''Limiting the dataset to {args.limit} entries.''' ) SCREAMING_SNAKE_CASE : Any = AutoTokenizer.from_pretrained(args.tokenizer_name_or_path ) # Handle output directory creation. # For serializing into a Google Cloud Storage Bucket, one needs to first # create a bucket. if "gs" not in args.output_dir: if not os.path.exists(args.output_dir ): os.makedirs(args.output_dir ) SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(args.output_dir , args.split ) if not os.path.exists(lowercase ): os.makedirs(lowercase ) else: SCREAMING_SNAKE_CASE : List[str] = os.path.join(args.output_dir , args.split ) # Tokenize the whole dataset at once. SCREAMING_SNAKE_CASE : Union[str, Any] = tokenize_function(lowercase ) SCREAMING_SNAKE_CASE : List[Any] = dataset.map(lowercase , batched=lowercase , num_proc=4 , remove_columns=["text"] ) # We need to concatenate all our texts together, and then split the result # into chunks of a fixed size, which we will call block_size. To do this, we # will use the map method again, with the option batched=True. When we use batched=True, # the function we pass to map() will be passed multiple inputs at once, allowing us # to group them into more or fewer examples than we had in the input. # This allows us to create our new fixed-length samples. The advantage of this # method is that we don't lose a whole lot of content from the dataset compared to the # case where we simply tokenize with a pre-defined max_length. def group_texts(lowercase ): # Concatenate all texts. SCREAMING_SNAKE_CASE : Union[str, Any] = {k: sum(examples[k] , [] ) for k in examples.keys()} SCREAMING_SNAKE_CASE : List[Any] = len(concatenated_examples[list(examples.keys() )[0]] ) # We drop the small remainder, though you could add padding instead if the model supports it # In this, as in all things, we advise you to follow your heart 🫀 SCREAMING_SNAKE_CASE : List[Any] = (total_length // args.max_length) * args.max_length # Split by chunks of max_len. SCREAMING_SNAKE_CASE : List[Any] = { k: [t[i : i + args.max_length] for i in range(0 , lowercase , args.max_length )] for k, t in concatenated_examples.items() } return result SCREAMING_SNAKE_CASE : Tuple = dataset_tokenized.map(lowercase , batched=lowercase , batch_size=1000 , num_proc=4 ) SCREAMING_SNAKE_CASE : str = 0 SCREAMING_SNAKE_CASE : Optional[int] = 0 for shard in range(0 , len(lowercase ) , args.shard_size ): SCREAMING_SNAKE_CASE : int = grouped_dataset[shard : shard + args.shard_size] SCREAMING_SNAKE_CASE : Dict = len(dataset_snapshot["input_ids"] ) SCREAMING_SNAKE_CASE : Any = os.path.join(lowercase , F'''dataset-{shard_count}-{records_containing}.tfrecord''' ) SCREAMING_SNAKE_CASE : Any = get_serialized_examples(lowercase ) with tf.io.TFRecordWriter(lowercase ) as out_file: for i in range(len(lowercase ) ): SCREAMING_SNAKE_CASE : List[Any] = serialized_examples[i] out_file.write(lowercase ) print("Wrote file {} containing {} records".format(lowercase , lowercase ) ) shard_count += 1 total_records += records_containing with open(F'''split-{args.split}-records-count.txt''' , "w" ) as f: print(F'''Total {args.split} records: {total_records}''' , file=lowercase ) if __name__ == "__main__": snake_case = parse_args() main(args)
62
import math import flax.linen as nn import jax.numpy as jnp def lowerCamelCase__ ( lowercase , lowercase , lowercase = 1 , lowercase = 1 , lowercase = 1.0E4 , lowercase = False , lowercase = 1.0 , ): """simple docstring""" assert timesteps.ndim == 1, "Timesteps should be a 1d-array" assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even''' SCREAMING_SNAKE_CASE : Union[str, Any] = float(embedding_dim // 2 ) SCREAMING_SNAKE_CASE : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift) SCREAMING_SNAKE_CASE : Optional[Any] = min_timescale * jnp.exp(jnp.arange(lowercase , dtype=jnp.floataa ) * -log_timescale_increment ) SCREAMING_SNAKE_CASE : Optional[int] = jnp.expand_dims(lowercase , 1 ) * jnp.expand_dims(lowercase , 0 ) # scale embeddings SCREAMING_SNAKE_CASE : Optional[int] = scale * emb if flip_sin_to_cos: SCREAMING_SNAKE_CASE : List[Any] = jnp.concatenate([jnp.cos(lowercase ), jnp.sin(lowercase )] , axis=1 ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = jnp.concatenate([jnp.sin(lowercase ), jnp.cos(lowercase )] , axis=1 ) SCREAMING_SNAKE_CASE : Tuple = jnp.reshape(lowercase , [jnp.shape(lowercase )[0], embedding_dim] ) return signal class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : jnp.dtype = jnp.floataa @nn.compact def __call__( self : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_1" )(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.silu(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_2" )(UpperCAmelCase_ ) return temb class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : bool = False UpperCamelCase_ : float = 1 @nn.compact def __call__( self : Optional[int] , UpperCAmelCase_ : int ): return get_sinusoidal_embeddings( UpperCAmelCase_ , embedding_dim=self.dim , flip_sin_to_cos=self.flip_sin_to_cos , freq_shift=self.freq_shift )
62
1
import numpy as np from cva import destroyAllWindows, imread, imshow, waitKey class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Dict , UpperCAmelCase_ : str , UpperCAmelCase_ : int , UpperCAmelCase_ : int ): if dst_width < 0 or dst_height < 0: raise ValueError("Destination width/height should be > 0" ) SCREAMING_SNAKE_CASE : List[Any] = img SCREAMING_SNAKE_CASE : Any = img.shape[1] SCREAMING_SNAKE_CASE : Any = img.shape[0] SCREAMING_SNAKE_CASE : Tuple = dst_width SCREAMING_SNAKE_CASE : str = dst_height SCREAMING_SNAKE_CASE : Tuple = self.src_w / self.dst_w SCREAMING_SNAKE_CASE : int = self.src_h / self.dst_h SCREAMING_SNAKE_CASE : str = ( np.ones((self.dst_h, self.dst_w, 3) , np.uinta ) * 255 ) def _A ( self : List[Any] ): for i in range(self.dst_h ): for j in range(self.dst_w ): SCREAMING_SNAKE_CASE : Optional[int] = self.img[self.get_y(UpperCAmelCase_ )][self.get_x(UpperCAmelCase_ )] def _A ( self : int , UpperCAmelCase_ : int ): return int(self.ratio_x * x ) def _A ( self : Optional[int] , UpperCAmelCase_ : int ): return int(self.ratio_y * y ) if __name__ == "__main__": snake_case , snake_case = 800, 600 snake_case = imread("""image_data/lena.jpg""", 1) snake_case = NearestNeighbour(im, dst_w, dst_h) n.process() imshow( F"""Image resized from: {im.shape[1]}x{im.shape[0]} to {dst_w}x{dst_h}""", n.output ) waitKey(0) destroyAllWindows()
62
# Copyright 2023 The HuggingFace Inc. team. All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import numpy as np import torch from ..models.clipseg import CLIPSegForImageSegmentation from ..utils import is_vision_available, requires_backends from .base import PipelineTool if is_vision_available(): from PIL import Image class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = ( '''This is a tool that creates a segmentation mask of an image according to a label. It cannot create an image.''' '''It takes two arguments named `image` which should be the original image, and `label` which should be a text ''' '''describing the elements what should be identified in the segmentation mask. The tool returns the mask.''' ) UpperCamelCase_ : Union[str, Any] = '''CIDAS/clipseg-rd64-refined''' UpperCamelCase_ : Any = '''image_segmenter''' UpperCamelCase_ : int = CLIPSegForImageSegmentation UpperCamelCase_ : Optional[Any] = ['''image''', '''text'''] UpperCamelCase_ : int = ['''image'''] def __init__( self : str , *UpperCAmelCase_ : Union[str, Any] , **UpperCAmelCase_ : Union[str, Any] ): requires_backends(self , ["vision"] ) super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : "Image" , UpperCAmelCase_ : str ): return self.pre_processor(text=[label] , images=[image] , padding=UpperCAmelCase_ , return_tensors="pt" ) def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): with torch.no_grad(): SCREAMING_SNAKE_CASE : str = self.model(**UpperCAmelCase_ ).logits return logits def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = outputs.cpu().detach().numpy() SCREAMING_SNAKE_CASE : Optional[int] = 0 SCREAMING_SNAKE_CASE : str = 1 return Image.fromarray((array * 255).astype(np.uinta ) )
62
1
import os from tempfile import TemporaryDirectory from unittest import TestCase import pytest from absl.testing import parameterized from datasets import config from datasets.arrow_reader import HF_GCP_BASE_URL from datasets.builder import DatasetBuilder from datasets.dataset_dict import IterableDatasetDict from datasets.iterable_dataset import IterableDataset from datasets.load import dataset_module_factory, import_main_class from datasets.utils.file_utils import cached_path snake_case = [ {"""dataset""": """wikipedia""", """config_name""": """20220301.de"""}, {"""dataset""": """wikipedia""", """config_name""": """20220301.en"""}, {"""dataset""": """wikipedia""", """config_name""": """20220301.fr"""}, {"""dataset""": """wikipedia""", """config_name""": """20220301.frr"""}, {"""dataset""": """wikipedia""", """config_name""": """20220301.it"""}, {"""dataset""": """wikipedia""", """config_name""": """20220301.simple"""}, {"""dataset""": """snli""", """config_name""": """plain_text"""}, {"""dataset""": """eli5""", """config_name""": """LFQA_reddit"""}, {"""dataset""": """wiki40b""", """config_name""": """en"""}, {"""dataset""": """wiki_dpr""", """config_name""": """psgs_w100.nq.compressed"""}, {"""dataset""": """wiki_dpr""", """config_name""": """psgs_w100.nq.no_index"""}, {"""dataset""": """wiki_dpr""", """config_name""": """psgs_w100.multiset.no_index"""}, {"""dataset""": """natural_questions""", """config_name""": """default"""}, ] def lowerCamelCase__ ( lowercase=True ): """simple docstring""" if with_config: return [ { "testcase_name": d["dataset"] + "/" + d["config_name"], "dataset": d["dataset"], "config_name": d["config_name"], } for d in DATASETS_ON_HF_GCP ] else: return [ {"testcase_name": dataset, "dataset": dataset} for dataset in {d["dataset"] for d in DATASETS_ON_HF_GCP} ] @parameterized.named_parameters(list_datasets_on_hf_gcp_parameters(with_config=lowerCAmelCase ) ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Union[str, Any] = None UpperCamelCase_ : Union[str, Any] = None def _A ( self : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Union[str, Any] ): with TemporaryDirectory() as tmp_dir: SCREAMING_SNAKE_CASE : List[str] = dataset_module_factory(UpperCAmelCase_ , cache_dir=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = import_main_class(dataset_module.module_path , dataset=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : DatasetBuilder = builder_cls( cache_dir=UpperCAmelCase_ , config_name=UpperCAmelCase_ , hash=dataset_module.hash , ) SCREAMING_SNAKE_CASE : int = "/".join( [ HF_GCP_BASE_URL, builder_instance._relative_data_dir(with_hash=UpperCAmelCase_ ).replace(os.sep , "/" ), config.DATASET_INFO_FILENAME, ] ) SCREAMING_SNAKE_CASE : int = cached_path(UpperCAmelCase_ , cache_dir=UpperCAmelCase_ ) self.assertTrue(os.path.exists(UpperCAmelCase_ ) ) @pytest.mark.integration def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : List[Any] = tmp_path_factory.mktemp("test_hf_gcp" ) / "test_wikipedia_simple" SCREAMING_SNAKE_CASE : str = dataset_module_factory("wikipedia" , cache_dir=lowercase ) SCREAMING_SNAKE_CASE : Tuple = import_main_class(dataset_module.module_path ) SCREAMING_SNAKE_CASE : DatasetBuilder = builder_cls( cache_dir=lowercase , config_name="20220301.frr" , hash=dataset_module.hash , ) # use the HF cloud storage, not the original download_and_prepare that uses apache-beam SCREAMING_SNAKE_CASE : Any = None builder_instance.download_and_prepare() SCREAMING_SNAKE_CASE : Optional[Any] = builder_instance.as_dataset() assert ds @pytest.mark.integration def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = dataset_module_factory("wikipedia" , cache_dir=lowercase ) SCREAMING_SNAKE_CASE : Union[str, Any] = import_main_class(dataset_module.module_path , dataset=lowercase ) SCREAMING_SNAKE_CASE : DatasetBuilder = builder_cls( cache_dir=lowercase , config_name="20220301.frr" , hash=dataset_module.hash , ) SCREAMING_SNAKE_CASE : List[Any] = builder_instance.as_streaming_dataset() assert ds assert isinstance(lowercase , lowercase ) assert "train" in ds assert isinstance(ds["train"] , lowercase ) assert next(iter(ds["train"] ) )
62
import json import os import unittest from transformers.models.blenderbot_small.tokenization_blenderbot_small import ( VOCAB_FILES_NAMES, BlenderbotSmallTokenizer, ) from ...test_tokenization_common import TokenizerTesterMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = BlenderbotSmallTokenizer UpperCamelCase_ : int = False def _A ( self : Union[str, Any] ): super().setUp() SCREAMING_SNAKE_CASE : List[Any] = ["__start__", "adapt", "act", "ap@@", "te", "__end__", "__unk__"] SCREAMING_SNAKE_CASE : Optional[Any] = dict(zip(UpperCAmelCase_ , range(len(UpperCAmelCase_ ) ) ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = ["#version: 0.2", "a p", "t e</w>", "ap t</w>", "a d", "ad apt</w>", "a c", "ac t</w>", ""] SCREAMING_SNAKE_CASE : int = {"unk_token": "__unk__", "bos_token": "__start__", "eos_token": "__end__"} SCREAMING_SNAKE_CASE : Union[str, Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["vocab_file"] ) SCREAMING_SNAKE_CASE : str = 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(UpperCAmelCase_ ) + "\n" ) with open(self.merges_file , "w" , encoding="utf-8" ) as fp: fp.write("\n".join(UpperCAmelCase_ ) ) def _A ( self : List[Any] , **UpperCAmelCase_ : str ): kwargs.update(self.special_tokens_map ) return BlenderbotSmallTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] , UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : int = "adapt act apte" return input_text, output_text def _A ( self : str ): SCREAMING_SNAKE_CASE : int = BlenderbotSmallTokenizer(self.vocab_file , self.merges_file , **self.special_tokens_map ) SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : List[str] = ["adapt", "act", "ap@@", "te"] SCREAMING_SNAKE_CASE : Optional[Any] = tokenizer.tokenize(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = [tokenizer.bos_token] + tokens + [tokenizer.eos_token] SCREAMING_SNAKE_CASE : Tuple = [0, 1, 2, 3, 4, 5] self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , UpperCAmelCase_ ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Union[str, Any] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) assert tok("sam" ).input_ids == [1384] SCREAMING_SNAKE_CASE : str = "I am a small frog." SCREAMING_SNAKE_CASE : List[Any] = tok([src_text] , padding=UpperCAmelCase_ , truncation=UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : int = tok.batch_decode(UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ , clean_up_tokenization_spaces=UpperCAmelCase_ )[0] assert src_text != decoded # I wish it did! assert decoded == "i am a small frog ." def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : List[str] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) SCREAMING_SNAKE_CASE : Tuple = "I am a small frog ." SCREAMING_SNAKE_CASE : Optional[int] = "." SCREAMING_SNAKE_CASE : Dict = tok(UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : Optional[Any] = tok(UpperCAmelCase_ )["input_ids"] assert encoded[-1] == encoded_dot[0]
62
1
import argparse import os import re import packaging.version snake_case = """examples/""" snake_case = { """examples""": (re.compile(r"""^check_min_version\(\"[^\"]+\"\)\s*$""", re.MULTILINE), """check_min_version(\"VERSION\")\n"""), """init""": (re.compile(r"""^__version__\s+=\s+\"([^\"]+)\"\s*$""", re.MULTILINE), """__version__ = \"VERSION\"\n"""), """setup""": (re.compile(r"""^(\s*)version\s*=\s*\"[^\"]+\",""", re.MULTILINE), r"""\1version=\"VERSION\","""), """doc""": (re.compile(r"""^(\s*)release\s*=\s*\"[^\"]+\"$""", re.MULTILINE), """release = \"VERSION\"\n"""), } snake_case = { """init""": """src/transformers/__init__.py""", """setup""": """setup.py""", } snake_case = """README.md""" def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" with open(lowercase , "r" , encoding="utf-8" , newline="\n" ) as f: SCREAMING_SNAKE_CASE : str = f.read() SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = REPLACE_PATTERNS[pattern] SCREAMING_SNAKE_CASE : List[str] = replace.replace("VERSION" , lowercase ) SCREAMING_SNAKE_CASE : Any = re_pattern.sub(lowercase , lowercase ) with open(lowercase , "w" , encoding="utf-8" , newline="\n" ) as f: f.write(lowercase ) def lowerCamelCase__ ( lowercase ): """simple docstring""" for folder, directories, fnames in os.walk(lowercase ): # Removing some of the folders with non-actively maintained examples from the walk if "research_projects" in directories: directories.remove("research_projects" ) if "legacy" in directories: directories.remove("legacy" ) for fname in fnames: if fname.endswith(".py" ): update_version_in_file(os.path.join(lowercase , lowercase ) , lowercase , pattern="examples" ) def lowerCamelCase__ ( lowercase , lowercase=False ): """simple docstring""" for pattern, fname in REPLACE_FILES.items(): update_version_in_file(lowercase , lowercase , lowercase ) if not patch: update_version_in_examples(lowercase ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = "🤗 Transformers currently provides the following architectures" SCREAMING_SNAKE_CASE : Optional[int] = "1. Want to contribute a new model?" with open(lowercase , "r" , encoding="utf-8" , newline="\n" ) as f: SCREAMING_SNAKE_CASE : str = f.readlines() # Find the start of the list. SCREAMING_SNAKE_CASE : List[Any] = 0 while not lines[start_index].startswith(_start_prompt ): start_index += 1 start_index += 1 SCREAMING_SNAKE_CASE : List[Any] = start_index # Update the lines in the model list. while not lines[index].startswith(_end_prompt ): if lines[index].startswith("1." ): SCREAMING_SNAKE_CASE : int = lines[index].replace( "https://huggingface.co/docs/transformers/main/model_doc" , "https://huggingface.co/docs/transformers/model_doc" , ) index += 1 with open(lowercase , "w" , encoding="utf-8" , newline="\n" ) as f: f.writelines(lowercase ) def lowerCamelCase__ ( ): """simple docstring""" with open(REPLACE_FILES["init"] , "r" ) as f: SCREAMING_SNAKE_CASE : int = f.read() SCREAMING_SNAKE_CASE : str = REPLACE_PATTERNS["init"][0].search(lowercase ).groups()[0] return packaging.version.parse(lowercase ) def lowerCamelCase__ ( lowercase=False ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = get_version() if patch and default_version.is_devrelease: raise ValueError("Can't create a patch version from the dev branch, checkout a released version!" ) if default_version.is_devrelease: SCREAMING_SNAKE_CASE : Dict = default_version.base_version elif patch: SCREAMING_SNAKE_CASE : str = F'''{default_version.major}.{default_version.minor}.{default_version.micro + 1}''' else: SCREAMING_SNAKE_CASE : str = F'''{default_version.major}.{default_version.minor + 1}.0''' # Now let's ask nicely if that's the right one. SCREAMING_SNAKE_CASE : str = input(F'''Which version are you releasing? [{default_version}]''' ) if len(lowercase ) == 0: SCREAMING_SNAKE_CASE : List[Any] = default_version print(F'''Updating version to {version}.''' ) global_version_update(lowercase , patch=lowercase ) if not patch: print("Cleaning main README, don't forget to run `make fix-copies`." ) clean_main_ref_in_model_list() def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[Any] = get_version() SCREAMING_SNAKE_CASE : Dict = F'''{current_version.major}.{current_version.minor + 1}.0.dev0''' SCREAMING_SNAKE_CASE : Union[str, Any] = current_version.base_version # Check with the user we got that right. SCREAMING_SNAKE_CASE : List[Any] = input(F'''Which version are we developing now? [{dev_version}]''' ) if len(lowercase ) == 0: SCREAMING_SNAKE_CASE : Optional[int] = dev_version print(F'''Updating version to {version}.''' ) global_version_update(lowercase ) print("Cleaning main README, don't forget to run `make fix-copies`." ) clean_main_ref_in_model_list() if __name__ == "__main__": snake_case = argparse.ArgumentParser() parser.add_argument("""--post_release""", action="""store_true""", help="""Whether this is pre or post release.""") parser.add_argument("""--patch""", action="""store_true""", help="""Whether or not this is a patch release.""") snake_case = parser.parse_args() if not args.post_release: pre_release_work(patch=args.patch) elif args.patch: print("""Nothing to do after a patch :-)""") else: post_release_work()
62
import argparse from collections import OrderedDict from pathlib import Path import torch from transformers import ( VisualBertConfig, VisualBertForMultipleChoice, VisualBertForPreTraining, VisualBertForQuestionAnswering, VisualBertForVisualReasoning, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) snake_case = [ ("""bert.bert""", """visual_bert"""), ("""bert.cls""", """cls"""), ("""bert.classifier""", """cls"""), ("""token_type_embeddings_visual""", """visual_token_type_embeddings"""), ("""position_embeddings_visual""", """visual_position_embeddings"""), ("""projection""", """visual_projection"""), ] snake_case = [ """nlvr2_coco_pre_trained.th""", """nlvr2_fine_tuned.th""", """nlvr2_pre_trained.th""", """vcr_coco_pre_train.th""", """vcr_fine_tune.th""", """vcr_pre_train.th""", """vqa_coco_pre_trained.th""", """vqa_fine_tuned.th""", """vqa_pre_trained.th""", ] def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = torch.load(lowercase , map_location="cpu" ) return sd def lowerCamelCase__ ( lowercase , lowercase , lowercase=rename_keys_prefix ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = OrderedDict() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.arange(config.max_position_embeddings ).expand((1, -1) ) # detector_d = OrderedDict() for key in d: if "detector" in key: # detector_d[key.replace('detector.','')] = d[key] continue SCREAMING_SNAKE_CASE : Union[str, Any] = key for name_pair in rename_keys_prefix: SCREAMING_SNAKE_CASE : Any = new_key.replace(name_pair[0] , name_pair[1] ) SCREAMING_SNAKE_CASE : Dict = d[key] if key == "bert.cls.predictions.decoder.weight": # Old bert code didn't have `decoder.bias`, but was added separately SCREAMING_SNAKE_CASE : Union[str, Any] = new_d["cls.predictions.bias"] return new_d @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( checkpoint_path.split("/" )[-1] in ACCEPTABLE_CHECKPOINTS ), F'''The checkpoint provided must be in {ACCEPTABLE_CHECKPOINTS}.''' # Get Config if "pre" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = "pretraining" if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = {"visual_embedding_dim": 512} elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = {"visual_embedding_dim": 2048} elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Any = {"visual_embedding_dim": 2048} elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 1024} else: raise NotImplementedError(F'''No implementation found for `{checkpoint_path}`.''' ) else: if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 512} SCREAMING_SNAKE_CASE : Tuple = "multichoice" elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[str] = {"visual_embedding_dim": 2048} SCREAMING_SNAKE_CASE : str = "vqa_advanced" elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Tuple = {"visual_embedding_dim": 2048, "num_labels": 3129} SCREAMING_SNAKE_CASE : Optional[Any] = "vqa" elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = { "visual_embedding_dim": 1024, "num_labels": 2, } SCREAMING_SNAKE_CASE : Tuple = "nlvr" SCREAMING_SNAKE_CASE : List[Any] = VisualBertConfig(**lowercase ) # Load State Dict SCREAMING_SNAKE_CASE : List[str] = load_state_dict(lowercase ) SCREAMING_SNAKE_CASE : Any = get_new_dict(lowercase , lowercase ) if model_type == "pretraining": SCREAMING_SNAKE_CASE : int = VisualBertForPreTraining(lowercase ) elif model_type == "vqa": SCREAMING_SNAKE_CASE : Tuple = VisualBertForQuestionAnswering(lowercase ) elif model_type == "nlvr": SCREAMING_SNAKE_CASE : List[Any] = VisualBertForVisualReasoning(lowercase ) elif model_type == "multichoice": SCREAMING_SNAKE_CASE : Optional[Any] = VisualBertForMultipleChoice(lowercase ) model.load_state_dict(lowercase ) # Save Checkpoints Path(lowercase ).mkdir(exist_ok=lowercase ) model.save_pretrained(lowercase ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument("""orig_checkpoint_path""", type=str, help="""A path to .th on local filesystem.""") parser.add_argument("""pytorch_dump_folder_path""", type=str, help="""Path to the output PyTorch model.""") snake_case = parser.parse_args() convert_visual_bert_checkpoint(args.orig_checkpoint_path, args.pytorch_dump_folder_path)
62
1
import argparse import json import logging import os import shutil import sys import tempfile import unittest from unittest import mock import torch from accelerate.utils import write_basic_config from transformers.testing_utils import TestCasePlus, get_gpu_count, run_command, slow, torch_device from transformers.utils import is_apex_available logging.basicConfig(level=logging.DEBUG) snake_case = logging.getLogger() def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = argparse.ArgumentParser() parser.add_argument("-f" ) SCREAMING_SNAKE_CASE : List[str] = parser.parse_args() return args.f def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = {} SCREAMING_SNAKE_CASE : Tuple = os.path.join(lowercase , "all_results.json" ) if os.path.exists(lowercase ): with open(lowercase , "r" ) as f: SCREAMING_SNAKE_CASE : List[Any] = json.load(lowercase ) else: raise ValueError(F'''can\'t find {path}''' ) return results def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = torch.cuda.is_available() and torch_device == "cuda" return is_using_cuda and is_apex_available() snake_case = logging.StreamHandler(sys.stdout) logger.addHandler(stream_handler) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @classmethod def _A ( cls : List[str] ): # Write Accelerate config, will pick up on CPU, GPU, and multi-GPU SCREAMING_SNAKE_CASE : Union[str, Any] = tempfile.mkdtemp() SCREAMING_SNAKE_CASE : str = os.path.join(cls.tmpdir , "default_config.yml" ) write_basic_config(save_location=cls.configPath ) SCREAMING_SNAKE_CASE : Optional[int] = ["accelerate", "launch", "--config_file", cls.configPath] @classmethod def _A ( cls : int ): shutil.rmtree(cls.tmpdir ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : str = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : List[str] = f''' {self.examples_dir}/pytorch/text-classification/run_glue_no_trainer.py --model_name_or_path distilbert-base-uncased --output_dir {tmp_dir} --train_file ./tests/fixtures/tests_samples/MRPC/train.csv --validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv --per_device_train_batch_size=2 --per_device_eval_batch_size=1 --learning_rate=1e-4 --seed=42 --checkpointing_steps epoch --with_tracking '''.split() if is_cuda_and_apex_available(): testargs.append("--fp16" ) run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : str = get_results(UpperCAmelCase_ ) self.assertGreaterEqual(result["eval_accuracy"] , 0.75 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "glue_no_trainer" ) ) ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Tuple = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Optional[Any] = f''' {self.examples_dir}/pytorch/language-modeling/run_clm_no_trainer.py --model_name_or_path distilgpt2 --train_file ./tests/fixtures/sample_text.txt --validation_file ./tests/fixtures/sample_text.txt --block_size 128 --per_device_train_batch_size 5 --per_device_eval_batch_size 5 --num_train_epochs 2 --output_dir {tmp_dir} --checkpointing_steps epoch --with_tracking '''.split() if torch.cuda.device_count() > 1: # Skipping because there are not enough batches to train the model + would need a drop_last to work. return run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : str = get_results(UpperCAmelCase_ ) self.assertLess(result["perplexity"] , 100 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "clm_no_trainer" ) ) ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : int ): SCREAMING_SNAKE_CASE : List[Any] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Tuple = f''' {self.examples_dir}/pytorch/language-modeling/run_mlm_no_trainer.py --model_name_or_path distilroberta-base --train_file ./tests/fixtures/sample_text.txt --validation_file ./tests/fixtures/sample_text.txt --output_dir {tmp_dir} --num_train_epochs=1 --checkpointing_steps epoch --with_tracking '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Union[str, Any] = get_results(UpperCAmelCase_ ) self.assertLess(result["perplexity"] , 42 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "mlm_no_trainer" ) ) ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Any ): # with so little data distributed training needs more epochs to get the score on par with 0/1 gpu SCREAMING_SNAKE_CASE : str = 7 if get_gpu_count() > 1 else 2 SCREAMING_SNAKE_CASE : Optional[int] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Any = f''' {self.examples_dir}/pytorch/token-classification/run_ner_no_trainer.py --model_name_or_path bert-base-uncased --train_file tests/fixtures/tests_samples/conll/sample.json --validation_file tests/fixtures/tests_samples/conll/sample.json --output_dir {tmp_dir} --learning_rate=2e-4 --per_device_train_batch_size=2 --per_device_eval_batch_size=2 --num_train_epochs={epochs} --seed 7 --checkpointing_steps epoch --with_tracking '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Optional[Any] = get_results(UpperCAmelCase_ ) self.assertGreaterEqual(result["eval_accuracy"] , 0.75 ) self.assertLess(result["train_loss"] , 0.5 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "ner_no_trainer" ) ) ) @unittest.skip(reason="Fix me @muellerzr" ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : int = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Optional[Any] = f''' {self.examples_dir}/pytorch/question-answering/run_qa_no_trainer.py --model_name_or_path bert-base-uncased --version_2_with_negative --train_file tests/fixtures/tests_samples/SQUAD/sample.json --validation_file tests/fixtures/tests_samples/SQUAD/sample.json --output_dir {tmp_dir} --seed=42 --max_train_steps=10 --num_warmup_steps=2 --learning_rate=2e-4 --per_device_train_batch_size=2 --per_device_eval_batch_size=1 --checkpointing_steps epoch --with_tracking '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : List[str] = get_results(UpperCAmelCase_ ) # Because we use --version_2_with_negative the testing script uses SQuAD v2 metrics. self.assertGreaterEqual(result["eval_f1"] , 28 ) self.assertGreaterEqual(result["eval_exact"] , 28 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "qa_no_trainer" ) ) ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : List[str] = f''' {self.examples_dir}/pytorch/multiple-choice/run_swag_no_trainer.py --model_name_or_path bert-base-uncased --train_file tests/fixtures/tests_samples/swag/sample.json --validation_file tests/fixtures/tests_samples/swag/sample.json --output_dir {tmp_dir} --max_train_steps=20 --num_warmup_steps=2 --learning_rate=2e-4 --per_device_train_batch_size=2 --per_device_eval_batch_size=1 --with_tracking '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Tuple = get_results(UpperCAmelCase_ ) self.assertGreaterEqual(result["eval_accuracy"] , 0.8 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "swag_no_trainer" ) ) ) @slow @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Dict = f''' {self.examples_dir}/pytorch/summarization/run_summarization_no_trainer.py --model_name_or_path t5-small --train_file tests/fixtures/tests_samples/xsum/sample.json --validation_file tests/fixtures/tests_samples/xsum/sample.json --output_dir {tmp_dir} --max_train_steps=50 --num_warmup_steps=8 --learning_rate=2e-4 --per_device_train_batch_size=2 --per_device_eval_batch_size=1 --checkpointing_steps epoch --with_tracking '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Any = get_results(UpperCAmelCase_ ) self.assertGreaterEqual(result["eval_rouge1"] , 10 ) self.assertGreaterEqual(result["eval_rouge2"] , 2 ) self.assertGreaterEqual(result["eval_rougeL"] , 7 ) self.assertGreaterEqual(result["eval_rougeLsum"] , 7 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "summarization_no_trainer" ) ) ) @slow @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Optional[Any] = f''' {self.examples_dir}/pytorch/translation/run_translation_no_trainer.py --model_name_or_path sshleifer/student_marian_en_ro_6_1 --source_lang en --target_lang ro --train_file tests/fixtures/tests_samples/wmt16/sample.json --validation_file tests/fixtures/tests_samples/wmt16/sample.json --output_dir {tmp_dir} --max_train_steps=50 --num_warmup_steps=8 --num_beams=6 --learning_rate=3e-3 --per_device_train_batch_size=2 --per_device_eval_batch_size=1 --source_lang en_XX --target_lang ro_RO --checkpointing_steps epoch --with_tracking '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Any = get_results(UpperCAmelCase_ ) self.assertGreaterEqual(result["eval_bleu"] , 30 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "epoch_0" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "translation_no_trainer" ) ) ) @slow def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Optional[int] = logging.StreamHandler(sys.stdout ) logger.addHandler(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Dict = f''' {self.examples_dir}/pytorch/semantic-segmentation/run_semantic_segmentation_no_trainer.py --dataset_name huggingface/semantic-segmentation-test-sample --output_dir {tmp_dir} --max_train_steps=10 --num_warmup_steps=2 --learning_rate=2e-4 --per_device_train_batch_size=2 --per_device_eval_batch_size=1 --checkpointing_steps epoch '''.split() run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Dict = get_results(UpperCAmelCase_ ) self.assertGreaterEqual(result["eval_overall_accuracy"] , 0.10 ) @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Dict = f''' {self.examples_dir}/pytorch/image-classification/run_image_classification_no_trainer.py --model_name_or_path google/vit-base-patch16-224-in21k --dataset_name hf-internal-testing/cats_vs_dogs_sample --learning_rate 1e-4 --per_device_train_batch_size 2 --per_device_eval_batch_size 1 --max_train_steps 2 --train_val_split 0.1 --seed 42 --output_dir {tmp_dir} --with_tracking --checkpointing_steps 1 '''.split() if is_cuda_and_apex_available(): testargs.append("--fp16" ) run_command(self._launch_args + testargs ) SCREAMING_SNAKE_CASE : Union[str, Any] = get_results(UpperCAmelCase_ ) # The base model scores a 25% self.assertGreaterEqual(result["eval_accuracy"] , 0.6 ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "step_1" ) ) ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "image_classification_no_trainer" ) ) )
62
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available snake_case = { """configuration_jukebox""": [ """JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP""", """JukeboxConfig""", """JukeboxPriorConfig""", """JukeboxVQVAEConfig""", ], """tokenization_jukebox""": ["""JukeboxTokenizer"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST""", """JukeboxModel""", """JukeboxPreTrainedModel""", """JukeboxVQVAE""", """JukeboxPrior""", ] if TYPE_CHECKING: from .configuration_jukebox import ( JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP, JukeboxConfig, JukeboxPriorConfig, JukeboxVQVAEConfig, ) from .tokenization_jukebox import JukeboxTokenizer try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_jukebox import ( JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST, JukeboxModel, JukeboxPreTrainedModel, JukeboxPrior, JukeboxVQVAE, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
1
from __future__ import annotations def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" if len(lowercase ) <= 1 or n <= 1: return insert_next(lowercase , n - 1 ) rec_insertion_sort(lowercase , n - 1 ) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" if index >= len(lowercase ) or collection[index - 1] <= collection[index]: return # Swaps adjacent elements since they are not in ascending order SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = ( collection[index], collection[index - 1], ) insert_next(lowercase , index + 1 ) if __name__ == "__main__": snake_case = input("""Enter integers separated by spaces: """) snake_case = [int(num) for num in numbers.split()] rec_insertion_sort(number_list, len(number_list)) print(number_list)
62
from typing import Dict, List from nltk.translate import gleu_score import datasets from datasets import MetricInfo snake_case = """\ @misc{wu2016googles, title={Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation}, author={Yonghui Wu and Mike Schuster and Zhifeng Chen and Quoc V. Le and Mohammad Norouzi and Wolfgang Macherey and Maxim Krikun and Yuan Cao and Qin Gao and Klaus Macherey and Jeff Klingner and Apurva Shah and Melvin Johnson and Xiaobing Liu and Łukasz Kaiser and Stephan Gouws and Yoshikiyo Kato and Taku Kudo and Hideto Kazawa and Keith Stevens and George Kurian and Nishant Patil and Wei Wang and Cliff Young and Jason Smith and Jason Riesa and Alex Rudnick and Oriol Vinyals and Greg Corrado and Macduff Hughes and Jeffrey Dean}, year={2016}, eprint={1609.08144}, archivePrefix={arXiv}, primaryClass={cs.CL} } """ snake_case = """\ The BLEU score has some undesirable properties when used for single sentences, as it was designed to be a corpus measure. We therefore use a slightly different score for our RL experiments which we call the 'GLEU score'. For the GLEU score, we record all sub-sequences of 1, 2, 3 or 4 tokens in output and target sequence (n-grams). We then compute a recall, which is the ratio of the number of matching n-grams to the number of total n-grams in the target (ground truth) sequence, and a precision, which is the ratio of the number of matching n-grams to the number of total n-grams in the generated output sequence. Then GLEU score is simply the minimum of recall and precision. This GLEU score's range is always between 0 (no matches) and 1 (all match) and it is symmetrical when switching output and target. According to our experiments, GLEU score correlates quite well with the BLEU metric on a corpus level but does not have its drawbacks for our per sentence reward objective. """ snake_case = """\ Computes corpus-level Google BLEU (GLEU) score of translated segments against one or more references. Instead of averaging the sentence level GLEU scores (i.e. macro-average precision), Wu et al. (2016) sum up the matching tokens and the max of hypothesis and reference tokens for each sentence, then compute using the aggregate values. Args: predictions (list of str): list of translations to score. Each translation should be tokenized into a list of tokens. references (list of list of str): list of lists of references for each translation. Each reference should be tokenized into a list of tokens. min_len (int): The minimum order of n-gram this function should extract. Defaults to 1. max_len (int): The maximum order of n-gram this function should extract. Defaults to 4. Returns: 'google_bleu': google_bleu score Examples: Example 1: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.44 Example 2: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.61 Example 3: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references, min_len=2) >>> print(round(results[\"google_bleu\"], 2)) 0.53 Example 4: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses,references=list_of_references, min_len=2, max_len=6) >>> print(round(results[\"google_bleu\"], 2)) 0.4 """ @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : Optional[Any] ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { "predictions": datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ), "references": datasets.Sequence( datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ) , id="references" ), } ) , ) def _A ( self : List[Any] , UpperCAmelCase_ : List[List[List[str]]] , UpperCAmelCase_ : List[List[str]] , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : int = 4 , ): return { "google_bleu": gleu_score.corpus_gleu( list_of_references=UpperCAmelCase_ , hypotheses=UpperCAmelCase_ , min_len=UpperCAmelCase_ , max_len=UpperCAmelCase_ ) }
62
1
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = len(lowercase ) print("The following activities are selected:" ) # The first activity is always selected SCREAMING_SNAKE_CASE : Optional[int] = 0 print(lowercase , end="," ) # Consider rest of the activities for j in range(lowercase ): # If this activity has start time greater than # or equal to the finish time of previously # selected activity, then select it if start[j] >= finish[i]: print(lowercase , end="," ) SCREAMING_SNAKE_CASE : List[str] = j if __name__ == "__main__": import doctest doctest.testmod() snake_case = [1, 3, 0, 5, 8, 5] snake_case = [2, 4, 6, 7, 9, 9] print_max_activities(start, finish)
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : int , UpperCAmelCase_ : float = 0 ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = row, column SCREAMING_SNAKE_CASE : Optional[Any] = [[default_value for c in range(UpperCAmelCase_ )] for r in range(UpperCAmelCase_ )] def __str__( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = f'''Matrix consist of {self.row} rows and {self.column} columns\n''' # Make string identifier SCREAMING_SNAKE_CASE : Dict = 0 for row_vector in self.array: for obj in row_vector: SCREAMING_SNAKE_CASE : Optional[Any] = max(UpperCAmelCase_ , len(str(UpperCAmelCase_ ) ) ) SCREAMING_SNAKE_CASE : Optional[int] = f'''%{max_element_length}s''' # Make string and return def single_line(UpperCAmelCase_ : list[float] ) -> str: nonlocal string_format_identifier SCREAMING_SNAKE_CASE : Optional[int] = "[" line += ", ".join(string_format_identifier % (obj,) for obj in row_vector ) line += "]" return line s += "\n".join(single_line(UpperCAmelCase_ ) for row_vector in self.array ) return s def __repr__( self : Dict ): return str(self ) def _A ( self : Optional[int] , UpperCAmelCase_ : tuple[int, int] ): if not (isinstance(UpperCAmelCase_ , (list, tuple) ) and len(UpperCAmelCase_ ) == 2): return False elif not (0 <= loc[0] < self.row and 0 <= loc[1] < self.column): return False else: return True def __getitem__( self : Any , UpperCAmelCase_ : tuple[int, int] ): assert self.validate_indicies(UpperCAmelCase_ ) return self.array[loc[0]][loc[1]] def __setitem__( self : int , UpperCAmelCase_ : tuple[int, int] , UpperCAmelCase_ : float ): assert self.validate_indicies(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = value def __add__( self : Optional[int] , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == another.row and self.column == another.column # Add SCREAMING_SNAKE_CASE : str = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : Any = self[r, c] + another[r, c] return result def __neg__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = -self[r, c] return result def __sub__( self : Optional[Any] , UpperCAmelCase_ : Matrix ): return self + (-another) def __mul__( self : Dict , UpperCAmelCase_ : int | float | Matrix ): if isinstance(UpperCAmelCase_ , (int, float) ): # Scalar multiplication SCREAMING_SNAKE_CASE : Any = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = self[r, c] * another return result elif isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): # Matrix multiplication assert self.column == another.row SCREAMING_SNAKE_CASE : Any = Matrix(self.row , another.column ) for r in range(self.row ): for c in range(another.column ): for i in range(self.column ): result[r, c] += self[r, i] * another[i, c] return result else: SCREAMING_SNAKE_CASE : List[str] = f'''Unsupported type given for another ({type(UpperCAmelCase_ )})''' raise TypeError(UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : List[str] = Matrix(self.column , self.row ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : List[str] = self[r, c] return result def _A ( self : Union[str, Any] , UpperCAmelCase_ : Matrix , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) and isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == self.column == u.row == v.row # u, v should be column vector assert u.column == v.column == 1 # u, v should be column vector # Calculate SCREAMING_SNAKE_CASE : Tuple = v.transpose() SCREAMING_SNAKE_CASE : int = (v_t * self * u)[0, 0] + 1 if numerator_factor == 0: return None # It's not invertable return self - ((self * u) * (v_t * self) * (1.0 / numerator_factor)) # Testing if __name__ == "__main__": def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 3 , 0 ) for i in range(3 ): SCREAMING_SNAKE_CASE : str = 1 print(F'''a^(-1) is {ainv}''' ) # u, v SCREAMING_SNAKE_CASE : Optional[int] = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = 1, 2, -3 SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = 4, -2, 5 print(F'''u is {u}''' ) print(F'''v is {v}''' ) print(F'''uv^T is {u * v.transpose()}''' ) # Sherman Morrison print(F'''(a + uv^T)^(-1) is {ainv.sherman_morrison(lowercase , lowercase )}''' ) def lowerCamelCase__ ( ): """simple docstring""" import doctest doctest.testmod() testa()
62
1
import unittest import torch from diffusers import DDIMScheduler, DDPMScheduler, UNetaDModel from diffusers.training_utils import set_seed from diffusers.utils.testing_utils import slow snake_case = False class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Optional[int] , UpperCAmelCase_ : Any=32 ): set_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = UNetaDModel(sample_size=UpperCAmelCase_ , in_channels=3 , out_channels=3 ) SCREAMING_SNAKE_CASE : Tuple = torch.optim.SGD(model.parameters() , lr=0.0_001 ) return model, optimizer @slow def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : str = "cpu" # ensure full determinism without setting the CUBLAS_WORKSPACE_CONFIG env variable SCREAMING_SNAKE_CASE : Union[str, Any] = DDPMScheduler( num_train_timesteps=1000 , beta_start=0.0_001 , beta_end=0.02 , beta_schedule="linear" , clip_sample=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : List[str] = DDIMScheduler( num_train_timesteps=1000 , beta_start=0.0_001 , beta_end=0.02 , beta_schedule="linear" , clip_sample=UpperCAmelCase_ , ) assert ddpm_scheduler.config.num_train_timesteps == ddim_scheduler.config.num_train_timesteps # shared batches for DDPM and DDIM set_seed(0 ) SCREAMING_SNAKE_CASE : List[Any] = [torch.randn((4, 3, 32, 32) ).clip(-1 , 1 ).to(UpperCAmelCase_ ) for _ in range(4 )] SCREAMING_SNAKE_CASE : Union[str, Any] = [torch.randn((4, 3, 32, 32) ).to(UpperCAmelCase_ ) for _ in range(4 )] SCREAMING_SNAKE_CASE : Any = [torch.randint(0 , 1000 , (4,) ).long().to(UpperCAmelCase_ ) for _ in range(4 )] # train with a DDPM scheduler SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = self.get_model_optimizer(resolution=32 ) model.train().to(UpperCAmelCase_ ) for i in range(4 ): optimizer.zero_grad() SCREAMING_SNAKE_CASE : Union[str, Any] = ddpm_scheduler.add_noise(clean_images[i] , noise[i] , timesteps[i] ) SCREAMING_SNAKE_CASE : str = model(UpperCAmelCase_ , timesteps[i] ).sample SCREAMING_SNAKE_CASE : str = torch.nn.functional.mse_loss(UpperCAmelCase_ , noise[i] ) loss.backward() optimizer.step() del model, optimizer # recreate the model and optimizer, and retry with DDIM SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : str = self.get_model_optimizer(resolution=32 ) model.train().to(UpperCAmelCase_ ) for i in range(4 ): optimizer.zero_grad() SCREAMING_SNAKE_CASE : Optional[int] = ddim_scheduler.add_noise(clean_images[i] , noise[i] , timesteps[i] ) SCREAMING_SNAKE_CASE : List[Any] = model(UpperCAmelCase_ , timesteps[i] ).sample SCREAMING_SNAKE_CASE : int = torch.nn.functional.mse_loss(UpperCAmelCase_ , noise[i] ) loss.backward() optimizer.step() del model, optimizer self.assertTrue(torch.allclose(UpperCAmelCase_ , UpperCAmelCase_ , atol=1E-5 ) ) self.assertTrue(torch.allclose(UpperCAmelCase_ , UpperCAmelCase_ , atol=1E-5 ) )
62
import json from typing import List, Optional, Tuple from tokenizers import normalizers from tokenizers.pre_tokenizers import BertPreTokenizer, PreTokenizer from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging from .tokenization_roformer import RoFormerTokenizer from .tokenization_utils import JiebaPreTokenizer snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """vocab.txt""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """junnyu/roformer_chinese_small""": """https://huggingface.co/junnyu/roformer_chinese_small/resolve/main/vocab.txt""", """junnyu/roformer_chinese_base""": """https://huggingface.co/junnyu/roformer_chinese_base/resolve/main/vocab.txt""", """junnyu/roformer_chinese_char_small""": ( """https://huggingface.co/junnyu/roformer_chinese_char_small/resolve/main/vocab.txt""" ), """junnyu/roformer_chinese_char_base""": ( """https://huggingface.co/junnyu/roformer_chinese_char_base/resolve/main/vocab.txt""" ), """junnyu/roformer_small_discriminator""": ( """https://huggingface.co/junnyu/roformer_small_discriminator/resolve/main/vocab.txt""" ), """junnyu/roformer_small_generator""": ( """https://huggingface.co/junnyu/roformer_small_generator/resolve/main/vocab.txt""" ), } } snake_case = { """junnyu/roformer_chinese_small""": 1_536, """junnyu/roformer_chinese_base""": 1_536, """junnyu/roformer_chinese_char_small""": 512, """junnyu/roformer_chinese_char_base""": 512, """junnyu/roformer_small_discriminator""": 128, """junnyu/roformer_small_generator""": 128, } snake_case = { """junnyu/roformer_chinese_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_base""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_base""": {"""do_lower_case""": True}, """junnyu/roformer_small_discriminator""": {"""do_lower_case""": True}, """junnyu/roformer_small_generator""": {"""do_lower_case""": True}, } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = VOCAB_FILES_NAMES UpperCamelCase_ : int = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Tuple = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : List[Any] = PRETRAINED_INIT_CONFIGURATION UpperCamelCase_ : Any = RoFormerTokenizer def __init__( self : Tuple , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : List[str]="[UNK]" , UpperCAmelCase_ : Any="[SEP]" , UpperCAmelCase_ : Any="[PAD]" , UpperCAmelCase_ : List[str]="[CLS]" , UpperCAmelCase_ : str="[MASK]" , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : Optional[Any]=None , **UpperCAmelCase_ : List[str] , ): super().__init__( UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , do_lower_case=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , tokenize_chinese_chars=UpperCAmelCase_ , strip_accents=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = json.loads(self.backend_tokenizer.normalizer.__getstate__() ) if ( pre_tok_state.get("lowercase" , UpperCAmelCase_ ) != do_lower_case or pre_tok_state.get("strip_accents" , UpperCAmelCase_ ) != strip_accents ): SCREAMING_SNAKE_CASE : Optional[Any] = getattr(UpperCAmelCase_ , pre_tok_state.pop("type" ) ) SCREAMING_SNAKE_CASE : Any = do_lower_case SCREAMING_SNAKE_CASE : List[str] = strip_accents SCREAMING_SNAKE_CASE : Tuple = pre_tok_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = do_lower_case def __getstate__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = self.__dict__.copy() SCREAMING_SNAKE_CASE : Optional[Any] = BertPreTokenizer() return state def __setstate__( self : Tuple , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = d SCREAMING_SNAKE_CASE : Dict = self.__dict__["_tokenizer"].get_vocab() SCREAMING_SNAKE_CASE : Any = PreTokenizer.custom(JiebaPreTokenizer(UpperCAmelCase_ ) ) def _A ( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Union[str, Any]=None ): SCREAMING_SNAKE_CASE : List[Any] = [self.cls_token_id] + token_ids_a + [self.sep_token_id] if token_ids_a: output += token_ids_a + [self.sep_token_id] return output def _A ( self : Tuple , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : List[Any] = [self.sep_token_id] SCREAMING_SNAKE_CASE : Optional[int] = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1] def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): SCREAMING_SNAKE_CASE : Optional[int] = self._tokenizer.model.save(UpperCAmelCase_ , name=UpperCAmelCase_ ) return tuple(UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Any=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : Union[str, Any] = BertPreTokenizer() return super().save_pretrained(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ )
62
1
import argparse import torch from diffusers.pipelines.stable_diffusion.convert_from_ckpt import download_from_original_stable_diffusion_ckpt if __name__ == "__main__": snake_case = argparse.ArgumentParser() parser.add_argument( """--checkpoint_path""", default=None, type=str, required=True, help="""Path to the checkpoint to convert.""" ) # !wget https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml parser.add_argument( """--original_config_file""", default=None, type=str, help="""The YAML config file corresponding to the original architecture.""", ) parser.add_argument( """--num_in_channels""", default=None, type=int, help="""The number of input channels. If `None` number of input channels will be automatically inferred.""", ) parser.add_argument( """--scheduler_type""", default="""pndm""", type=str, help="""Type of scheduler to use. Should be one of ['pndm', 'lms', 'ddim', 'euler', 'euler-ancestral', 'dpm']""", ) parser.add_argument( """--pipeline_type""", default=None, type=str, help=( """The pipeline type. One of 'FrozenOpenCLIPEmbedder', 'FrozenCLIPEmbedder', 'PaintByExample'""" """. If `None` pipeline will be automatically inferred.""" ), ) parser.add_argument( """--image_size""", default=None, type=int, help=( """The image size that the model was trained on. Use 512 for Stable Diffusion v1.X and Stable Siffusion v2""" """ Base. Use 768 for Stable Diffusion v2.""" ), ) parser.add_argument( """--prediction_type""", default=None, type=str, help=( """The prediction type that the model was trained on. Use 'epsilon' for Stable Diffusion v1.X and Stable""" """ Diffusion v2 Base. Use 'v_prediction' for Stable Diffusion v2.""" ), ) parser.add_argument( """--extract_ema""", action="""store_true""", help=( """Only relevant for checkpoints that have both EMA and non-EMA weights. Whether to extract the EMA weights""" """ or not. Defaults to `False`. Add `--extract_ema` to extract the EMA weights. EMA weights usually yield""" """ higher quality images for inference. Non-EMA weights are usually better to continue fine-tuning.""" ), ) parser.add_argument( """--upcast_attention""", action="""store_true""", help=( """Whether the attention computation should always be upcasted. This is necessary when running stable""" """ diffusion 2.1.""" ), ) parser.add_argument( """--from_safetensors""", action="""store_true""", help="""If `--checkpoint_path` is in `safetensors` format, load checkpoint with safetensors instead of PyTorch.""", ) parser.add_argument( """--to_safetensors""", action="""store_true""", help="""Whether to store pipeline in safetensors format or not.""", ) parser.add_argument("""--dump_path""", default=None, type=str, required=True, help="""Path to the output model.""") parser.add_argument("""--device""", type=str, help="""Device to use (e.g. cpu, cuda:0, cuda:1, etc.)""") parser.add_argument( """--stable_unclip""", type=str, default=None, required=False, help="""Set if this is a stable unCLIP model. One of 'txt2img' or 'img2img'.""", ) parser.add_argument( """--stable_unclip_prior""", type=str, default=None, required=False, help="""Set if this is a stable unCLIP txt2img model. Selects which prior to use. If `--stable_unclip` is set to `txt2img`, the karlo prior (https://huggingface.co/kakaobrain/karlo-v1-alpha/tree/main/prior) is selected by default.""", ) parser.add_argument( """--clip_stats_path""", type=str, help="""Path to the clip stats file. Only required if the stable unclip model's config specifies `model.params.noise_aug_config.params.clip_stats_path`.""", required=False, ) parser.add_argument( """--controlnet""", action="""store_true""", default=None, help="""Set flag if this is a controlnet checkpoint.""" ) parser.add_argument("""--half""", action="""store_true""", help="""Save weights in half precision.""") parser.add_argument( """--vae_path""", type=str, default=None, required=False, help="""Set to a path, hub id to an already converted vae to not convert it again.""", ) snake_case = parser.parse_args() snake_case = download_from_original_stable_diffusion_ckpt( checkpoint_path=args.checkpoint_path, original_config_file=args.original_config_file, image_size=args.image_size, prediction_type=args.prediction_type, model_type=args.pipeline_type, extract_ema=args.extract_ema, scheduler_type=args.scheduler_type, num_in_channels=args.num_in_channels, upcast_attention=args.upcast_attention, from_safetensors=args.from_safetensors, device=args.device, stable_unclip=args.stable_unclip, stable_unclip_prior=args.stable_unclip_prior, clip_stats_path=args.clip_stats_path, controlnet=args.controlnet, vae_path=args.vae_path, ) if args.half: pipe.to(torch_dtype=torch.floataa) if args.controlnet: # only save the controlnet model pipe.controlnet.save_pretrained(args.dump_path, safe_serialization=args.to_safetensors) else: pipe.save_pretrained(args.dump_path, safe_serialization=args.to_safetensors)
62
def lowerCamelCase__ ( lowercase ): """simple docstring""" if not isinstance(lowercase , lowercase ): raise TypeError("only integers accepted as input" ) else: SCREAMING_SNAKE_CASE : Optional[int] = str(abs(lowercase ) ) SCREAMING_SNAKE_CASE : str = [list(lowercase ) for char in range(len(lowercase ) )] for index in range(len(lowercase ) ): num_transpositions[index].pop(lowercase ) return max( int("".join(list(lowercase ) ) ) for transposition in num_transpositions ) if __name__ == "__main__": __import__("""doctest""").testmod()
62
1
import requests from bsa import BeautifulSoup def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = BeautifulSoup(requests.get(lowercase , params=lowercase ).content , "html.parser" ) SCREAMING_SNAKE_CASE : Union[str, Any] = soup.find("div" , attrs={"class": "gs_ri"} ) SCREAMING_SNAKE_CASE : str = div.find("div" , attrs={"class": "gs_fl"} ).find_all("a" ) return anchors[2].get_text() if __name__ == "__main__": snake_case = { """title""": ( """Precisely geometry controlled microsupercapacitors for ultrahigh areal """ """capacitance, volumetric capacitance, and energy density""" ), """journal""": """Chem. Mater.""", """volume""": 30, """pages""": """3979-3990""", """year""": 2_018, """hl""": """en""", } print(get_citation("""https://scholar.google.com/scholar_lookup""", params=params))
62
# This model implementation is heavily inspired by https://github.com/haofanwang/ControlNet-for-Diffusers/ import gc import random import tempfile import unittest import numpy as np import torch from PIL import Image from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, ControlNetModel, DDIMScheduler, StableDiffusionControlNetImgaImgPipeline, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_controlnet import MultiControlNetModel from diffusers.utils import floats_tensor, load_image, load_numpy, randn_tensor, slow, torch_device from diffusers.utils.import_utils import is_xformers_available from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..pipeline_params import ( IMAGE_TO_IMAGE_IMAGE_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_PARAMS, ) from ..test_pipelines_common import ( PipelineKarrasSchedulerTesterMixin, PipelineLatentTesterMixin, PipelineTesterMixin, ) enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : List[Any] = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[Any] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Tuple = IMAGE_TO_IMAGE_IMAGE_PARAMS.union({'''control_image'''} ) UpperCamelCase_ : Any = IMAGE_TO_IMAGE_IMAGE_PARAMS def _A ( self : List[str] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : str = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : str , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Any = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Optional[Any] = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = 2 SCREAMING_SNAKE_CASE : Union[str, Any] = randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ) SCREAMING_SNAKE_CASE : Tuple = floats_tensor(control_image.shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : str = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : List[str] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : int ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : Union[str, Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : str = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[int] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Dict = frozenset([] ) # TO_DO: add image_params once refactored VaeImageProcessor.preprocess def _A ( self : Optional[Any] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) def init_weights(UpperCAmelCase_ : List[Any] ): if isinstance(UpperCAmelCase_ , torch.nn.Convad ): torch.nn.init.normal(m.weight ) m.bias.data.fill_(1.0 ) SCREAMING_SNAKE_CASE : List[str] = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Dict = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Any = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : Tuple = MultiControlNetModel([controlneta, controlneta] ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Dict = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = 2 SCREAMING_SNAKE_CASE : Tuple = [ randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), ] SCREAMING_SNAKE_CASE : Optional[int] = floats_tensor(control_image[0].shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : Optional[Any] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : Any = self.get_dummy_components() SCREAMING_SNAKE_CASE : str = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = 10.0 SCREAMING_SNAKE_CASE : Any = 4 SCREAMING_SNAKE_CASE : Optional[int] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : Optional[int] = pipe(**UpperCAmelCase_ )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : Any = scale SCREAMING_SNAKE_CASE : List[str] = pipe(**UpperCAmelCase_ , control_guidance_start=0.1 , control_guidance_end=0.2 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : List[Any] = pipe(**UpperCAmelCase_ , control_guidance_start=[0.1, 0.3] , control_guidance_end=[0.2, 0.7] )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = steps SCREAMING_SNAKE_CASE : Dict = scale SCREAMING_SNAKE_CASE : Dict = pipe(**UpperCAmelCase_ , control_guidance_start=0.4 , control_guidance_end=[0.5, 0.8] )[0] # make sure that all outputs are different assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 def _A ( self : Union[str, Any] ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : List[Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = self.get_dummy_components() SCREAMING_SNAKE_CASE : Dict = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) with tempfile.TemporaryDirectory() as tmpdir: try: # save_pretrained is not implemented for Multi-ControlNet pipe.save_pretrained(UpperCAmelCase_ ) except NotImplementedError: pass @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Optional[Any] ): super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny" ) SCREAMING_SNAKE_CASE : Union[str, Any] = StableDiffusionControlNetImgaImgPipeline.from_pretrained( "runwayml/stable-diffusion-v1-5" , safety_checker=UpperCAmelCase_ , controlnet=UpperCAmelCase_ ) pipe.enable_model_cpu_offload() pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.Generator(device="cpu" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : str = "evil space-punk bird" SCREAMING_SNAKE_CASE : Optional[Any] = load_image( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/bird_canny.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : Optional[int] = load_image( "https://huggingface.co/lllyasviel/sd-controlnet-canny/resolve/main/images/bird.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : str = pipe( UpperCAmelCase_ , UpperCAmelCase_ , control_image=UpperCAmelCase_ , generator=UpperCAmelCase_ , output_type="np" , num_inference_steps=50 , strength=0.6 , ) SCREAMING_SNAKE_CASE : int = output.images[0] assert image.shape == (512, 512, 3) SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/img2img.npy" ) assert np.abs(expected_image - image ).max() < 9E-2
62
1
snake_case = """ # Installazione di Transformers ! pip install transformers datasets # Per installare dalla fonte invece dell'ultima versione rilasciata, commenta il comando sopra e # rimuovi la modalità commento al comando seguente. # ! pip install git+https://github.com/huggingface/transformers.git """ snake_case = [{"""type""": """code""", """content""": INSTALL_CONTENT}] snake_case = { """{processor_class}""": """FakeProcessorClass""", """{model_class}""": """FakeModelClass""", """{object_class}""": """FakeObjectClass""", }
62
import argparse import json from pathlib import Path import requests import torch from huggingface_hub import hf_hub_download from PIL import Image from transformers import ( MobileViTConfig, MobileViTForImageClassification, MobileViTForSemanticSegmentation, MobileViTImageProcessor, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = MobileViTConfig() # size of the architecture if "mobilevit_s" in mobilevit_name: SCREAMING_SNAKE_CASE : List[Any] = [144, 192, 240] SCREAMING_SNAKE_CASE : Tuple = [16, 32, 64, 96, 128, 160, 640] elif "mobilevit_xs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [96, 120, 144] SCREAMING_SNAKE_CASE : Dict = [16, 32, 48, 64, 80, 96, 384] elif "mobilevit_xxs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [64, 80, 96] SCREAMING_SNAKE_CASE : List[str] = [16, 16, 24, 48, 64, 80, 320] SCREAMING_SNAKE_CASE : int = 0.05 SCREAMING_SNAKE_CASE : int = 2.0 if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : str = 512 SCREAMING_SNAKE_CASE : List[str] = 16 SCREAMING_SNAKE_CASE : Union[str, Any] = 21 SCREAMING_SNAKE_CASE : Dict = "pascal-voc-id2label.json" else: SCREAMING_SNAKE_CASE : Optional[Any] = 1000 SCREAMING_SNAKE_CASE : Optional[Any] = "imagenet-1k-id2label.json" SCREAMING_SNAKE_CASE : Any = "huggingface/label-files" SCREAMING_SNAKE_CASE : Tuple = json.load(open(hf_hub_download(lowercase , lowercase , repo_type="dataset" ) , "r" ) ) SCREAMING_SNAKE_CASE : List[str] = {int(lowercase ): v for k, v in idalabel.items()} SCREAMING_SNAKE_CASE : Optional[Any] = idalabel SCREAMING_SNAKE_CASE : str = {v: k for k, v in idalabel.items()} return config def lowerCamelCase__ ( lowercase , lowercase=False ): """simple docstring""" for i in range(1 , 6 ): if F'''layer_{i}.''' in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(F'''layer_{i}.''' , F'''encoder.layer.{i - 1}.''' ) if "conv_1." in name: SCREAMING_SNAKE_CASE : Dict = name.replace("conv_1." , "conv_stem." ) if ".block." in name: SCREAMING_SNAKE_CASE : List[str] = name.replace(".block." , "." ) if "exp_1x1" in name: SCREAMING_SNAKE_CASE : str = name.replace("exp_1x1" , "expand_1x1" ) if "red_1x1" in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("red_1x1" , "reduce_1x1" ) if ".local_rep.conv_3x3." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".local_rep.conv_3x3." , ".conv_kxk." ) if ".local_rep.conv_1x1." in name: SCREAMING_SNAKE_CASE : int = name.replace(".local_rep.conv_1x1." , ".conv_1x1." ) if ".norm." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".norm." , ".normalization." ) if ".conv." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".conv." , ".convolution." ) if ".conv_proj." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".conv_proj." , ".conv_projection." ) for i in range(0 , 2 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.layer.{j}.''' ) for i in range(2 , 6 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.''' ) if "expand_1x1" in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace("expand_1x1" , "downsampling_layer.expand_1x1" ) if "conv_3x3" in name: SCREAMING_SNAKE_CASE : str = name.replace("conv_3x3" , "downsampling_layer.conv_3x3" ) if "reduce_1x1" in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("reduce_1x1" , "downsampling_layer.reduce_1x1" ) for i in range(2 , 5 ): if F'''.global_rep.{i}.weight''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.global_rep.{i}.weight''' , ".layernorm.weight" ) if F'''.global_rep.{i}.bias''' in name: SCREAMING_SNAKE_CASE : str = name.replace(F'''.global_rep.{i}.bias''' , ".layernorm.bias" ) if ".global_rep." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".global_rep." , ".transformer." ) if ".pre_norm_mha.0." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_mha.0." , ".layernorm_before." ) if ".pre_norm_mha.1.out_proj." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_mha.1.out_proj." , ".attention.output.dense." ) if ".pre_norm_ffn.0." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(".pre_norm_ffn.0." , ".layernorm_after." ) if ".pre_norm_ffn.1." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_ffn.1." , ".intermediate.dense." ) if ".pre_norm_ffn.4." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_ffn.4." , ".output.dense." ) if ".transformer." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".transformer." , ".transformer.layer." ) if ".aspp_layer." in name: SCREAMING_SNAKE_CASE : int = name.replace(".aspp_layer." , "." ) if ".aspp_pool." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".aspp_pool." , "." ) if "seg_head." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("seg_head." , "segmentation_head." ) if "segmentation_head.classifier.classifier." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace("segmentation_head.classifier.classifier." , "segmentation_head.classifier." ) if "classifier.fc." in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("classifier.fc." , "classifier." ) elif (not base_model) and ("segmentation_head." not in name): SCREAMING_SNAKE_CASE : List[Any] = "mobilevit." + name return name def lowerCamelCase__ ( lowercase , lowercase , lowercase=False ): """simple docstring""" if base_model: SCREAMING_SNAKE_CASE : Optional[int] = "" else: SCREAMING_SNAKE_CASE : Any = "mobilevit." for key in orig_state_dict.copy().keys(): SCREAMING_SNAKE_CASE : Union[str, Any] = orig_state_dict.pop(lowercase ) if key[:8] == "encoder.": SCREAMING_SNAKE_CASE : int = key[8:] if "qkv" in key: SCREAMING_SNAKE_CASE : Optional[int] = key.split("." ) SCREAMING_SNAKE_CASE : Any = int(key_split[0][6:] ) - 1 SCREAMING_SNAKE_CASE : List[Any] = int(key_split[3] ) SCREAMING_SNAKE_CASE : List[Any] = model.get_submodule(F'''{model_prefix}encoder.layer.{layer_num}''' ) SCREAMING_SNAKE_CASE : int = layer.transformer.layer[transformer_num].attention.attention.all_head_size SCREAMING_SNAKE_CASE : Union[str, Any] = ( F'''{model_prefix}encoder.layer.{layer_num}.transformer.layer.{transformer_num}.attention.attention.''' ) if "weight" in key: SCREAMING_SNAKE_CASE : Optional[int] = val[:dim, :] SCREAMING_SNAKE_CASE : Tuple = val[dim : dim * 2, :] SCREAMING_SNAKE_CASE : Dict = val[-dim:, :] else: SCREAMING_SNAKE_CASE : str = val[:dim] SCREAMING_SNAKE_CASE : Union[str, Any] = val[dim : dim * 2] SCREAMING_SNAKE_CASE : Union[str, Any] = val[-dim:] else: SCREAMING_SNAKE_CASE : List[Any] = val return orig_state_dict def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = "http://images.cocodataset.org/val2017/000000039769.jpg" SCREAMING_SNAKE_CASE : Union[str, Any] = Image.open(requests.get(lowercase , stream=lowercase ).raw ) return im @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase=False ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = get_mobilevit_config(lowercase ) # load original state_dict SCREAMING_SNAKE_CASE : Union[str, Any] = torch.load(lowercase , map_location="cpu" ) # load 🤗 model if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : List[str] = MobileViTForSemanticSegmentation(lowercase ).eval() else: SCREAMING_SNAKE_CASE : str = MobileViTForImageClassification(lowercase ).eval() SCREAMING_SNAKE_CASE : Any = convert_state_dict(lowercase , lowercase ) model.load_state_dict(lowercase ) # Check outputs on an image, prepared by MobileViTImageProcessor SCREAMING_SNAKE_CASE : List[Any] = MobileViTImageProcessor(crop_size=config.image_size , size=config.image_size + 32 ) SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=prepare_img() , return_tensors="pt" ) SCREAMING_SNAKE_CASE : List[Any] = model(**lowercase ) SCREAMING_SNAKE_CASE : str = outputs.logits if mobilevit_name.startswith("deeplabv3_" ): assert logits.shape == (1, 21, 32, 32) if mobilevit_name == "deeplabv3_mobilevit_s": SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor( [ [[6.2065, 6.1292, 6.2070], [6.1079, 6.1254, 6.1747], [6.0042, 6.1071, 6.1034]], [[-6.9253, -6.8653, -7.0398], [-7.3218, -7.3983, -7.3670], [-7.1961, -7.2482, -7.1569]], [[-4.4723, -4.4348, -4.3769], [-5.3629, -5.4632, -5.4598], [-5.1587, -5.3402, -5.5059]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor( [ [[5.4449, 5.5733, 5.6314], [5.1815, 5.3930, 5.5963], [5.1656, 5.4333, 5.4853]], [[-9.4423, -9.7766, -9.6714], [-9.1581, -9.5720, -9.5519], [-9.1006, -9.6458, -9.5703]], [[-7.7721, -7.3716, -7.1583], [-8.4599, -8.0624, -7.7944], [-8.4172, -7.8366, -7.5025]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xxs": SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [ [[6.9811, 6.9743, 7.3123], [7.1777, 7.1931, 7.3938], [7.5633, 7.8050, 7.8901]], [[-10.5536, -10.2332, -10.2924], [-10.2336, -9.8624, -9.5964], [-10.8840, -10.8158, -10.6659]], [[-3.4938, -3.0631, -2.8620], [-3.4205, -2.8135, -2.6875], [-3.4179, -2.7945, -2.8750]], ] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3, :3, :3] , lowercase , atol=1E-4 ) else: assert logits.shape == (1, 1000) if mobilevit_name == "mobilevit_s": SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([-0.9866, 0.2392, -1.1241] ) elif mobilevit_name == "mobilevit_xs": SCREAMING_SNAKE_CASE : Dict = torch.tensor([-2.4761, -0.9399, -1.9587] ) elif mobilevit_name == "mobilevit_xxs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor([-1.9364, -1.2327, -0.4653] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3] , lowercase , atol=1E-4 ) Path(lowercase ).mkdir(exist_ok=lowercase ) print(F'''Saving model {mobilevit_name} to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowercase ) print(F'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowercase ) if push_to_hub: SCREAMING_SNAKE_CASE : List[str] = { "mobilevit_s": "mobilevit-small", "mobilevit_xs": "mobilevit-x-small", "mobilevit_xxs": "mobilevit-xx-small", "deeplabv3_mobilevit_s": "deeplabv3-mobilevit-small", "deeplabv3_mobilevit_xs": "deeplabv3-mobilevit-x-small", "deeplabv3_mobilevit_xxs": "deeplabv3-mobilevit-xx-small", } print("Pushing to the hub..." ) SCREAMING_SNAKE_CASE : int = model_mapping[mobilevit_name] image_processor.push_to_hub(lowercase , organization="apple" ) model.push_to_hub(lowercase , organization="apple" ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument( """--mobilevit_name""", default="""mobilevit_s""", type=str, help=( """Name of the MobileViT model you'd like to convert. Should be one of 'mobilevit_s', 'mobilevit_xs',""" """ 'mobilevit_xxs', 'deeplabv3_mobilevit_s', 'deeplabv3_mobilevit_xs', 'deeplabv3_mobilevit_xxs'.""" ), ) parser.add_argument( """--checkpoint_path""", required=True, type=str, help="""Path to the original state dict (.pt file).""" ) parser.add_argument( """--pytorch_dump_folder_path""", required=True, type=str, help="""Path to the output PyTorch model directory.""" ) parser.add_argument( """--push_to_hub""", action="""store_true""", help="""Whether or not to push the converted model to the 🤗 hub.""" ) snake_case = parser.parse_args() convert_movilevit_checkpoint( args.mobilevit_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub )
62
1
from collections import OrderedDict from typing import Any, List, Mapping, Optional from ... import PreTrainedTokenizer, TensorType, is_torch_available from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfigWithPast, PatchingSpec from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """EleutherAI/gpt-j-6B""": """https://huggingface.co/EleutherAI/gpt-j-6B/resolve/main/config.json""", # See all GPT-J models at https://huggingface.co/models?filter=gpt_j } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Tuple = '''gptj''' UpperCamelCase_ : int = { '''max_position_embeddings''': '''n_positions''', '''hidden_size''': '''n_embd''', '''num_attention_heads''': '''n_head''', '''num_hidden_layers''': '''n_layer''', } def __init__( self : Optional[int] , UpperCAmelCase_ : Any=5_0400 , UpperCAmelCase_ : Union[str, Any]=2048 , UpperCAmelCase_ : Optional[Any]=4096 , UpperCAmelCase_ : Tuple=28 , UpperCAmelCase_ : Union[str, Any]=16 , UpperCAmelCase_ : Tuple=64 , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Tuple="gelu_new" , UpperCAmelCase_ : List[str]=0.0 , UpperCAmelCase_ : List[str]=0.0 , UpperCAmelCase_ : Optional[int]=0.0 , UpperCAmelCase_ : Optional[int]=1E-5 , UpperCAmelCase_ : Union[str, Any]=0.02 , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Tuple=5_0256 , UpperCAmelCase_ : Tuple=5_0256 , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : str = vocab_size SCREAMING_SNAKE_CASE : List[str] = n_positions SCREAMING_SNAKE_CASE : Optional[int] = n_embd SCREAMING_SNAKE_CASE : List[str] = n_layer SCREAMING_SNAKE_CASE : List[str] = n_head SCREAMING_SNAKE_CASE : List[Any] = n_inner SCREAMING_SNAKE_CASE : Dict = rotary_dim SCREAMING_SNAKE_CASE : Any = activation_function SCREAMING_SNAKE_CASE : Union[str, Any] = resid_pdrop SCREAMING_SNAKE_CASE : Optional[int] = embd_pdrop SCREAMING_SNAKE_CASE : Optional[int] = attn_pdrop SCREAMING_SNAKE_CASE : Union[str, Any] = layer_norm_epsilon SCREAMING_SNAKE_CASE : Union[str, Any] = initializer_range SCREAMING_SNAKE_CASE : Tuple = use_cache SCREAMING_SNAKE_CASE : Union[str, Any] = bos_token_id SCREAMING_SNAKE_CASE : List[str] = eos_token_id super().__init__( bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , tie_word_embeddings=UpperCAmelCase_ , **UpperCAmelCase_ ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : Tuple , UpperCAmelCase_ : PretrainedConfig , UpperCAmelCase_ : str = "default" , UpperCAmelCase_ : List[PatchingSpec] = None , UpperCAmelCase_ : bool = False , ): super().__init__(UpperCAmelCase_ , task=UpperCAmelCase_ , patching_specs=UpperCAmelCase_ , use_past=UpperCAmelCase_ ) if not getattr(self._config , "pad_token_id" , UpperCAmelCase_ ): # TODO: how to do that better? SCREAMING_SNAKE_CASE : List[str] = 0 @property def _A ( self : str ): SCREAMING_SNAKE_CASE : Tuple = OrderedDict({"input_ids": {0: "batch", 1: "sequence"}} ) if self.use_past: self.fill_with_past_key_values_(UpperCAmelCase_ , direction="inputs" ) SCREAMING_SNAKE_CASE : Tuple = {0: "batch", 1: "past_sequence + sequence"} else: SCREAMING_SNAKE_CASE : Any = {0: "batch", 1: "sequence"} return common_inputs @property def _A ( self : Optional[Any] ): return self._config.n_layer @property def _A ( self : Optional[Any] ): return self._config.n_head def _A ( self : Dict , UpperCAmelCase_ : PreTrainedTokenizer , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : int = -1 , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : Optional[TensorType] = None , ): SCREAMING_SNAKE_CASE : List[Any] = super(UpperCAmelCase_ , self ).generate_dummy_inputs( UpperCAmelCase_ , batch_size=UpperCAmelCase_ , seq_length=UpperCAmelCase_ , is_pair=UpperCAmelCase_ , framework=UpperCAmelCase_ ) # We need to order the input in the way they appears in the forward() SCREAMING_SNAKE_CASE : Optional[Any] = 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 SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = common_inputs["input_ids"].shape # Not using the same length for past_key_values SCREAMING_SNAKE_CASE : Dict = seqlen + 2 SCREAMING_SNAKE_CASE : str = ( batch, self.num_attention_heads, past_key_values_length, self._config.hidden_size // self.num_attention_heads, ) SCREAMING_SNAKE_CASE : Union[str, Any] = [ (torch.zeros(UpperCAmelCase_ ), torch.zeros(UpperCAmelCase_ )) for _ in range(self.num_layers ) ] SCREAMING_SNAKE_CASE : Tuple = common_inputs["attention_mask"] if self.use_past: SCREAMING_SNAKE_CASE : Optional[Any] = ordered_inputs["attention_mask"].dtype SCREAMING_SNAKE_CASE : List[Any] = torch.cat( [ordered_inputs["attention_mask"], torch.ones(UpperCAmelCase_ , UpperCAmelCase_ , dtype=UpperCAmelCase_ )] , dim=1 ) return ordered_inputs @property def _A ( self : List[Any] ): return 13
62
from collections import Counter import numpy as np from sklearn import datasets from sklearn.model_selection import train_test_split snake_case = datasets.load_iris() snake_case = np.array(data["""data"""]) snake_case = np.array(data["""target"""]) snake_case = data["""target_names"""] snake_case , snake_case , snake_case , snake_case = train_test_split(X, y) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return np.linalg.norm(np.array(lowercase ) - np.array(lowercase ) ) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase=5 ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = zip(lowercase , lowercase ) # List of distances of all points from the point to be classified SCREAMING_SNAKE_CASE : Optional[int] = [] for data_point in data: SCREAMING_SNAKE_CASE : Tuple = euclidean_distance(data_point[0] , lowercase ) distances.append((distance, data_point[1]) ) # Choosing 'k' points with the least distances. SCREAMING_SNAKE_CASE : List[Any] = [i[1] for i in sorted(lowercase )[:k]] # Most commonly occurring class among them # is the class into which the point is classified SCREAMING_SNAKE_CASE : List[Any] = Counter(lowercase ).most_common(1 )[0][0] return classes[result] if __name__ == "__main__": print(classifier(X_train, y_train, classes, [4.4, 3.1, 1.3, 1.4]))
62
1
from __future__ import annotations import csv import requests from bsa import BeautifulSoup def lowerCamelCase__ ( lowercase = "" ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = url or "https://www.imdb.com/chart/top/?ref_=nv_mv_250" SCREAMING_SNAKE_CASE : int = BeautifulSoup(requests.get(lowercase ).text , "html.parser" ) SCREAMING_SNAKE_CASE : Any = soup.find_all("td" , attrs="titleColumn" ) SCREAMING_SNAKE_CASE : Optional[int] = soup.find_all("td" , class_="ratingColumn imdbRating" ) return { title.a.text: float(rating.strong.text ) for title, rating in zip(lowercase , lowercase ) } def lowerCamelCase__ ( lowercase = "IMDb_Top_250_Movies.csv" ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = get_imdb_top_aaa_movies() with open(lowercase , "w" , newline="" ) as out_file: SCREAMING_SNAKE_CASE : Optional[Any] = csv.writer(lowercase ) writer.writerow(["Movie title", "IMDb rating"] ) for title, rating in movies.items(): writer.writerow([title, rating] ) if __name__ == "__main__": write_movies()
62
from typing import Callable, Optional, Union from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """microsoft/xprophetnet-large-wiki100-cased""": ( """https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Tuple = '''xlm-prophetnet''' UpperCamelCase_ : Tuple = ['''past_key_values'''] UpperCamelCase_ : int = { '''num_attention_heads''': '''num_encoder_attention_heads''', } def __init__( self : Dict , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[Union[str, Callable]] = "gelu" , UpperCAmelCase_ : Optional[int] = 3_0522 , UpperCAmelCase_ : Optional[int] = 1024 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[int] = 512 , UpperCAmelCase_ : Optional[float] = 0.02 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 2 , UpperCAmelCase_ : Optional[int] = 32 , UpperCAmelCase_ : Optional[int] = 128 , UpperCAmelCase_ : Optional[bool] = False , UpperCAmelCase_ : Optional[float] = 0.0 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 1 , UpperCAmelCase_ : Optional[int] = 2 , **UpperCAmelCase_ : Dict , ): SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : str = hidden_size SCREAMING_SNAKE_CASE : Tuple = encoder_ffn_dim SCREAMING_SNAKE_CASE : List[Any] = num_encoder_layers SCREAMING_SNAKE_CASE : Any = num_encoder_attention_heads SCREAMING_SNAKE_CASE : Dict = decoder_ffn_dim SCREAMING_SNAKE_CASE : List[str] = num_decoder_layers SCREAMING_SNAKE_CASE : List[Any] = num_decoder_attention_heads SCREAMING_SNAKE_CASE : List[Any] = max_position_embeddings SCREAMING_SNAKE_CASE : Dict = init_std # Normal(0, this parameter) SCREAMING_SNAKE_CASE : Union[str, Any] = activation_function # parameters for xlmprophetnet SCREAMING_SNAKE_CASE : Dict = ngram SCREAMING_SNAKE_CASE : Any = num_buckets SCREAMING_SNAKE_CASE : str = relative_max_distance SCREAMING_SNAKE_CASE : str = disable_ngram_loss SCREAMING_SNAKE_CASE : Dict = eps # 3 Types of Dropout SCREAMING_SNAKE_CASE : Union[str, Any] = attention_dropout SCREAMING_SNAKE_CASE : Optional[int] = activation_dropout SCREAMING_SNAKE_CASE : List[Any] = dropout SCREAMING_SNAKE_CASE : int = use_cache super().__init__( pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , is_encoder_decoder=UpperCAmelCase_ , add_cross_attention=UpperCAmelCase_ , decoder_start_token_id=UpperCAmelCase_ , **UpperCAmelCase_ , ) @property def _A ( self : int ): return self.num_encoder_layers + self.num_decoder_layers @num_hidden_layers.setter def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): raise NotImplementedError( "This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and" " `num_decoder_layers`." )
62
1
from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """transfo-xl-wt103""": """https://huggingface.co/transfo-xl-wt103/resolve/main/config.json""", } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : List[str] = '''transfo-xl''' UpperCamelCase_ : List[str] = ['''mems'''] UpperCamelCase_ : Optional[Any] = { '''n_token''': '''vocab_size''', '''hidden_size''': '''d_model''', '''num_attention_heads''': '''n_head''', '''num_hidden_layers''': '''n_layer''', } def __init__( self : Tuple , UpperCAmelCase_ : List[str]=26_7735 , UpperCAmelCase_ : str=[2_0000, 4_0000, 20_0000] , UpperCAmelCase_ : Optional[int]=1024 , UpperCAmelCase_ : int=1024 , UpperCAmelCase_ : str=16 , UpperCAmelCase_ : int=64 , UpperCAmelCase_ : Union[str, Any]=4096 , UpperCAmelCase_ : Optional[int]=4 , UpperCAmelCase_ : Any=False , UpperCAmelCase_ : List[str]=18 , UpperCAmelCase_ : List[str]=1600 , UpperCAmelCase_ : Optional[Any]=1000 , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : List[str]=0 , UpperCAmelCase_ : int=-1 , UpperCAmelCase_ : str=True , UpperCAmelCase_ : List[str]=0.1 , UpperCAmelCase_ : Union[str, Any]=0.0 , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : Union[str, Any]="normal" , UpperCAmelCase_ : Optional[Any]=0.01 , UpperCAmelCase_ : Any=0.01 , UpperCAmelCase_ : int=0.02 , UpperCAmelCase_ : Dict=1E-5 , UpperCAmelCase_ : int=0 , **UpperCAmelCase_ : List[Any] , ): SCREAMING_SNAKE_CASE : List[Any] = vocab_size SCREAMING_SNAKE_CASE : List[Any] = [] self.cutoffs.extend(UpperCAmelCase_ ) if proj_share_all_but_first: SCREAMING_SNAKE_CASE : List[Any] = [False] + [True] * len(self.cutoffs ) else: SCREAMING_SNAKE_CASE : Dict = [False] + [False] * len(self.cutoffs ) SCREAMING_SNAKE_CASE : Dict = d_model SCREAMING_SNAKE_CASE : Optional[int] = d_embed SCREAMING_SNAKE_CASE : Optional[Any] = d_head SCREAMING_SNAKE_CASE : Any = d_inner SCREAMING_SNAKE_CASE : str = div_val SCREAMING_SNAKE_CASE : str = pre_lnorm SCREAMING_SNAKE_CASE : Optional[int] = n_layer SCREAMING_SNAKE_CASE : Optional[int] = n_head SCREAMING_SNAKE_CASE : List[str] = mem_len SCREAMING_SNAKE_CASE : Optional[int] = same_length SCREAMING_SNAKE_CASE : Optional[int] = attn_type SCREAMING_SNAKE_CASE : List[Any] = clamp_len SCREAMING_SNAKE_CASE : Dict = sample_softmax SCREAMING_SNAKE_CASE : List[str] = adaptive SCREAMING_SNAKE_CASE : Union[str, Any] = dropout SCREAMING_SNAKE_CASE : Optional[int] = dropatt SCREAMING_SNAKE_CASE : str = untie_r SCREAMING_SNAKE_CASE : str = init SCREAMING_SNAKE_CASE : List[Any] = init_range SCREAMING_SNAKE_CASE : Union[str, Any] = proj_init_std SCREAMING_SNAKE_CASE : Optional[Any] = init_std SCREAMING_SNAKE_CASE : Any = layer_norm_epsilon super().__init__(eos_token_id=UpperCAmelCase_ , **UpperCAmelCase_ ) @property def _A ( self : Any ): # Message copied from Transformer-XL documentation logger.info(f'''The model {self.model_type} is one of the few models that has no sequence length limit.''' ) return -1 @max_position_embeddings.setter def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[Any] ): # Message copied from Transformer-XL documentation raise NotImplementedError( f'''The model {self.model_type} is one of the few models that has no sequence length limit.''' )
62
import gc import random import unittest import numpy as np import torch from diffusers import DDIMScheduler, KandinskyVaaPipeline, KandinskyVaaPriorPipeline, UNetaDConditionModel, VQModel from diffusers.utils import floats_tensor, load_numpy, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = KandinskyVaaPipeline UpperCamelCase_ : List[Any] = [ '''image_embeds''', '''negative_image_embeds''', ] UpperCamelCase_ : Tuple = ['''image_embeds''', '''negative_image_embeds'''] UpperCamelCase_ : Any = [ '''generator''', '''height''', '''width''', '''latents''', '''guidance_scale''', '''num_inference_steps''', '''return_dict''', '''guidance_scale''', '''num_images_per_prompt''', '''output_type''', '''return_dict''', ] UpperCamelCase_ : List[str] = False @property def _A ( self : List[Any] ): return 32 @property def _A ( self : List[Any] ): return 32 @property def _A ( self : Any ): return self.time_input_dim @property def _A ( self : Union[str, Any] ): return self.time_input_dim * 4 @property def _A ( self : Tuple ): return 100 @property def _A ( self : Optional[int] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = { "in_channels": 4, # Out channels is double in channels because predicts mean and variance "out_channels": 8, "addition_embed_type": "image", "down_block_types": ("ResnetDownsampleBlock2D", "SimpleCrossAttnDownBlock2D"), "up_block_types": ("SimpleCrossAttnUpBlock2D", "ResnetUpsampleBlock2D"), "mid_block_type": "UNetMidBlock2DSimpleCrossAttn", "block_out_channels": (self.block_out_channels_a, self.block_out_channels_a * 2), "layers_per_block": 1, "encoder_hid_dim": self.text_embedder_hidden_size, "encoder_hid_dim_type": "image_proj", "cross_attention_dim": self.cross_attention_dim, "attention_head_dim": 4, "resnet_time_scale_shift": "scale_shift", "class_embed_type": None, } SCREAMING_SNAKE_CASE : str = UNetaDConditionModel(**UpperCAmelCase_ ) return model @property def _A ( self : int ): return { "block_out_channels": [32, 64], "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"], "in_channels": 3, "latent_channels": 4, "layers_per_block": 1, "norm_num_groups": 8, "norm_type": "spatial", "num_vq_embeddings": 12, "out_channels": 3, "up_block_types": [ "AttnUpDecoderBlock2D", "UpDecoderBlock2D", ], "vq_embed_dim": 4, } @property def _A ( self : Any ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : List[str] = VQModel(**self.dummy_movq_kwargs ) return model def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.dummy_unet SCREAMING_SNAKE_CASE : str = self.dummy_movq SCREAMING_SNAKE_CASE : Optional[int] = DDIMScheduler( num_train_timesteps=1000 , beta_schedule="linear" , beta_start=0.00_085 , beta_end=0.012 , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , steps_offset=1 , prediction_type="epsilon" , thresholding=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "scheduler": scheduler, "movq": movq, } return components def _A ( self : List[str] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any]=0 ): SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(seed + 1 ) ).to( UpperCAmelCase_ ) if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : List[Any] = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = { "image_embeds": image_embeds, "negative_image_embeds": negative_image_embeds, "generator": generator, "height": 64, "width": 64, "guidance_scale": 4.0, "num_inference_steps": 2, "output_type": "np", } return inputs def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "cpu" SCREAMING_SNAKE_CASE : Tuple = self.get_dummy_components() SCREAMING_SNAKE_CASE : List[str] = self.pipeline_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(**self.get_dummy_inputs(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : str = output.images SCREAMING_SNAKE_CASE : Tuple = pipe( **self.get_dummy_inputs(UpperCAmelCase_ ) , return_dict=UpperCAmelCase_ , )[0] SCREAMING_SNAKE_CASE : str = image[0, -3:, -3:, -1] SCREAMING_SNAKE_CASE : Any = image_from_tuple[0, -3:, -3:, -1] assert image.shape == (1, 64, 64, 3) SCREAMING_SNAKE_CASE : List[Any] = np.array( [0.6_237_976, 1.0, 0.36_441_332, 1.0, 0.70_639_634, 0.29_877_186, 0.85_652_125, 0.5_216_843, 0.54_454_046] ) assert ( np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_slice.flatten()}''' assert ( np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_from_tuple_slice.flatten()}''' @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main" "/kandinskyv22/kandinskyv22_text2img_cat_fp16.npy" ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPriorPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-prior" , torch_dtype=torch.floataa ) pipe_prior.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-decoder" , torch_dtype=torch.floataa ) SCREAMING_SNAKE_CASE : Tuple = pipeline.to(UpperCAmelCase_ ) pipeline.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = "red cat, 4k photo" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = pipe_prior( UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=5 , negative_prompt="" , ).to_tuple() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = pipeline( image_embeds=UpperCAmelCase_ , negative_image_embeds=UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=100 , output_type="np" , ) SCREAMING_SNAKE_CASE : List[Any] = output.images[0] assert image.shape == (512, 512, 3) assert_mean_pixel_difference(UpperCAmelCase_ , UpperCAmelCase_ )
62
1
import re import warnings from contextlib import contextmanager from ...processing_utils import ProcessorMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : str = ['''image_processor''', '''tokenizer'''] UpperCamelCase_ : Tuple = '''AutoImageProcessor''' UpperCamelCase_ : Dict = '''AutoTokenizer''' def __init__( self : List[str] , UpperCAmelCase_ : Union[str, Any]=None , UpperCAmelCase_ : Optional[int]=None , **UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Union[str, Any] = None if "feature_extractor" in kwargs: warnings.warn( "The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`" " instead." , UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : List[str] = kwargs.pop("feature_extractor" ) SCREAMING_SNAKE_CASE : Any = 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__(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.image_processor SCREAMING_SNAKE_CASE : Any = False def __call__( self : str , *UpperCAmelCase_ : Tuple , **UpperCAmelCase_ : str ): # For backward compatibility if self._in_target_context_manager: return self.current_processor(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = kwargs.pop("images" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = kwargs.pop("text" , UpperCAmelCase_ ) if len(UpperCAmelCase_ ) > 0: SCREAMING_SNAKE_CASE : Tuple = args[0] SCREAMING_SNAKE_CASE : str = args[1:] if images is None and text is None: raise ValueError("You need to specify either an `images` or `text` input to process." ) if images is not None: SCREAMING_SNAKE_CASE : Dict = self.image_processor(UpperCAmelCase_ , *UpperCAmelCase_ , **UpperCAmelCase_ ) if text is not None: SCREAMING_SNAKE_CASE : str = self.tokenizer(UpperCAmelCase_ , **UpperCAmelCase_ ) if text is None: return inputs elif images is None: return encodings else: SCREAMING_SNAKE_CASE : str = encodings["input_ids"] return inputs def _A ( self : Optional[Any] , *UpperCAmelCase_ : List[Any] , **UpperCAmelCase_ : str ): return self.tokenizer.batch_decode(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Dict , *UpperCAmelCase_ : Optional[Any] , **UpperCAmelCase_ : List[Any] ): return self.tokenizer.decode(*UpperCAmelCase_ , **UpperCAmelCase_ ) @contextmanager def _A ( self : str ): 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 images inputs, or in a separate call." ) SCREAMING_SNAKE_CASE : Union[str, Any] = True SCREAMING_SNAKE_CASE : List[str] = self.tokenizer yield SCREAMING_SNAKE_CASE : int = self.image_processor SCREAMING_SNAKE_CASE : Tuple = False def _A ( self : Union[str, Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Optional[Any]=False , UpperCAmelCase_ : Tuple=None ): if added_vocab is None: SCREAMING_SNAKE_CASE : Any = self.tokenizer.get_added_vocab() SCREAMING_SNAKE_CASE : str = {} while tokens: SCREAMING_SNAKE_CASE : Dict = re.search(r"<s_(.*?)>" , UpperCAmelCase_ , re.IGNORECASE ) if start_token is None: break SCREAMING_SNAKE_CASE : Optional[Any] = start_token.group(1 ) SCREAMING_SNAKE_CASE : Union[str, Any] = re.search(rf'''</s_{key}>''' , UpperCAmelCase_ , re.IGNORECASE ) SCREAMING_SNAKE_CASE : Any = start_token.group() if end_token is None: SCREAMING_SNAKE_CASE : List[str] = tokens.replace(UpperCAmelCase_ , "" ) else: SCREAMING_SNAKE_CASE : List[str] = end_token.group() SCREAMING_SNAKE_CASE : Tuple = re.escape(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = re.escape(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = re.search(f'''{start_token_escaped}(.*?){end_token_escaped}''' , UpperCAmelCase_ , re.IGNORECASE ) if content is not None: SCREAMING_SNAKE_CASE : str = content.group(1 ).strip() if r"<s_" in content and r"</s_" in content: # non-leaf node SCREAMING_SNAKE_CASE : Optional[int] = self.tokenajson(UpperCAmelCase_ , is_inner_value=UpperCAmelCase_ , added_vocab=UpperCAmelCase_ ) if value: if len(UpperCAmelCase_ ) == 1: SCREAMING_SNAKE_CASE : List[Any] = value[0] SCREAMING_SNAKE_CASE : Dict = value else: # leaf nodes SCREAMING_SNAKE_CASE : Union[str, Any] = [] for leaf in content.split(r"<sep/>" ): SCREAMING_SNAKE_CASE : List[Any] = leaf.strip() if leaf in added_vocab and leaf[0] == "<" and leaf[-2:] == "/>": SCREAMING_SNAKE_CASE : int = leaf[1:-2] # for categorical special tokens output[key].append(UpperCAmelCase_ ) if len(output[key] ) == 1: SCREAMING_SNAKE_CASE : Tuple = output[key][0] SCREAMING_SNAKE_CASE : List[str] = tokens[tokens.find(UpperCAmelCase_ ) + len(UpperCAmelCase_ ) :].strip() if tokens[:6] == r"<sep/>": # non-leaf nodes return [output] + self.tokenajson(tokens[6:] , is_inner_value=UpperCAmelCase_ , added_vocab=UpperCAmelCase_ ) if len(UpperCAmelCase_ ): return [output] if is_inner_value else output else: return [] if is_inner_value else {"text_sequence": tokens} @property def _A ( self : Optional[int] ): warnings.warn( "`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead." , UpperCAmelCase_ , ) return self.image_processor_class @property def _A ( self : Optional[int] ): warnings.warn( "`feature_extractor` is deprecated and will be removed in v5. Use `image_processor` instead." , UpperCAmelCase_ , ) return self.image_processor
62
import logging import os import quant_trainer import torch from torch.utils.data import DataLoader from transformers import Trainer, is_torch_tpu_available from transformers.trainer_utils import PredictionOutput snake_case = logging.getLogger(__name__) if is_torch_tpu_available(check_device=False): import torch_xla.core.xla_model as xm import torch_xla.debug.metrics as met class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , *UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Tuple=None , **UpperCAmelCase_ : Optional[Any] ): super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = eval_examples SCREAMING_SNAKE_CASE : List[Any] = post_process_function SCREAMING_SNAKE_CASE : Any = quant_trainer_args SCREAMING_SNAKE_CASE : Optional[Any] = 128 # default number of calibration samples def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple=None ): if calib_dataset is None and self.calib_dataset is None: raise ValueError("Trainer: calibration requires an calib_dataset." ) SCREAMING_SNAKE_CASE : str = calib_dataset if calib_dataset is not None else self.calib_dataset SCREAMING_SNAKE_CASE : str = self._remove_unused_columns(UpperCAmelCase_ , description="Calibration" ) return DataLoader( UpperCAmelCase_ , batch_size=self.args.eval_batch_size , collate_fn=self.data_collator , drop_last=self.args.dataloader_drop_last , num_workers=self.args.dataloader_num_workers , pin_memory=self.args.dataloader_pin_memory , shuffle=UpperCAmelCase_ , ) def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[int]=None ): SCREAMING_SNAKE_CASE : Any = self.train_dataset if calib_dataset is None else calib_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_calib_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args , calib=UpperCAmelCase_ ) model.eval() quant_trainer.enable_calibration(UpperCAmelCase_ ) logger.info("***** Running calibration *****" ) logger.info(f''' Num examples = {self.calib_num}''' ) logger.info(f''' Batch size = {calib_dataloader.batch_size}''' ) for step, inputs in enumerate(UpperCAmelCase_ ): # Prediction step SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = self.prediction_step(UpperCAmelCase_ , UpperCAmelCase_ , prediction_loss_only=UpperCAmelCase_ ) if (step + 1) * calib_dataloader.batch_size >= self.calib_num: break quant_trainer.finish_calibration(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : Optional[int] = model def _A ( self : List[Any] , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : str = "eval" ): SCREAMING_SNAKE_CASE : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset SCREAMING_SNAKE_CASE : Tuple = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.eval_examples if eval_examples is None else eval_examples # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Dict = self.compute_metrics SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : int = eval_loop( UpperCAmelCase_ , description="Evaluation" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : int = compute_metrics if self.post_process_function is not None and self.compute_metrics is not None: SCREAMING_SNAKE_CASE : List[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions ) SCREAMING_SNAKE_CASE : Any = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : Union[str, Any] = metrics.pop(UpperCAmelCase_ ) self.log(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[Any] = {} if self.args.tpu_metrics_debug or self.args.debug: # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.) xm.master_print(met.metrics_report() ) SCREAMING_SNAKE_CASE : List[str] = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCAmelCase_ ) return metrics def _A ( self : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : str = "test" ): SCREAMING_SNAKE_CASE : str = self.get_test_dataloader(UpperCAmelCase_ ) # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Tuple = self.compute_metrics SCREAMING_SNAKE_CASE : Union[str, Any] = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : str = eval_loop( UpperCAmelCase_ , description="Prediction" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : Union[str, Any] = compute_metrics if self.post_process_function is None or self.compute_metrics is None: return output SCREAMING_SNAKE_CASE : Optional[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions , "predict" ) SCREAMING_SNAKE_CASE : str = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : str = metrics.pop(UpperCAmelCase_ ) return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : int="./" ): SCREAMING_SNAKE_CASE : List[Any] = self.eval_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = next(iter(UpperCAmelCase_ ) ) # saving device - to make it consistent SCREAMING_SNAKE_CASE : int = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) # convert to tuple SCREAMING_SNAKE_CASE : Tuple = tuple(v.to(UpperCAmelCase_ ) for k, v in batch.items() ) logger.info("Converting model to be onnx compatible" ) from pytorch_quantization.nn import TensorQuantizer SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Any = self.model.to(UpperCAmelCase_ ) model.eval() model.float() SCREAMING_SNAKE_CASE : str = model.module if hasattr(UpperCAmelCase_ , "module" ) else model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : List[Any] = os.path.join(UpperCAmelCase_ , "model.onnx" ) logger.info(f'''exporting model to {output_model_file}''' ) SCREAMING_SNAKE_CASE : int = {0: "batch_size", 1: "seq_len"} torch.onnx.export( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , export_params=UpperCAmelCase_ , opset_version=13 , do_constant_folding=UpperCAmelCase_ , input_names=["input_ids", "attention_mask", "token_type_ids"] , output_names=["output_start_logits", "output_end_logits"] , dynamic_axes={ "input_ids": axes, "attention_mask": axes, "token_type_ids": axes, "output_start_logits": axes, "output_end_logits": axes, } , verbose=UpperCAmelCase_ , ) logger.info("onnx export finished" )
62
1
import os import tempfile import unittest from transformers import is_torch_available from transformers.testing_utils import require_torch if is_torch_available(): import torch from torch import nn from transformers import ( Adafactor, AdamW, get_constant_schedule, get_constant_schedule_with_warmup, get_cosine_schedule_with_warmup, get_cosine_with_hard_restarts_schedule_with_warmup, get_inverse_sqrt_schedule, get_linear_schedule_with_warmup, get_polynomial_decay_schedule_with_warmup, ) def lowerCamelCase__ ( lowercase , lowercase=10 ): """simple docstring""" SCREAMING_SNAKE_CASE : str = [] for _ in range(lowercase ): lrs.append(scheduler.get_lr()[0] ) scheduler.step() return lrs def lowerCamelCase__ ( lowercase , lowercase=10 ): """simple docstring""" SCREAMING_SNAKE_CASE : str = [] for step in range(lowercase ): lrs.append(scheduler.get_lr()[0] ) scheduler.step() if step == num_steps // 2: with tempfile.TemporaryDirectory() as tmpdirname: SCREAMING_SNAKE_CASE : Dict = os.path.join(lowercase , "schedule.bin" ) torch.save(scheduler.state_dict() , lowercase ) SCREAMING_SNAKE_CASE : List[Any] = torch.load(lowercase ) scheduler.load_state_dict(lowercase ) return lrs @require_torch class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any , UpperCAmelCase_ : Any , UpperCAmelCase_ : Any , UpperCAmelCase_ : Dict ): self.assertEqual(len(UpperCAmelCase_ ) , len(UpperCAmelCase_ ) ) for a, b in zip(UpperCAmelCase_ , UpperCAmelCase_ ): self.assertAlmostEqual(UpperCAmelCase_ , UpperCAmelCase_ , delta=UpperCAmelCase_ ) def _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = torch.tensor([0.1, -0.2, -0.1] , requires_grad=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = torch.tensor([0.4, 0.2, -0.5] ) SCREAMING_SNAKE_CASE : str = nn.MSELoss() # No warmup, constant schedule, no gradient clipping SCREAMING_SNAKE_CASE : Dict = AdamW(params=[w] , lr=2E-1 , weight_decay=0.0 ) for _ in range(100 ): SCREAMING_SNAKE_CASE : Optional[Any] = criterion(UpperCAmelCase_ , UpperCAmelCase_ ) loss.backward() optimizer.step() w.grad.detach_() # No zero_grad() function on simple tensors. we do it ourselves. w.grad.zero_() self.assertListAlmostEqual(w.tolist() , [0.4, 0.2, -0.5] , tol=1E-2 ) def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = torch.tensor([0.1, -0.2, -0.1] , requires_grad=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor([0.4, 0.2, -0.5] ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.MSELoss() # No warmup, constant schedule, no gradient clipping SCREAMING_SNAKE_CASE : List[str] = Adafactor( params=[w] , lr=1E-2 , eps=(1E-30, 1E-3) , clip_threshold=1.0 , decay_rate=-0.8 , betaa=UpperCAmelCase_ , weight_decay=0.0 , relative_step=UpperCAmelCase_ , scale_parameter=UpperCAmelCase_ , warmup_init=UpperCAmelCase_ , ) for _ in range(1000 ): SCREAMING_SNAKE_CASE : Union[str, Any] = criterion(UpperCAmelCase_ , UpperCAmelCase_ ) loss.backward() optimizer.step() w.grad.detach_() # No zero_grad() function on simple tensors. we do it ourselves. w.grad.zero_() self.assertListAlmostEqual(w.tolist() , [0.4, 0.2, -0.5] , tol=1E-2 ) @require_torch class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Dict = nn.Linear(5_0 , 5_0 ) if is_torch_available() else None UpperCamelCase_ : Tuple = AdamW(m.parameters() , lr=10.0 ) if is_torch_available() else None UpperCamelCase_ : str = 1_0 def _A ( self : Optional[Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int , UpperCAmelCase_ : Union[str, Any]=None ): self.assertEqual(len(UpperCAmelCase_ ) , len(UpperCAmelCase_ ) ) for a, b in zip(UpperCAmelCase_ , UpperCAmelCase_ ): self.assertAlmostEqual(UpperCAmelCase_ , UpperCAmelCase_ , delta=UpperCAmelCase_ , msg=UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : List[Any] = {"num_warmup_steps": 2, "num_training_steps": 10} # schedulers doct format # function: (sched_args_dict, expected_learning_rates) SCREAMING_SNAKE_CASE : Union[str, Any] = { get_constant_schedule: ({}, [10.0] * self.num_steps), get_constant_schedule_with_warmup: ( {"num_warmup_steps": 4}, [0.0, 2.5, 5.0, 7.5, 10.0, 10.0, 10.0, 10.0, 10.0, 10.0], ), get_linear_schedule_with_warmup: ( {**common_kwargs}, [0.0, 5.0, 10.0, 8.75, 7.5, 6.25, 5.0, 3.75, 2.5, 1.25], ), get_cosine_schedule_with_warmup: ( {**common_kwargs}, [0.0, 5.0, 10.0, 9.61, 8.53, 6.91, 5.0, 3.08, 1.46, 0.38], ), get_cosine_with_hard_restarts_schedule_with_warmup: ( {**common_kwargs, "num_cycles": 2}, [0.0, 5.0, 10.0, 8.53, 5.0, 1.46, 10.0, 8.53, 5.0, 1.46], ), get_polynomial_decay_schedule_with_warmup: ( {**common_kwargs, "power": 2.0, "lr_end": 1E-7}, [0.0, 5.0, 10.0, 7.656, 5.625, 3.906, 2.5, 1.406, 0.625, 0.156], ), get_inverse_sqrt_schedule: ( {"num_warmup_steps": 2}, [0.0, 5.0, 10.0, 8.165, 7.071, 6.325, 5.774, 5.345, 5.0, 4.714], ), } for scheduler_func, data in scheds.items(): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = data SCREAMING_SNAKE_CASE : Union[str, Any] = scheduler_func(self.optimizer , **UpperCAmelCase_ ) self.assertEqual(len([scheduler.get_lr()[0]] ) , 1 ) SCREAMING_SNAKE_CASE : Optional[int] = unwrap_schedule(UpperCAmelCase_ , self.num_steps ) self.assertListAlmostEqual( UpperCAmelCase_ , UpperCAmelCase_ , tol=1E-2 , msg=f'''failed for {scheduler_func} in normal scheduler''' , ) SCREAMING_SNAKE_CASE : Optional[int] = scheduler_func(self.optimizer , **UpperCAmelCase_ ) if scheduler_func.__name__ != "get_constant_schedule": LambdaScheduleWrapper.wrap_scheduler(UpperCAmelCase_ ) # wrap to test picklability of the schedule SCREAMING_SNAKE_CASE : Optional[Any] = unwrap_and_save_reload_schedule(UpperCAmelCase_ , self.num_steps ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ , msg=f'''failed for {scheduler_func} in save and reload''' ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Any , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : Union[str, Any] = fn def __call__( self : Any , *UpperCAmelCase_ : str , **UpperCAmelCase_ : Dict ): return self.fn(*UpperCAmelCase_ , **UpperCAmelCase_ ) @classmethod def _A ( self : Tuple , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = list(map(self , scheduler.lr_lambdas ) )
62
import os import unittest from transformers import LayoutLMTokenizer, LayoutLMTokenizerFast from transformers.models.layoutlm.tokenization_layoutlm import VOCAB_FILES_NAMES from transformers.testing_utils import require_tokenizers from ...test_tokenization_common import TokenizerTesterMixin @require_tokenizers class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = LayoutLMTokenizer UpperCamelCase_ : str = LayoutLMTokenizerFast UpperCamelCase_ : Any = True UpperCamelCase_ : Optional[Any] = True def _A ( self : Any ): super().setUp() SCREAMING_SNAKE_CASE : Optional[Any] = [ "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest", ] SCREAMING_SNAKE_CASE : Optional[Any] = 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 _A ( self : str , **UpperCAmelCase_ : Optional[int] ): return LayoutLMTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Any = "UNwant\u00E9d,running" SCREAMING_SNAKE_CASE : Union[str, Any] = "unwanted, running" return input_text, output_text def _A ( self : int ): SCREAMING_SNAKE_CASE : Optional[int] = self.tokenizer_class(self.vocab_file ) SCREAMING_SNAKE_CASE : Dict = tokenizer.tokenize("UNwant\u00E9d,running" ) self.assertListEqual(UpperCAmelCase_ , ["un", "##want", "##ed", ",", "runn", "##ing"] ) self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , [7, 4, 5, 10, 8, 9] ) def _A ( self : List[str] ): pass
62
1
import argparse import glob import logging import os import sys import time from collections import defaultdict from pathlib import Path from typing import Dict, List, Tuple import numpy as np import pytorch_lightning as pl import torch from callbacks import SeqaSeqLoggingCallback, get_checkpoint_callback, get_early_stopping_callback from torch import nn from torch.utils.data import DataLoader from transformers import MBartTokenizer, TaForConditionalGeneration from transformers.models.bart.modeling_bart import shift_tokens_right from utils import ( ROUGE_KEYS, LegacySeqaSeqDataset, SeqaSeqDataset, assert_all_frozen, calculate_bleu, calculate_rouge, check_output_dir, flatten_list, freeze_embeds, freeze_params, get_git_info, label_smoothed_nll_loss, lmap, pickle_save, save_git_info, save_json, use_task_specific_params, ) # need the parent dir module sys.path.insert(2, str(Path(__file__).resolve().parents[1])) from lightning_base import BaseTransformer, add_generic_args, generic_train # noqa snake_case = logging.getLogger(__name__) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[int] = '''summarization''' UpperCamelCase_ : Any = ['''loss'''] UpperCamelCase_ : int = ROUGE_KEYS UpperCamelCase_ : Tuple = '''rouge2''' def __init__( self : List[Any] , UpperCAmelCase_ : Optional[int] , **UpperCAmelCase_ : Optional[int] ): if hparams.sortish_sampler and hparams.gpus > 1: SCREAMING_SNAKE_CASE : Tuple = False elif hparams.max_tokens_per_batch is not None: if hparams.gpus > 1: raise NotImplementedError("Dynamic Batch size does not work for multi-gpu training" ) if hparams.sortish_sampler: raise ValueError("--sortish_sampler and --max_tokens_per_batch may not be used simultaneously" ) super().__init__(UpperCAmelCase_ , num_labels=UpperCAmelCase_ , mode=self.mode , **UpperCAmelCase_ ) use_task_specific_params(self.model , "summarization" ) save_git_info(self.hparams.output_dir ) SCREAMING_SNAKE_CASE : int = Path(self.output_dir ) / "metrics.json" SCREAMING_SNAKE_CASE : Union[str, Any] = Path(self.output_dir ) / "hparams.pkl" pickle_save(self.hparams , self.hparams_save_path ) SCREAMING_SNAKE_CASE : List[Any] = 0 SCREAMING_SNAKE_CASE : Tuple = defaultdict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.config.model_type SCREAMING_SNAKE_CASE : Any = self.config.tgt_vocab_size if self.model_type == "fsmt" else self.config.vocab_size SCREAMING_SNAKE_CASE : dict = { "data_dir": self.hparams.data_dir, "max_source_length": self.hparams.max_source_length, "prefix": self.model.config.prefix or "", } SCREAMING_SNAKE_CASE : Tuple = { "train": self.hparams.n_train, "val": self.hparams.n_val, "test": self.hparams.n_test, } SCREAMING_SNAKE_CASE : Union[str, Any] = {k: v if v >= 0 else None for k, v in n_observations_per_split.items()} SCREAMING_SNAKE_CASE : List[Any] = { "train": self.hparams.max_target_length, "val": self.hparams.val_max_target_length, "test": self.hparams.test_max_target_length, } assert self.target_lens["train"] <= self.target_lens["val"], f'''target_lens: {self.target_lens}''' assert self.target_lens["train"] <= self.target_lens["test"], f'''target_lens: {self.target_lens}''' if self.hparams.freeze_embeds: freeze_embeds(self.model ) if self.hparams.freeze_encoder: freeze_params(self.model.get_encoder() ) assert_all_frozen(self.model.get_encoder() ) SCREAMING_SNAKE_CASE : int = get_git_info()["repo_sha"] SCREAMING_SNAKE_CASE : Any = hparams.num_workers SCREAMING_SNAKE_CASE : Dict = None # default to config if self.model.config.decoder_start_token_id is None and isinstance(self.tokenizer , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Tuple = self.tokenizer.lang_code_to_id[hparams.tgt_lang] SCREAMING_SNAKE_CASE : str = self.decoder_start_token_id SCREAMING_SNAKE_CASE : Optional[int] = ( SeqaSeqDataset if hasattr(self.tokenizer , "prepare_seq2seq_batch" ) else LegacySeqaSeqDataset ) SCREAMING_SNAKE_CASE : Any = False SCREAMING_SNAKE_CASE : Optional[int] = self.model.config.num_beams if self.hparams.eval_beams is None else self.hparams.eval_beams if self.hparams.eval_max_gen_length is not None: SCREAMING_SNAKE_CASE : Dict = self.hparams.eval_max_gen_length else: SCREAMING_SNAKE_CASE : List[str] = self.model.config.max_length SCREAMING_SNAKE_CASE : str = self.default_val_metric if self.hparams.val_metric is None else self.hparams.val_metric def _A ( self : Any , UpperCAmelCase_ : Dict[str, torch.Tensor] ): SCREAMING_SNAKE_CASE : List[str] = { k: self.tokenizer.batch_decode(v.tolist() ) if "mask" not in k else v.shape for k, v in batch.items() } save_json(UpperCAmelCase_ , Path(self.output_dir ) / "text_batch.json" ) save_json({k: v.tolist() for k, v in batch.items()} , Path(self.output_dir ) / "tok_batch.json" ) SCREAMING_SNAKE_CASE : int = True return readable_batch def _A ( self : List[str] , UpperCAmelCase_ : List[Any] , **UpperCAmelCase_ : Any ): return self.model(UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[int] ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.tokenizer.batch_decode( UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ , clean_up_tokenization_spaces=UpperCAmelCase_ ) return lmap(str.strip , UpperCAmelCase_ ) def _A ( self : str , UpperCAmelCase_ : dict ): SCREAMING_SNAKE_CASE : str = self.tokenizer.pad_token_id SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = batch["input_ids"], batch["attention_mask"] SCREAMING_SNAKE_CASE : List[str] = batch["labels"] if isinstance(self.model , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : int = self.model._shift_right(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = shift_tokens_right(UpperCAmelCase_ , UpperCAmelCase_ ) if not self.already_saved_batch: # This would be slightly better if it only happened on rank zero SCREAMING_SNAKE_CASE : Optional[int] = decoder_input_ids self.save_readable_batch(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self(UpperCAmelCase_ , attention_mask=UpperCAmelCase_ , decoder_input_ids=UpperCAmelCase_ , use_cache=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = outputs["logits"] if self.hparams.label_smoothing == 0: # Same behavior as modeling_bart.py, besides ignoring pad_token_id SCREAMING_SNAKE_CASE : List[str] = nn.CrossEntropyLoss(ignore_index=UpperCAmelCase_ ) assert lm_logits.shape[-1] == self.vocab_size SCREAMING_SNAKE_CASE : Optional[Any] = ce_loss_fct(lm_logits.view(-1 , lm_logits.shape[-1] ) , tgt_ids.view(-1 ) ) else: SCREAMING_SNAKE_CASE : List[Any] = nn.functional.log_softmax(UpperCAmelCase_ , dim=-1 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = label_smoothed_nll_loss( UpperCAmelCase_ , UpperCAmelCase_ , self.hparams.label_smoothing , ignore_index=UpperCAmelCase_ ) return (loss,) @property def _A ( self : int ): return self.tokenizer.pad_token_id def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = self._step(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = dict(zip(self.loss_names , UpperCAmelCase_ ) ) # tokens per batch SCREAMING_SNAKE_CASE : Dict = batch["input_ids"].ne(self.pad ).sum() + batch["labels"].ne(self.pad ).sum() SCREAMING_SNAKE_CASE : Tuple = batch["input_ids"].shape[0] SCREAMING_SNAKE_CASE : Any = batch["input_ids"].eq(self.pad ).sum() SCREAMING_SNAKE_CASE : Optional[Any] = batch["input_ids"].eq(self.pad ).float().mean() # TODO(SS): make a wandb summary metric for this return {"loss": loss_tensors[0], "log": logs} def _A ( self : List[str] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int ): return self._generative_step(UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : Any , UpperCAmelCase_ : str="val" ): self.step_count += 1 SCREAMING_SNAKE_CASE : int = {k: torch.stack([x[k] for x in outputs] ).mean() for k in self.loss_names} SCREAMING_SNAKE_CASE : Union[str, Any] = losses["loss"] SCREAMING_SNAKE_CASE : Tuple = { k: np.array([x[k] for x in outputs] ).mean() for k in self.metric_names + ["gen_time", "gen_len"] } SCREAMING_SNAKE_CASE : str = ( generative_metrics[self.val_metric] if self.val_metric in generative_metrics else losses[self.val_metric] ) SCREAMING_SNAKE_CASE : torch.FloatTensor = torch.tensor(UpperCAmelCase_ ).type_as(UpperCAmelCase_ ) generative_metrics.update({k: v.item() for k, v in losses.items()} ) losses.update(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = {f'''{prefix}_avg_{k}''': x for k, x in losses.items()} SCREAMING_SNAKE_CASE : Tuple = self.step_count self.metrics[prefix].append(UpperCAmelCase_ ) # callback writes this to self.metrics_save_path SCREAMING_SNAKE_CASE : Any = flatten_list([x["preds"] for x in outputs] ) return { "log": all_metrics, "preds": preds, f'''{prefix}_loss''': loss, f'''{prefix}_{self.val_metric}''': metric_tensor, } def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any] ): return calculate_rouge(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[str] , UpperCAmelCase_ : dict ): SCREAMING_SNAKE_CASE : Union[str, Any] = time.time() # parser.add_argument('--eval_max_gen_length', type=int, default=None, help='never generate more than n tokens') SCREAMING_SNAKE_CASE : Union[str, Any] = self.model.generate( batch["input_ids"] , attention_mask=batch["attention_mask"] , use_cache=UpperCAmelCase_ , decoder_start_token_id=self.decoder_start_token_id , num_beams=self.eval_beams , max_length=self.eval_max_length , ) SCREAMING_SNAKE_CASE : Any = (time.time() - ta) / batch["input_ids"].shape[0] SCREAMING_SNAKE_CASE : List[str] = self.ids_to_clean_text(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = self.ids_to_clean_text(batch["labels"] ) SCREAMING_SNAKE_CASE : Dict = self._step(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = dict(zip(self.loss_names , UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : Dict = self.calc_generative_metrics(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = np.mean(lmap(UpperCAmelCase_ , UpperCAmelCase_ ) ) base_metrics.update(gen_time=UpperCAmelCase_ , gen_len=UpperCAmelCase_ , preds=UpperCAmelCase_ , target=UpperCAmelCase_ , **UpperCAmelCase_ ) return base_metrics def _A ( self : Dict , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Any ): return self._generative_step(UpperCAmelCase_ ) def _A ( self : str , UpperCAmelCase_ : int ): return self.validation_epoch_end(UpperCAmelCase_ , prefix="test" ) def _A ( self : Optional[int] , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.n_obs[type_path] SCREAMING_SNAKE_CASE : int = self.target_lens[type_path] SCREAMING_SNAKE_CASE : Any = self.dataset_class( self.tokenizer , type_path=UpperCAmelCase_ , n_obs=UpperCAmelCase_ , max_target_length=UpperCAmelCase_ , **self.dataset_kwargs , ) return dataset def _A ( self : Dict , UpperCAmelCase_ : str , UpperCAmelCase_ : int , UpperCAmelCase_ : bool = False ): SCREAMING_SNAKE_CASE : int = self.get_dataset(UpperCAmelCase_ ) if self.hparams.sortish_sampler and type_path != "test" and type_path != "val": SCREAMING_SNAKE_CASE : int = dataset.make_sortish_sampler(UpperCAmelCase_ , distributed=self.hparams.gpus > 1 ) return DataLoader( UpperCAmelCase_ , batch_size=UpperCAmelCase_ , collate_fn=dataset.collate_fn , shuffle=UpperCAmelCase_ , num_workers=self.num_workers , sampler=UpperCAmelCase_ , ) elif self.hparams.max_tokens_per_batch is not None and type_path != "test" and type_path != "val": SCREAMING_SNAKE_CASE : Union[str, Any] = dataset.make_dynamic_sampler( self.hparams.max_tokens_per_batch , distributed=self.hparams.gpus > 1 ) return DataLoader( UpperCAmelCase_ , batch_sampler=UpperCAmelCase_ , collate_fn=dataset.collate_fn , num_workers=self.num_workers , ) else: return DataLoader( UpperCAmelCase_ , batch_size=UpperCAmelCase_ , collate_fn=dataset.collate_fn , shuffle=UpperCAmelCase_ , num_workers=self.num_workers , sampler=UpperCAmelCase_ , ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = self.get_dataloader("train" , batch_size=self.hparams.train_batch_size , shuffle=UpperCAmelCase_ ) return dataloader def _A ( self : str ): return self.get_dataloader("val" , batch_size=self.hparams.eval_batch_size ) def _A ( self : int ): return self.get_dataloader("test" , batch_size=self.hparams.eval_batch_size ) @staticmethod def _A ( UpperCAmelCase_ : Any , UpperCAmelCase_ : List[Any] ): BaseTransformer.add_model_specific_args(UpperCAmelCase_ , UpperCAmelCase_ ) add_generic_args(UpperCAmelCase_ , UpperCAmelCase_ ) parser.add_argument( "--max_source_length" , default=1024 , type=UpperCAmelCase_ , help=( "The maximum total input sequence length after tokenization. Sequences longer " "than this will be truncated, sequences shorter will be padded." ) , ) parser.add_argument( "--max_target_length" , default=56 , type=UpperCAmelCase_ , help=( "The maximum total input sequence length after tokenization. Sequences longer " "than this will be truncated, sequences shorter will be padded." ) , ) parser.add_argument( "--val_max_target_length" , default=142 , type=UpperCAmelCase_ , help=( "The maximum total input sequence length after tokenization. Sequences longer " "than this will be truncated, sequences shorter will be padded." ) , ) parser.add_argument( "--test_max_target_length" , default=142 , type=UpperCAmelCase_ , help=( "The maximum total input sequence length after tokenization. Sequences longer " "than this will be truncated, sequences shorter will be padded." ) , ) parser.add_argument("--freeze_encoder" , action="store_true" ) parser.add_argument("--freeze_embeds" , action="store_true" ) parser.add_argument("--sortish_sampler" , action="store_true" , default=UpperCAmelCase_ ) parser.add_argument("--overwrite_output_dir" , action="store_true" , default=UpperCAmelCase_ ) parser.add_argument("--max_tokens_per_batch" , type=UpperCAmelCase_ , default=UpperCAmelCase_ ) parser.add_argument("--logger_name" , type=UpperCAmelCase_ , choices=["default", "wandb", "wandb_shared"] , default="default" ) parser.add_argument("--n_train" , type=UpperCAmelCase_ , default=-1 , required=UpperCAmelCase_ , help="# examples. -1 means use all." ) parser.add_argument("--n_val" , type=UpperCAmelCase_ , default=500 , required=UpperCAmelCase_ , help="# examples. -1 means use all." ) parser.add_argument("--n_test" , type=UpperCAmelCase_ , default=-1 , required=UpperCAmelCase_ , help="# examples. -1 means use all." ) parser.add_argument( "--task" , type=UpperCAmelCase_ , default="summarization" , required=UpperCAmelCase_ , help="# examples. -1 means use all." ) parser.add_argument("--label_smoothing" , type=UpperCAmelCase_ , default=0.0 , required=UpperCAmelCase_ ) parser.add_argument("--src_lang" , type=UpperCAmelCase_ , default="" , required=UpperCAmelCase_ ) parser.add_argument("--tgt_lang" , type=UpperCAmelCase_ , default="" , required=UpperCAmelCase_ ) parser.add_argument("--eval_beams" , type=UpperCAmelCase_ , default=UpperCAmelCase_ , required=UpperCAmelCase_ ) parser.add_argument( "--val_metric" , type=UpperCAmelCase_ , default=UpperCAmelCase_ , required=UpperCAmelCase_ , choices=["bleu", "rouge2", "loss", None] ) parser.add_argument("--eval_max_gen_length" , type=UpperCAmelCase_ , default=UpperCAmelCase_ , help="never generate more than n tokens" ) parser.add_argument("--save_top_k" , type=UpperCAmelCase_ , default=1 , required=UpperCAmelCase_ , help="How many checkpoints to save" ) parser.add_argument( "--early_stopping_patience" , type=UpperCAmelCase_ , default=-1 , required=UpperCAmelCase_ , help=( "-1 means never early stop. early_stopping_patience is measured in validation checks, not epochs. So" " val_check_interval will effect it." ) , ) return parser class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Union[str, Any] = '''translation''' UpperCamelCase_ : Union[str, Any] = ['''loss'''] UpperCamelCase_ : List[Any] = ['''bleu'''] UpperCamelCase_ : Dict = '''bleu''' def __init__( self : int , UpperCAmelCase_ : List[str] , **UpperCAmelCase_ : Optional[int] ): super().__init__(UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = hparams.src_lang SCREAMING_SNAKE_CASE : Tuple = hparams.tgt_lang def _A ( self : Dict , UpperCAmelCase_ : int , UpperCAmelCase_ : Union[str, Any] ): return calculate_bleu(UpperCAmelCase_ , UpperCAmelCase_ ) def lowerCamelCase__ ( lowercase , lowercase=None ): """simple docstring""" Path(args.output_dir ).mkdir(exist_ok=lowercase ) check_output_dir(lowercase , expected_items=3 ) if model is None: if "summarization" in args.task: SCREAMING_SNAKE_CASE : SummarizationModule = SummarizationModule(lowercase ) else: SCREAMING_SNAKE_CASE : SummarizationModule = TranslationModule(lowercase ) SCREAMING_SNAKE_CASE : Dict = Path(args.data_dir ).name if ( args.logger_name == "default" or args.fast_dev_run or str(args.output_dir ).startswith("/tmp" ) or str(args.output_dir ).startswith("/var" ) ): SCREAMING_SNAKE_CASE : Dict = True # don't pollute wandb logs unnecessarily elif args.logger_name == "wandb": from pytorch_lightning.loggers import WandbLogger SCREAMING_SNAKE_CASE : Union[str, Any] = os.environ.get("WANDB_PROJECT" , lowercase ) SCREAMING_SNAKE_CASE : Any = WandbLogger(name=model.output_dir.name , project=lowercase ) elif args.logger_name == "wandb_shared": from pytorch_lightning.loggers import WandbLogger SCREAMING_SNAKE_CASE : Any = WandbLogger(name=model.output_dir.name , project=F'''hf_{dataset}''' ) if args.early_stopping_patience >= 0: SCREAMING_SNAKE_CASE : str = get_early_stopping_callback(model.val_metric , args.early_stopping_patience ) else: SCREAMING_SNAKE_CASE : int = False SCREAMING_SNAKE_CASE : Tuple = args.val_metric == "loss" SCREAMING_SNAKE_CASE : pl.Trainer = generic_train( lowercase , lowercase , logging_callback=SeqaSeqLoggingCallback() , checkpoint_callback=get_checkpoint_callback( args.output_dir , model.val_metric , args.save_top_k , lowercase ) , early_stopping_callback=lowercase , logger=lowercase , ) pickle_save(model.hparams , model.output_dir / "hparams.pkl" ) if not args.do_predict: return model SCREAMING_SNAKE_CASE : str = "" SCREAMING_SNAKE_CASE : List[str] = sorted(glob.glob(os.path.join(args.output_dir , "*.ckpt" ) , recursive=lowercase ) ) if checkpoints: SCREAMING_SNAKE_CASE : Tuple = checkpoints[-1] SCREAMING_SNAKE_CASE : int = checkpoints[-1] trainer.logger.log_hyperparams(model.hparams ) # test() without a model tests using the best checkpoint automatically trainer.test() return model if __name__ == "__main__": snake_case = argparse.ArgumentParser() snake_case = pl.Trainer.add_argparse_args(parser) snake_case = SummarizationModule.add_model_specific_args(parser, os.getcwd()) snake_case = parser.parse_args() main(args)
62
import argparse import json import os from collections import OrderedDict import numpy as np import tensorflow as tf import torch def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = os.path.join(args.tf_model_dir , "parameters.json" ) SCREAMING_SNAKE_CASE : Union[str, Any] = json.loads(open(lowercase ).read() ) if not params: raise ValueError( F'''It seems that the json file at {parameter_file} is empty. Make sure you have a correct json file.''' ) if not args.output.endswith(".pt" ): SCREAMING_SNAKE_CASE : Optional[int] = args.output + ".pt" SCREAMING_SNAKE_CASE : Any = OrderedDict() with tf.device("/CPU:0" ): SCREAMING_SNAKE_CASE : Union[str, Any] = tf.train.load_checkpoint(args.tf_model_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = reader.get_variable_to_shape_map() for key_name in shapes.keys(): SCREAMING_SNAKE_CASE : Any = reader.get_tensor(lowercase ).astype(np.floataa ) if key_name.endswith("/adam_m" ) or key_name.endswith("/adam_v" ): continue if key_name.startswith("pasts/" ): if key_name.startswith("pasts/mlp" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9] ) elif key_name.startswith("pasts/out" ): SCREAMING_SNAKE_CASE : Optional[int] = 8 SCREAMING_SNAKE_CASE : List[Any] = "model.sqout.%d.weight" % (player * 2) # enter to nn.Sequencial with Tanh, so 2 at a time SCREAMING_SNAKE_CASE : Optional[int] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.startswith("model/moe" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/switch_gating/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.router.classifier.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/softmlp/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.soft_bypass_mlp.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/wo/kernel" ) or key_name.endswith("/wi/kernel" ): SCREAMING_SNAKE_CASE : Optional[int] = key_name[-9:-7] for i in range(16 ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.feed_forward.mlp.experts.expert_%d.%s.weight" % (player, i, nlayer) SCREAMING_SNAKE_CASE : List[str] = ( vnp[i].transpose([1, 0] ).copy() ) # In Mesh-Tensorflow, it is one array, so it is divided SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name.startswith("model/mlp" ): SCREAMING_SNAKE_CASE : str = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/p1/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.mlp.wi.weight" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/p1/bias" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.wi.bias" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/kernel" ): SCREAMING_SNAKE_CASE : str = "model.blocks.%d.feed_forward.mlp.wo.weight" % player SCREAMING_SNAKE_CASE : List[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/bias" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.feed_forward.mlp.wo.bias" % player SCREAMING_SNAKE_CASE : str = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) elif key_name.startswith("model/ln" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.norm.bias" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : List[str] = "model.blocks.%d.feed_forward.norm.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/att" ): SCREAMING_SNAKE_CASE : Optional[int] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/qkv/kernel" ): SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # Compute same dimension as Mesh-tensorflow using einsum SCREAMING_SNAKE_CASE : List[str] = state[:, 0, :, :] SCREAMING_SNAKE_CASE : Tuple = state[:, 1, :, :] SCREAMING_SNAKE_CASE : List[Any] = state[:, 2, :, :] SCREAMING_SNAKE_CASE : Tuple = ( state_q.reshape([state_q.shape[0], state_q.shape[1] * state_q.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = ( state_k.reshape([state_k.shape[0], state_k.shape[1] * state_k.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = ( state_v.reshape([state_v.shape[0], state_v.shape[1] * state_v.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : int = "model.blocks.%d.self_attn.self_attn.q_proj.weight" % player SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = "model.blocks.%d.self_attn.self_attn.k_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = "model.blocks.%d.self_attn.self_attn.v_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/o/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.self_attn.self_attn.out_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = ( vnp.reshape([vnp.shape[0] * vnp.shape[1], vnp.shape[2]] ).transpose([1, 0] ).copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif key_name.startswith("model/an" ): SCREAMING_SNAKE_CASE : int = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.self_attn.norm.bias" % player SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[str] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.self_attn.norm.weight" % player SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif ( key_name.startswith("model/wte" ) or key_name.startswith("model/wpe" ) or key_name.startswith("model/ete" ) ): SCREAMING_SNAKE_CASE : str = {"wte": "embed_tokens", "wpe": "position_embeddings", "ete": "extra_position_embeddings"}[ key_name[-3:] ] SCREAMING_SNAKE_CASE : List[str] = "model.%s.weight" % nlayer SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) if key_name.startswith("model/wte" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "lm_head.weight" SCREAMING_SNAKE_CASE : List[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/wob" ): SCREAMING_SNAKE_CASE : List[Any] = "final_logits_bias" SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : List[str] = state.reshape((1, -1) ) SCREAMING_SNAKE_CASE : int = torch.tensor(lowercase ) elif key_name == "model/dense/kernel": SCREAMING_SNAKE_CASE : Optional[int] = "model.last_project.weight" SCREAMING_SNAKE_CASE : Dict = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name == "model/dense_1/bias": SCREAMING_SNAKE_CASE : str = "model.last_project.bias" SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) torch.save(lowercase , args.output ) if __name__ == "__main__": snake_case = argparse.ArgumentParser( description="""model converter.""", formatter_class=argparse.ArgumentDefaultsHelpFormatter ) parser.add_argument("""--tf_model_dir""", metavar="""PATH""", type=str, required=True, help="""import model""") parser.add_argument("""--output""", metavar="""PATH""", type=str, required=True, help="""output model""") snake_case = parser.parse_args() convert_tf_gptsan_to_pt(args)
62
1
import copy import json import os import tempfile from transformers import is_torch_available from .test_configuration_utils import config_common_kwargs class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : Union[str, Any] , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Dict=None , UpperCAmelCase_ : Dict=True , UpperCAmelCase_ : List[Any]=None , **UpperCAmelCase_ : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = parent SCREAMING_SNAKE_CASE : Dict = config_class SCREAMING_SNAKE_CASE : Optional[Any] = has_text_modality SCREAMING_SNAKE_CASE : List[str] = kwargs SCREAMING_SNAKE_CASE : List[Any] = common_properties def _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : int = self.config_class(**self.inputs_dict ) SCREAMING_SNAKE_CASE : Dict = ( ["hidden_size", "num_attention_heads", "num_hidden_layers"] if self.common_properties is None else self.common_properties ) # Add common fields for text models if self.has_text_modality: common_properties.extend(["vocab_size"] ) # Test that config has the common properties as getters for prop in common_properties: self.parent.assertTrue(hasattr(UpperCAmelCase_ , UpperCAmelCase_ ) , msg=f'''`{prop}` does not exist''' ) # Test that config has the common properties as setter for idx, name in enumerate(UpperCAmelCase_ ): try: setattr(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) self.parent.assertEqual( getattr(UpperCAmelCase_ , UpperCAmelCase_ ) , UpperCAmelCase_ , msg=f'''`{name} value {idx} expected, but was {getattr(UpperCAmelCase_ , UpperCAmelCase_ )}''' ) except NotImplementedError: # Some models might not be able to implement setters for common_properties # In that case, a NotImplementedError is raised pass # Test if config class can be called with Config(prop_name=..) for idx, name in enumerate(UpperCAmelCase_ ): try: SCREAMING_SNAKE_CASE : List[str] = self.config_class(**{name: idx} ) self.parent.assertEqual( getattr(UpperCAmelCase_ , UpperCAmelCase_ ) , UpperCAmelCase_ , msg=f'''`{name} value {idx} expected, but was {getattr(UpperCAmelCase_ , UpperCAmelCase_ )}''' ) except NotImplementedError: # Some models might not be able to implement setters for common_properties # In that case, a NotImplementedError is raised pass def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Optional[int] = self.config_class(**self.inputs_dict ) SCREAMING_SNAKE_CASE : Any = json.loads(config.to_json_string() ) for key, value in self.inputs_dict.items(): self.parent.assertEqual(obj[key] , UpperCAmelCase_ ) def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : Optional[Any] = self.config_class(**self.inputs_dict ) with tempfile.TemporaryDirectory() as tmpdirname: SCREAMING_SNAKE_CASE : Dict = os.path.join(UpperCAmelCase_ , "config.json" ) config_first.to_json_file(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.config_class.from_json_file(UpperCAmelCase_ ) self.parent.assertEqual(config_second.to_dict() , config_first.to_dict() ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Tuple = self.config_class(**self.inputs_dict ) with tempfile.TemporaryDirectory() as tmpdirname: config_first.save_pretrained(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = self.config_class.from_pretrained(UpperCAmelCase_ ) self.parent.assertEqual(config_second.to_dict() , config_first.to_dict() ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.config_class(**self.inputs_dict ) SCREAMING_SNAKE_CASE : Dict = "test" with tempfile.TemporaryDirectory() as tmpdirname: SCREAMING_SNAKE_CASE : Tuple = os.path.join(UpperCAmelCase_ , UpperCAmelCase_ ) config_first.save_pretrained(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = self.config_class.from_pretrained(UpperCAmelCase_ , subfolder=UpperCAmelCase_ ) self.parent.assertEqual(config_second.to_dict() , config_first.to_dict() ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.config_class(**self.inputs_dict , num_labels=5 ) self.parent.assertEqual(len(config.idalabel ) , 5 ) self.parent.assertEqual(len(config.labelaid ) , 5 ) SCREAMING_SNAKE_CASE : List[Any] = 3 self.parent.assertEqual(len(config.idalabel ) , 3 ) self.parent.assertEqual(len(config.labelaid ) , 3 ) def _A ( self : str ): if self.config_class.is_composition: return SCREAMING_SNAKE_CASE : Optional[int] = self.config_class() self.parent.assertIsNotNone(UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = copy.deepcopy(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = self.config_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = [] for key, value in config_common_kwargs.items(): if key == "torch_dtype": if not is_torch_available(): continue else: import torch if config.torch_dtype != torch.floataa: wrong_values.append(("torch_dtype", config.torch_dtype, torch.floataa) ) elif getattr(UpperCAmelCase_ , UpperCAmelCase_ ) != value: wrong_values.append((key, getattr(UpperCAmelCase_ , UpperCAmelCase_ ), value) ) if len(UpperCAmelCase_ ) > 0: SCREAMING_SNAKE_CASE : Any = "\n".join([f'''- {v[0]}: got {v[1]} instead of {v[2]}''' for v in wrong_values] ) raise ValueError(f'''The following keys were not properly set in the config:\n{errors}''' ) def _A ( self : Tuple ): self.create_and_test_config_common_properties() self.create_and_test_config_to_json_string() self.create_and_test_config_to_json_file() self.create_and_test_config_from_and_save_pretrained() self.create_and_test_config_from_and_save_pretrained_subfolder() self.create_and_test_config_with_num_labels() self.check_config_can_be_init_without_params() self.check_config_arguments_init()
62
import json import os import shutil import tempfile import unittest import numpy as np from transformers import BertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer from transformers.testing_utils import require_tokenizers, require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor @require_tokenizers @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = tempfile.mkdtemp() # fmt: off SCREAMING_SNAKE_CASE : Any = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest"] # fmt: on SCREAMING_SNAKE_CASE : Dict = 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] ) ) SCREAMING_SNAKE_CASE : Optional[int] = { "do_resize": True, "size": {"height": 18, "width": 18}, "do_normalize": True, "image_mean": [0.5, 0.5, 0.5], "image_std": [0.5, 0.5, 0.5], } SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(self.tmpdirname , UpperCAmelCase_ ) with open(self.image_processor_file , "w" , encoding="utf-8" ) as fp: json.dump(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] , **UpperCAmelCase_ : List[str] ): return BertTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , **UpperCAmelCase_ : Any ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] ): shutil.rmtree(self.tmpdirname ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] SCREAMING_SNAKE_CASE : Union[str, Any] = [Image.fromarray(np.moveaxis(UpperCAmelCase_ , 0 , -1 ) ) for x in image_inputs] return image_inputs def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : int = VisionTextDualEncoderProcessor( tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer(bos_token="(BOS)" , eos_token="(EOS)" ) SCREAMING_SNAKE_CASE : Any = self.get_image_processor(do_normalize=UpperCAmelCase_ , padding_value=1.0 ) SCREAMING_SNAKE_CASE : str = VisionTextDualEncoderProcessor.from_pretrained( self.tmpdirname , bos_token="(BOS)" , eos_token="(EOS)" , do_normalize=UpperCAmelCase_ , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : int = image_processor(UpperCAmelCase_ , return_tensors="np" ) SCREAMING_SNAKE_CASE : Optional[int] = processor(images=UpperCAmelCase_ , 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 _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = "lower newer" SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = tokenizer(UpperCAmelCase_ ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : int = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer() SCREAMING_SNAKE_CASE : List[str] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = "lower newer" SCREAMING_SNAKE_CASE : Union[str, Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Optional[Any] = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , ["input_ids", "token_type_ids", "attention_mask", "pixel_values"] ) # test if it raises when no input is passed with self.assertRaises(UpperCAmelCase_ ): processor() def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] SCREAMING_SNAKE_CASE : Optional[Any] = processor.batch_decode(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer.batch_decode(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = "lower newer" SCREAMING_SNAKE_CASE : Any = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
62
1
def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = set() # edges = list of graph's edges SCREAMING_SNAKE_CASE : List[Any] = get_edges(lowercase ) # While there are still elements in edges list, take an arbitrary edge # (from_node, to_node) and add his extremity to chosen_vertices and then # remove all arcs adjacent to the from_node and to_node while edges: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = edges.pop() chosen_vertices.add(lowercase ) chosen_vertices.add(lowercase ) for edge in edges.copy(): if from_node in edge or to_node in edge: edges.discard(lowercase ) return chosen_vertices def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = set() for from_node, to_nodes in graph.items(): for to_node in to_nodes: edges.add((from_node, to_node) ) return edges if __name__ == "__main__": import doctest doctest.testmod() # graph = {0: [1, 3], 1: [0, 3], 2: [0, 3, 4], 3: [0, 1, 2], 4: [2, 3]} # print(f"Matching vertex cover:\n{matching_min_vertex_cover(graph)}")
62
import inspect import unittest from transformers import MobileNetVaConfig from transformers.testing_utils import require_torch, require_vision, slow, torch_device from transformers.utils import cached_property, is_torch_available, is_vision_available from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor from ...test_pipeline_mixin import PipelineTesterMixin if is_torch_available(): import torch from transformers import MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation, MobileNetVaModel from transformers.models.mobilenet_va.modeling_mobilenet_va import MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST if is_vision_available(): from PIL import Image from transformers import MobileNetVaImageProcessor class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.config_class(**self.inputs_dict ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "tf_padding" ) ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "depth_multiplier" ) ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : List[str]=13 , UpperCAmelCase_ : Any=3 , UpperCAmelCase_ : Union[str, Any]=32 , UpperCAmelCase_ : int=0.25 , UpperCAmelCase_ : Union[str, Any]=8 , UpperCAmelCase_ : Dict=8 , UpperCAmelCase_ : Optional[int]=6 , UpperCAmelCase_ : Optional[Any]=32 , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : str=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : str="relu6" , UpperCAmelCase_ : List[str]=1280 , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : Any=0.02 , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=10 , UpperCAmelCase_ : Optional[Any]=None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = parent SCREAMING_SNAKE_CASE : int = batch_size SCREAMING_SNAKE_CASE : Dict = num_channels SCREAMING_SNAKE_CASE : Dict = image_size SCREAMING_SNAKE_CASE : int = depth_multiplier SCREAMING_SNAKE_CASE : str = depth_divisible_by SCREAMING_SNAKE_CASE : Union[str, Any] = min_depth SCREAMING_SNAKE_CASE : int = expand_ratio SCREAMING_SNAKE_CASE : Tuple = tf_padding SCREAMING_SNAKE_CASE : List[str] = output_stride SCREAMING_SNAKE_CASE : Optional[int] = first_layer_is_expansion SCREAMING_SNAKE_CASE : Any = finegrained_output SCREAMING_SNAKE_CASE : Optional[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[int] = last_hidden_size if finegrained_output else int(last_hidden_size * depth_multiplier ) SCREAMING_SNAKE_CASE : Any = classifier_dropout_prob SCREAMING_SNAKE_CASE : Dict = use_labels SCREAMING_SNAKE_CASE : int = is_training SCREAMING_SNAKE_CASE : Dict = num_labels SCREAMING_SNAKE_CASE : Optional[Any] = initializer_range SCREAMING_SNAKE_CASE : Union[str, Any] = scope def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] ) SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : List[Any] = None if self.use_labels: SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size] , self.num_labels ) SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size, self.image_size, self.image_size] , self.num_labels ) SCREAMING_SNAKE_CASE : Tuple = self.get_config() return config, pixel_values, labels, pixel_labels def _A ( self : Optional[int] ): return MobileNetVaConfig( num_channels=self.num_channels , image_size=self.image_size , depth_multiplier=self.depth_multiplier , depth_divisible_by=self.depth_divisible_by , min_depth=self.min_depth , expand_ratio=self.expand_ratio , output_stride=self.output_stride , first_layer_is_expansion=self.first_layer_is_expansion , finegrained_output=self.finegrained_output , hidden_act=self.hidden_act , tf_padding=self.tf_padding , classifier_dropout_prob=self.classifier_dropout_prob , initializer_range=self.initializer_range , ) def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaModel(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ ) self.parent.assertEqual( result.last_hidden_state.shape , ( self.batch_size, self.last_hidden_size, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) self.parent.assertEqual( result.pooler_output.shape , (self.batch_size, self.last_hidden_size) , ) def _A ( self : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : str = self.num_labels SCREAMING_SNAKE_CASE : List[Any] = MobileNetVaForImageClassification(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Dict = MobileNetVaForSemanticSegmentation(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[int] = model(UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : List[Any] = self.prepare_config_and_inputs() SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = config_and_inputs SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": pixel_values} return config, inputs_dict @require_torch class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Any = ( (MobileNetVaModel, MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation) if is_torch_available() else () ) UpperCamelCase_ : List[Any] = ( { '''feature-extraction''': MobileNetVaModel, '''image-classification''': MobileNetVaForImageClassification, '''image-segmentation''': MobileNetVaForSemanticSegmentation, } if is_torch_available() else {} ) UpperCamelCase_ : Any = False UpperCamelCase_ : List[str] = False UpperCamelCase_ : int = False UpperCamelCase_ : str = False def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = MobileNetVaModelTester(self ) SCREAMING_SNAKE_CASE : Optional[Any] = MobileNetVaConfigTester(self , config_class=UpperCAmelCase_ , has_text_modality=UpperCAmelCase_ ) def _A ( self : Optional[Any] ): self.config_tester.run_common_tests() @unittest.skip(reason="MobileNetV2 does not use inputs_embeds" ) def _A ( self : List[Any] ): pass @unittest.skip(reason="MobileNetV2 does not support input and output embeddings" ) def _A ( self : Dict ): pass @unittest.skip(reason="MobileNetV2 does not output attentions" ) def _A ( self : Union[str, Any] ): pass def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : Tuple = model_class(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic SCREAMING_SNAKE_CASE : Dict = [*signature.parameters.keys()] SCREAMING_SNAKE_CASE : Union[str, Any] = ["pixel_values"] self.assertListEqual(arg_names[:1] , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*UpperCAmelCase_ ) def _A ( self : List[Any] ): def check_hidden_states_output(UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = model_class(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**self._prepare_for_class(UpperCAmelCase_ , UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : List[Any] = outputs.hidden_states SCREAMING_SNAKE_CASE : Any = 16 self.assertEqual(len(UpperCAmelCase_ ) , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : str = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) # check that output_hidden_states also work using config del inputs_dict["output_hidden_states"] SCREAMING_SNAKE_CASE : List[Any] = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Tuple = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_image_classification(*UpperCAmelCase_ ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_semantic_segmentation(*UpperCAmelCase_ ) @slow def _A ( self : Optional[Any] ): for model_name in MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: SCREAMING_SNAKE_CASE : int = MobileNetVaModel.from_pretrained(UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : int = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" ) return image @require_torch @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @cached_property def _A ( self : Optional[int] ): return ( MobileNetVaImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224" ) if is_vision_available() else None ) @slow def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaForImageClassification.from_pretrained("google/mobilenet_v2_1.0_224" ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.default_image_processor SCREAMING_SNAKE_CASE : Optional[int] = prepare_img() SCREAMING_SNAKE_CASE : Optional[int] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**UpperCAmelCase_ ) # verify the logits SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Size((1, 1001) ) self.assertEqual(outputs.logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = torch.tensor([0.2_445, -1.1_993, 0.1_905] ).to(UpperCAmelCase_ ) self.assertTrue(torch.allclose(outputs.logits[0, :3] , UpperCAmelCase_ , atol=1E-4 ) ) @slow def _A ( self : str ): SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaForSemanticSegmentation.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : int = model.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaImageProcessor.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : Optional[Any] = prepare_img() SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = outputs.logits # verify the logits SCREAMING_SNAKE_CASE : Dict = torch.Size((1, 21, 65, 65) ) self.assertEqual(logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.tensor( [ [[17.5_790, 17.7_581, 18.3_355], [18.3_257, 18.4_230, 18.8_973], [18.6_169, 18.8_650, 19.2_187]], [[-2.1_595, -2.0_977, -2.3_741], [-2.4_226, -2.3_028, -2.6_835], [-2.7_819, -2.5_991, -2.7_706]], [[4.2_058, 4.8_317, 4.7_638], [4.4_136, 5.0_361, 4.9_383], [4.5_028, 4.9_644, 4.8_734]], ] , device=UpperCAmelCase_ , ) self.assertTrue(torch.allclose(logits[0, :3, :3, :3] , UpperCAmelCase_ , atol=1E-4 ) )
62
1
import argparse import glob import logging import os import time from argparse import Namespace import numpy as np import torch from lightning_base import BaseTransformer, add_generic_args, generic_train from torch.utils.data import DataLoader, TensorDataset from transformers import glue_compute_metrics as compute_metrics from transformers import glue_convert_examples_to_features as convert_examples_to_features from transformers import glue_output_modes, glue_tasks_num_labels from transformers import glue_processors as processors snake_case = logging.getLogger(__name__) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : List[str] = '''sequence-classification''' def __init__( self : Optional[int] , UpperCAmelCase_ : Tuple ): if type(UpperCAmelCase_ ) == dict: SCREAMING_SNAKE_CASE : List[str] = Namespace(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = glue_output_modes[hparams.task] SCREAMING_SNAKE_CASE : Union[str, Any] = glue_tasks_num_labels[hparams.task] super().__init__(UpperCAmelCase_ , UpperCAmelCase_ , self.mode ) def _A ( self : List[str] , **UpperCAmelCase_ : str ): return self.model(**UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = {"input_ids": batch[0], "attention_mask": batch[1], "labels": batch[3]} if self.config.model_type not in ["distilbert", "bart"]: SCREAMING_SNAKE_CASE : Union[str, Any] = batch[2] if self.config.model_type in ["bert", "xlnet", "albert"] else None SCREAMING_SNAKE_CASE : Union[str, Any] = self(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = outputs[0] SCREAMING_SNAKE_CASE : List[Any] = self.trainer.lr_schedulers[0]["scheduler"] SCREAMING_SNAKE_CASE : List[str] = {"loss": loss, "rate": lr_scheduler.get_last_lr()[-1]} return {"loss": loss, "log": tensorboard_logs} def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = self.hparams SCREAMING_SNAKE_CASE : Tuple = processors[args.task]() SCREAMING_SNAKE_CASE : Optional[int] = processor.get_labels() for mode in ["train", "dev"]: SCREAMING_SNAKE_CASE : str = self._feature_file(UpperCAmelCase_ ) if os.path.exists(UpperCAmelCase_ ) and not args.overwrite_cache: logger.info("Loading features from cached file %s" , UpperCAmelCase_ ) else: logger.info("Creating features from dataset file at %s" , args.data_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = ( processor.get_dev_examples(args.data_dir ) if mode == "dev" else processor.get_train_examples(args.data_dir ) ) SCREAMING_SNAKE_CASE : List[str] = convert_examples_to_features( UpperCAmelCase_ , self.tokenizer , max_length=args.max_seq_length , label_list=self.labels , output_mode=args.glue_output_mode , ) logger.info("Saving features into cached file %s" , UpperCAmelCase_ ) torch.save(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : int , UpperCAmelCase_ : bool = False ): SCREAMING_SNAKE_CASE : Dict = "dev" if mode == "test" else mode SCREAMING_SNAKE_CASE : List[Any] = self._feature_file(UpperCAmelCase_ ) logger.info("Loading features from cached file %s" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = torch.load(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = torch.tensor([f.input_ids for f in features] , dtype=torch.long ) SCREAMING_SNAKE_CASE : Dict = torch.tensor([f.attention_mask for f in features] , dtype=torch.long ) SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor([f.token_type_ids for f in features] , dtype=torch.long ) if self.hparams.glue_output_mode == "classification": SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor([f.label for f in features] , dtype=torch.long ) elif self.hparams.glue_output_mode == "regression": SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor([f.label for f in features] , dtype=torch.float ) return DataLoader( TensorDataset(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) , batch_size=UpperCAmelCase_ , shuffle=UpperCAmelCase_ , ) def _A ( self : str , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : Dict = {"input_ids": batch[0], "attention_mask": batch[1], "labels": batch[3]} if self.config.model_type not in ["distilbert", "bart"]: SCREAMING_SNAKE_CASE : List[str] = batch[2] if self.config.model_type in ["bert", "xlnet", "albert"] else None SCREAMING_SNAKE_CASE : str = self(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = outputs[:2] SCREAMING_SNAKE_CASE : Union[str, Any] = logits.detach().cpu().numpy() SCREAMING_SNAKE_CASE : Optional[int] = inputs["labels"].detach().cpu().numpy() return {"val_loss": tmp_eval_loss.detach().cpu(), "pred": preds, "target": out_label_ids} def _A ( self : List[Any] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : Union[str, Any] = torch.stack([x["val_loss"] for x in outputs] ).mean().detach().cpu().item() SCREAMING_SNAKE_CASE : Tuple = np.concatenate([x["pred"] for x in outputs] , axis=0 ) if self.hparams.glue_output_mode == "classification": SCREAMING_SNAKE_CASE : List[str] = np.argmax(UpperCAmelCase_ , axis=1 ) elif self.hparams.glue_output_mode == "regression": SCREAMING_SNAKE_CASE : str = np.squeeze(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = np.concatenate([x["target"] for x in outputs] , axis=0 ) SCREAMING_SNAKE_CASE : Dict = [[] for _ in range(out_label_ids.shape[0] )] SCREAMING_SNAKE_CASE : List[Any] = [[] for _ in range(out_label_ids.shape[0] )] SCREAMING_SNAKE_CASE : Any = {**{"val_loss": val_loss_mean}, **compute_metrics(self.hparams.task , UpperCAmelCase_ , UpperCAmelCase_ )} SCREAMING_SNAKE_CASE : Union[str, Any] = dict(results.items() ) SCREAMING_SNAKE_CASE : Optional[Any] = results return ret, preds_list, out_label_list def _A ( self : Union[str, Any] , UpperCAmelCase_ : list ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = self._eval_end(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = ret["log"] return {"val_loss": logs["val_loss"], "log": logs, "progress_bar": logs} def _A ( self : List[str] , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = self._eval_end(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = ret["log"] # `val_loss` is the key returned by `self._eval_end()` but actually refers to `test_loss` return {"avg_test_loss": logs["val_loss"], "log": logs, "progress_bar": logs} @staticmethod def _A ( UpperCAmelCase_ : str , UpperCAmelCase_ : int ): BaseTransformer.add_model_specific_args(UpperCAmelCase_ , UpperCAmelCase_ ) parser.add_argument( "--max_seq_length" , default=128 , type=UpperCAmelCase_ , help=( "The maximum total input sequence length after tokenization. Sequences longer " "than this will be truncated, sequences shorter will be padded." ) , ) parser.add_argument( "--task" , default="" , type=UpperCAmelCase_ , required=UpperCAmelCase_ , help="The GLUE task to run" , ) parser.add_argument( "--gpus" , default=0 , type=UpperCAmelCase_ , help="The number of GPUs allocated for this, it is by default 0 meaning none" , ) parser.add_argument( "--overwrite_cache" , action="store_true" , help="Overwrite the cached training and evaluation sets" ) return parser def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = argparse.ArgumentParser() add_generic_args(lowercase , os.getcwd() ) SCREAMING_SNAKE_CASE : int = GLUETransformer.add_model_specific_args(lowercase , os.getcwd() ) SCREAMING_SNAKE_CASE : List[Any] = parser.parse_args() # If output_dir not provided, a folder will be generated in pwd if args.output_dir is None: SCREAMING_SNAKE_CASE : List[str] = os.path.join( "./results" , F'''{args.task}_{time.strftime('%Y%m%d_%H%M%S' )}''' , ) os.makedirs(args.output_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = GLUETransformer(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = generic_train(lowercase , lowercase ) # Optionally, predict on dev set and write to output_dir if args.do_predict: SCREAMING_SNAKE_CASE : Optional[Any] = sorted(glob.glob(os.path.join(args.output_dir , "checkpoint-epoch=*.ckpt" ) , recursive=lowercase ) ) SCREAMING_SNAKE_CASE : Dict = model.load_from_checkpoint(checkpoints[-1] ) return trainer.test(lowercase ) if __name__ == "__main__": main()
62
import math from typing import Dict, Iterable, List, Optional, Tuple, Union import numpy as np from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format from ...image_utils import ( IMAGENET_STANDARD_MEAN, IMAGENET_STANDARD_STD, ChannelDimension, ImageInput, PILImageResampling, get_image_size, is_torch_available, is_torch_tensor, make_list_of_images, to_numpy_array, valid_images, ) from ...utils import TensorType, is_vision_available, logging if is_torch_available(): import torch if is_vision_available(): import PIL snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" def constraint_to_multiple_of(lowercase , lowercase , lowercase=0 , lowercase=None ): SCREAMING_SNAKE_CASE : Any = round(val / multiple ) * multiple if max_val is not None and x > max_val: SCREAMING_SNAKE_CASE : Optional[int] = math.floor(val / multiple ) * multiple if x < min_val: SCREAMING_SNAKE_CASE : int = math.ceil(val / multiple ) * multiple return x SCREAMING_SNAKE_CASE : Tuple = (output_size, output_size) if isinstance(lowercase , lowercase ) else output_size SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = get_image_size(lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = output_size # determine new height and width SCREAMING_SNAKE_CASE : Tuple = output_height / input_height SCREAMING_SNAKE_CASE : List[Any] = output_width / input_width if keep_aspect_ratio: # scale as little as possible if abs(1 - scale_width ) < abs(1 - scale_height ): # fit width SCREAMING_SNAKE_CASE : List[str] = scale_width else: # fit height SCREAMING_SNAKE_CASE : Optional[Any] = scale_height SCREAMING_SNAKE_CASE : int = constraint_to_multiple_of(scale_height * input_height , multiple=lowercase ) SCREAMING_SNAKE_CASE : Dict = constraint_to_multiple_of(scale_width * input_width , multiple=lowercase ) return (new_height, new_width) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Any = ['''pixel_values'''] def __init__( self : Any , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Dict[str, int] = None , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BILINEAR , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Union[int, float] = 1 / 255 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , **UpperCAmelCase_ : str , ): super().__init__(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = size if size is not None else {"height": 384, "width": 384} SCREAMING_SNAKE_CASE : List[Any] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = do_resize SCREAMING_SNAKE_CASE : Optional[Any] = size SCREAMING_SNAKE_CASE : str = keep_aspect_ratio SCREAMING_SNAKE_CASE : int = ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample SCREAMING_SNAKE_CASE : List[str] = do_rescale SCREAMING_SNAKE_CASE : Tuple = rescale_factor SCREAMING_SNAKE_CASE : Union[str, Any] = do_normalize SCREAMING_SNAKE_CASE : Union[str, Any] = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN SCREAMING_SNAKE_CASE : int = image_std if image_std is not None else IMAGENET_STANDARD_STD def _A ( self : Tuple , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Dict[str, int] , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : int , ): SCREAMING_SNAKE_CASE : List[str] = get_size_dict(UpperCAmelCase_ ) if "height" not in size or "width" not in size: raise ValueError(f'''The size dictionary must contain the keys \'height\' and \'width\'. Got {size.keys()}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = get_resize_output_image_size( UpperCAmelCase_ , output_size=(size["height"], size["width"]) , keep_aspect_ratio=UpperCAmelCase_ , multiple=UpperCAmelCase_ , ) return resize(UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[int, float] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : Dict , ): return rescale(UpperCAmelCase_ , scale=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Dict , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : List[str] , ): return normalize(UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : ImageInput , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : PILImageResampling = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : float = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[str, TensorType]] = None , UpperCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **UpperCAmelCase_ : Optional[Any] , ): SCREAMING_SNAKE_CASE : Optional[int] = do_resize if do_resize is not None else self.do_resize SCREAMING_SNAKE_CASE : Dict = size if size is not None else self.size SCREAMING_SNAKE_CASE : Optional[int] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = keep_aspect_ratio if keep_aspect_ratio is not None else self.keep_aspect_ratio SCREAMING_SNAKE_CASE : Optional[Any] = ensure_multiple_of if ensure_multiple_of is not None else self.ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample if resample is not None else self.resample SCREAMING_SNAKE_CASE : Dict = do_rescale if do_rescale is not None else self.do_rescale SCREAMING_SNAKE_CASE : Optional[Any] = rescale_factor if rescale_factor is not None else self.rescale_factor SCREAMING_SNAKE_CASE : Optional[int] = do_normalize if do_normalize is not None else self.do_normalize SCREAMING_SNAKE_CASE : int = image_mean if image_mean is not None else self.image_mean SCREAMING_SNAKE_CASE : Tuple = image_std if image_std is not None else self.image_std SCREAMING_SNAKE_CASE : List[str] = make_list_of_images(UpperCAmelCase_ ) if not valid_images(UpperCAmelCase_ ): raise ValueError( "Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, " "torch.Tensor, tf.Tensor or jax.ndarray." ) if do_resize and size is None or resample is None: raise ValueError("Size and resample must be specified if do_resize is True." ) if do_rescale and rescale_factor is None: raise ValueError("Rescale factor must be specified if do_rescale is True." ) if do_normalize and (image_mean is None or image_std is None): raise ValueError("Image mean and std must be specified if do_normalize is True." ) # All transformations expect numpy arrays. SCREAMING_SNAKE_CASE : Any = [to_numpy_array(UpperCAmelCase_ ) for image in images] if do_resize: SCREAMING_SNAKE_CASE : Tuple = [self.resize(image=UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ ) for image in images] if do_rescale: SCREAMING_SNAKE_CASE : Tuple = [self.rescale(image=UpperCAmelCase_ , scale=UpperCAmelCase_ ) for image in images] if do_normalize: SCREAMING_SNAKE_CASE : Optional[Any] = [self.normalize(image=UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = [to_channel_dimension_format(UpperCAmelCase_ , UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": images} return BatchFeature(data=UpperCAmelCase_ , tensor_type=UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Tuple] = None ): SCREAMING_SNAKE_CASE : int = outputs.logits # Resize logits and compute semantic segmentation maps if target_sizes is not None: if len(UpperCAmelCase_ ) != len(UpperCAmelCase_ ): raise ValueError( "Make sure that you pass in as many target sizes as the batch dimension of the logits" ) if is_torch_tensor(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = target_sizes.numpy() SCREAMING_SNAKE_CASE : str = [] for idx in range(len(UpperCAmelCase_ ) ): SCREAMING_SNAKE_CASE : Dict = torch.nn.functional.interpolate( logits[idx].unsqueeze(dim=0 ) , size=target_sizes[idx] , mode="bilinear" , align_corners=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = resized_logits[0].argmax(dim=0 ) semantic_segmentation.append(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = logits.argmax(dim=1 ) SCREAMING_SNAKE_CASE : str = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )] return semantic_segmentation
62
1
from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """s-JoL/Open-Llama-V1""": """https://huggingface.co/s-JoL/Open-Llama-V1/blob/main/config.json""", } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : List[str] = '''open-llama''' def __init__( self : Optional[Any] , UpperCAmelCase_ : Union[str, Any]=10_0000 , UpperCAmelCase_ : List[str]=4096 , UpperCAmelCase_ : Any=1_1008 , UpperCAmelCase_ : Optional[Any]=32 , UpperCAmelCase_ : List[Any]=32 , UpperCAmelCase_ : Dict="silu" , UpperCAmelCase_ : Dict=2048 , UpperCAmelCase_ : Optional[int]=0.02 , UpperCAmelCase_ : str=1E-6 , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=0 , UpperCAmelCase_ : Dict=1 , UpperCAmelCase_ : List[Any]=2 , UpperCAmelCase_ : Optional[Any]=False , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : Tuple=0.1 , UpperCAmelCase_ : int=True , UpperCAmelCase_ : List[str]=True , UpperCAmelCase_ : Any=None , **UpperCAmelCase_ : Optional[int] , ): SCREAMING_SNAKE_CASE : int = vocab_size SCREAMING_SNAKE_CASE : Any = max_position_embeddings SCREAMING_SNAKE_CASE : List[Any] = hidden_size SCREAMING_SNAKE_CASE : Union[str, Any] = intermediate_size SCREAMING_SNAKE_CASE : List[Any] = num_hidden_layers SCREAMING_SNAKE_CASE : Optional[int] = num_attention_heads SCREAMING_SNAKE_CASE : List[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[int] = initializer_range SCREAMING_SNAKE_CASE : Tuple = rms_norm_eps SCREAMING_SNAKE_CASE : Optional[int] = use_cache SCREAMING_SNAKE_CASE : Optional[int] = kwargs.pop( "use_memorry_efficient_attention" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = hidden_dropout_prob SCREAMING_SNAKE_CASE : Dict = attention_dropout_prob SCREAMING_SNAKE_CASE : Optional[Any] = use_stable_embedding SCREAMING_SNAKE_CASE : Optional[Any] = shared_input_output_embedding SCREAMING_SNAKE_CASE : Union[str, Any] = rope_scaling self._rope_scaling_validation() super().__init__( pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , tie_word_embeddings=UpperCAmelCase_ , **UpperCAmelCase_ , ) def _A ( self : List[str] ): if self.rope_scaling is None: return if not isinstance(self.rope_scaling , UpperCAmelCase_ ) or len(self.rope_scaling ) != 2: raise ValueError( "`rope_scaling` must be a dictionary with with two fields, `name` and `factor`, " f'''got {self.rope_scaling}''' ) SCREAMING_SNAKE_CASE : List[str] = self.rope_scaling.get("type" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.rope_scaling.get("factor" , UpperCAmelCase_ ) if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic"]: raise ValueError( f'''`rope_scaling`\'s name field must be one of [\'linear\', \'dynamic\'], got {rope_scaling_type}''' ) if rope_scaling_factor is None or not isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) or rope_scaling_factor <= 1.0: raise ValueError(f'''`rope_scaling`\'s factor field must be an float > 1, got {rope_scaling_factor}''' )
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : List[str] , UpperCAmelCase_ : int = 6 ): SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None self.create_linked_list(UpperCAmelCase_ ) def _A ( self : List[Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[int] = Node() SCREAMING_SNAKE_CASE : str = current_node SCREAMING_SNAKE_CASE : Optional[int] = current_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node for _ in range(1 , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Tuple = Node() SCREAMING_SNAKE_CASE : Dict = current_node SCREAMING_SNAKE_CASE : Optional[Any] = previous_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node SCREAMING_SNAKE_CASE : Union[str, Any] = self.front SCREAMING_SNAKE_CASE : List[str] = previous_node def _A ( self : Union[str, Any] ): return ( self.front == self.rear and self.front is not None and self.front.data is None ) def _A ( self : Optional[int] ): self.check_can_perform_operation() return self.front.data if self.front else None def _A ( self : Optional[int] , UpperCAmelCase_ : Any ): if self.rear is None: return self.check_is_full() if not self.is_empty(): SCREAMING_SNAKE_CASE : List[str] = self.rear.next if self.rear: SCREAMING_SNAKE_CASE : Dict = data def _A ( self : List[str] ): self.check_can_perform_operation() if self.rear is None or self.front is None: return None if self.front == self.rear: SCREAMING_SNAKE_CASE : List[str] = self.front.data SCREAMING_SNAKE_CASE : Optional[int] = None return data SCREAMING_SNAKE_CASE : List[str] = self.front SCREAMING_SNAKE_CASE : List[str] = old_front.next SCREAMING_SNAKE_CASE : Optional[int] = old_front.data SCREAMING_SNAKE_CASE : List[str] = None return data def _A ( self : Any ): if self.is_empty(): raise Exception("Empty Queue" ) def _A ( self : Optional[Any] ): if self.rear and self.rear.next == self.front: raise Exception("Full Queue" ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any | None = None SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None if __name__ == "__main__": import doctest doctest.testmod()
62
1
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments from transformers.testing_utils import TestCasePlus, require_torch, slow from transformers.utils import is_datasets_available if is_datasets_available(): import datasets class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @slow @require_torch def _A ( self : Dict ): SCREAMING_SNAKE_CASE : str = EncoderDecoderModel.from_encoder_decoder_pretrained("prajjwal1/bert-tiny" , "prajjwal1/bert-tiny" ) SCREAMING_SNAKE_CASE : List[Any] = BertTokenizer.from_pretrained("bert-base-uncased" ) SCREAMING_SNAKE_CASE : List[str] = bertabert.config.encoder.vocab_size SCREAMING_SNAKE_CASE : List[Any] = tokenizer.sep_token_id SCREAMING_SNAKE_CASE : List[Any] = tokenizer.cls_token_id SCREAMING_SNAKE_CASE : List[Any] = 128 SCREAMING_SNAKE_CASE : Dict = datasets.load_dataset("cnn_dailymail" , "3.0.0" , split="train[:1%]" ) SCREAMING_SNAKE_CASE : List[str] = datasets.load_dataset("cnn_dailymail" , "3.0.0" , split="validation[:1%]" ) SCREAMING_SNAKE_CASE : List[str] = train_dataset.select(range(32 ) ) SCREAMING_SNAKE_CASE : int = val_dataset.select(range(16 ) ) SCREAMING_SNAKE_CASE : Optional[Any] = 4 def _map_to_encoder_decoder_inputs(UpperCAmelCase_ : Optional[Any] ): # Tokenizer will automatically set [BOS] <text> [EOS] SCREAMING_SNAKE_CASE : Optional[int] = tokenizer(batch["article"] , padding="max_length" , truncation=UpperCAmelCase_ , max_length=512 ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer(batch["highlights"] , padding="max_length" , truncation=UpperCAmelCase_ , max_length=128 ) SCREAMING_SNAKE_CASE : List[str] = inputs.input_ids SCREAMING_SNAKE_CASE : List[Any] = inputs.attention_mask SCREAMING_SNAKE_CASE : Union[str, Any] = outputs.input_ids SCREAMING_SNAKE_CASE : List[str] = outputs.input_ids.copy() SCREAMING_SNAKE_CASE : List[Any] = [ [-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch["labels"] ] SCREAMING_SNAKE_CASE : str = outputs.attention_mask assert all(len(UpperCAmelCase_ ) == 512 for x in inputs.input_ids ) assert all(len(UpperCAmelCase_ ) == 128 for x in outputs.input_ids ) return batch def _compute_metrics(UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = pred.label_ids SCREAMING_SNAKE_CASE : Any = pred.predictions # all unnecessary tokens are removed SCREAMING_SNAKE_CASE : str = tokenizer.batch_decode(UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = tokenizer.batch_decode(UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = sum([int(pred_str[i] == label_str[i] ) for i in range(len(UpperCAmelCase_ ) )] ) / len(UpperCAmelCase_ ) return {"accuracy": accuracy} # map train dataset SCREAMING_SNAKE_CASE : Dict = train_dataset.map( _map_to_encoder_decoder_inputs , batched=UpperCAmelCase_ , batch_size=UpperCAmelCase_ , remove_columns=["article", "highlights"] , ) train_dataset.set_format( type="torch" , columns=["input_ids", "attention_mask", "decoder_input_ids", "decoder_attention_mask", "labels"] , ) # same for validation dataset SCREAMING_SNAKE_CASE : int = val_dataset.map( _map_to_encoder_decoder_inputs , batched=UpperCAmelCase_ , batch_size=UpperCAmelCase_ , remove_columns=["article", "highlights"] , ) val_dataset.set_format( type="torch" , columns=["input_ids", "attention_mask", "decoder_input_ids", "decoder_attention_mask", "labels"] , ) SCREAMING_SNAKE_CASE : Any = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : str = SeqaSeqTrainingArguments( output_dir=UpperCAmelCase_ , per_device_train_batch_size=UpperCAmelCase_ , per_device_eval_batch_size=UpperCAmelCase_ , predict_with_generate=UpperCAmelCase_ , evaluation_strategy="steps" , do_train=UpperCAmelCase_ , do_eval=UpperCAmelCase_ , warmup_steps=0 , eval_steps=2 , logging_steps=2 , ) # instantiate trainer SCREAMING_SNAKE_CASE : List[str] = SeqaSeqTrainer( model=UpperCAmelCase_ , args=UpperCAmelCase_ , compute_metrics=_compute_metrics , train_dataset=UpperCAmelCase_ , eval_dataset=UpperCAmelCase_ , tokenizer=UpperCAmelCase_ , ) # start training trainer.train()
62
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return 1 if input_a == input_a else 0 def lowerCamelCase__ ( ): """simple docstring""" assert xnor_gate(0 , 0 ) == 1 assert xnor_gate(0 , 1 ) == 0 assert xnor_gate(1 , 0 ) == 0 assert xnor_gate(1 , 1 ) == 1 if __name__ == "__main__": print(xnor_gate(0, 0)) print(xnor_gate(0, 1)) print(xnor_gate(1, 0)) print(xnor_gate(1, 1))
62
1
import logging import os import quant_trainer import torch from torch.utils.data import DataLoader from transformers import Trainer, is_torch_tpu_available from transformers.trainer_utils import PredictionOutput snake_case = logging.getLogger(__name__) if is_torch_tpu_available(check_device=False): import torch_xla.core.xla_model as xm import torch_xla.debug.metrics as met class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , *UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Tuple=None , **UpperCAmelCase_ : Optional[Any] ): super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = eval_examples SCREAMING_SNAKE_CASE : List[Any] = post_process_function SCREAMING_SNAKE_CASE : Any = quant_trainer_args SCREAMING_SNAKE_CASE : Optional[Any] = 128 # default number of calibration samples def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple=None ): if calib_dataset is None and self.calib_dataset is None: raise ValueError("Trainer: calibration requires an calib_dataset." ) SCREAMING_SNAKE_CASE : str = calib_dataset if calib_dataset is not None else self.calib_dataset SCREAMING_SNAKE_CASE : str = self._remove_unused_columns(UpperCAmelCase_ , description="Calibration" ) return DataLoader( UpperCAmelCase_ , batch_size=self.args.eval_batch_size , collate_fn=self.data_collator , drop_last=self.args.dataloader_drop_last , num_workers=self.args.dataloader_num_workers , pin_memory=self.args.dataloader_pin_memory , shuffle=UpperCAmelCase_ , ) def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[int]=None ): SCREAMING_SNAKE_CASE : Any = self.train_dataset if calib_dataset is None else calib_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_calib_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args , calib=UpperCAmelCase_ ) model.eval() quant_trainer.enable_calibration(UpperCAmelCase_ ) logger.info("***** Running calibration *****" ) logger.info(f''' Num examples = {self.calib_num}''' ) logger.info(f''' Batch size = {calib_dataloader.batch_size}''' ) for step, inputs in enumerate(UpperCAmelCase_ ): # Prediction step SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = self.prediction_step(UpperCAmelCase_ , UpperCAmelCase_ , prediction_loss_only=UpperCAmelCase_ ) if (step + 1) * calib_dataloader.batch_size >= self.calib_num: break quant_trainer.finish_calibration(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : Optional[int] = model def _A ( self : List[Any] , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : str = "eval" ): SCREAMING_SNAKE_CASE : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset SCREAMING_SNAKE_CASE : Tuple = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.eval_examples if eval_examples is None else eval_examples # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Dict = self.compute_metrics SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : int = eval_loop( UpperCAmelCase_ , description="Evaluation" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : int = compute_metrics if self.post_process_function is not None and self.compute_metrics is not None: SCREAMING_SNAKE_CASE : List[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions ) SCREAMING_SNAKE_CASE : Any = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : Union[str, Any] = metrics.pop(UpperCAmelCase_ ) self.log(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[Any] = {} if self.args.tpu_metrics_debug or self.args.debug: # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.) xm.master_print(met.metrics_report() ) SCREAMING_SNAKE_CASE : List[str] = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCAmelCase_ ) return metrics def _A ( self : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : str = "test" ): SCREAMING_SNAKE_CASE : str = self.get_test_dataloader(UpperCAmelCase_ ) # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Tuple = self.compute_metrics SCREAMING_SNAKE_CASE : Union[str, Any] = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : str = eval_loop( UpperCAmelCase_ , description="Prediction" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : Union[str, Any] = compute_metrics if self.post_process_function is None or self.compute_metrics is None: return output SCREAMING_SNAKE_CASE : Optional[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions , "predict" ) SCREAMING_SNAKE_CASE : str = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : str = metrics.pop(UpperCAmelCase_ ) return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : int="./" ): SCREAMING_SNAKE_CASE : List[Any] = self.eval_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = next(iter(UpperCAmelCase_ ) ) # saving device - to make it consistent SCREAMING_SNAKE_CASE : int = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) # convert to tuple SCREAMING_SNAKE_CASE : Tuple = tuple(v.to(UpperCAmelCase_ ) for k, v in batch.items() ) logger.info("Converting model to be onnx compatible" ) from pytorch_quantization.nn import TensorQuantizer SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Any = self.model.to(UpperCAmelCase_ ) model.eval() model.float() SCREAMING_SNAKE_CASE : str = model.module if hasattr(UpperCAmelCase_ , "module" ) else model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : List[Any] = os.path.join(UpperCAmelCase_ , "model.onnx" ) logger.info(f'''exporting model to {output_model_file}''' ) SCREAMING_SNAKE_CASE : int = {0: "batch_size", 1: "seq_len"} torch.onnx.export( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , export_params=UpperCAmelCase_ , opset_version=13 , do_constant_folding=UpperCAmelCase_ , input_names=["input_ids", "attention_mask", "token_type_ids"] , output_names=["output_start_logits", "output_end_logits"] , dynamic_axes={ "input_ids": axes, "attention_mask": axes, "token_type_ids": axes, "output_start_logits": axes, "output_end_logits": axes, } , verbose=UpperCAmelCase_ , ) logger.info("onnx export finished" )
62
import math import flax.linen as nn import jax.numpy as jnp def lowerCamelCase__ ( lowercase , lowercase , lowercase = 1 , lowercase = 1 , lowercase = 1.0E4 , lowercase = False , lowercase = 1.0 , ): """simple docstring""" assert timesteps.ndim == 1, "Timesteps should be a 1d-array" assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even''' SCREAMING_SNAKE_CASE : Union[str, Any] = float(embedding_dim // 2 ) SCREAMING_SNAKE_CASE : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift) SCREAMING_SNAKE_CASE : Optional[Any] = min_timescale * jnp.exp(jnp.arange(lowercase , dtype=jnp.floataa ) * -log_timescale_increment ) SCREAMING_SNAKE_CASE : Optional[int] = jnp.expand_dims(lowercase , 1 ) * jnp.expand_dims(lowercase , 0 ) # scale embeddings SCREAMING_SNAKE_CASE : Optional[int] = scale * emb if flip_sin_to_cos: SCREAMING_SNAKE_CASE : List[Any] = jnp.concatenate([jnp.cos(lowercase ), jnp.sin(lowercase )] , axis=1 ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = jnp.concatenate([jnp.sin(lowercase ), jnp.cos(lowercase )] , axis=1 ) SCREAMING_SNAKE_CASE : Tuple = jnp.reshape(lowercase , [jnp.shape(lowercase )[0], embedding_dim] ) return signal class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : jnp.dtype = jnp.floataa @nn.compact def __call__( self : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_1" )(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.silu(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_2" )(UpperCAmelCase_ ) return temb class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : bool = False UpperCamelCase_ : float = 1 @nn.compact def __call__( self : Optional[int] , UpperCAmelCase_ : int ): return get_sinusoidal_embeddings( UpperCAmelCase_ , embedding_dim=self.dim , flip_sin_to_cos=self.flip_sin_to_cos , freq_shift=self.freq_shift )
62
1
def lowerCamelCase__ ( lowercase ): """simple docstring""" if bit_count < 0: raise ValueError("The given input must be positive" ) # get the generated string sequence SCREAMING_SNAKE_CASE : Optional[int] = gray_code_sequence_string(lowercase ) # # convert them to integers for i in range(len(lowercase ) ): SCREAMING_SNAKE_CASE : Optional[int] = int(sequence[i] , 2 ) return sequence def lowerCamelCase__ ( lowercase ): """simple docstring""" if bit_count == 0: return ["0"] if bit_count == 1: return ["0", "1"] SCREAMING_SNAKE_CASE : Optional[int] = 1 << bit_count # defines the length of the sequence # 1<< n is equivalent to 2^n # recursive answer will generate answer for n-1 bits SCREAMING_SNAKE_CASE : str = gray_code_sequence_string(bit_count - 1 ) SCREAMING_SNAKE_CASE : Any = [] # append 0 to first half of the smaller sequence generated for i in range(seq_len // 2 ): SCREAMING_SNAKE_CASE : Tuple = "0" + smaller_sequence[i] sequence.append(lowercase ) # append 1 to second half ... start from the end of the list for i in reversed(range(seq_len // 2 ) ): SCREAMING_SNAKE_CASE : Dict = "1" + smaller_sequence[i] sequence.append(lowercase ) return sequence if __name__ == "__main__": import doctest doctest.testmod()
62
# Copyright 2023 The HuggingFace Inc. team. All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import numpy as np import torch from ..models.clipseg import CLIPSegForImageSegmentation from ..utils import is_vision_available, requires_backends from .base import PipelineTool if is_vision_available(): from PIL import Image class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = ( '''This is a tool that creates a segmentation mask of an image according to a label. It cannot create an image.''' '''It takes two arguments named `image` which should be the original image, and `label` which should be a text ''' '''describing the elements what should be identified in the segmentation mask. The tool returns the mask.''' ) UpperCamelCase_ : Union[str, Any] = '''CIDAS/clipseg-rd64-refined''' UpperCamelCase_ : Any = '''image_segmenter''' UpperCamelCase_ : int = CLIPSegForImageSegmentation UpperCamelCase_ : Optional[Any] = ['''image''', '''text'''] UpperCamelCase_ : int = ['''image'''] def __init__( self : str , *UpperCAmelCase_ : Union[str, Any] , **UpperCAmelCase_ : Union[str, Any] ): requires_backends(self , ["vision"] ) super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : "Image" , UpperCAmelCase_ : str ): return self.pre_processor(text=[label] , images=[image] , padding=UpperCAmelCase_ , return_tensors="pt" ) def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): with torch.no_grad(): SCREAMING_SNAKE_CASE : str = self.model(**UpperCAmelCase_ ).logits return logits def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = outputs.cpu().detach().numpy() SCREAMING_SNAKE_CASE : Optional[int] = 0 SCREAMING_SNAKE_CASE : str = 1 return Image.fromarray((array * 255).astype(np.uinta ) )
62
1
def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" global f # a global dp table for knapsack if f[i][j] < 0: if j < wt[i - 1]: SCREAMING_SNAKE_CASE : Optional[int] = mf_knapsack(i - 1 , lowercase , lowercase , lowercase ) else: SCREAMING_SNAKE_CASE : Optional[Any] = max( mf_knapsack(i - 1 , lowercase , lowercase , lowercase ) , mf_knapsack(i - 1 , lowercase , lowercase , j - wt[i - 1] ) + val[i - 1] , ) SCREAMING_SNAKE_CASE : Optional[Any] = val return f[i][j] def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = [[0] * (w + 1) for _ in range(n + 1 )] for i in range(1 , n + 1 ): for w_ in range(1 , w + 1 ): if wt[i - 1] <= w_: SCREAMING_SNAKE_CASE : Tuple = max(val[i - 1] + dp[i - 1][w_ - wt[i - 1]] , dp[i - 1][w_] ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = dp[i - 1][w_] return dp[n][w_], dp def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if not (isinstance(lowercase , (list, tuple) ) and isinstance(lowercase , (list, tuple) )): raise ValueError( "Both the weights and values vectors must be either lists or tuples" ) SCREAMING_SNAKE_CASE : Dict = len(lowercase ) if num_items != len(lowercase ): SCREAMING_SNAKE_CASE : Optional[int] = ( "The number of weights must be the same as the number of values.\n" F'''But got {num_items} weights and {len(lowercase )} values''' ) raise ValueError(lowercase ) for i in range(lowercase ): if not isinstance(wt[i] , lowercase ): SCREAMING_SNAKE_CASE : Optional[int] = ( "All weights must be integers but got weight of " F'''type {type(wt[i] )} at index {i}''' ) raise TypeError(lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = knapsack(lowercase , lowercase , lowercase , lowercase ) SCREAMING_SNAKE_CASE : set = set() _construct_solution(lowercase , lowercase , lowercase , lowercase , lowercase ) return optimal_val, example_optional_set def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase ): """simple docstring""" if i > 0 and j > 0: if dp[i - 1][j] == dp[i][j]: _construct_solution(lowercase , lowercase , i - 1 , lowercase , lowercase ) else: optimal_set.add(lowercase ) _construct_solution(lowercase , lowercase , i - 1 , j - wt[i - 1] , lowercase ) if __name__ == "__main__": snake_case = [3, 2, 4, 4] snake_case = [4, 3, 2, 3] snake_case = 4 snake_case = 6 snake_case = [[0] * (w + 1)] + [[0] + [-1] * (w + 1) for _ in range(n + 1)] snake_case , snake_case = knapsack(w, wt, val, n) print(optimal_solution) print(mf_knapsack(n, wt, val, w)) # switched the n and w # testing the dynamic programming problem with example # the optimal subset for the above example are items 3 and 4 snake_case , snake_case = knapsack_with_example_solution(w, wt, val) assert optimal_solution == 8 assert optimal_subset == {3, 4} print("""optimal_value = """, optimal_solution) print("""An optimal subset corresponding to the optimal value""", optimal_subset)
62
import json import os import unittest from transformers.models.blenderbot_small.tokenization_blenderbot_small import ( VOCAB_FILES_NAMES, BlenderbotSmallTokenizer, ) from ...test_tokenization_common import TokenizerTesterMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = BlenderbotSmallTokenizer UpperCamelCase_ : int = False def _A ( self : Union[str, Any] ): super().setUp() SCREAMING_SNAKE_CASE : List[Any] = ["__start__", "adapt", "act", "ap@@", "te", "__end__", "__unk__"] SCREAMING_SNAKE_CASE : Optional[Any] = dict(zip(UpperCAmelCase_ , range(len(UpperCAmelCase_ ) ) ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = ["#version: 0.2", "a p", "t e</w>", "ap t</w>", "a d", "ad apt</w>", "a c", "ac t</w>", ""] SCREAMING_SNAKE_CASE : int = {"unk_token": "__unk__", "bos_token": "__start__", "eos_token": "__end__"} SCREAMING_SNAKE_CASE : Union[str, Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["vocab_file"] ) SCREAMING_SNAKE_CASE : str = 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(UpperCAmelCase_ ) + "\n" ) with open(self.merges_file , "w" , encoding="utf-8" ) as fp: fp.write("\n".join(UpperCAmelCase_ ) ) def _A ( self : List[Any] , **UpperCAmelCase_ : str ): kwargs.update(self.special_tokens_map ) return BlenderbotSmallTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] , UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : int = "adapt act apte" return input_text, output_text def _A ( self : str ): SCREAMING_SNAKE_CASE : int = BlenderbotSmallTokenizer(self.vocab_file , self.merges_file , **self.special_tokens_map ) SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : List[str] = ["adapt", "act", "ap@@", "te"] SCREAMING_SNAKE_CASE : Optional[Any] = tokenizer.tokenize(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = [tokenizer.bos_token] + tokens + [tokenizer.eos_token] SCREAMING_SNAKE_CASE : Tuple = [0, 1, 2, 3, 4, 5] self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , UpperCAmelCase_ ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Union[str, Any] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) assert tok("sam" ).input_ids == [1384] SCREAMING_SNAKE_CASE : str = "I am a small frog." SCREAMING_SNAKE_CASE : List[Any] = tok([src_text] , padding=UpperCAmelCase_ , truncation=UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : int = tok.batch_decode(UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ , clean_up_tokenization_spaces=UpperCAmelCase_ )[0] assert src_text != decoded # I wish it did! assert decoded == "i am a small frog ." def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : List[str] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) SCREAMING_SNAKE_CASE : Tuple = "I am a small frog ." SCREAMING_SNAKE_CASE : Optional[int] = "." SCREAMING_SNAKE_CASE : Dict = tok(UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : Optional[Any] = tok(UpperCAmelCase_ )["input_ids"] assert encoded[-1] == encoded_dot[0]
62
1
from collections import OrderedDict from typing import Mapping from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """kssteven/ibert-roberta-base""": """https://huggingface.co/kssteven/ibert-roberta-base/resolve/main/config.json""", """kssteven/ibert-roberta-large""": """https://huggingface.co/kssteven/ibert-roberta-large/resolve/main/config.json""", """kssteven/ibert-roberta-large-mnli""": ( """https://huggingface.co/kssteven/ibert-roberta-large-mnli/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[int] = '''ibert''' def __init__( self : List[str] , UpperCAmelCase_ : List[Any]=3_0522 , UpperCAmelCase_ : Tuple=768 , UpperCAmelCase_ : str=12 , UpperCAmelCase_ : Dict=12 , UpperCAmelCase_ : int=3072 , UpperCAmelCase_ : Dict="gelu" , UpperCAmelCase_ : Tuple=0.1 , UpperCAmelCase_ : str=0.1 , UpperCAmelCase_ : List[Any]=512 , UpperCAmelCase_ : int=2 , UpperCAmelCase_ : Union[str, Any]=0.02 , UpperCAmelCase_ : List[Any]=1E-12 , UpperCAmelCase_ : Any=1 , UpperCAmelCase_ : Any=0 , UpperCAmelCase_ : Union[str, Any]=2 , UpperCAmelCase_ : List[str]="absolute" , UpperCAmelCase_ : Optional[int]=False , UpperCAmelCase_ : List[str]="none" , **UpperCAmelCase_ : Dict , ): super().__init__(pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = vocab_size SCREAMING_SNAKE_CASE : Dict = hidden_size SCREAMING_SNAKE_CASE : str = num_hidden_layers SCREAMING_SNAKE_CASE : Any = num_attention_heads SCREAMING_SNAKE_CASE : Any = hidden_act SCREAMING_SNAKE_CASE : Union[str, Any] = intermediate_size SCREAMING_SNAKE_CASE : Union[str, Any] = hidden_dropout_prob SCREAMING_SNAKE_CASE : Tuple = attention_probs_dropout_prob SCREAMING_SNAKE_CASE : Tuple = max_position_embeddings SCREAMING_SNAKE_CASE : Optional[int] = type_vocab_size SCREAMING_SNAKE_CASE : List[str] = initializer_range SCREAMING_SNAKE_CASE : Dict = layer_norm_eps SCREAMING_SNAKE_CASE : List[str] = position_embedding_type SCREAMING_SNAKE_CASE : Optional[int] = quant_mode SCREAMING_SNAKE_CASE : str = force_dequant class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @property def _A ( self : List[str] ): if self.task == "multiple-choice": SCREAMING_SNAKE_CASE : List[str] = {0: "batch", 1: "choice", 2: "sequence"} else: SCREAMING_SNAKE_CASE : Tuple = {0: "batch", 1: "sequence"} return OrderedDict( [ ("input_ids", dynamic_axis), ("attention_mask", dynamic_axis), ] )
62
import argparse from collections import OrderedDict from pathlib import Path import torch from transformers import ( VisualBertConfig, VisualBertForMultipleChoice, VisualBertForPreTraining, VisualBertForQuestionAnswering, VisualBertForVisualReasoning, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) snake_case = [ ("""bert.bert""", """visual_bert"""), ("""bert.cls""", """cls"""), ("""bert.classifier""", """cls"""), ("""token_type_embeddings_visual""", """visual_token_type_embeddings"""), ("""position_embeddings_visual""", """visual_position_embeddings"""), ("""projection""", """visual_projection"""), ] snake_case = [ """nlvr2_coco_pre_trained.th""", """nlvr2_fine_tuned.th""", """nlvr2_pre_trained.th""", """vcr_coco_pre_train.th""", """vcr_fine_tune.th""", """vcr_pre_train.th""", """vqa_coco_pre_trained.th""", """vqa_fine_tuned.th""", """vqa_pre_trained.th""", ] def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = torch.load(lowercase , map_location="cpu" ) return sd def lowerCamelCase__ ( lowercase , lowercase , lowercase=rename_keys_prefix ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = OrderedDict() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.arange(config.max_position_embeddings ).expand((1, -1) ) # detector_d = OrderedDict() for key in d: if "detector" in key: # detector_d[key.replace('detector.','')] = d[key] continue SCREAMING_SNAKE_CASE : Union[str, Any] = key for name_pair in rename_keys_prefix: SCREAMING_SNAKE_CASE : Any = new_key.replace(name_pair[0] , name_pair[1] ) SCREAMING_SNAKE_CASE : Dict = d[key] if key == "bert.cls.predictions.decoder.weight": # Old bert code didn't have `decoder.bias`, but was added separately SCREAMING_SNAKE_CASE : Union[str, Any] = new_d["cls.predictions.bias"] return new_d @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( checkpoint_path.split("/" )[-1] in ACCEPTABLE_CHECKPOINTS ), F'''The checkpoint provided must be in {ACCEPTABLE_CHECKPOINTS}.''' # Get Config if "pre" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = "pretraining" if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = {"visual_embedding_dim": 512} elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = {"visual_embedding_dim": 2048} elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Any = {"visual_embedding_dim": 2048} elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 1024} else: raise NotImplementedError(F'''No implementation found for `{checkpoint_path}`.''' ) else: if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 512} SCREAMING_SNAKE_CASE : Tuple = "multichoice" elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[str] = {"visual_embedding_dim": 2048} SCREAMING_SNAKE_CASE : str = "vqa_advanced" elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Tuple = {"visual_embedding_dim": 2048, "num_labels": 3129} SCREAMING_SNAKE_CASE : Optional[Any] = "vqa" elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = { "visual_embedding_dim": 1024, "num_labels": 2, } SCREAMING_SNAKE_CASE : Tuple = "nlvr" SCREAMING_SNAKE_CASE : List[Any] = VisualBertConfig(**lowercase ) # Load State Dict SCREAMING_SNAKE_CASE : List[str] = load_state_dict(lowercase ) SCREAMING_SNAKE_CASE : Any = get_new_dict(lowercase , lowercase ) if model_type == "pretraining": SCREAMING_SNAKE_CASE : int = VisualBertForPreTraining(lowercase ) elif model_type == "vqa": SCREAMING_SNAKE_CASE : Tuple = VisualBertForQuestionAnswering(lowercase ) elif model_type == "nlvr": SCREAMING_SNAKE_CASE : List[Any] = VisualBertForVisualReasoning(lowercase ) elif model_type == "multichoice": SCREAMING_SNAKE_CASE : Optional[Any] = VisualBertForMultipleChoice(lowercase ) model.load_state_dict(lowercase ) # Save Checkpoints Path(lowercase ).mkdir(exist_ok=lowercase ) model.save_pretrained(lowercase ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument("""orig_checkpoint_path""", type=str, help="""A path to .th on local filesystem.""") parser.add_argument("""pytorch_dump_folder_path""", type=str, help="""Path to the output PyTorch model.""") snake_case = parser.parse_args() convert_visual_bert_checkpoint(args.orig_checkpoint_path, args.pytorch_dump_folder_path)
62
1
from typing import Dict, List from nltk.translate import gleu_score import datasets from datasets import MetricInfo snake_case = """\ @misc{wu2016googles, title={Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation}, author={Yonghui Wu and Mike Schuster and Zhifeng Chen and Quoc V. Le and Mohammad Norouzi and Wolfgang Macherey and Maxim Krikun and Yuan Cao and Qin Gao and Klaus Macherey and Jeff Klingner and Apurva Shah and Melvin Johnson and Xiaobing Liu and Łukasz Kaiser and Stephan Gouws and Yoshikiyo Kato and Taku Kudo and Hideto Kazawa and Keith Stevens and George Kurian and Nishant Patil and Wei Wang and Cliff Young and Jason Smith and Jason Riesa and Alex Rudnick and Oriol Vinyals and Greg Corrado and Macduff Hughes and Jeffrey Dean}, year={2016}, eprint={1609.08144}, archivePrefix={arXiv}, primaryClass={cs.CL} } """ snake_case = """\ The BLEU score has some undesirable properties when used for single sentences, as it was designed to be a corpus measure. We therefore use a slightly different score for our RL experiments which we call the 'GLEU score'. For the GLEU score, we record all sub-sequences of 1, 2, 3 or 4 tokens in output and target sequence (n-grams). We then compute a recall, which is the ratio of the number of matching n-grams to the number of total n-grams in the target (ground truth) sequence, and a precision, which is the ratio of the number of matching n-grams to the number of total n-grams in the generated output sequence. Then GLEU score is simply the minimum of recall and precision. This GLEU score's range is always between 0 (no matches) and 1 (all match) and it is symmetrical when switching output and target. According to our experiments, GLEU score correlates quite well with the BLEU metric on a corpus level but does not have its drawbacks for our per sentence reward objective. """ snake_case = """\ Computes corpus-level Google BLEU (GLEU) score of translated segments against one or more references. Instead of averaging the sentence level GLEU scores (i.e. macro-average precision), Wu et al. (2016) sum up the matching tokens and the max of hypothesis and reference tokens for each sentence, then compute using the aggregate values. Args: predictions (list of str): list of translations to score. Each translation should be tokenized into a list of tokens. references (list of list of str): list of lists of references for each translation. Each reference should be tokenized into a list of tokens. min_len (int): The minimum order of n-gram this function should extract. Defaults to 1. max_len (int): The maximum order of n-gram this function should extract. Defaults to 4. Returns: 'google_bleu': google_bleu score Examples: Example 1: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.44 Example 2: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.61 Example 3: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references, min_len=2) >>> print(round(results[\"google_bleu\"], 2)) 0.53 Example 4: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses,references=list_of_references, min_len=2, max_len=6) >>> print(round(results[\"google_bleu\"], 2)) 0.4 """ @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : Optional[Any] ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { "predictions": datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ), "references": datasets.Sequence( datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ) , id="references" ), } ) , ) def _A ( self : List[Any] , UpperCAmelCase_ : List[List[List[str]]] , UpperCAmelCase_ : List[List[str]] , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : int = 4 , ): return { "google_bleu": gleu_score.corpus_gleu( list_of_references=UpperCAmelCase_ , hypotheses=UpperCAmelCase_ , min_len=UpperCAmelCase_ , max_len=UpperCAmelCase_ ) }
62
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available snake_case = { """configuration_jukebox""": [ """JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP""", """JukeboxConfig""", """JukeboxPriorConfig""", """JukeboxVQVAEConfig""", ], """tokenization_jukebox""": ["""JukeboxTokenizer"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST""", """JukeboxModel""", """JukeboxPreTrainedModel""", """JukeboxVQVAE""", """JukeboxPrior""", ] if TYPE_CHECKING: from .configuration_jukebox import ( JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP, JukeboxConfig, JukeboxPriorConfig, JukeboxVQVAEConfig, ) from .tokenization_jukebox import JukeboxTokenizer try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_jukebox import ( JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST, JukeboxModel, JukeboxPreTrainedModel, JukeboxPrior, JukeboxVQVAE, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
1
import datetime import platform import subprocess from typing import Optional, Tuple, Union import numpy as np def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : List[Any] = F'''{sampling_rate}''' SCREAMING_SNAKE_CASE : Any = "1" SCREAMING_SNAKE_CASE : str = "f32le" SCREAMING_SNAKE_CASE : Any = [ "ffmpeg", "-i", "pipe:0", "-ac", ac, "-ar", ar, "-f", format_for_conversion, "-hide_banner", "-loglevel", "quiet", "pipe:1", ] try: with subprocess.Popen(lowercase , stdin=subprocess.PIPE , stdout=subprocess.PIPE ) as ffmpeg_process: SCREAMING_SNAKE_CASE : List[Any] = ffmpeg_process.communicate(lowercase ) except FileNotFoundError as error: raise ValueError("ffmpeg was not found but is required to load audio files from filename" ) from error SCREAMING_SNAKE_CASE : Dict = output_stream[0] SCREAMING_SNAKE_CASE : List[Any] = np.frombuffer(lowercase , np.floataa ) if audio.shape[0] == 0: raise ValueError("Malformed soundfile" ) return audio def lowerCamelCase__ ( lowercase , lowercase , lowercase = "f32le" , ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = F'''{sampling_rate}''' SCREAMING_SNAKE_CASE : List[str] = "1" if format_for_conversion == "s16le": SCREAMING_SNAKE_CASE : int = 2 elif format_for_conversion == "f32le": SCREAMING_SNAKE_CASE : int = 4 else: raise ValueError(F'''Unhandled format `{format_for_conversion}`. Please use `s16le` or `f32le`''' ) SCREAMING_SNAKE_CASE : int = platform.system() if system == "Linux": SCREAMING_SNAKE_CASE : Dict = "alsa" SCREAMING_SNAKE_CASE : List[str] = "default" elif system == "Darwin": SCREAMING_SNAKE_CASE : List[Any] = "avfoundation" SCREAMING_SNAKE_CASE : Tuple = ":0" elif system == "Windows": SCREAMING_SNAKE_CASE : Optional[Any] = "dshow" SCREAMING_SNAKE_CASE : Dict = "default" SCREAMING_SNAKE_CASE : Dict = [ "ffmpeg", "-f", format_, "-i", input_, "-ac", ac, "-ar", ar, "-f", format_for_conversion, "-fflags", "nobuffer", "-hide_banner", "-loglevel", "quiet", "pipe:1", ] SCREAMING_SNAKE_CASE : List[str] = int(round(sampling_rate * chunk_length_s ) ) * size_of_sample SCREAMING_SNAKE_CASE : Any = _ffmpeg_stream(lowercase , lowercase ) for item in iterator: yield item def lowerCamelCase__ ( lowercase , lowercase , lowercase = None , lowercase = None , lowercase = "f32le" , ): """simple docstring""" if stream_chunk_s is not None: SCREAMING_SNAKE_CASE : List[Any] = stream_chunk_s else: SCREAMING_SNAKE_CASE : Dict = chunk_length_s SCREAMING_SNAKE_CASE : Tuple = ffmpeg_microphone(lowercase , lowercase , format_for_conversion=lowercase ) if format_for_conversion == "s16le": SCREAMING_SNAKE_CASE : Union[str, Any] = np.intaa SCREAMING_SNAKE_CASE : str = 2 elif format_for_conversion == "f32le": SCREAMING_SNAKE_CASE : Optional[int] = np.floataa SCREAMING_SNAKE_CASE : Any = 4 else: raise ValueError(F'''Unhandled format `{format_for_conversion}`. Please use `s16le` or `f32le`''' ) if stride_length_s is None: SCREAMING_SNAKE_CASE : List[str] = chunk_length_s / 6 SCREAMING_SNAKE_CASE : str = int(round(sampling_rate * chunk_length_s ) ) * size_of_sample if isinstance(lowercase , (int, float) ): SCREAMING_SNAKE_CASE : Dict = [stride_length_s, stride_length_s] SCREAMING_SNAKE_CASE : Union[str, Any] = int(round(sampling_rate * stride_length_s[0] ) ) * size_of_sample SCREAMING_SNAKE_CASE : List[Any] = int(round(sampling_rate * stride_length_s[1] ) ) * size_of_sample SCREAMING_SNAKE_CASE : int = datetime.datetime.now() SCREAMING_SNAKE_CASE : Dict = datetime.timedelta(seconds=lowercase ) for item in chunk_bytes_iter(lowercase , lowercase , stride=(stride_left, stride_right) , stream=lowercase ): # Put everything back in numpy scale SCREAMING_SNAKE_CASE : int = np.frombuffer(item["raw"] , dtype=lowercase ) SCREAMING_SNAKE_CASE : Dict = ( item["stride"][0] // size_of_sample, item["stride"][1] // size_of_sample, ) SCREAMING_SNAKE_CASE : Optional[Any] = sampling_rate audio_time += delta if datetime.datetime.now() > audio_time + 10 * delta: # We're late !! SKIP continue yield item def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase = False ): """simple docstring""" SCREAMING_SNAKE_CASE : int = B"" SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = stride if stride_left + stride_right >= chunk_len: raise ValueError( F'''Stride needs to be strictly smaller than chunk_len: ({stride_left}, {stride_right}) vs {chunk_len}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = 0 for raw in iterator: acc += raw if stream and len(lowercase ) < chunk_len: SCREAMING_SNAKE_CASE : Any = (_stride_left, 0) yield {"raw": acc[:chunk_len], "stride": stride, "partial": True} else: while len(lowercase ) >= chunk_len: # We are flushing the accumulator SCREAMING_SNAKE_CASE : int = (_stride_left, stride_right) SCREAMING_SNAKE_CASE : Any = {"raw": acc[:chunk_len], "stride": stride} if stream: SCREAMING_SNAKE_CASE : List[Any] = False yield item SCREAMING_SNAKE_CASE : int = stride_left SCREAMING_SNAKE_CASE : str = acc[chunk_len - stride_left - stride_right :] # Last chunk if len(lowercase ) > stride_left: SCREAMING_SNAKE_CASE : Union[str, Any] = {"raw": acc, "stride": (_stride_left, 0)} if stream: SCREAMING_SNAKE_CASE : Union[str, Any] = False yield item def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = 2**24 # 16Mo try: with subprocess.Popen(lowercase , stdout=subprocess.PIPE , bufsize=lowercase ) as ffmpeg_process: while True: SCREAMING_SNAKE_CASE : Optional[int] = ffmpeg_process.stdout.read(lowercase ) if raw == b"": break yield raw except FileNotFoundError as error: raise ValueError("ffmpeg was not found but is required to stream audio files from filename" ) from error
62
from typing import Dict, List from nltk.translate import gleu_score import datasets from datasets import MetricInfo snake_case = """\ @misc{wu2016googles, title={Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation}, author={Yonghui Wu and Mike Schuster and Zhifeng Chen and Quoc V. Le and Mohammad Norouzi and Wolfgang Macherey and Maxim Krikun and Yuan Cao and Qin Gao and Klaus Macherey and Jeff Klingner and Apurva Shah and Melvin Johnson and Xiaobing Liu and Łukasz Kaiser and Stephan Gouws and Yoshikiyo Kato and Taku Kudo and Hideto Kazawa and Keith Stevens and George Kurian and Nishant Patil and Wei Wang and Cliff Young and Jason Smith and Jason Riesa and Alex Rudnick and Oriol Vinyals and Greg Corrado and Macduff Hughes and Jeffrey Dean}, year={2016}, eprint={1609.08144}, archivePrefix={arXiv}, primaryClass={cs.CL} } """ snake_case = """\ The BLEU score has some undesirable properties when used for single sentences, as it was designed to be a corpus measure. We therefore use a slightly different score for our RL experiments which we call the 'GLEU score'. For the GLEU score, we record all sub-sequences of 1, 2, 3 or 4 tokens in output and target sequence (n-grams). We then compute a recall, which is the ratio of the number of matching n-grams to the number of total n-grams in the target (ground truth) sequence, and a precision, which is the ratio of the number of matching n-grams to the number of total n-grams in the generated output sequence. Then GLEU score is simply the minimum of recall and precision. This GLEU score's range is always between 0 (no matches) and 1 (all match) and it is symmetrical when switching output and target. According to our experiments, GLEU score correlates quite well with the BLEU metric on a corpus level but does not have its drawbacks for our per sentence reward objective. """ snake_case = """\ Computes corpus-level Google BLEU (GLEU) score of translated segments against one or more references. Instead of averaging the sentence level GLEU scores (i.e. macro-average precision), Wu et al. (2016) sum up the matching tokens and the max of hypothesis and reference tokens for each sentence, then compute using the aggregate values. Args: predictions (list of str): list of translations to score. Each translation should be tokenized into a list of tokens. references (list of list of str): list of lists of references for each translation. Each reference should be tokenized into a list of tokens. min_len (int): The minimum order of n-gram this function should extract. Defaults to 1. max_len (int): The maximum order of n-gram this function should extract. Defaults to 4. Returns: 'google_bleu': google_bleu score Examples: Example 1: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.44 Example 2: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.61 Example 3: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references, min_len=2) >>> print(round(results[\"google_bleu\"], 2)) 0.53 Example 4: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses,references=list_of_references, min_len=2, max_len=6) >>> print(round(results[\"google_bleu\"], 2)) 0.4 """ @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : Optional[Any] ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { "predictions": datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ), "references": datasets.Sequence( datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ) , id="references" ), } ) , ) def _A ( self : List[Any] , UpperCAmelCase_ : List[List[List[str]]] , UpperCAmelCase_ : List[List[str]] , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : int = 4 , ): return { "google_bleu": gleu_score.corpus_gleu( list_of_references=UpperCAmelCase_ , hypotheses=UpperCAmelCase_ , min_len=UpperCAmelCase_ , max_len=UpperCAmelCase_ ) }
62
1
import argparse import logging import os from pathlib import Path from typing import Any, Dict import pytorch_lightning as pl from pytorch_lightning.utilities import rank_zero_info from transformers import ( AdamW, AutoConfig, AutoModel, AutoModelForPreTraining, AutoModelForQuestionAnswering, AutoModelForSeqaSeqLM, AutoModelForSequenceClassification, AutoModelForTokenClassification, AutoModelWithLMHead, AutoTokenizer, PretrainedConfig, PreTrainedTokenizer, ) from transformers.optimization import ( Adafactor, get_cosine_schedule_with_warmup, get_cosine_with_hard_restarts_schedule_with_warmup, get_linear_schedule_with_warmup, get_polynomial_decay_schedule_with_warmup, ) from transformers.utils.versions import require_version snake_case = logging.getLogger(__name__) require_version("""pytorch_lightning>=1.0.4""") snake_case = { """base""": AutoModel, """sequence-classification""": AutoModelForSequenceClassification, """question-answering""": AutoModelForQuestionAnswering, """pretraining""": AutoModelForPreTraining, """token-classification""": AutoModelForTokenClassification, """language-modeling""": AutoModelWithLMHead, """summarization""": AutoModelForSeqaSeqLM, """translation""": AutoModelForSeqaSeqLM, } # update this and the import above to support new schedulers from transformers.optimization snake_case = { """linear""": get_linear_schedule_with_warmup, """cosine""": get_cosine_schedule_with_warmup, """cosine_w_restarts""": get_cosine_with_hard_restarts_schedule_with_warmup, """polynomial""": get_polynomial_decay_schedule_with_warmup, # '': get_constant_schedule, # not supported for now # '': get_constant_schedule_with_warmup, # not supported for now } snake_case = sorted(arg_to_scheduler.keys()) snake_case = """{""" + """, """.join(arg_to_scheduler_choices) + """}""" class SCREAMING_SNAKE_CASE ( pl.LightningModule ): '''simple docstring''' def __init__( self : Any , UpperCAmelCase_ : argparse.Namespace , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Optional[int]="base" , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , **UpperCAmelCase_ : List[Any] , ): super().__init__() # TODO: move to self.save_hyperparameters() # self.save_hyperparameters() # can also expand arguments into trainer signature for easier reading self.save_hyperparameters(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = 0 SCREAMING_SNAKE_CASE : List[str] = Path(self.hparams.output_dir ) SCREAMING_SNAKE_CASE : Optional[Any] = self.hparams.cache_dir if self.hparams.cache_dir else None if config is None: SCREAMING_SNAKE_CASE : Dict = AutoConfig.from_pretrained( self.hparams.config_name if self.hparams.config_name else self.hparams.model_name_or_path , **({"num_labels": num_labels} if num_labels is not None else {}) , cache_dir=UpperCAmelCase_ , **UpperCAmelCase_ , ) else: SCREAMING_SNAKE_CASE : PretrainedConfig = config SCREAMING_SNAKE_CASE : Optional[int] = ("encoder_layerdrop", "decoder_layerdrop", "dropout", "attention_dropout") for p in extra_model_params: if getattr(self.hparams , UpperCAmelCase_ , UpperCAmelCase_ ): assert hasattr(self.config , UpperCAmelCase_ ), f'''model config doesn\'t have a `{p}` attribute''' setattr(self.config , UpperCAmelCase_ , getattr(self.hparams , UpperCAmelCase_ ) ) if tokenizer is None: SCREAMING_SNAKE_CASE : Any = AutoTokenizer.from_pretrained( self.hparams.tokenizer_name if self.hparams.tokenizer_name else self.hparams.model_name_or_path , cache_dir=UpperCAmelCase_ , ) else: SCREAMING_SNAKE_CASE : PreTrainedTokenizer = tokenizer SCREAMING_SNAKE_CASE : Union[str, Any] = MODEL_MODES[mode] if model is None: SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_type.from_pretrained( self.hparams.model_name_or_path , from_tf=bool(".ckpt" in self.hparams.model_name_or_path ) , config=self.config , cache_dir=UpperCAmelCase_ , ) else: SCREAMING_SNAKE_CASE : int = model def _A ( self : Optional[Any] , *UpperCAmelCase_ : Optional[Any] , **UpperCAmelCase_ : List[Any] ): SCREAMING_SNAKE_CASE : str = self.model_type.from_pretrained(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Any = arg_to_scheduler[self.hparams.lr_scheduler] SCREAMING_SNAKE_CASE : str = get_schedule_func( self.opt , num_warmup_steps=self.hparams.warmup_steps , num_training_steps=self.total_steps() ) SCREAMING_SNAKE_CASE : str = {"scheduler": scheduler, "interval": "step", "frequency": 1} return scheduler def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.model SCREAMING_SNAKE_CASE : Optional[int] = ["bias", "LayerNorm.weight"] SCREAMING_SNAKE_CASE : List[Any] = [ { "params": [ p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay ) ], # check this named paramters "weight_decay": self.hparams.weight_decay, }, { "params": [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay )], "weight_decay": 0.0, }, ] if self.hparams.adafactor: SCREAMING_SNAKE_CASE : Optional[int] = Adafactor( UpperCAmelCase_ , lr=self.hparams.learning_rate , scale_parameter=UpperCAmelCase_ , relative_step=UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = AdamW( UpperCAmelCase_ , lr=self.hparams.learning_rate , eps=self.hparams.adam_epsilon ) SCREAMING_SNAKE_CASE : Optional[int] = optimizer SCREAMING_SNAKE_CASE : Any = self.get_lr_scheduler() return [optimizer], [scheduler] def _A ( self : Any , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : str ): return self.validation_step(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : int ): return self.validation_end(UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = max(1 , self.hparams.gpus ) # TODO: consider num_tpu_cores SCREAMING_SNAKE_CASE : int = self.hparams.train_batch_size * self.hparams.accumulate_grad_batches * num_devices return (self.dataset_size / effective_batch_size) * self.hparams.max_epochs def _A ( self : int , UpperCAmelCase_ : List[str] ): if stage == "test": SCREAMING_SNAKE_CASE : Any = len(self.test_dataloader().dataset ) else: SCREAMING_SNAKE_CASE : Any = self.get_dataloader("train" , self.hparams.train_batch_size , shuffle=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = len(self.train_dataloader().dataset ) def _A ( self : Any , UpperCAmelCase_ : str , UpperCAmelCase_ : int , UpperCAmelCase_ : bool = False ): raise NotImplementedError("You must implement this for your task" ) def _A ( self : List[str] ): return self.train_loader def _A ( self : List[Any] ): return self.get_dataloader("dev" , self.hparams.eval_batch_size , shuffle=UpperCAmelCase_ ) def _A ( self : Union[str, Any] ): return self.get_dataloader("test" , self.hparams.eval_batch_size , shuffle=UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : List[str] ): return os.path.join( self.hparams.data_dir , "cached_{}_{}_{}".format( UpperCAmelCase_ , list(filter(UpperCAmelCase_ , self.hparams.model_name_or_path.split("/" ) ) ).pop() , str(self.hparams.max_seq_length ) , ) , ) @pl.utilities.rank_zero_only def _A ( self : List[str] , UpperCAmelCase_ : Dict[str, Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.output_dir.joinpath("best_tfmr" ) SCREAMING_SNAKE_CASE : Optional[int] = self.step_count self.model.save_pretrained(UpperCAmelCase_ ) self.tokenizer.save_pretrained(UpperCAmelCase_ ) @staticmethod def _A ( UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Dict ): parser.add_argument( "--model_name_or_path" , default=UpperCAmelCase_ , type=UpperCAmelCase_ , required=UpperCAmelCase_ , help="Path to pretrained model or model identifier from huggingface.co/models" , ) parser.add_argument( "--config_name" , default="" , type=UpperCAmelCase_ , help="Pretrained config name or path if not the same as model_name" ) parser.add_argument( "--tokenizer_name" , default=UpperCAmelCase_ , type=UpperCAmelCase_ , help="Pretrained tokenizer name or path if not the same as model_name" , ) parser.add_argument( "--cache_dir" , default=str(Path(UpperCAmelCase_ ).parent / "test_run" / "cache" ) , type=UpperCAmelCase_ , help="Where do you want to store the pre-trained models downloaded from huggingface.co" , ) parser.add_argument( "--encoder_layerdrop" , type=UpperCAmelCase_ , help="Encoder layer dropout probability (Optional). Goes into model.config" , ) parser.add_argument( "--decoder_layerdrop" , type=UpperCAmelCase_ , help="Decoder layer dropout probability (Optional). Goes into model.config" , ) parser.add_argument( "--dropout" , type=UpperCAmelCase_ , help="Dropout probability (Optional). Goes into model.config" , ) parser.add_argument( "--attention_dropout" , type=UpperCAmelCase_ , help="Attention dropout probability (Optional). Goes into model.config" , ) parser.add_argument("--learning_rate" , default=5E-5 , type=UpperCAmelCase_ , help="The initial learning rate for Adam." ) parser.add_argument( "--lr_scheduler" , default="linear" , choices=UpperCAmelCase_ , metavar=UpperCAmelCase_ , type=UpperCAmelCase_ , help="Learning rate scheduler" , ) parser.add_argument("--weight_decay" , default=0.0 , type=UpperCAmelCase_ , help="Weight decay if we apply some." ) parser.add_argument("--adam_epsilon" , default=1E-8 , type=UpperCAmelCase_ , help="Epsilon for Adam optimizer." ) parser.add_argument("--warmup_steps" , default=0 , type=UpperCAmelCase_ , help="Linear warmup over warmup_steps." ) parser.add_argument("--num_workers" , default=4 , type=UpperCAmelCase_ , help="kwarg passed to DataLoader" ) parser.add_argument("--num_train_epochs" , dest="max_epochs" , default=3 , type=UpperCAmelCase_ ) parser.add_argument("--train_batch_size" , default=32 , type=UpperCAmelCase_ ) parser.add_argument("--eval_batch_size" , default=32 , type=UpperCAmelCase_ ) parser.add_argument("--adafactor" , action="store_true" ) class SCREAMING_SNAKE_CASE ( pl.Callback ): '''simple docstring''' def _A ( self : str , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[int] ): if ( trainer.is_global_zero and trainer.global_rank == 0 ): # we initialize the retriever only on master worker with RAY. In new pytorch-lightning accelorators are removed. pl_module.model.rag.retriever.init_retrieval() # better to use hook functions. class SCREAMING_SNAKE_CASE ( pl.Callback ): '''simple docstring''' def _A ( self : int , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str ): # print(pl_module.model.rag) for name, param in pl_module.model.rag.named_parameters(): if param.grad is None: print(UpperCAmelCase_ ) class SCREAMING_SNAKE_CASE ( pl.Callback ): '''simple docstring''' def _A ( self : Tuple , UpperCAmelCase_ : int , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = trainer.lr_schedulers[0]["scheduler"] SCREAMING_SNAKE_CASE : Optional[Any] = {f'''lr_group_{i}''': lr for i, lr in enumerate(lr_scheduler.get_lr() )} pl_module.logger.log_metrics(UpperCAmelCase_ ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : pl.Trainer , UpperCAmelCase_ : pl.LightningModule ): rank_zero_info("***** Validation results *****" ) SCREAMING_SNAKE_CASE : Optional[Any] = trainer.callback_metrics # Log results for key in sorted(UpperCAmelCase_ ): if key not in ["log", "progress_bar"]: rank_zero_info("{} = {}\n".format(UpperCAmelCase_ , str(metrics[key] ) ) ) def _A ( self : Dict , UpperCAmelCase_ : pl.Trainer , UpperCAmelCase_ : pl.LightningModule ): rank_zero_info("***** Test results *****" ) SCREAMING_SNAKE_CASE : List[str] = trainer.callback_metrics # Log and save results to file SCREAMING_SNAKE_CASE : Optional[int] = os.path.join(pl_module.hparams.output_dir , "test_results.txt" ) with open(UpperCAmelCase_ , "w" ) as writer: for key in sorted(UpperCAmelCase_ ): if key not in ["log", "progress_bar"]: rank_zero_info("{} = {}\n".format(UpperCAmelCase_ , str(metrics[key] ) ) ) writer.write("{} = {}\n".format(UpperCAmelCase_ , str(metrics[key] ) ) ) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" parser.add_argument( "--output_dir" , default=str(Path(lowercase ).parent / "test_run" / "model_checkpoints" ) , type=lowercase , help="The output directory where the model predictions and checkpoints will be written." , ) parser.add_argument( "--fp16" , action="store_true" , help="Whether to use 16-bit (mixed) precision (through NVIDIA apex) instead of 32-bit" , ) parser.add_argument( "--fp16_opt_level" , type=lowercase , default="O2" , help=( "For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']." "See details at https://nvidia.github.io/apex/amp.html" ) , ) parser.add_argument("--n_tpu_cores" , dest="tpu_cores" , type=lowercase ) parser.add_argument("--max_grad_norm" , dest="gradient_clip_val" , default=1.0 , type=lowercase , help="Max gradient norm" ) parser.add_argument("--do_train" , action="store_true" , help="Whether to run training." ) parser.add_argument("--do_predict" , action="store_true" , help="Whether to run predictions on the test set." ) parser.add_argument( "--gradient_accumulation_steps" , dest="accumulate_grad_batches" , type=lowercase , default=1 , help="Number of updates steps to accumulate before performing a backward/update pass." , ) parser.add_argument("--seed" , type=lowercase , default=42 , help="random seed for initialization" ) parser.add_argument( "--data_dir" , default=str(Path(lowercase ).parent / "test_run" / "dummy-train-data" ) , type=lowercase , help="The input data dir. Should contain the training files for the CoNLL-2003 NER task." , ) def lowerCamelCase__ ( lowercase , lowercase , lowercase=None , lowercase=True , lowercase=[] , lowercase=None , lowercase=None , **lowercase , ): """simple docstring""" pl.seed_everything(args.seed ) # init model SCREAMING_SNAKE_CASE : Optional[int] = Path(model.hparams.output_dir ) odir.mkdir(exist_ok=lowercase ) # add custom checkpoints if checkpoint_callback is None: SCREAMING_SNAKE_CASE : Optional[int] = pl.callbacks.ModelCheckpoint( filepath=args.output_dir , prefix="checkpoint" , monitor="val_loss" , mode="min" , save_top_k=1 ) if early_stopping_callback: extra_callbacks.append(lowercase ) if logging_callback is None: SCREAMING_SNAKE_CASE : Optional[Any] = LoggingCallback() SCREAMING_SNAKE_CASE : List[Any] = {} if args.fpaa: SCREAMING_SNAKE_CASE : Optional[Any] = 16 if args.gpus > 1: SCREAMING_SNAKE_CASE : Optional[int] = "auto" SCREAMING_SNAKE_CASE : Tuple = "ddp" SCREAMING_SNAKE_CASE : str = args.accumulate_grad_batches SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : str = "auto" SCREAMING_SNAKE_CASE : Tuple = pl.Trainer.from_argparse_args( lowercase , weights_summary=lowercase , callbacks=[logging_callback] + extra_callbacks + [InitCallback()] + [checkpoint_callback] , logger=lowercase , val_check_interval=1 , num_sanity_val_steps=2 , **lowercase , ) if args.do_train: trainer.fit(lowercase ) else: print("RAG modeling tests with new set functions successfuly executed!" ) return trainer
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : int , UpperCAmelCase_ : float = 0 ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = row, column SCREAMING_SNAKE_CASE : Optional[Any] = [[default_value for c in range(UpperCAmelCase_ )] for r in range(UpperCAmelCase_ )] def __str__( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = f'''Matrix consist of {self.row} rows and {self.column} columns\n''' # Make string identifier SCREAMING_SNAKE_CASE : Dict = 0 for row_vector in self.array: for obj in row_vector: SCREAMING_SNAKE_CASE : Optional[Any] = max(UpperCAmelCase_ , len(str(UpperCAmelCase_ ) ) ) SCREAMING_SNAKE_CASE : Optional[int] = f'''%{max_element_length}s''' # Make string and return def single_line(UpperCAmelCase_ : list[float] ) -> str: nonlocal string_format_identifier SCREAMING_SNAKE_CASE : Optional[int] = "[" line += ", ".join(string_format_identifier % (obj,) for obj in row_vector ) line += "]" return line s += "\n".join(single_line(UpperCAmelCase_ ) for row_vector in self.array ) return s def __repr__( self : Dict ): return str(self ) def _A ( self : Optional[int] , UpperCAmelCase_ : tuple[int, int] ): if not (isinstance(UpperCAmelCase_ , (list, tuple) ) and len(UpperCAmelCase_ ) == 2): return False elif not (0 <= loc[0] < self.row and 0 <= loc[1] < self.column): return False else: return True def __getitem__( self : Any , UpperCAmelCase_ : tuple[int, int] ): assert self.validate_indicies(UpperCAmelCase_ ) return self.array[loc[0]][loc[1]] def __setitem__( self : int , UpperCAmelCase_ : tuple[int, int] , UpperCAmelCase_ : float ): assert self.validate_indicies(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = value def __add__( self : Optional[int] , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == another.row and self.column == another.column # Add SCREAMING_SNAKE_CASE : str = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : Any = self[r, c] + another[r, c] return result def __neg__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = -self[r, c] return result def __sub__( self : Optional[Any] , UpperCAmelCase_ : Matrix ): return self + (-another) def __mul__( self : Dict , UpperCAmelCase_ : int | float | Matrix ): if isinstance(UpperCAmelCase_ , (int, float) ): # Scalar multiplication SCREAMING_SNAKE_CASE : Any = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = self[r, c] * another return result elif isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): # Matrix multiplication assert self.column == another.row SCREAMING_SNAKE_CASE : Any = Matrix(self.row , another.column ) for r in range(self.row ): for c in range(another.column ): for i in range(self.column ): result[r, c] += self[r, i] * another[i, c] return result else: SCREAMING_SNAKE_CASE : List[str] = f'''Unsupported type given for another ({type(UpperCAmelCase_ )})''' raise TypeError(UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : List[str] = Matrix(self.column , self.row ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : List[str] = self[r, c] return result def _A ( self : Union[str, Any] , UpperCAmelCase_ : Matrix , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) and isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == self.column == u.row == v.row # u, v should be column vector assert u.column == v.column == 1 # u, v should be column vector # Calculate SCREAMING_SNAKE_CASE : Tuple = v.transpose() SCREAMING_SNAKE_CASE : int = (v_t * self * u)[0, 0] + 1 if numerator_factor == 0: return None # It's not invertable return self - ((self * u) * (v_t * self) * (1.0 / numerator_factor)) # Testing if __name__ == "__main__": def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 3 , 0 ) for i in range(3 ): SCREAMING_SNAKE_CASE : str = 1 print(F'''a^(-1) is {ainv}''' ) # u, v SCREAMING_SNAKE_CASE : Optional[int] = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = 1, 2, -3 SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = 4, -2, 5 print(F'''u is {u}''' ) print(F'''v is {v}''' ) print(F'''uv^T is {u * v.transpose()}''' ) # Sherman Morrison print(F'''(a + uv^T)^(-1) is {ainv.sherman_morrison(lowercase , lowercase )}''' ) def lowerCamelCase__ ( ): """simple docstring""" import doctest doctest.testmod() testa()
62
1
# Copyright 2021 The HuggingFace Team. All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from argparse import ArgumentParser from accelerate.commands.config import get_config_parser from accelerate.commands.env import env_command_parser from accelerate.commands.launch import launch_command_parser from accelerate.commands.test import test_command_parser from accelerate.commands.tpu import tpu_command_parser def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = ArgumentParser("Accelerate CLI tool" , usage="accelerate <command> [<args>]" , allow_abbrev=lowercase ) SCREAMING_SNAKE_CASE : Union[str, Any] = parser.add_subparsers(help="accelerate command helpers" ) # Register commands get_config_parser(subparsers=lowercase ) env_command_parser(subparsers=lowercase ) launch_command_parser(subparsers=lowercase ) tpu_command_parser(subparsers=lowercase ) test_command_parser(subparsers=lowercase ) # Let's go SCREAMING_SNAKE_CASE : int = parser.parse_args() if not hasattr(lowercase , "func" ): parser.print_help() exit(1 ) # Run args.func(lowercase ) if __name__ == "__main__": main()
62
import json from typing import List, Optional, Tuple from tokenizers import normalizers from tokenizers.pre_tokenizers import BertPreTokenizer, PreTokenizer from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging from .tokenization_roformer import RoFormerTokenizer from .tokenization_utils import JiebaPreTokenizer snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """vocab.txt""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """junnyu/roformer_chinese_small""": """https://huggingface.co/junnyu/roformer_chinese_small/resolve/main/vocab.txt""", """junnyu/roformer_chinese_base""": """https://huggingface.co/junnyu/roformer_chinese_base/resolve/main/vocab.txt""", """junnyu/roformer_chinese_char_small""": ( """https://huggingface.co/junnyu/roformer_chinese_char_small/resolve/main/vocab.txt""" ), """junnyu/roformer_chinese_char_base""": ( """https://huggingface.co/junnyu/roformer_chinese_char_base/resolve/main/vocab.txt""" ), """junnyu/roformer_small_discriminator""": ( """https://huggingface.co/junnyu/roformer_small_discriminator/resolve/main/vocab.txt""" ), """junnyu/roformer_small_generator""": ( """https://huggingface.co/junnyu/roformer_small_generator/resolve/main/vocab.txt""" ), } } snake_case = { """junnyu/roformer_chinese_small""": 1_536, """junnyu/roformer_chinese_base""": 1_536, """junnyu/roformer_chinese_char_small""": 512, """junnyu/roformer_chinese_char_base""": 512, """junnyu/roformer_small_discriminator""": 128, """junnyu/roformer_small_generator""": 128, } snake_case = { """junnyu/roformer_chinese_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_base""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_base""": {"""do_lower_case""": True}, """junnyu/roformer_small_discriminator""": {"""do_lower_case""": True}, """junnyu/roformer_small_generator""": {"""do_lower_case""": True}, } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = VOCAB_FILES_NAMES UpperCamelCase_ : int = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Tuple = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : List[Any] = PRETRAINED_INIT_CONFIGURATION UpperCamelCase_ : Any = RoFormerTokenizer def __init__( self : Tuple , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : List[str]="[UNK]" , UpperCAmelCase_ : Any="[SEP]" , UpperCAmelCase_ : Any="[PAD]" , UpperCAmelCase_ : List[str]="[CLS]" , UpperCAmelCase_ : str="[MASK]" , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : Optional[Any]=None , **UpperCAmelCase_ : List[str] , ): super().__init__( UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , do_lower_case=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , tokenize_chinese_chars=UpperCAmelCase_ , strip_accents=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = json.loads(self.backend_tokenizer.normalizer.__getstate__() ) if ( pre_tok_state.get("lowercase" , UpperCAmelCase_ ) != do_lower_case or pre_tok_state.get("strip_accents" , UpperCAmelCase_ ) != strip_accents ): SCREAMING_SNAKE_CASE : Optional[Any] = getattr(UpperCAmelCase_ , pre_tok_state.pop("type" ) ) SCREAMING_SNAKE_CASE : Any = do_lower_case SCREAMING_SNAKE_CASE : List[str] = strip_accents SCREAMING_SNAKE_CASE : Tuple = pre_tok_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = do_lower_case def __getstate__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = self.__dict__.copy() SCREAMING_SNAKE_CASE : Optional[Any] = BertPreTokenizer() return state def __setstate__( self : Tuple , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = d SCREAMING_SNAKE_CASE : Dict = self.__dict__["_tokenizer"].get_vocab() SCREAMING_SNAKE_CASE : Any = PreTokenizer.custom(JiebaPreTokenizer(UpperCAmelCase_ ) ) def _A ( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Union[str, Any]=None ): SCREAMING_SNAKE_CASE : List[Any] = [self.cls_token_id] + token_ids_a + [self.sep_token_id] if token_ids_a: output += token_ids_a + [self.sep_token_id] return output def _A ( self : Tuple , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : List[Any] = [self.sep_token_id] SCREAMING_SNAKE_CASE : Optional[int] = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1] def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): SCREAMING_SNAKE_CASE : Optional[int] = self._tokenizer.model.save(UpperCAmelCase_ , name=UpperCAmelCase_ ) return tuple(UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Any=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : Union[str, Any] = BertPreTokenizer() return super().save_pretrained(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ )
62
1
class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Any , UpperCAmelCase_ : list[int] ): SCREAMING_SNAKE_CASE : List[str] = len(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = [0] * len_array if len_array > 0: SCREAMING_SNAKE_CASE : List[str] = array[0] for i in range(1 , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.prefix_sum[i - 1] + array[i] def _A ( self : List[str] , UpperCAmelCase_ : int , UpperCAmelCase_ : int ): if start == 0: return self.prefix_sum[end] return self.prefix_sum[end] - self.prefix_sum[start - 1] def _A ( self : Any , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[Any] = {0} for sum_item in self.prefix_sum: if sum_item - target_sum in sums: return True sums.add(UpperCAmelCase_ ) return False if __name__ == "__main__": import doctest doctest.testmod()
62
def lowerCamelCase__ ( lowercase ): """simple docstring""" if not isinstance(lowercase , lowercase ): raise TypeError("only integers accepted as input" ) else: SCREAMING_SNAKE_CASE : Optional[int] = str(abs(lowercase ) ) SCREAMING_SNAKE_CASE : str = [list(lowercase ) for char in range(len(lowercase ) )] for index in range(len(lowercase ) ): num_transpositions[index].pop(lowercase ) return max( int("".join(list(lowercase ) ) ) for transposition in num_transpositions ) if __name__ == "__main__": __import__("""doctest""").testmod()
62
1
import gc import unittest import torch from parameterized import parameterized from diffusers import AutoencoderKL from diffusers.utils import floats_tensor, load_hf_numpy, require_torch_gpu, slow, torch_all_close, torch_device from diffusers.utils.import_utils import is_xformers_available from diffusers.utils.testing_utils import enable_full_determinism from .test_modeling_common import ModelTesterMixin, UNetTesterMixin enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[Any] = AutoencoderKL UpperCamelCase_ : Any = '''sample''' UpperCamelCase_ : int = 1e-2 @property def _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Tuple = 4 SCREAMING_SNAKE_CASE : Tuple = 3 SCREAMING_SNAKE_CASE : Optional[Any] = (32, 32) SCREAMING_SNAKE_CASE : Dict = floats_tensor((batch_size, num_channels) + sizes ).to(UpperCAmelCase_ ) return {"sample": image} @property def _A ( self : Any ): return (3, 32, 32) @property def _A ( self : str ): return (3, 32, 32) def _A ( self : int ): SCREAMING_SNAKE_CASE : Union[str, Any] = { "block_out_channels": [32, 64], "in_channels": 3, "out_channels": 3, "down_block_types": ["DownEncoderBlock2D", "DownEncoderBlock2D"], "up_block_types": ["UpDecoderBlock2D", "UpDecoderBlock2D"], "latent_channels": 4, } SCREAMING_SNAKE_CASE : List[Any] = self.dummy_input return init_dict, inputs_dict def _A ( self : Optional[int] ): pass def _A ( self : Union[str, Any] ): pass @unittest.skipIf(torch_device == "mps" , "Gradient checkpointing skipped on MPS" ) def _A ( self : Optional[int] ): # enable deterministic behavior for gradient checkpointing SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = self.prepare_init_args_and_inputs_for_common() SCREAMING_SNAKE_CASE : Optional[Any] = self.model_class(**UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) assert not model.is_gradient_checkpointing and model.training SCREAMING_SNAKE_CASE : List[Any] = model(**UpperCAmelCase_ ).sample # run the backwards pass on the model. For backwards pass, for simplicity purpose, # we won't calculate the loss and rather backprop on out.sum() model.zero_grad() SCREAMING_SNAKE_CASE : Optional[Any] = torch.randn_like(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = (out - labels).mean() loss.backward() # re-instantiate the model now enabling gradient checkpointing SCREAMING_SNAKE_CASE : Any = self.model_class(**UpperCAmelCase_ ) # clone model model_a.load_state_dict(model.state_dict() ) model_a.to(UpperCAmelCase_ ) model_a.enable_gradient_checkpointing() assert model_a.is_gradient_checkpointing and model_a.training SCREAMING_SNAKE_CASE : List[str] = model_a(**UpperCAmelCase_ ).sample # run the backwards pass on the model. For backwards pass, for simplicity purpose, # we won't calculate the loss and rather backprop on out.sum() model_a.zero_grad() SCREAMING_SNAKE_CASE : Optional[int] = (out_a - labels).mean() loss_a.backward() # compare the output and parameters gradients self.assertTrue((loss - loss_a).abs() < 1E-5 ) SCREAMING_SNAKE_CASE : Any = dict(model.named_parameters() ) SCREAMING_SNAKE_CASE : Any = dict(model_a.named_parameters() ) for name, param in named_params.items(): self.assertTrue(torch_all_close(param.grad.data , named_params_a[name].grad.data , atol=5E-5 ) ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = AutoencoderKL.from_pretrained("fusing/autoencoder-kl-dummy" , output_loading_info=UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) self.assertEqual(len(loading_info["missing_keys"] ) , 0 ) model.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = model(**self.dummy_input ) assert image is not None, "Make sure output is not None" def _A ( self : int ): SCREAMING_SNAKE_CASE : List[Any] = AutoencoderKL.from_pretrained("fusing/autoencoder-kl-dummy" ) SCREAMING_SNAKE_CASE : Optional[int] = model.to(UpperCAmelCase_ ) model.eval() if torch_device == "mps": SCREAMING_SNAKE_CASE : Any = torch.manual_seed(0 ) else: SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device=UpperCAmelCase_ ).manual_seed(0 ) SCREAMING_SNAKE_CASE : int = torch.randn( 1 , model.config.in_channels , model.config.sample_size , model.config.sample_size , generator=torch.manual_seed(0 ) , ) SCREAMING_SNAKE_CASE : Any = image.to(UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[int] = model(UpperCAmelCase_ , sample_posterior=UpperCAmelCase_ , generator=UpperCAmelCase_ ).sample SCREAMING_SNAKE_CASE : List[Any] = output[0, -1, -3:, -3:].flatten().cpu() # Since the VAE Gaussian prior's generator is seeded on the appropriate device, # the expected output slices are not the same for CPU and GPU. if torch_device == "mps": SCREAMING_SNAKE_CASE : List[str] = torch.tensor( [ -4.0_078E-01, -3.8_323E-04, -1.2_681E-01, -1.1_462E-01, 2.0_095E-01, 1.0_893E-01, -8.8_247E-02, -3.0_361E-01, -9.8_644E-03, ] ) elif torch_device == "cpu": SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [-0.1_352, 0.0_878, 0.0_419, -0.0_818, -0.1_069, 0.0_688, -0.1_458, -0.4_446, -0.0_026] ) else: SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor( [-0.2_421, 0.4_642, 0.2_507, -0.0_438, 0.0_682, 0.3_160, -0.2_018, -0.0_727, 0.2_485] ) self.assertTrue(torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , rtol=1E-2 ) ) @slow class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : List[str] , UpperCAmelCase_ : Any , UpperCAmelCase_ : Dict ): return f'''gaussian_noise_s={seed}_shape={'_'.join([str(UpperCAmelCase_ ) for s in shape] )}.npy''' def _A ( self : Union[str, Any] ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : Tuple , UpperCAmelCase_ : List[Any]=0 , UpperCAmelCase_ : Dict=(4, 3, 512, 512) , UpperCAmelCase_ : Any=False ): SCREAMING_SNAKE_CASE : Optional[int] = torch.floataa if fpaa else torch.floataa SCREAMING_SNAKE_CASE : Dict = torch.from_numpy(load_hf_numpy(self.get_file_format(UpperCAmelCase_ , UpperCAmelCase_ ) ) ).to(UpperCAmelCase_ ).to(UpperCAmelCase_ ) return image def _A ( self : Union[str, Any] , UpperCAmelCase_ : Any="CompVis/stable-diffusion-v1-4" , UpperCAmelCase_ : List[str]=False ): SCREAMING_SNAKE_CASE : int = "fp16" if fpaa else None SCREAMING_SNAKE_CASE : List[str] = torch.floataa if fpaa else torch.floataa SCREAMING_SNAKE_CASE : Any = AutoencoderKL.from_pretrained( UpperCAmelCase_ , subfolder="vae" , torch_dtype=UpperCAmelCase_ , revision=UpperCAmelCase_ , ) model.to(UpperCAmelCase_ ).eval() return model def _A ( self : Optional[Any] , UpperCAmelCase_ : int=0 ): if torch_device == "mps": return torch.manual_seed(UpperCAmelCase_ ) return torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) @parameterized.expand( [ # fmt: off [33, [-0.1_603, 0.9_878, -0.0_495, -0.0_790, -0.2_709, 0.8_375, -0.2_060, -0.0_824], [-0.2_395, 0.0_098, 0.0_102, -0.0_709, -0.2_840, -0.0_274, -0.0_718, -0.1_824]], [47, [-0.2_376, 0.1_168, 0.1_332, -0.4_840, -0.2_508, -0.0_791, -0.0_493, -0.4_089], [0.0_350, 0.0_847, 0.0_467, 0.0_344, -0.0_842, -0.0_547, -0.0_633, -0.1_131]], # fmt: on ] ) def _A ( self : List[str] , UpperCAmelCase_ : Dict , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Any] ): SCREAMING_SNAKE_CASE : Tuple = self.get_sd_vae_model() SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_sd_image(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = self.get_generator(UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : List[str] = model(UpperCAmelCase_ , generator=UpperCAmelCase_ , sample_posterior=UpperCAmelCase_ ).sample assert sample.shape == image.shape SCREAMING_SNAKE_CASE : List[Any] = sample[-1, -2:, -2:, :2].flatten().float().cpu() SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(expected_slice_mps if torch_device == "mps" else expected_slice ) assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=3E-3 ) @parameterized.expand( [ # fmt: off [33, [-0.0_513, 0.0_289, 1.3_799, 0.2_166, -0.2_573, -0.0_871, 0.5_103, -0.0_999]], [47, [-0.4_128, -0.1_320, -0.3_704, 0.1_965, -0.4_116, -0.2_332, -0.3_340, 0.2_247]], # fmt: on ] ) @require_torch_gpu def _A ( self : int , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : str = self.get_sd_vae_model(fpaa=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = self.get_sd_image(UpperCAmelCase_ , fpaa=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = self.get_generator(UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , generator=UpperCAmelCase_ , sample_posterior=UpperCAmelCase_ ).sample assert sample.shape == image.shape SCREAMING_SNAKE_CASE : List[Any] = sample[-1, -2:, :2, -2:].flatten().float().cpu() SCREAMING_SNAKE_CASE : str = torch.tensor(UpperCAmelCase_ ) assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=1E-2 ) @parameterized.expand( [ # fmt: off [33, [-0.1_609, 0.9_866, -0.0_487, -0.0_777, -0.2_716, 0.8_368, -0.2_055, -0.0_814], [-0.2_395, 0.0_098, 0.0_102, -0.0_709, -0.2_840, -0.0_274, -0.0_718, -0.1_824]], [47, [-0.2_377, 0.1_147, 0.1_333, -0.4_841, -0.2_506, -0.0_805, -0.0_491, -0.4_085], [0.0_350, 0.0_847, 0.0_467, 0.0_344, -0.0_842, -0.0_547, -0.0_633, -0.1_131]], # fmt: on ] ) def _A ( self : List[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_sd_vae_model() SCREAMING_SNAKE_CASE : Tuple = self.get_sd_image(UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : List[Any] = model(UpperCAmelCase_ ).sample assert sample.shape == image.shape SCREAMING_SNAKE_CASE : str = sample[-1, -2:, -2:, :2].flatten().float().cpu() SCREAMING_SNAKE_CASE : str = torch.tensor(expected_slice_mps if torch_device == "mps" else expected_slice ) assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=3E-3 ) @parameterized.expand( [ # fmt: off [13, [-0.2_051, -0.1_803, -0.2_311, -0.2_114, -0.3_292, -0.3_574, -0.2_953, -0.3_323]], [37, [-0.2_632, -0.2_625, -0.2_199, -0.2_741, -0.4_539, -0.4_990, -0.3_720, -0.4_925]], # fmt: on ] ) @require_torch_gpu def _A ( self : int , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : Dict = self.get_sd_vae_model() SCREAMING_SNAKE_CASE : Any = self.get_sd_image(UpperCAmelCase_ , shape=(3, 4, 64, 64) ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model.decode(UpperCAmelCase_ ).sample assert list(sample.shape ) == [3, 3, 512, 512] SCREAMING_SNAKE_CASE : Dict = sample[-1, -2:, :2, -2:].flatten().cpu() SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(UpperCAmelCase_ ) assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=1E-3 ) @parameterized.expand( [ # fmt: off [27, [-0.0_369, 0.0_207, -0.0_776, -0.0_682, -0.1_747, -0.1_930, -0.1_465, -0.2_039]], [16, [-0.1_628, -0.2_134, -0.2_747, -0.2_642, -0.3_774, -0.4_404, -0.3_687, -0.4_277]], # fmt: on ] ) @require_torch_gpu def _A ( self : Union[str, Any] , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : str = self.get_sd_vae_model(fpaa=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.get_sd_image(UpperCAmelCase_ , shape=(3, 4, 64, 64) , fpaa=UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model.decode(UpperCAmelCase_ ).sample assert list(sample.shape ) == [3, 3, 512, 512] SCREAMING_SNAKE_CASE : int = sample[-1, -2:, :2, -2:].flatten().float().cpu() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(UpperCAmelCase_ ) assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=5E-3 ) @parameterized.expand([(13,), (16,), (27,)] ) @require_torch_gpu @unittest.skipIf(not is_xformers_available() , reason="xformers is not required when using PyTorch 2.0." ) def _A ( self : Optional[int] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : Dict = self.get_sd_vae_model(fpaa=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = self.get_sd_image(UpperCAmelCase_ , shape=(3, 4, 64, 64) , fpaa=UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model.decode(UpperCAmelCase_ ).sample model.enable_xformers_memory_efficient_attention() with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model.decode(UpperCAmelCase_ ).sample assert list(sample.shape ) == [3, 3, 512, 512] assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=1E-1 ) @parameterized.expand([(13,), (16,), (37,)] ) @require_torch_gpu @unittest.skipIf(not is_xformers_available() , reason="xformers is not required when using PyTorch 2.0." ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : Tuple = self.get_sd_vae_model() SCREAMING_SNAKE_CASE : List[str] = self.get_sd_image(UpperCAmelCase_ , shape=(3, 4, 64, 64) ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Dict = model.decode(UpperCAmelCase_ ).sample model.enable_xformers_memory_efficient_attention() with torch.no_grad(): SCREAMING_SNAKE_CASE : int = model.decode(UpperCAmelCase_ ).sample assert list(sample.shape ) == [3, 3, 512, 512] assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=1E-2 ) @parameterized.expand( [ # fmt: off [33, [-0.3_001, 0.0_918, -2.6_984, -3.9_720, -3.2_099, -5.0_353, 1.7_338, -0.2_065, 3.4_267]], [47, [-1.5_030, -4.3_871, -6.0_355, -9.1_157, -1.6_661, -2.7_853, 2.1_607, -5.0_823, 2.5_633]], # fmt: on ] ) def _A ( self : int , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[int] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_sd_vae_model() SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_sd_image(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = self.get_generator(UpperCAmelCase_ ) with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model.encode(UpperCAmelCase_ ).latent_dist SCREAMING_SNAKE_CASE : List[str] = dist.sample(generator=UpperCAmelCase_ ) assert list(sample.shape ) == [image.shape[0], 4] + [i // 8 for i in image.shape[2:]] SCREAMING_SNAKE_CASE : List[str] = sample[0, -1, -3:, -3:].flatten().cpu() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = 3E-3 if torch_device != "mps" else 1E-2 assert torch_all_close(UpperCAmelCase_ , UpperCAmelCase_ , atol=UpperCAmelCase_ )
62
# This model implementation is heavily inspired by https://github.com/haofanwang/ControlNet-for-Diffusers/ import gc import random import tempfile import unittest import numpy as np import torch from PIL import Image from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, ControlNetModel, DDIMScheduler, StableDiffusionControlNetImgaImgPipeline, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_controlnet import MultiControlNetModel from diffusers.utils import floats_tensor, load_image, load_numpy, randn_tensor, slow, torch_device from diffusers.utils.import_utils import is_xformers_available from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..pipeline_params import ( IMAGE_TO_IMAGE_IMAGE_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_PARAMS, ) from ..test_pipelines_common import ( PipelineKarrasSchedulerTesterMixin, PipelineLatentTesterMixin, PipelineTesterMixin, ) enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : List[Any] = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[Any] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Tuple = IMAGE_TO_IMAGE_IMAGE_PARAMS.union({'''control_image'''} ) UpperCamelCase_ : Any = IMAGE_TO_IMAGE_IMAGE_PARAMS def _A ( self : List[str] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : str = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : str , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Any = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Optional[Any] = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = 2 SCREAMING_SNAKE_CASE : Union[str, Any] = randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ) SCREAMING_SNAKE_CASE : Tuple = floats_tensor(control_image.shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : str = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : List[str] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : int ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : Union[str, Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : str = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[int] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Dict = frozenset([] ) # TO_DO: add image_params once refactored VaeImageProcessor.preprocess def _A ( self : Optional[Any] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) def init_weights(UpperCAmelCase_ : List[Any] ): if isinstance(UpperCAmelCase_ , torch.nn.Convad ): torch.nn.init.normal(m.weight ) m.bias.data.fill_(1.0 ) SCREAMING_SNAKE_CASE : List[str] = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Dict = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Any = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : Tuple = MultiControlNetModel([controlneta, controlneta] ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Dict = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = 2 SCREAMING_SNAKE_CASE : Tuple = [ randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), ] SCREAMING_SNAKE_CASE : Optional[int] = floats_tensor(control_image[0].shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : Optional[Any] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : Any = self.get_dummy_components() SCREAMING_SNAKE_CASE : str = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = 10.0 SCREAMING_SNAKE_CASE : Any = 4 SCREAMING_SNAKE_CASE : Optional[int] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : Optional[int] = pipe(**UpperCAmelCase_ )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : Any = scale SCREAMING_SNAKE_CASE : List[str] = pipe(**UpperCAmelCase_ , control_guidance_start=0.1 , control_guidance_end=0.2 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : List[Any] = pipe(**UpperCAmelCase_ , control_guidance_start=[0.1, 0.3] , control_guidance_end=[0.2, 0.7] )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = steps SCREAMING_SNAKE_CASE : Dict = scale SCREAMING_SNAKE_CASE : Dict = pipe(**UpperCAmelCase_ , control_guidance_start=0.4 , control_guidance_end=[0.5, 0.8] )[0] # make sure that all outputs are different assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 def _A ( self : Union[str, Any] ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : List[Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = self.get_dummy_components() SCREAMING_SNAKE_CASE : Dict = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) with tempfile.TemporaryDirectory() as tmpdir: try: # save_pretrained is not implemented for Multi-ControlNet pipe.save_pretrained(UpperCAmelCase_ ) except NotImplementedError: pass @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Optional[Any] ): super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny" ) SCREAMING_SNAKE_CASE : Union[str, Any] = StableDiffusionControlNetImgaImgPipeline.from_pretrained( "runwayml/stable-diffusion-v1-5" , safety_checker=UpperCAmelCase_ , controlnet=UpperCAmelCase_ ) pipe.enable_model_cpu_offload() pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.Generator(device="cpu" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : str = "evil space-punk bird" SCREAMING_SNAKE_CASE : Optional[Any] = load_image( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/bird_canny.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : Optional[int] = load_image( "https://huggingface.co/lllyasviel/sd-controlnet-canny/resolve/main/images/bird.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : str = pipe( UpperCAmelCase_ , UpperCAmelCase_ , control_image=UpperCAmelCase_ , generator=UpperCAmelCase_ , output_type="np" , num_inference_steps=50 , strength=0.6 , ) SCREAMING_SNAKE_CASE : int = output.images[0] assert image.shape == (512, 512, 3) SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/img2img.npy" ) assert np.abs(expected_image - image ).max() < 9E-2
62
1
import json import os import shutil import tempfile import unittest import numpy as np from transformers import BertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer from transformers.testing_utils import require_tokenizers, require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor @require_tokenizers @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = tempfile.mkdtemp() # fmt: off SCREAMING_SNAKE_CASE : Any = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest"] # fmt: on SCREAMING_SNAKE_CASE : Dict = 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] ) ) SCREAMING_SNAKE_CASE : Optional[int] = { "do_resize": True, "size": {"height": 18, "width": 18}, "do_normalize": True, "image_mean": [0.5, 0.5, 0.5], "image_std": [0.5, 0.5, 0.5], } SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(self.tmpdirname , UpperCAmelCase_ ) with open(self.image_processor_file , "w" , encoding="utf-8" ) as fp: json.dump(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] , **UpperCAmelCase_ : List[str] ): return BertTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , **UpperCAmelCase_ : Any ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] ): shutil.rmtree(self.tmpdirname ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] SCREAMING_SNAKE_CASE : Union[str, Any] = [Image.fromarray(np.moveaxis(UpperCAmelCase_ , 0 , -1 ) ) for x in image_inputs] return image_inputs def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : int = VisionTextDualEncoderProcessor( tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer(bos_token="(BOS)" , eos_token="(EOS)" ) SCREAMING_SNAKE_CASE : Any = self.get_image_processor(do_normalize=UpperCAmelCase_ , padding_value=1.0 ) SCREAMING_SNAKE_CASE : str = VisionTextDualEncoderProcessor.from_pretrained( self.tmpdirname , bos_token="(BOS)" , eos_token="(EOS)" , do_normalize=UpperCAmelCase_ , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : int = image_processor(UpperCAmelCase_ , return_tensors="np" ) SCREAMING_SNAKE_CASE : Optional[int] = processor(images=UpperCAmelCase_ , 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 _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = "lower newer" SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = tokenizer(UpperCAmelCase_ ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : int = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer() SCREAMING_SNAKE_CASE : List[str] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = "lower newer" SCREAMING_SNAKE_CASE : Union[str, Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Optional[Any] = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , ["input_ids", "token_type_ids", "attention_mask", "pixel_values"] ) # test if it raises when no input is passed with self.assertRaises(UpperCAmelCase_ ): processor() def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] SCREAMING_SNAKE_CASE : Optional[Any] = processor.batch_decode(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer.batch_decode(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = "lower newer" SCREAMING_SNAKE_CASE : Any = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
62
import argparse import json from pathlib import Path import requests import torch from huggingface_hub import hf_hub_download from PIL import Image from transformers import ( MobileViTConfig, MobileViTForImageClassification, MobileViTForSemanticSegmentation, MobileViTImageProcessor, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = MobileViTConfig() # size of the architecture if "mobilevit_s" in mobilevit_name: SCREAMING_SNAKE_CASE : List[Any] = [144, 192, 240] SCREAMING_SNAKE_CASE : Tuple = [16, 32, 64, 96, 128, 160, 640] elif "mobilevit_xs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [96, 120, 144] SCREAMING_SNAKE_CASE : Dict = [16, 32, 48, 64, 80, 96, 384] elif "mobilevit_xxs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [64, 80, 96] SCREAMING_SNAKE_CASE : List[str] = [16, 16, 24, 48, 64, 80, 320] SCREAMING_SNAKE_CASE : int = 0.05 SCREAMING_SNAKE_CASE : int = 2.0 if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : str = 512 SCREAMING_SNAKE_CASE : List[str] = 16 SCREAMING_SNAKE_CASE : Union[str, Any] = 21 SCREAMING_SNAKE_CASE : Dict = "pascal-voc-id2label.json" else: SCREAMING_SNAKE_CASE : Optional[Any] = 1000 SCREAMING_SNAKE_CASE : Optional[Any] = "imagenet-1k-id2label.json" SCREAMING_SNAKE_CASE : Any = "huggingface/label-files" SCREAMING_SNAKE_CASE : Tuple = json.load(open(hf_hub_download(lowercase , lowercase , repo_type="dataset" ) , "r" ) ) SCREAMING_SNAKE_CASE : List[str] = {int(lowercase ): v for k, v in idalabel.items()} SCREAMING_SNAKE_CASE : Optional[Any] = idalabel SCREAMING_SNAKE_CASE : str = {v: k for k, v in idalabel.items()} return config def lowerCamelCase__ ( lowercase , lowercase=False ): """simple docstring""" for i in range(1 , 6 ): if F'''layer_{i}.''' in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(F'''layer_{i}.''' , F'''encoder.layer.{i - 1}.''' ) if "conv_1." in name: SCREAMING_SNAKE_CASE : Dict = name.replace("conv_1." , "conv_stem." ) if ".block." in name: SCREAMING_SNAKE_CASE : List[str] = name.replace(".block." , "." ) if "exp_1x1" in name: SCREAMING_SNAKE_CASE : str = name.replace("exp_1x1" , "expand_1x1" ) if "red_1x1" in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("red_1x1" , "reduce_1x1" ) if ".local_rep.conv_3x3." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".local_rep.conv_3x3." , ".conv_kxk." ) if ".local_rep.conv_1x1." in name: SCREAMING_SNAKE_CASE : int = name.replace(".local_rep.conv_1x1." , ".conv_1x1." ) if ".norm." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".norm." , ".normalization." ) if ".conv." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".conv." , ".convolution." ) if ".conv_proj." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".conv_proj." , ".conv_projection." ) for i in range(0 , 2 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.layer.{j}.''' ) for i in range(2 , 6 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.''' ) if "expand_1x1" in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace("expand_1x1" , "downsampling_layer.expand_1x1" ) if "conv_3x3" in name: SCREAMING_SNAKE_CASE : str = name.replace("conv_3x3" , "downsampling_layer.conv_3x3" ) if "reduce_1x1" in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("reduce_1x1" , "downsampling_layer.reduce_1x1" ) for i in range(2 , 5 ): if F'''.global_rep.{i}.weight''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.global_rep.{i}.weight''' , ".layernorm.weight" ) if F'''.global_rep.{i}.bias''' in name: SCREAMING_SNAKE_CASE : str = name.replace(F'''.global_rep.{i}.bias''' , ".layernorm.bias" ) if ".global_rep." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".global_rep." , ".transformer." ) if ".pre_norm_mha.0." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_mha.0." , ".layernorm_before." ) if ".pre_norm_mha.1.out_proj." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_mha.1.out_proj." , ".attention.output.dense." ) if ".pre_norm_ffn.0." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(".pre_norm_ffn.0." , ".layernorm_after." ) if ".pre_norm_ffn.1." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_ffn.1." , ".intermediate.dense." ) if ".pre_norm_ffn.4." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_ffn.4." , ".output.dense." ) if ".transformer." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".transformer." , ".transformer.layer." ) if ".aspp_layer." in name: SCREAMING_SNAKE_CASE : int = name.replace(".aspp_layer." , "." ) if ".aspp_pool." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".aspp_pool." , "." ) if "seg_head." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("seg_head." , "segmentation_head." ) if "segmentation_head.classifier.classifier." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace("segmentation_head.classifier.classifier." , "segmentation_head.classifier." ) if "classifier.fc." in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("classifier.fc." , "classifier." ) elif (not base_model) and ("segmentation_head." not in name): SCREAMING_SNAKE_CASE : List[Any] = "mobilevit." + name return name def lowerCamelCase__ ( lowercase , lowercase , lowercase=False ): """simple docstring""" if base_model: SCREAMING_SNAKE_CASE : Optional[int] = "" else: SCREAMING_SNAKE_CASE : Any = "mobilevit." for key in orig_state_dict.copy().keys(): SCREAMING_SNAKE_CASE : Union[str, Any] = orig_state_dict.pop(lowercase ) if key[:8] == "encoder.": SCREAMING_SNAKE_CASE : int = key[8:] if "qkv" in key: SCREAMING_SNAKE_CASE : Optional[int] = key.split("." ) SCREAMING_SNAKE_CASE : Any = int(key_split[0][6:] ) - 1 SCREAMING_SNAKE_CASE : List[Any] = int(key_split[3] ) SCREAMING_SNAKE_CASE : List[Any] = model.get_submodule(F'''{model_prefix}encoder.layer.{layer_num}''' ) SCREAMING_SNAKE_CASE : int = layer.transformer.layer[transformer_num].attention.attention.all_head_size SCREAMING_SNAKE_CASE : Union[str, Any] = ( F'''{model_prefix}encoder.layer.{layer_num}.transformer.layer.{transformer_num}.attention.attention.''' ) if "weight" in key: SCREAMING_SNAKE_CASE : Optional[int] = val[:dim, :] SCREAMING_SNAKE_CASE : Tuple = val[dim : dim * 2, :] SCREAMING_SNAKE_CASE : Dict = val[-dim:, :] else: SCREAMING_SNAKE_CASE : str = val[:dim] SCREAMING_SNAKE_CASE : Union[str, Any] = val[dim : dim * 2] SCREAMING_SNAKE_CASE : Union[str, Any] = val[-dim:] else: SCREAMING_SNAKE_CASE : List[Any] = val return orig_state_dict def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = "http://images.cocodataset.org/val2017/000000039769.jpg" SCREAMING_SNAKE_CASE : Union[str, Any] = Image.open(requests.get(lowercase , stream=lowercase ).raw ) return im @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase=False ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = get_mobilevit_config(lowercase ) # load original state_dict SCREAMING_SNAKE_CASE : Union[str, Any] = torch.load(lowercase , map_location="cpu" ) # load 🤗 model if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : List[str] = MobileViTForSemanticSegmentation(lowercase ).eval() else: SCREAMING_SNAKE_CASE : str = MobileViTForImageClassification(lowercase ).eval() SCREAMING_SNAKE_CASE : Any = convert_state_dict(lowercase , lowercase ) model.load_state_dict(lowercase ) # Check outputs on an image, prepared by MobileViTImageProcessor SCREAMING_SNAKE_CASE : List[Any] = MobileViTImageProcessor(crop_size=config.image_size , size=config.image_size + 32 ) SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=prepare_img() , return_tensors="pt" ) SCREAMING_SNAKE_CASE : List[Any] = model(**lowercase ) SCREAMING_SNAKE_CASE : str = outputs.logits if mobilevit_name.startswith("deeplabv3_" ): assert logits.shape == (1, 21, 32, 32) if mobilevit_name == "deeplabv3_mobilevit_s": SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor( [ [[6.2065, 6.1292, 6.2070], [6.1079, 6.1254, 6.1747], [6.0042, 6.1071, 6.1034]], [[-6.9253, -6.8653, -7.0398], [-7.3218, -7.3983, -7.3670], [-7.1961, -7.2482, -7.1569]], [[-4.4723, -4.4348, -4.3769], [-5.3629, -5.4632, -5.4598], [-5.1587, -5.3402, -5.5059]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor( [ [[5.4449, 5.5733, 5.6314], [5.1815, 5.3930, 5.5963], [5.1656, 5.4333, 5.4853]], [[-9.4423, -9.7766, -9.6714], [-9.1581, -9.5720, -9.5519], [-9.1006, -9.6458, -9.5703]], [[-7.7721, -7.3716, -7.1583], [-8.4599, -8.0624, -7.7944], [-8.4172, -7.8366, -7.5025]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xxs": SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [ [[6.9811, 6.9743, 7.3123], [7.1777, 7.1931, 7.3938], [7.5633, 7.8050, 7.8901]], [[-10.5536, -10.2332, -10.2924], [-10.2336, -9.8624, -9.5964], [-10.8840, -10.8158, -10.6659]], [[-3.4938, -3.0631, -2.8620], [-3.4205, -2.8135, -2.6875], [-3.4179, -2.7945, -2.8750]], ] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3, :3, :3] , lowercase , atol=1E-4 ) else: assert logits.shape == (1, 1000) if mobilevit_name == "mobilevit_s": SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([-0.9866, 0.2392, -1.1241] ) elif mobilevit_name == "mobilevit_xs": SCREAMING_SNAKE_CASE : Dict = torch.tensor([-2.4761, -0.9399, -1.9587] ) elif mobilevit_name == "mobilevit_xxs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor([-1.9364, -1.2327, -0.4653] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3] , lowercase , atol=1E-4 ) Path(lowercase ).mkdir(exist_ok=lowercase ) print(F'''Saving model {mobilevit_name} to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowercase ) print(F'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowercase ) if push_to_hub: SCREAMING_SNAKE_CASE : List[str] = { "mobilevit_s": "mobilevit-small", "mobilevit_xs": "mobilevit-x-small", "mobilevit_xxs": "mobilevit-xx-small", "deeplabv3_mobilevit_s": "deeplabv3-mobilevit-small", "deeplabv3_mobilevit_xs": "deeplabv3-mobilevit-x-small", "deeplabv3_mobilevit_xxs": "deeplabv3-mobilevit-xx-small", } print("Pushing to the hub..." ) SCREAMING_SNAKE_CASE : int = model_mapping[mobilevit_name] image_processor.push_to_hub(lowercase , organization="apple" ) model.push_to_hub(lowercase , organization="apple" ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument( """--mobilevit_name""", default="""mobilevit_s""", type=str, help=( """Name of the MobileViT model you'd like to convert. Should be one of 'mobilevit_s', 'mobilevit_xs',""" """ 'mobilevit_xxs', 'deeplabv3_mobilevit_s', 'deeplabv3_mobilevit_xs', 'deeplabv3_mobilevit_xxs'.""" ), ) parser.add_argument( """--checkpoint_path""", required=True, type=str, help="""Path to the original state dict (.pt file).""" ) parser.add_argument( """--pytorch_dump_folder_path""", required=True, type=str, help="""Path to the output PyTorch model directory.""" ) parser.add_argument( """--push_to_hub""", action="""store_true""", help="""Whether or not to push the converted model to the 🤗 hub.""" ) snake_case = parser.parse_args() convert_movilevit_checkpoint( args.mobilevit_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub )
62
1
from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """facebook/dpr-ctx_encoder-single-nq-base""": ( """https://huggingface.co/facebook/dpr-ctx_encoder-single-nq-base/resolve/main/config.json""" ), """facebook/dpr-question_encoder-single-nq-base""": ( """https://huggingface.co/facebook/dpr-question_encoder-single-nq-base/resolve/main/config.json""" ), """facebook/dpr-reader-single-nq-base""": ( """https://huggingface.co/facebook/dpr-reader-single-nq-base/resolve/main/config.json""" ), """facebook/dpr-ctx_encoder-multiset-base""": ( """https://huggingface.co/facebook/dpr-ctx_encoder-multiset-base/resolve/main/config.json""" ), """facebook/dpr-question_encoder-multiset-base""": ( """https://huggingface.co/facebook/dpr-question_encoder-multiset-base/resolve/main/config.json""" ), """facebook/dpr-reader-multiset-base""": ( """https://huggingface.co/facebook/dpr-reader-multiset-base/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : List[str] = '''dpr''' def __init__( self : int , UpperCAmelCase_ : Optional[Any]=3_0522 , UpperCAmelCase_ : Optional[Any]=768 , UpperCAmelCase_ : List[Any]=12 , UpperCAmelCase_ : Tuple=12 , UpperCAmelCase_ : int=3072 , UpperCAmelCase_ : int="gelu" , UpperCAmelCase_ : Optional[Any]=0.1 , UpperCAmelCase_ : Any=0.1 , UpperCAmelCase_ : Union[str, Any]=512 , UpperCAmelCase_ : int=2 , UpperCAmelCase_ : List[Any]=0.02 , UpperCAmelCase_ : List[str]=1E-12 , UpperCAmelCase_ : Optional[Any]=0 , UpperCAmelCase_ : Tuple="absolute" , UpperCAmelCase_ : int = 0 , **UpperCAmelCase_ : str , ): super().__init__(pad_token_id=UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : Optional[int] = hidden_size SCREAMING_SNAKE_CASE : int = num_hidden_layers SCREAMING_SNAKE_CASE : Optional[int] = num_attention_heads SCREAMING_SNAKE_CASE : List[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[Any] = intermediate_size SCREAMING_SNAKE_CASE : str = hidden_dropout_prob SCREAMING_SNAKE_CASE : Any = attention_probs_dropout_prob SCREAMING_SNAKE_CASE : Optional[int] = max_position_embeddings SCREAMING_SNAKE_CASE : int = type_vocab_size SCREAMING_SNAKE_CASE : Tuple = initializer_range SCREAMING_SNAKE_CASE : List[Any] = layer_norm_eps SCREAMING_SNAKE_CASE : List[str] = projection_dim SCREAMING_SNAKE_CASE : List[str] = position_embedding_type
62
from collections import Counter import numpy as np from sklearn import datasets from sklearn.model_selection import train_test_split snake_case = datasets.load_iris() snake_case = np.array(data["""data"""]) snake_case = np.array(data["""target"""]) snake_case = data["""target_names"""] snake_case , snake_case , snake_case , snake_case = train_test_split(X, y) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return np.linalg.norm(np.array(lowercase ) - np.array(lowercase ) ) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase=5 ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = zip(lowercase , lowercase ) # List of distances of all points from the point to be classified SCREAMING_SNAKE_CASE : Optional[int] = [] for data_point in data: SCREAMING_SNAKE_CASE : Tuple = euclidean_distance(data_point[0] , lowercase ) distances.append((distance, data_point[1]) ) # Choosing 'k' points with the least distances. SCREAMING_SNAKE_CASE : List[Any] = [i[1] for i in sorted(lowercase )[:k]] # Most commonly occurring class among them # is the class into which the point is classified SCREAMING_SNAKE_CASE : List[Any] = Counter(lowercase ).most_common(1 )[0][0] return classes[result] if __name__ == "__main__": print(classifier(X_train, y_train, classes, [4.4, 3.1, 1.3, 1.4]))
62
1
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return (pointa[0] - pointa[0]) ** 2 + (pointa[1] - pointa[1]) ** 2 def lowerCamelCase__ ( lowercase , lowercase=0 ): """simple docstring""" return sorted(lowercase , key=lambda lowercase : x[column] ) def lowerCamelCase__ ( lowercase , lowercase , lowercase=float("inf" ) ): """simple docstring""" for i in range(points_counts - 1 ): for j in range(i + 1 , lowercase ): SCREAMING_SNAKE_CASE : Any = euclidean_distance_sqr(points[i] , points[j] ) if current_dis < min_dis: SCREAMING_SNAKE_CASE : Dict = current_dis return min_dis def lowerCamelCase__ ( lowercase , lowercase , lowercase=float("inf" ) ): """simple docstring""" for i in range(min(6 , points_counts - 1 ) , lowercase ): for j in range(max(0 , i - 6 ) , lowercase ): SCREAMING_SNAKE_CASE : Union[str, Any] = euclidean_distance_sqr(points[i] , points[j] ) if current_dis < min_dis: SCREAMING_SNAKE_CASE : str = current_dis return min_dis def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if points_counts <= 3: return dis_between_closest_pair(lowercase , lowercase ) # recursion SCREAMING_SNAKE_CASE : str = points_counts // 2 SCREAMING_SNAKE_CASE : str = closest_pair_of_points_sqr( lowercase , points_sorted_on_y[:mid] , lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = closest_pair_of_points_sqr( lowercase , points_sorted_on_y[mid:] , points_counts - mid ) SCREAMING_SNAKE_CASE : int = min(lowercase , lowercase ) SCREAMING_SNAKE_CASE : Tuple = [] for point in points_sorted_on_x: if abs(point[0] - points_sorted_on_x[mid][0] ) < closest_pair_dis: cross_strip.append(lowercase ) SCREAMING_SNAKE_CASE : List[str] = dis_between_closest_in_strip( lowercase , len(lowercase ) , lowercase ) return min(lowercase , lowercase ) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = column_based_sort(lowercase , column=0 ) SCREAMING_SNAKE_CASE : Union[str, Any] = column_based_sort(lowercase , column=1 ) return ( closest_pair_of_points_sqr( lowercase , lowercase , lowercase ) ) ** 0.5 if __name__ == "__main__": snake_case = [(2, 3), (12, 30), (40, 50), (5, 1), (12, 10), (3, 4)] print("""Distance:""", closest_pair_of_points(points, len(points)))
62
from typing import Callable, Optional, Union from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """microsoft/xprophetnet-large-wiki100-cased""": ( """https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Tuple = '''xlm-prophetnet''' UpperCamelCase_ : Tuple = ['''past_key_values'''] UpperCamelCase_ : int = { '''num_attention_heads''': '''num_encoder_attention_heads''', } def __init__( self : Dict , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[Union[str, Callable]] = "gelu" , UpperCAmelCase_ : Optional[int] = 3_0522 , UpperCAmelCase_ : Optional[int] = 1024 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[int] = 512 , UpperCAmelCase_ : Optional[float] = 0.02 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 2 , UpperCAmelCase_ : Optional[int] = 32 , UpperCAmelCase_ : Optional[int] = 128 , UpperCAmelCase_ : Optional[bool] = False , UpperCAmelCase_ : Optional[float] = 0.0 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 1 , UpperCAmelCase_ : Optional[int] = 2 , **UpperCAmelCase_ : Dict , ): SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : str = hidden_size SCREAMING_SNAKE_CASE : Tuple = encoder_ffn_dim SCREAMING_SNAKE_CASE : List[Any] = num_encoder_layers SCREAMING_SNAKE_CASE : Any = num_encoder_attention_heads SCREAMING_SNAKE_CASE : Dict = decoder_ffn_dim SCREAMING_SNAKE_CASE : List[str] = num_decoder_layers SCREAMING_SNAKE_CASE : List[Any] = num_decoder_attention_heads SCREAMING_SNAKE_CASE : List[Any] = max_position_embeddings SCREAMING_SNAKE_CASE : Dict = init_std # Normal(0, this parameter) SCREAMING_SNAKE_CASE : Union[str, Any] = activation_function # parameters for xlmprophetnet SCREAMING_SNAKE_CASE : Dict = ngram SCREAMING_SNAKE_CASE : Any = num_buckets SCREAMING_SNAKE_CASE : str = relative_max_distance SCREAMING_SNAKE_CASE : str = disable_ngram_loss SCREAMING_SNAKE_CASE : Dict = eps # 3 Types of Dropout SCREAMING_SNAKE_CASE : Union[str, Any] = attention_dropout SCREAMING_SNAKE_CASE : Optional[int] = activation_dropout SCREAMING_SNAKE_CASE : List[Any] = dropout SCREAMING_SNAKE_CASE : int = use_cache super().__init__( pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , is_encoder_decoder=UpperCAmelCase_ , add_cross_attention=UpperCAmelCase_ , decoder_start_token_id=UpperCAmelCase_ , **UpperCAmelCase_ , ) @property def _A ( self : int ): return self.num_encoder_layers + self.num_decoder_layers @num_hidden_layers.setter def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): raise NotImplementedError( "This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and" " `num_decoder_layers`." )
62
1
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available snake_case = { """configuration_timesformer""": ["""TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP""", """TimesformerConfig"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST""", """TimesformerModel""", """TimesformerForVideoClassification""", """TimesformerPreTrainedModel""", ] if TYPE_CHECKING: from .configuration_timesformer import TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, TimesformerConfig try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_timesformer import ( TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST, TimesformerForVideoClassification, TimesformerModel, TimesformerPreTrainedModel, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
import gc import random import unittest import numpy as np import torch from diffusers import DDIMScheduler, KandinskyVaaPipeline, KandinskyVaaPriorPipeline, UNetaDConditionModel, VQModel from diffusers.utils import floats_tensor, load_numpy, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = KandinskyVaaPipeline UpperCamelCase_ : List[Any] = [ '''image_embeds''', '''negative_image_embeds''', ] UpperCamelCase_ : Tuple = ['''image_embeds''', '''negative_image_embeds'''] UpperCamelCase_ : Any = [ '''generator''', '''height''', '''width''', '''latents''', '''guidance_scale''', '''num_inference_steps''', '''return_dict''', '''guidance_scale''', '''num_images_per_prompt''', '''output_type''', '''return_dict''', ] UpperCamelCase_ : List[str] = False @property def _A ( self : List[Any] ): return 32 @property def _A ( self : List[Any] ): return 32 @property def _A ( self : Any ): return self.time_input_dim @property def _A ( self : Union[str, Any] ): return self.time_input_dim * 4 @property def _A ( self : Tuple ): return 100 @property def _A ( self : Optional[int] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = { "in_channels": 4, # Out channels is double in channels because predicts mean and variance "out_channels": 8, "addition_embed_type": "image", "down_block_types": ("ResnetDownsampleBlock2D", "SimpleCrossAttnDownBlock2D"), "up_block_types": ("SimpleCrossAttnUpBlock2D", "ResnetUpsampleBlock2D"), "mid_block_type": "UNetMidBlock2DSimpleCrossAttn", "block_out_channels": (self.block_out_channels_a, self.block_out_channels_a * 2), "layers_per_block": 1, "encoder_hid_dim": self.text_embedder_hidden_size, "encoder_hid_dim_type": "image_proj", "cross_attention_dim": self.cross_attention_dim, "attention_head_dim": 4, "resnet_time_scale_shift": "scale_shift", "class_embed_type": None, } SCREAMING_SNAKE_CASE : str = UNetaDConditionModel(**UpperCAmelCase_ ) return model @property def _A ( self : int ): return { "block_out_channels": [32, 64], "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"], "in_channels": 3, "latent_channels": 4, "layers_per_block": 1, "norm_num_groups": 8, "norm_type": "spatial", "num_vq_embeddings": 12, "out_channels": 3, "up_block_types": [ "AttnUpDecoderBlock2D", "UpDecoderBlock2D", ], "vq_embed_dim": 4, } @property def _A ( self : Any ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : List[str] = VQModel(**self.dummy_movq_kwargs ) return model def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.dummy_unet SCREAMING_SNAKE_CASE : str = self.dummy_movq SCREAMING_SNAKE_CASE : Optional[int] = DDIMScheduler( num_train_timesteps=1000 , beta_schedule="linear" , beta_start=0.00_085 , beta_end=0.012 , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , steps_offset=1 , prediction_type="epsilon" , thresholding=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "scheduler": scheduler, "movq": movq, } return components def _A ( self : List[str] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any]=0 ): SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(seed + 1 ) ).to( UpperCAmelCase_ ) if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : List[Any] = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = { "image_embeds": image_embeds, "negative_image_embeds": negative_image_embeds, "generator": generator, "height": 64, "width": 64, "guidance_scale": 4.0, "num_inference_steps": 2, "output_type": "np", } return inputs def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "cpu" SCREAMING_SNAKE_CASE : Tuple = self.get_dummy_components() SCREAMING_SNAKE_CASE : List[str] = self.pipeline_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(**self.get_dummy_inputs(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : str = output.images SCREAMING_SNAKE_CASE : Tuple = pipe( **self.get_dummy_inputs(UpperCAmelCase_ ) , return_dict=UpperCAmelCase_ , )[0] SCREAMING_SNAKE_CASE : str = image[0, -3:, -3:, -1] SCREAMING_SNAKE_CASE : Any = image_from_tuple[0, -3:, -3:, -1] assert image.shape == (1, 64, 64, 3) SCREAMING_SNAKE_CASE : List[Any] = np.array( [0.6_237_976, 1.0, 0.36_441_332, 1.0, 0.70_639_634, 0.29_877_186, 0.85_652_125, 0.5_216_843, 0.54_454_046] ) assert ( np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_slice.flatten()}''' assert ( np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_from_tuple_slice.flatten()}''' @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main" "/kandinskyv22/kandinskyv22_text2img_cat_fp16.npy" ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPriorPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-prior" , torch_dtype=torch.floataa ) pipe_prior.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-decoder" , torch_dtype=torch.floataa ) SCREAMING_SNAKE_CASE : Tuple = pipeline.to(UpperCAmelCase_ ) pipeline.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = "red cat, 4k photo" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = pipe_prior( UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=5 , negative_prompt="" , ).to_tuple() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = pipeline( image_embeds=UpperCAmelCase_ , negative_image_embeds=UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=100 , output_type="np" , ) SCREAMING_SNAKE_CASE : List[Any] = output.images[0] assert image.shape == (512, 512, 3) assert_mean_pixel_difference(UpperCAmelCase_ , UpperCAmelCase_ )
62
1
import argparse import torch from transformers import GPTaLMHeadModel, RobertaForMaskedLM if __name__ == "__main__": snake_case = argparse.ArgumentParser( description=( """Extraction some layers of the full RobertaForMaskedLM or GPT2LMHeadModel for Transfer Learned""" """ Distillation""" ) ) parser.add_argument("""--model_type""", default="""roberta""", choices=["""roberta""", """gpt2"""]) parser.add_argument("""--model_name""", default="""roberta-large""", type=str) parser.add_argument("""--dump_checkpoint""", default="""serialization_dir/tf_roberta_048131723.pth""", type=str) parser.add_argument("""--vocab_transform""", action="""store_true""") snake_case = parser.parse_args() if args.model_type == "roberta": snake_case = RobertaForMaskedLM.from_pretrained(args.model_name) snake_case = """roberta""" elif args.model_type == "gpt2": snake_case = GPTaLMHeadModel.from_pretrained(args.model_name) snake_case = """transformer""" snake_case = model.state_dict() snake_case = {} # Embeddings # if args.model_type == "gpt2": for param_name in ["wte.weight", "wpe.weight"]: snake_case = state_dict[F"""{prefix}.{param_name}"""] else: for w in ["word_embeddings", "position_embeddings", "token_type_embeddings"]: snake_case = F"""{prefix}.embeddings.{w}.weight""" snake_case = state_dict[param_name] for w in ["weight", "bias"]: snake_case = F"""{prefix}.embeddings.LayerNorm.{w}""" snake_case = state_dict[param_name] # Transformer Blocks # snake_case = 0 for teacher_idx in [0, 2, 4, 7, 9, 11]: if args.model_type == "gpt2": for layer in ["ln_1", "attn.c_attn", "attn.c_proj", "ln_2", "mlp.c_fc", "mlp.c_proj"]: for w in ["weight", "bias"]: snake_case = state_dict[ F"""{prefix}.h.{teacher_idx}.{layer}.{w}""" ] snake_case = state_dict[F"""{prefix}.h.{teacher_idx}.attn.bias"""] else: for layer in [ "attention.self.query", "attention.self.key", "attention.self.value", "attention.output.dense", "attention.output.LayerNorm", "intermediate.dense", "output.dense", "output.LayerNorm", ]: for w in ["weight", "bias"]: snake_case = state_dict[ F"""{prefix}.encoder.layer.{teacher_idx}.{layer}.{w}""" ] std_idx += 1 # Language Modeling Head ###s if args.model_type == "roberta": for layer in ["lm_head.decoder.weight", "lm_head.bias"]: snake_case = state_dict[F"""{layer}"""] if args.vocab_transform: for w in ["weight", "bias"]: snake_case = state_dict[F"""lm_head.dense.{w}"""] snake_case = state_dict[F"""lm_head.layer_norm.{w}"""] elif args.model_type == "gpt2": for w in ["weight", "bias"]: snake_case = state_dict[F"""{prefix}.ln_f.{w}"""] snake_case = state_dict["""lm_head.weight"""] print(F"""N layers selected for distillation: {std_idx}""") print(F"""Number of params transferred for distillation: {len(compressed_sd.keys())}""") print(F"""Save transferred checkpoint to {args.dump_checkpoint}.""") torch.save(compressed_sd, args.dump_checkpoint)
62
import logging import os import quant_trainer import torch from torch.utils.data import DataLoader from transformers import Trainer, is_torch_tpu_available from transformers.trainer_utils import PredictionOutput snake_case = logging.getLogger(__name__) if is_torch_tpu_available(check_device=False): import torch_xla.core.xla_model as xm import torch_xla.debug.metrics as met class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , *UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Tuple=None , **UpperCAmelCase_ : Optional[Any] ): super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = eval_examples SCREAMING_SNAKE_CASE : List[Any] = post_process_function SCREAMING_SNAKE_CASE : Any = quant_trainer_args SCREAMING_SNAKE_CASE : Optional[Any] = 128 # default number of calibration samples def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple=None ): if calib_dataset is None and self.calib_dataset is None: raise ValueError("Trainer: calibration requires an calib_dataset." ) SCREAMING_SNAKE_CASE : str = calib_dataset if calib_dataset is not None else self.calib_dataset SCREAMING_SNAKE_CASE : str = self._remove_unused_columns(UpperCAmelCase_ , description="Calibration" ) return DataLoader( UpperCAmelCase_ , batch_size=self.args.eval_batch_size , collate_fn=self.data_collator , drop_last=self.args.dataloader_drop_last , num_workers=self.args.dataloader_num_workers , pin_memory=self.args.dataloader_pin_memory , shuffle=UpperCAmelCase_ , ) def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[int]=None ): SCREAMING_SNAKE_CASE : Any = self.train_dataset if calib_dataset is None else calib_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_calib_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args , calib=UpperCAmelCase_ ) model.eval() quant_trainer.enable_calibration(UpperCAmelCase_ ) logger.info("***** Running calibration *****" ) logger.info(f''' Num examples = {self.calib_num}''' ) logger.info(f''' Batch size = {calib_dataloader.batch_size}''' ) for step, inputs in enumerate(UpperCAmelCase_ ): # Prediction step SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = self.prediction_step(UpperCAmelCase_ , UpperCAmelCase_ , prediction_loss_only=UpperCAmelCase_ ) if (step + 1) * calib_dataloader.batch_size >= self.calib_num: break quant_trainer.finish_calibration(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : Optional[int] = model def _A ( self : List[Any] , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : str = "eval" ): SCREAMING_SNAKE_CASE : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset SCREAMING_SNAKE_CASE : Tuple = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.eval_examples if eval_examples is None else eval_examples # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Dict = self.compute_metrics SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : int = eval_loop( UpperCAmelCase_ , description="Evaluation" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : int = compute_metrics if self.post_process_function is not None and self.compute_metrics is not None: SCREAMING_SNAKE_CASE : List[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions ) SCREAMING_SNAKE_CASE : Any = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : Union[str, Any] = metrics.pop(UpperCAmelCase_ ) self.log(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[Any] = {} if self.args.tpu_metrics_debug or self.args.debug: # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.) xm.master_print(met.metrics_report() ) SCREAMING_SNAKE_CASE : List[str] = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCAmelCase_ ) return metrics def _A ( self : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : str = "test" ): SCREAMING_SNAKE_CASE : str = self.get_test_dataloader(UpperCAmelCase_ ) # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Tuple = self.compute_metrics SCREAMING_SNAKE_CASE : Union[str, Any] = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : str = eval_loop( UpperCAmelCase_ , description="Prediction" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : Union[str, Any] = compute_metrics if self.post_process_function is None or self.compute_metrics is None: return output SCREAMING_SNAKE_CASE : Optional[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions , "predict" ) SCREAMING_SNAKE_CASE : str = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : str = metrics.pop(UpperCAmelCase_ ) return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : int="./" ): SCREAMING_SNAKE_CASE : List[Any] = self.eval_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = next(iter(UpperCAmelCase_ ) ) # saving device - to make it consistent SCREAMING_SNAKE_CASE : int = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) # convert to tuple SCREAMING_SNAKE_CASE : Tuple = tuple(v.to(UpperCAmelCase_ ) for k, v in batch.items() ) logger.info("Converting model to be onnx compatible" ) from pytorch_quantization.nn import TensorQuantizer SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Any = self.model.to(UpperCAmelCase_ ) model.eval() model.float() SCREAMING_SNAKE_CASE : str = model.module if hasattr(UpperCAmelCase_ , "module" ) else model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : List[Any] = os.path.join(UpperCAmelCase_ , "model.onnx" ) logger.info(f'''exporting model to {output_model_file}''' ) SCREAMING_SNAKE_CASE : int = {0: "batch_size", 1: "seq_len"} torch.onnx.export( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , export_params=UpperCAmelCase_ , opset_version=13 , do_constant_folding=UpperCAmelCase_ , input_names=["input_ids", "attention_mask", "token_type_ids"] , output_names=["output_start_logits", "output_end_logits"] , dynamic_axes={ "input_ids": axes, "attention_mask": axes, "token_type_ids": axes, "output_start_logits": axes, "output_end_logits": axes, } , verbose=UpperCAmelCase_ , ) logger.info("onnx export finished" )
62
1
import argparse import os import sys from unittest.mock import patch import pytorch_lightning as pl import timeout_decorator import torch from distillation import SummarizationDistiller, distill_main from finetune import SummarizationModule, main from transformers import MarianMTModel from transformers.file_utils import cached_path from transformers.testing_utils import TestCasePlus, require_torch_gpu, slow from utils import load_json snake_case = """sshleifer/mar_enro_6_3_student""" class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[str] ): super().setUp() SCREAMING_SNAKE_CASE : Union[str, Any] = cached_path( "https://cdn-datasets.huggingface.co/translation/wmt_en_ro-tr40k-va0.5k-te0.5k.tar.gz" , extract_compressed_file=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : List[str] = f'''{data_cached}/wmt_en_ro-tr40k-va0.5k-te0.5k''' @slow @require_torch_gpu def _A ( self : List[Any] ): MarianMTModel.from_pretrained(UpperCAmelCase_ ) @slow @require_torch_gpu def _A ( self : int ): SCREAMING_SNAKE_CASE : Dict = { "$MAX_LEN": 64, "$BS": 64, "$GAS": 1, "$ENRO_DIR": self.data_dir, "facebook/mbart-large-cc25": MARIAN_MODEL, # "val_check_interval=0.25": "val_check_interval=1.0", "--learning_rate=3e-5": "--learning_rate 3e-4", "--num_train_epochs 6": "--num_train_epochs 1", } # Clean up bash script SCREAMING_SNAKE_CASE : Dict = (self.test_file_dir / "train_mbart_cc25_enro.sh").open().read().split("finetune.py" )[1].strip() SCREAMING_SNAKE_CASE : List[Any] = bash_script.replace("\\\n" , "" ).strip().replace("\"$@\"" , "" ) for k, v in env_vars_to_replace.items(): SCREAMING_SNAKE_CASE : str = bash_script.replace(UpperCAmelCase_ , str(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : int = self.get_auto_remove_tmp_dir() # bash_script = bash_script.replace("--fp16 ", "") SCREAMING_SNAKE_CASE : str = f''' --output_dir {output_dir} --tokenizer_name Helsinki-NLP/opus-mt-en-ro --sortish_sampler --do_predict --gpus 1 --freeze_encoder --n_train 40000 --n_val 500 --n_test 500 --fp16_opt_level O1 --num_sanity_val_steps 0 --eval_beams 2 '''.split() # XXX: args.gpus > 1 : handle multi_gpu in the future SCREAMING_SNAKE_CASE : List[Any] = ["finetune.py"] + bash_script.split() + args with patch.object(UpperCAmelCase_ , "argv" , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = argparse.ArgumentParser() SCREAMING_SNAKE_CASE : Optional[int] = pl.Trainer.add_argparse_args(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = SummarizationModule.add_model_specific_args(UpperCAmelCase_ , os.getcwd() ) SCREAMING_SNAKE_CASE : Tuple = parser.parse_args() SCREAMING_SNAKE_CASE : Optional[int] = main(UpperCAmelCase_ ) # Check metrics SCREAMING_SNAKE_CASE : Union[str, Any] = load_json(model.metrics_save_path ) SCREAMING_SNAKE_CASE : Optional[Any] = metrics["val"][0] SCREAMING_SNAKE_CASE : str = metrics["val"][-1] self.assertEqual(len(metrics["val"] ) , (args.max_epochs / args.val_check_interval) ) assert isinstance(last_step_stats[f'''val_avg_{model.val_metric}'''] , UpperCAmelCase_ ) self.assertGreater(last_step_stats["val_avg_gen_time"] , 0.01 ) # model hanging on generate. Maybe bad config was saved. (XXX: old comment/assert?) self.assertLessEqual(last_step_stats["val_avg_gen_time"] , 1.0 ) # test learning requirements: # 1. BLEU improves over the course of training by more than 2 pts self.assertGreater(last_step_stats["val_avg_bleu"] - first_step_stats["val_avg_bleu"] , 2 ) # 2. BLEU finishes above 17 self.assertGreater(last_step_stats["val_avg_bleu"] , 17 ) # 3. test BLEU and val BLEU within ~1.1 pt. self.assertLess(abs(metrics["val"][-1]["val_avg_bleu"] - metrics["test"][-1]["test_avg_bleu"] ) , 1.1 ) # check lightning ckpt can be loaded and has a reasonable statedict SCREAMING_SNAKE_CASE : str = os.listdir(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [x for x in contents if x.endswith(".ckpt" )][0] SCREAMING_SNAKE_CASE : List[Any] = os.path.join(args.output_dir , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = torch.load(UpperCAmelCase_ , map_location="cpu" ) SCREAMING_SNAKE_CASE : List[Any] = "model.model.decoder.layers.0.encoder_attn_layer_norm.weight" assert expected_key in ckpt["state_dict"] assert ckpt["state_dict"]["model.model.decoder.layers.0.encoder_attn_layer_norm.weight"].dtype == torch.floataa # TODO: turn on args.do_predict when PL bug fixed. if args.do_predict: SCREAMING_SNAKE_CASE : Tuple = {os.path.basename(UpperCAmelCase_ ) for p in contents} assert "test_generations.txt" in contents assert "test_results.txt" in contents # assert len(metrics["val"]) == desired_n_evals assert len(metrics["test"] ) == 1 class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @timeout_decorator.timeout(600 ) @slow @require_torch_gpu def _A ( self : Dict ): SCREAMING_SNAKE_CASE : str = f'''{self.test_file_dir_str}/test_data/wmt_en_ro''' SCREAMING_SNAKE_CASE : Tuple = { "--fp16_opt_level=O1": "", "$MAX_LEN": 128, "$BS": 16, "$GAS": 1, "$ENRO_DIR": data_dir, "$m": "sshleifer/student_marian_en_ro_6_1", "val_check_interval=0.25": "val_check_interval=1.0", } # Clean up bash script SCREAMING_SNAKE_CASE : Union[str, Any] = ( (self.test_file_dir / "distil_marian_no_teacher.sh").open().read().split("distillation.py" )[1].strip() ) SCREAMING_SNAKE_CASE : Dict = bash_script.replace("\\\n" , "" ).strip().replace("\"$@\"" , "" ) SCREAMING_SNAKE_CASE : Optional[int] = bash_script.replace("--fp16 " , " " ) for k, v in env_vars_to_replace.items(): SCREAMING_SNAKE_CASE : str = bash_script.replace(UpperCAmelCase_ , str(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : List[str] = self.get_auto_remove_tmp_dir() SCREAMING_SNAKE_CASE : Dict = bash_script.replace("--fp16" , "" ) SCREAMING_SNAKE_CASE : Tuple = 6 SCREAMING_SNAKE_CASE : int = ( ["distillation.py"] + bash_script.split() + [ f'''--output_dir={output_dir}''', "--gpus=1", "--learning_rate=1e-3", f'''--num_train_epochs={epochs}''', "--warmup_steps=10", "--val_check_interval=1.0", "--do_predict", ] ) with patch.object(UpperCAmelCase_ , "argv" , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Union[str, Any] = argparse.ArgumentParser() SCREAMING_SNAKE_CASE : Optional[int] = pl.Trainer.add_argparse_args(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = SummarizationDistiller.add_model_specific_args(UpperCAmelCase_ , os.getcwd() ) SCREAMING_SNAKE_CASE : Dict = parser.parse_args() # assert args.gpus == gpus THIS BREAKS for multi_gpu SCREAMING_SNAKE_CASE : Union[str, Any] = distill_main(UpperCAmelCase_ ) # Check metrics SCREAMING_SNAKE_CASE : List[str] = load_json(model.metrics_save_path ) SCREAMING_SNAKE_CASE : Any = metrics["val"][0] SCREAMING_SNAKE_CASE : Any = metrics["val"][-1] assert len(metrics["val"] ) >= (args.max_epochs / args.val_check_interval) # +1 accounts for val_sanity_check assert last_step_stats["val_avg_gen_time"] >= 0.01 assert first_step_stats["val_avg_bleu"] < last_step_stats["val_avg_bleu"] # model learned nothing assert 1.0 >= last_step_stats["val_avg_gen_time"] # model hanging on generate. Maybe bad config was saved. assert isinstance(last_step_stats[f'''val_avg_{model.val_metric}'''] , UpperCAmelCase_ ) # check lightning ckpt can be loaded and has a reasonable statedict SCREAMING_SNAKE_CASE : int = os.listdir(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = [x for x in contents if x.endswith(".ckpt" )][0] SCREAMING_SNAKE_CASE : int = os.path.join(args.output_dir , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = torch.load(UpperCAmelCase_ , map_location="cpu" ) SCREAMING_SNAKE_CASE : List[Any] = "model.model.decoder.layers.0.encoder_attn_layer_norm.weight" assert expected_key in ckpt["state_dict"] assert ckpt["state_dict"]["model.model.decoder.layers.0.encoder_attn_layer_norm.weight"].dtype == torch.floataa # TODO: turn on args.do_predict when PL bug fixed. if args.do_predict: SCREAMING_SNAKE_CASE : Optional[Any] = {os.path.basename(UpperCAmelCase_ ) for p in contents} assert "test_generations.txt" in contents assert "test_results.txt" in contents # assert len(metrics["val"]) == desired_n_evals assert len(metrics["test"] ) == 1
62
import os import unittest from transformers import LayoutLMTokenizer, LayoutLMTokenizerFast from transformers.models.layoutlm.tokenization_layoutlm import VOCAB_FILES_NAMES from transformers.testing_utils import require_tokenizers from ...test_tokenization_common import TokenizerTesterMixin @require_tokenizers class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = LayoutLMTokenizer UpperCamelCase_ : str = LayoutLMTokenizerFast UpperCamelCase_ : Any = True UpperCamelCase_ : Optional[Any] = True def _A ( self : Any ): super().setUp() SCREAMING_SNAKE_CASE : Optional[Any] = [ "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest", ] SCREAMING_SNAKE_CASE : Optional[Any] = 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 _A ( self : str , **UpperCAmelCase_ : Optional[int] ): return LayoutLMTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Any = "UNwant\u00E9d,running" SCREAMING_SNAKE_CASE : Union[str, Any] = "unwanted, running" return input_text, output_text def _A ( self : int ): SCREAMING_SNAKE_CASE : Optional[int] = self.tokenizer_class(self.vocab_file ) SCREAMING_SNAKE_CASE : Dict = tokenizer.tokenize("UNwant\u00E9d,running" ) self.assertListEqual(UpperCAmelCase_ , ["un", "##want", "##ed", ",", "runn", "##ing"] ) self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , [7, 4, 5, 10, 8, 9] ) def _A ( self : List[str] ): pass
62
1
import json from typing import List, Optional, Tuple from tokenizers import normalizers from tokenizers.pre_tokenizers import BertPreTokenizer, PreTokenizer from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging from .tokenization_roformer import RoFormerTokenizer from .tokenization_utils import JiebaPreTokenizer snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """vocab.txt""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """junnyu/roformer_chinese_small""": """https://huggingface.co/junnyu/roformer_chinese_small/resolve/main/vocab.txt""", """junnyu/roformer_chinese_base""": """https://huggingface.co/junnyu/roformer_chinese_base/resolve/main/vocab.txt""", """junnyu/roformer_chinese_char_small""": ( """https://huggingface.co/junnyu/roformer_chinese_char_small/resolve/main/vocab.txt""" ), """junnyu/roformer_chinese_char_base""": ( """https://huggingface.co/junnyu/roformer_chinese_char_base/resolve/main/vocab.txt""" ), """junnyu/roformer_small_discriminator""": ( """https://huggingface.co/junnyu/roformer_small_discriminator/resolve/main/vocab.txt""" ), """junnyu/roformer_small_generator""": ( """https://huggingface.co/junnyu/roformer_small_generator/resolve/main/vocab.txt""" ), } } snake_case = { """junnyu/roformer_chinese_small""": 1_536, """junnyu/roformer_chinese_base""": 1_536, """junnyu/roformer_chinese_char_small""": 512, """junnyu/roformer_chinese_char_base""": 512, """junnyu/roformer_small_discriminator""": 128, """junnyu/roformer_small_generator""": 128, } snake_case = { """junnyu/roformer_chinese_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_base""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_base""": {"""do_lower_case""": True}, """junnyu/roformer_small_discriminator""": {"""do_lower_case""": True}, """junnyu/roformer_small_generator""": {"""do_lower_case""": True}, } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = VOCAB_FILES_NAMES UpperCamelCase_ : int = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Tuple = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : List[Any] = PRETRAINED_INIT_CONFIGURATION UpperCamelCase_ : Any = RoFormerTokenizer def __init__( self : Tuple , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : List[str]="[UNK]" , UpperCAmelCase_ : Any="[SEP]" , UpperCAmelCase_ : Any="[PAD]" , UpperCAmelCase_ : List[str]="[CLS]" , UpperCAmelCase_ : str="[MASK]" , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : Optional[Any]=None , **UpperCAmelCase_ : List[str] , ): super().__init__( UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , do_lower_case=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , tokenize_chinese_chars=UpperCAmelCase_ , strip_accents=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = json.loads(self.backend_tokenizer.normalizer.__getstate__() ) if ( pre_tok_state.get("lowercase" , UpperCAmelCase_ ) != do_lower_case or pre_tok_state.get("strip_accents" , UpperCAmelCase_ ) != strip_accents ): SCREAMING_SNAKE_CASE : Optional[Any] = getattr(UpperCAmelCase_ , pre_tok_state.pop("type" ) ) SCREAMING_SNAKE_CASE : Any = do_lower_case SCREAMING_SNAKE_CASE : List[str] = strip_accents SCREAMING_SNAKE_CASE : Tuple = pre_tok_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = do_lower_case def __getstate__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = self.__dict__.copy() SCREAMING_SNAKE_CASE : Optional[Any] = BertPreTokenizer() return state def __setstate__( self : Tuple , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = d SCREAMING_SNAKE_CASE : Dict = self.__dict__["_tokenizer"].get_vocab() SCREAMING_SNAKE_CASE : Any = PreTokenizer.custom(JiebaPreTokenizer(UpperCAmelCase_ ) ) def _A ( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Union[str, Any]=None ): SCREAMING_SNAKE_CASE : List[Any] = [self.cls_token_id] + token_ids_a + [self.sep_token_id] if token_ids_a: output += token_ids_a + [self.sep_token_id] return output def _A ( self : Tuple , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : List[Any] = [self.sep_token_id] SCREAMING_SNAKE_CASE : Optional[int] = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1] def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): SCREAMING_SNAKE_CASE : Optional[int] = self._tokenizer.model.save(UpperCAmelCase_ , name=UpperCAmelCase_ ) return tuple(UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Any=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : Union[str, Any] = BertPreTokenizer() return super().save_pretrained(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ )
62
import argparse import json import os from collections import OrderedDict import numpy as np import tensorflow as tf import torch def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = os.path.join(args.tf_model_dir , "parameters.json" ) SCREAMING_SNAKE_CASE : Union[str, Any] = json.loads(open(lowercase ).read() ) if not params: raise ValueError( F'''It seems that the json file at {parameter_file} is empty. Make sure you have a correct json file.''' ) if not args.output.endswith(".pt" ): SCREAMING_SNAKE_CASE : Optional[int] = args.output + ".pt" SCREAMING_SNAKE_CASE : Any = OrderedDict() with tf.device("/CPU:0" ): SCREAMING_SNAKE_CASE : Union[str, Any] = tf.train.load_checkpoint(args.tf_model_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = reader.get_variable_to_shape_map() for key_name in shapes.keys(): SCREAMING_SNAKE_CASE : Any = reader.get_tensor(lowercase ).astype(np.floataa ) if key_name.endswith("/adam_m" ) or key_name.endswith("/adam_v" ): continue if key_name.startswith("pasts/" ): if key_name.startswith("pasts/mlp" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9] ) elif key_name.startswith("pasts/out" ): SCREAMING_SNAKE_CASE : Optional[int] = 8 SCREAMING_SNAKE_CASE : List[Any] = "model.sqout.%d.weight" % (player * 2) # enter to nn.Sequencial with Tanh, so 2 at a time SCREAMING_SNAKE_CASE : Optional[int] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.startswith("model/moe" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/switch_gating/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.router.classifier.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/softmlp/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.soft_bypass_mlp.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/wo/kernel" ) or key_name.endswith("/wi/kernel" ): SCREAMING_SNAKE_CASE : Optional[int] = key_name[-9:-7] for i in range(16 ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.feed_forward.mlp.experts.expert_%d.%s.weight" % (player, i, nlayer) SCREAMING_SNAKE_CASE : List[str] = ( vnp[i].transpose([1, 0] ).copy() ) # In Mesh-Tensorflow, it is one array, so it is divided SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name.startswith("model/mlp" ): SCREAMING_SNAKE_CASE : str = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/p1/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.mlp.wi.weight" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/p1/bias" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.wi.bias" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/kernel" ): SCREAMING_SNAKE_CASE : str = "model.blocks.%d.feed_forward.mlp.wo.weight" % player SCREAMING_SNAKE_CASE : List[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/bias" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.feed_forward.mlp.wo.bias" % player SCREAMING_SNAKE_CASE : str = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) elif key_name.startswith("model/ln" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.norm.bias" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : List[str] = "model.blocks.%d.feed_forward.norm.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/att" ): SCREAMING_SNAKE_CASE : Optional[int] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/qkv/kernel" ): SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # Compute same dimension as Mesh-tensorflow using einsum SCREAMING_SNAKE_CASE : List[str] = state[:, 0, :, :] SCREAMING_SNAKE_CASE : Tuple = state[:, 1, :, :] SCREAMING_SNAKE_CASE : List[Any] = state[:, 2, :, :] SCREAMING_SNAKE_CASE : Tuple = ( state_q.reshape([state_q.shape[0], state_q.shape[1] * state_q.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = ( state_k.reshape([state_k.shape[0], state_k.shape[1] * state_k.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = ( state_v.reshape([state_v.shape[0], state_v.shape[1] * state_v.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : int = "model.blocks.%d.self_attn.self_attn.q_proj.weight" % player SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = "model.blocks.%d.self_attn.self_attn.k_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = "model.blocks.%d.self_attn.self_attn.v_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/o/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.self_attn.self_attn.out_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = ( vnp.reshape([vnp.shape[0] * vnp.shape[1], vnp.shape[2]] ).transpose([1, 0] ).copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif key_name.startswith("model/an" ): SCREAMING_SNAKE_CASE : int = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.self_attn.norm.bias" % player SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[str] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.self_attn.norm.weight" % player SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif ( key_name.startswith("model/wte" ) or key_name.startswith("model/wpe" ) or key_name.startswith("model/ete" ) ): SCREAMING_SNAKE_CASE : str = {"wte": "embed_tokens", "wpe": "position_embeddings", "ete": "extra_position_embeddings"}[ key_name[-3:] ] SCREAMING_SNAKE_CASE : List[str] = "model.%s.weight" % nlayer SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) if key_name.startswith("model/wte" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "lm_head.weight" SCREAMING_SNAKE_CASE : List[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/wob" ): SCREAMING_SNAKE_CASE : List[Any] = "final_logits_bias" SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : List[str] = state.reshape((1, -1) ) SCREAMING_SNAKE_CASE : int = torch.tensor(lowercase ) elif key_name == "model/dense/kernel": SCREAMING_SNAKE_CASE : Optional[int] = "model.last_project.weight" SCREAMING_SNAKE_CASE : Dict = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name == "model/dense_1/bias": SCREAMING_SNAKE_CASE : str = "model.last_project.bias" SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) torch.save(lowercase , args.output ) if __name__ == "__main__": snake_case = argparse.ArgumentParser( description="""model converter.""", formatter_class=argparse.ArgumentDefaultsHelpFormatter ) parser.add_argument("""--tf_model_dir""", metavar="""PATH""", type=str, required=True, help="""import model""") parser.add_argument("""--output""", metavar="""PATH""", type=str, required=True, help="""output model""") snake_case = parser.parse_args() convert_tf_gptsan_to_pt(args)
62
1
from __future__ import annotations class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[int] , UpperCAmelCase_ : list[list[int]] ): SCREAMING_SNAKE_CASE : List[str] = TypeError( "Matrices must be formed from a list of zero or more lists containing at " "least one and the same number of values, each of which must be of type " "int or float." ) if len(UpperCAmelCase_ ) != 0: SCREAMING_SNAKE_CASE : Optional[int] = len(rows[0] ) if cols == 0: raise error for row in rows: if len(UpperCAmelCase_ ) != cols: raise error for value in row: if not isinstance(UpperCAmelCase_ , (int, float) ): raise error SCREAMING_SNAKE_CASE : Optional[Any] = rows else: SCREAMING_SNAKE_CASE : Tuple = [] def _A ( self : Dict ): return [[row[i] for row in self.rows] for i in range(len(self.rows[0] ) )] @property def _A ( self : List[str] ): return len(self.rows ) @property def _A ( self : List[Any] ): return len(self.rows[0] ) @property def _A ( self : Optional[Any] ): return (self.num_rows, self.num_columns) @property def _A ( self : Optional[int] ): return self.order[0] == self.order[1] def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = [ [0 if column_num != row_num else 1 for column_num in range(self.num_rows )] for row_num in range(self.num_rows ) ] return Matrix(UpperCAmelCase_ ) def _A ( self : Union[str, Any] ): if not self.is_square: return 0 if self.order == (0, 0): return 1 if self.order == (1, 1): return int(self.rows[0][0] ) if self.order == (2, 2): return int( (self.rows[0][0] * self.rows[1][1]) - (self.rows[0][1] * self.rows[1][0]) ) else: return sum( self.rows[0][column] * self.cofactors().rows[0][column] for column in range(self.num_columns ) ) def _A ( self : List[str] ): return bool(self.determinant() ) def _A ( self : Optional[int] , UpperCAmelCase_ : int , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[Any] = [ [ self.rows[other_row][other_column] for other_column in range(self.num_columns ) if other_column != column ] for other_row in range(self.num_rows ) if other_row != row ] return Matrix(UpperCAmelCase_ ).determinant() def _A ( self : Union[str, Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : int ): if (row + column) % 2 == 0: return self.get_minor(UpperCAmelCase_ , UpperCAmelCase_ ) return -1 * self.get_minor(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[str] ): return Matrix( [ [self.get_minor(UpperCAmelCase_ , UpperCAmelCase_ ) for column in range(self.num_columns )] for row in range(self.num_rows ) ] ) def _A ( self : Dict ): return Matrix( [ [ self.minors().rows[row][column] if (row + column) % 2 == 0 else self.minors().rows[row][column] * -1 for column in range(self.minors().num_columns ) ] for row in range(self.minors().num_rows ) ] ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : int = [ [self.cofactors().rows[column][row] for column in range(self.num_columns )] for row in range(self.num_rows ) ] return Matrix(UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : Optional[int] = self.determinant() if not determinant: raise TypeError("Only matrices with a non-zero determinant have an inverse" ) return self.adjugate() * (1 / determinant) def __repr__( self : Optional[int] ): return str(self.rows ) def __str__( self : int ): if self.num_rows == 0: return "[]" if self.num_rows == 1: return "[[" + ". ".join(str(self.rows[0] ) ) + "]]" return ( "[" + "\n ".join( [ "[" + ". ".join([str(UpperCAmelCase_ ) for value in row] ) + ".]" for row in self.rows ] ) + "]" ) def _A ( self : int , UpperCAmelCase_ : list[int] , UpperCAmelCase_ : int | None = None ): SCREAMING_SNAKE_CASE : int = TypeError("Row must be a list containing all ints and/or floats" ) if not isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): raise type_error for value in row: if not isinstance(UpperCAmelCase_ , (int, float) ): raise type_error if len(UpperCAmelCase_ ) != self.num_columns: raise ValueError( "Row must be equal in length to the other rows in the matrix" ) if position is None: self.rows.append(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[str] = self.rows[0:position] + [row] + self.rows[position:] def _A ( self : Optional[Any] , UpperCAmelCase_ : list[int] , UpperCAmelCase_ : int | None = None ): SCREAMING_SNAKE_CASE : List[str] = TypeError( "Column must be a list containing all ints and/or floats" ) if not isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): raise type_error for value in column: if not isinstance(UpperCAmelCase_ , (int, float) ): raise type_error if len(UpperCAmelCase_ ) != self.num_rows: raise ValueError( "Column must be equal in length to the other columns in the matrix" ) if position is None: SCREAMING_SNAKE_CASE : Union[str, Any] = [self.rows[i] + [column[i]] for i in range(self.num_rows )] else: SCREAMING_SNAKE_CASE : List[str] = [ self.rows[i][0:position] + [column[i]] + self.rows[i][position:] for i in range(self.num_rows ) ] def __eq__( self : Any , UpperCAmelCase_ : object ): if not isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): return NotImplemented return self.rows == other.rows def __ne__( self : List[Any] , UpperCAmelCase_ : object ): return not self == other def __neg__( self : Tuple ): return self * -1 def __add__( self : str , UpperCAmelCase_ : Matrix ): if self.order != other.order: raise ValueError("Addition requires matrices of the same order" ) return Matrix( [ [self.rows[i][j] + other.rows[i][j] for j in range(self.num_columns )] for i in range(self.num_rows ) ] ) def __sub__( self : Any , UpperCAmelCase_ : Matrix ): if self.order != other.order: raise ValueError("Subtraction requires matrices of the same order" ) return Matrix( [ [self.rows[i][j] - other.rows[i][j] for j in range(self.num_columns )] for i in range(self.num_rows ) ] ) def __mul__( self : Any , UpperCAmelCase_ : Matrix | int | float ): if isinstance(UpperCAmelCase_ , (int, float) ): return Matrix( [[int(element * other ) for element in row] for row in self.rows] ) elif isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): if self.num_columns != other.num_rows: raise ValueError( "The number of columns in the first matrix must " "be equal to the number of rows in the second" ) return Matrix( [ [Matrix.dot_product(UpperCAmelCase_ , UpperCAmelCase_ ) for column in other.columns()] for row in self.rows ] ) else: raise TypeError( "A Matrix can only be multiplied by an int, float, or another matrix" ) def __pow__( self : Any , UpperCAmelCase_ : int ): if not isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): raise TypeError("A Matrix can only be raised to the power of an int" ) if not self.is_square: raise ValueError("Only square matrices can be raised to a power" ) if other == 0: return self.identity() if other < 0: if self.is_invertable(): return self.inverse() ** (-other) raise ValueError( "Only invertable matrices can be raised to a negative power" ) SCREAMING_SNAKE_CASE : Tuple = self for _ in range(other - 1 ): result *= self return result @classmethod def _A ( cls : List[Any] , UpperCAmelCase_ : list[int] , UpperCAmelCase_ : list[int] ): return sum(row[i] * column[i] for i in range(len(UpperCAmelCase_ ) ) ) if __name__ == "__main__": import doctest doctest.testmod()
62
import json import os import shutil import tempfile import unittest import numpy as np from transformers import BertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer from transformers.testing_utils import require_tokenizers, require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor @require_tokenizers @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = tempfile.mkdtemp() # fmt: off SCREAMING_SNAKE_CASE : Any = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest"] # fmt: on SCREAMING_SNAKE_CASE : Dict = 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] ) ) SCREAMING_SNAKE_CASE : Optional[int] = { "do_resize": True, "size": {"height": 18, "width": 18}, "do_normalize": True, "image_mean": [0.5, 0.5, 0.5], "image_std": [0.5, 0.5, 0.5], } SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(self.tmpdirname , UpperCAmelCase_ ) with open(self.image_processor_file , "w" , encoding="utf-8" ) as fp: json.dump(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] , **UpperCAmelCase_ : List[str] ): return BertTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , **UpperCAmelCase_ : Any ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] ): shutil.rmtree(self.tmpdirname ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] SCREAMING_SNAKE_CASE : Union[str, Any] = [Image.fromarray(np.moveaxis(UpperCAmelCase_ , 0 , -1 ) ) for x in image_inputs] return image_inputs def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : int = VisionTextDualEncoderProcessor( tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer(bos_token="(BOS)" , eos_token="(EOS)" ) SCREAMING_SNAKE_CASE : Any = self.get_image_processor(do_normalize=UpperCAmelCase_ , padding_value=1.0 ) SCREAMING_SNAKE_CASE : str = VisionTextDualEncoderProcessor.from_pretrained( self.tmpdirname , bos_token="(BOS)" , eos_token="(EOS)" , do_normalize=UpperCAmelCase_ , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : int = image_processor(UpperCAmelCase_ , return_tensors="np" ) SCREAMING_SNAKE_CASE : Optional[int] = processor(images=UpperCAmelCase_ , 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 _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = "lower newer" SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = tokenizer(UpperCAmelCase_ ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : int = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer() SCREAMING_SNAKE_CASE : List[str] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = "lower newer" SCREAMING_SNAKE_CASE : Union[str, Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Optional[Any] = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , ["input_ids", "token_type_ids", "attention_mask", "pixel_values"] ) # test if it raises when no input is passed with self.assertRaises(UpperCAmelCase_ ): processor() def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] SCREAMING_SNAKE_CASE : Optional[Any] = processor.batch_decode(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer.batch_decode(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = "lower newer" SCREAMING_SNAKE_CASE : Any = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
62
1
snake_case = 8.314462 # Unit - J mol-1 K-1 def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if moles < 0 or kelvin < 0 or volume < 0: raise ValueError("Invalid inputs. Enter positive value." ) return moles * kelvin * UNIVERSAL_GAS_CONSTANT / volume def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" if moles < 0 or kelvin < 0 or pressure < 0: raise ValueError("Invalid inputs. Enter positive value." ) return moles * kelvin * UNIVERSAL_GAS_CONSTANT / pressure if __name__ == "__main__": from doctest import testmod testmod()
62
import inspect import unittest from transformers import MobileNetVaConfig from transformers.testing_utils import require_torch, require_vision, slow, torch_device from transformers.utils import cached_property, is_torch_available, is_vision_available from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor from ...test_pipeline_mixin import PipelineTesterMixin if is_torch_available(): import torch from transformers import MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation, MobileNetVaModel from transformers.models.mobilenet_va.modeling_mobilenet_va import MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST if is_vision_available(): from PIL import Image from transformers import MobileNetVaImageProcessor class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.config_class(**self.inputs_dict ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "tf_padding" ) ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "depth_multiplier" ) ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : List[str]=13 , UpperCAmelCase_ : Any=3 , UpperCAmelCase_ : Union[str, Any]=32 , UpperCAmelCase_ : int=0.25 , UpperCAmelCase_ : Union[str, Any]=8 , UpperCAmelCase_ : Dict=8 , UpperCAmelCase_ : Optional[int]=6 , UpperCAmelCase_ : Optional[Any]=32 , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : str=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : str="relu6" , UpperCAmelCase_ : List[str]=1280 , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : Any=0.02 , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=10 , UpperCAmelCase_ : Optional[Any]=None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = parent SCREAMING_SNAKE_CASE : int = batch_size SCREAMING_SNAKE_CASE : Dict = num_channels SCREAMING_SNAKE_CASE : Dict = image_size SCREAMING_SNAKE_CASE : int = depth_multiplier SCREAMING_SNAKE_CASE : str = depth_divisible_by SCREAMING_SNAKE_CASE : Union[str, Any] = min_depth SCREAMING_SNAKE_CASE : int = expand_ratio SCREAMING_SNAKE_CASE : Tuple = tf_padding SCREAMING_SNAKE_CASE : List[str] = output_stride SCREAMING_SNAKE_CASE : Optional[int] = first_layer_is_expansion SCREAMING_SNAKE_CASE : Any = finegrained_output SCREAMING_SNAKE_CASE : Optional[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[int] = last_hidden_size if finegrained_output else int(last_hidden_size * depth_multiplier ) SCREAMING_SNAKE_CASE : Any = classifier_dropout_prob SCREAMING_SNAKE_CASE : Dict = use_labels SCREAMING_SNAKE_CASE : int = is_training SCREAMING_SNAKE_CASE : Dict = num_labels SCREAMING_SNAKE_CASE : Optional[Any] = initializer_range SCREAMING_SNAKE_CASE : Union[str, Any] = scope def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] ) SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : List[Any] = None if self.use_labels: SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size] , self.num_labels ) SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size, self.image_size, self.image_size] , self.num_labels ) SCREAMING_SNAKE_CASE : Tuple = self.get_config() return config, pixel_values, labels, pixel_labels def _A ( self : Optional[int] ): return MobileNetVaConfig( num_channels=self.num_channels , image_size=self.image_size , depth_multiplier=self.depth_multiplier , depth_divisible_by=self.depth_divisible_by , min_depth=self.min_depth , expand_ratio=self.expand_ratio , output_stride=self.output_stride , first_layer_is_expansion=self.first_layer_is_expansion , finegrained_output=self.finegrained_output , hidden_act=self.hidden_act , tf_padding=self.tf_padding , classifier_dropout_prob=self.classifier_dropout_prob , initializer_range=self.initializer_range , ) def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaModel(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ ) self.parent.assertEqual( result.last_hidden_state.shape , ( self.batch_size, self.last_hidden_size, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) self.parent.assertEqual( result.pooler_output.shape , (self.batch_size, self.last_hidden_size) , ) def _A ( self : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : str = self.num_labels SCREAMING_SNAKE_CASE : List[Any] = MobileNetVaForImageClassification(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Dict = MobileNetVaForSemanticSegmentation(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[int] = model(UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : List[Any] = self.prepare_config_and_inputs() SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = config_and_inputs SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": pixel_values} return config, inputs_dict @require_torch class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Any = ( (MobileNetVaModel, MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation) if is_torch_available() else () ) UpperCamelCase_ : List[Any] = ( { '''feature-extraction''': MobileNetVaModel, '''image-classification''': MobileNetVaForImageClassification, '''image-segmentation''': MobileNetVaForSemanticSegmentation, } if is_torch_available() else {} ) UpperCamelCase_ : Any = False UpperCamelCase_ : List[str] = False UpperCamelCase_ : int = False UpperCamelCase_ : str = False def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = MobileNetVaModelTester(self ) SCREAMING_SNAKE_CASE : Optional[Any] = MobileNetVaConfigTester(self , config_class=UpperCAmelCase_ , has_text_modality=UpperCAmelCase_ ) def _A ( self : Optional[Any] ): self.config_tester.run_common_tests() @unittest.skip(reason="MobileNetV2 does not use inputs_embeds" ) def _A ( self : List[Any] ): pass @unittest.skip(reason="MobileNetV2 does not support input and output embeddings" ) def _A ( self : Dict ): pass @unittest.skip(reason="MobileNetV2 does not output attentions" ) def _A ( self : Union[str, Any] ): pass def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : Tuple = model_class(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic SCREAMING_SNAKE_CASE : Dict = [*signature.parameters.keys()] SCREAMING_SNAKE_CASE : Union[str, Any] = ["pixel_values"] self.assertListEqual(arg_names[:1] , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*UpperCAmelCase_ ) def _A ( self : List[Any] ): def check_hidden_states_output(UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = model_class(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**self._prepare_for_class(UpperCAmelCase_ , UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : List[Any] = outputs.hidden_states SCREAMING_SNAKE_CASE : Any = 16 self.assertEqual(len(UpperCAmelCase_ ) , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : str = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) # check that output_hidden_states also work using config del inputs_dict["output_hidden_states"] SCREAMING_SNAKE_CASE : List[Any] = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Tuple = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_image_classification(*UpperCAmelCase_ ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_semantic_segmentation(*UpperCAmelCase_ ) @slow def _A ( self : Optional[Any] ): for model_name in MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: SCREAMING_SNAKE_CASE : int = MobileNetVaModel.from_pretrained(UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : int = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" ) return image @require_torch @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @cached_property def _A ( self : Optional[int] ): return ( MobileNetVaImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224" ) if is_vision_available() else None ) @slow def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaForImageClassification.from_pretrained("google/mobilenet_v2_1.0_224" ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.default_image_processor SCREAMING_SNAKE_CASE : Optional[int] = prepare_img() SCREAMING_SNAKE_CASE : Optional[int] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**UpperCAmelCase_ ) # verify the logits SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Size((1, 1001) ) self.assertEqual(outputs.logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = torch.tensor([0.2_445, -1.1_993, 0.1_905] ).to(UpperCAmelCase_ ) self.assertTrue(torch.allclose(outputs.logits[0, :3] , UpperCAmelCase_ , atol=1E-4 ) ) @slow def _A ( self : str ): SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaForSemanticSegmentation.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : int = model.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaImageProcessor.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : Optional[Any] = prepare_img() SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = outputs.logits # verify the logits SCREAMING_SNAKE_CASE : Dict = torch.Size((1, 21, 65, 65) ) self.assertEqual(logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.tensor( [ [[17.5_790, 17.7_581, 18.3_355], [18.3_257, 18.4_230, 18.8_973], [18.6_169, 18.8_650, 19.2_187]], [[-2.1_595, -2.0_977, -2.3_741], [-2.4_226, -2.3_028, -2.6_835], [-2.7_819, -2.5_991, -2.7_706]], [[4.2_058, 4.8_317, 4.7_638], [4.4_136, 5.0_361, 4.9_383], [4.5_028, 4.9_644, 4.8_734]], ] , device=UpperCAmelCase_ , ) self.assertTrue(torch.allclose(logits[0, :3, :3, :3] , UpperCAmelCase_ , atol=1E-4 ) )
62
1
def lowerCamelCase__ ( lowercase = 50 ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = [[0] * 3 for _ in range(length + 1 )] for row_length in range(length + 1 ): for tile_length in range(2 , 5 ): for tile_start in range(row_length - tile_length + 1 ): different_colour_ways_number[row_length][tile_length - 2] += ( different_colour_ways_number[row_length - tile_start - tile_length][ tile_length - 2 ] + 1 ) return sum(different_colour_ways_number[length] ) if __name__ == "__main__": print(F"""{solution() = }""")
62
import math from typing import Dict, Iterable, List, Optional, Tuple, Union import numpy as np from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format from ...image_utils import ( IMAGENET_STANDARD_MEAN, IMAGENET_STANDARD_STD, ChannelDimension, ImageInput, PILImageResampling, get_image_size, is_torch_available, is_torch_tensor, make_list_of_images, to_numpy_array, valid_images, ) from ...utils import TensorType, is_vision_available, logging if is_torch_available(): import torch if is_vision_available(): import PIL snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" def constraint_to_multiple_of(lowercase , lowercase , lowercase=0 , lowercase=None ): SCREAMING_SNAKE_CASE : Any = round(val / multiple ) * multiple if max_val is not None and x > max_val: SCREAMING_SNAKE_CASE : Optional[int] = math.floor(val / multiple ) * multiple if x < min_val: SCREAMING_SNAKE_CASE : int = math.ceil(val / multiple ) * multiple return x SCREAMING_SNAKE_CASE : Tuple = (output_size, output_size) if isinstance(lowercase , lowercase ) else output_size SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = get_image_size(lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = output_size # determine new height and width SCREAMING_SNAKE_CASE : Tuple = output_height / input_height SCREAMING_SNAKE_CASE : List[Any] = output_width / input_width if keep_aspect_ratio: # scale as little as possible if abs(1 - scale_width ) < abs(1 - scale_height ): # fit width SCREAMING_SNAKE_CASE : List[str] = scale_width else: # fit height SCREAMING_SNAKE_CASE : Optional[Any] = scale_height SCREAMING_SNAKE_CASE : int = constraint_to_multiple_of(scale_height * input_height , multiple=lowercase ) SCREAMING_SNAKE_CASE : Dict = constraint_to_multiple_of(scale_width * input_width , multiple=lowercase ) return (new_height, new_width) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Any = ['''pixel_values'''] def __init__( self : Any , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Dict[str, int] = None , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BILINEAR , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Union[int, float] = 1 / 255 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , **UpperCAmelCase_ : str , ): super().__init__(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = size if size is not None else {"height": 384, "width": 384} SCREAMING_SNAKE_CASE : List[Any] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = do_resize SCREAMING_SNAKE_CASE : Optional[Any] = size SCREAMING_SNAKE_CASE : str = keep_aspect_ratio SCREAMING_SNAKE_CASE : int = ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample SCREAMING_SNAKE_CASE : List[str] = do_rescale SCREAMING_SNAKE_CASE : Tuple = rescale_factor SCREAMING_SNAKE_CASE : Union[str, Any] = do_normalize SCREAMING_SNAKE_CASE : Union[str, Any] = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN SCREAMING_SNAKE_CASE : int = image_std if image_std is not None else IMAGENET_STANDARD_STD def _A ( self : Tuple , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Dict[str, int] , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : int , ): SCREAMING_SNAKE_CASE : List[str] = get_size_dict(UpperCAmelCase_ ) if "height" not in size or "width" not in size: raise ValueError(f'''The size dictionary must contain the keys \'height\' and \'width\'. Got {size.keys()}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = get_resize_output_image_size( UpperCAmelCase_ , output_size=(size["height"], size["width"]) , keep_aspect_ratio=UpperCAmelCase_ , multiple=UpperCAmelCase_ , ) return resize(UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[int, float] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : Dict , ): return rescale(UpperCAmelCase_ , scale=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Dict , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : List[str] , ): return normalize(UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : ImageInput , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : PILImageResampling = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : float = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[str, TensorType]] = None , UpperCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **UpperCAmelCase_ : Optional[Any] , ): SCREAMING_SNAKE_CASE : Optional[int] = do_resize if do_resize is not None else self.do_resize SCREAMING_SNAKE_CASE : Dict = size if size is not None else self.size SCREAMING_SNAKE_CASE : Optional[int] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = keep_aspect_ratio if keep_aspect_ratio is not None else self.keep_aspect_ratio SCREAMING_SNAKE_CASE : Optional[Any] = ensure_multiple_of if ensure_multiple_of is not None else self.ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample if resample is not None else self.resample SCREAMING_SNAKE_CASE : Dict = do_rescale if do_rescale is not None else self.do_rescale SCREAMING_SNAKE_CASE : Optional[Any] = rescale_factor if rescale_factor is not None else self.rescale_factor SCREAMING_SNAKE_CASE : Optional[int] = do_normalize if do_normalize is not None else self.do_normalize SCREAMING_SNAKE_CASE : int = image_mean if image_mean is not None else self.image_mean SCREAMING_SNAKE_CASE : Tuple = image_std if image_std is not None else self.image_std SCREAMING_SNAKE_CASE : List[str] = make_list_of_images(UpperCAmelCase_ ) if not valid_images(UpperCAmelCase_ ): raise ValueError( "Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, " "torch.Tensor, tf.Tensor or jax.ndarray." ) if do_resize and size is None or resample is None: raise ValueError("Size and resample must be specified if do_resize is True." ) if do_rescale and rescale_factor is None: raise ValueError("Rescale factor must be specified if do_rescale is True." ) if do_normalize and (image_mean is None or image_std is None): raise ValueError("Image mean and std must be specified if do_normalize is True." ) # All transformations expect numpy arrays. SCREAMING_SNAKE_CASE : Any = [to_numpy_array(UpperCAmelCase_ ) for image in images] if do_resize: SCREAMING_SNAKE_CASE : Tuple = [self.resize(image=UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ ) for image in images] if do_rescale: SCREAMING_SNAKE_CASE : Tuple = [self.rescale(image=UpperCAmelCase_ , scale=UpperCAmelCase_ ) for image in images] if do_normalize: SCREAMING_SNAKE_CASE : Optional[Any] = [self.normalize(image=UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = [to_channel_dimension_format(UpperCAmelCase_ , UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": images} return BatchFeature(data=UpperCAmelCase_ , tensor_type=UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Tuple] = None ): SCREAMING_SNAKE_CASE : int = outputs.logits # Resize logits and compute semantic segmentation maps if target_sizes is not None: if len(UpperCAmelCase_ ) != len(UpperCAmelCase_ ): raise ValueError( "Make sure that you pass in as many target sizes as the batch dimension of the logits" ) if is_torch_tensor(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = target_sizes.numpy() SCREAMING_SNAKE_CASE : str = [] for idx in range(len(UpperCAmelCase_ ) ): SCREAMING_SNAKE_CASE : Dict = torch.nn.functional.interpolate( logits[idx].unsqueeze(dim=0 ) , size=target_sizes[idx] , mode="bilinear" , align_corners=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = resized_logits[0].argmax(dim=0 ) semantic_segmentation.append(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = logits.argmax(dim=1 ) SCREAMING_SNAKE_CASE : str = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )] return semantic_segmentation
62
1
import datasets import faiss import numpy as np import streamlit as st import torch from elasticsearch import Elasticsearch from elia_utils import ( embed_questions_for_retrieval, make_qa_sas_model, qa_sas_generate, query_es_index, query_qa_dense_index, ) import transformers from transformers import AutoModel, AutoModelForSeqaSeqLM, AutoTokenizer snake_case = """bart""" snake_case = True @st.cache(allow_output_mutation=lowercase ) def lowerCamelCase__ ( ): """simple docstring""" if LOAD_DENSE_INDEX: SCREAMING_SNAKE_CASE : str = AutoTokenizer.from_pretrained("yjernite/retribert-base-uncased" ) SCREAMING_SNAKE_CASE : Optional[int] = AutoModel.from_pretrained("yjernite/retribert-base-uncased" ).to("cuda:0" ) SCREAMING_SNAKE_CASE : Union[str, Any] = qar_model.eval() else: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = (None, None) if MODEL_TYPE == "bart": SCREAMING_SNAKE_CASE : Any = AutoTokenizer.from_pretrained("yjernite/bart_eli5" ) SCREAMING_SNAKE_CASE : Dict = AutoModelForSeqaSeqLM.from_pretrained("yjernite/bart_eli5" ).to("cuda:0" ) SCREAMING_SNAKE_CASE : List[str] = torch.load("seq2seq_models/eli5_bart_model_blm_2.pth" ) sas_model.load_state_dict(save_dict["model"] ) SCREAMING_SNAKE_CASE : List[Any] = sas_model.eval() else: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : str = make_qa_sas_model( model_name="t5-small" , from_file="seq2seq_models/eli5_t5_model_1024_4.pth" , device="cuda:0" ) return (qar_tokenizer, qar_model, sas_tokenizer, sas_model) @st.cache(allow_output_mutation=lowercase ) def lowerCamelCase__ ( ): """simple docstring""" if LOAD_DENSE_INDEX: SCREAMING_SNAKE_CASE : str = faiss.StandardGpuResources() SCREAMING_SNAKE_CASE : Union[str, Any] = datasets.load_dataset(path="wiki_snippets" , name="wiki40b_en_100_0" )["train"] SCREAMING_SNAKE_CASE : int = np.memmap( "wiki40b_passages_reps_32_l-8_h-768_b-512-512.dat" , dtype="float32" , mode="r" , shape=(wikiaab_passages.num_rows, 128) , ) SCREAMING_SNAKE_CASE : Optional[int] = faiss.IndexFlatIP(128 ) SCREAMING_SNAKE_CASE : int = faiss.index_cpu_to_gpu(lowercase , 1 , lowercase ) wikiaab_gpu_index_flat.add(lowercase ) # TODO fix for larger GPU else: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = (None, None) SCREAMING_SNAKE_CASE : List[str] = Elasticsearch([{"host": "localhost", "port": "9200"}] ) return (wikiaab_passages, wikiaab_gpu_index_flat, es_client) @st.cache(allow_output_mutation=lowercase ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = datasets.load_dataset("eli5" , name="LFQA_reddit" ) SCREAMING_SNAKE_CASE : Any = elia["train_eli5"] SCREAMING_SNAKE_CASE : Optional[int] = np.memmap( "eli5_questions_reps.dat" , dtype="float32" , mode="r" , shape=(elia_train.num_rows, 128) ) SCREAMING_SNAKE_CASE : Union[str, Any] = faiss.IndexFlatIP(128 ) eli5_train_q_index.add(lowercase ) return (elia_train, eli5_train_q_index) snake_case , snake_case , snake_case = load_indexes() snake_case , snake_case , snake_case , snake_case = load_models() snake_case , snake_case = load_train_data() def lowerCamelCase__ ( lowercase , lowercase=10 ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = embed_questions_for_retrieval([question] , lowercase , lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = eli5_train_q_index.search(lowercase , lowercase ) SCREAMING_SNAKE_CASE : List[str] = [elia_train[int(lowercase )] for i in I[0]] return nn_examples def lowerCamelCase__ ( lowercase , lowercase="wiki40b" , lowercase="dense" , lowercase=10 ): """simple docstring""" if source == "none": SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = (" <P> ".join(["" for _ in range(11 )] ).strip(), []) else: if method == "dense": SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = query_qa_dense_index( lowercase , lowercase , lowercase , lowercase , lowercase , lowercase ) else: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = query_es_index( lowercase , lowercase , index_name="english_wiki40b_snippets_100w" , n_results=lowercase , ) SCREAMING_SNAKE_CASE : Union[str, Any] = [ (res["article_title"], res["section_title"].strip(), res["score"], res["passage_text"]) for res in hit_lst ] SCREAMING_SNAKE_CASE : Tuple = "question: {} context: {}".format(lowercase , lowercase ) return question_doc, support_list @st.cache( hash_funcs={ torch.Tensor: (lambda lowercase : None), transformers.models.bart.tokenization_bart.BartTokenizer: (lambda lowercase : None), } ) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase=64 , lowercase=256 , lowercase=False , lowercase=2 , lowercase=0.95 , lowercase=0.8 ): """simple docstring""" with torch.no_grad(): SCREAMING_SNAKE_CASE : List[Any] = qa_sas_generate( lowercase , lowercase , lowercase , num_answers=1 , num_beams=lowercase , min_len=lowercase , max_len=lowercase , do_sample=lowercase , temp=lowercase , top_p=lowercase , top_k=lowercase , max_input_length=1024 , device="cuda:0" , )[0] return (answer, support_list) st.title("""Long Form Question Answering with ELI5""") # Start sidebar snake_case = """<img src='https://huggingface.co/front/assets/huggingface_logo.svg'>""" snake_case = """ <html> <head> <style> .img-container { padding-left: 90px; padding-right: 90px; padding-top: 50px; padding-bottom: 50px; background-color: #f0f3f9; } </style> </head> <body> <span class=\"img-container\"> <!-- Inline parent element --> %s </span> </body> </html> """ % ( header_html, ) st.sidebar.markdown( header_full, unsafe_allow_html=True, ) # Long Form QA with ELI5 and Wikipedia snake_case = """ This demo presents a model trained to [provide long-form answers to open-domain questions](https://yjernite.github.io/lfqa.html). First, a document retriever fetches a set of relevant Wikipedia passages given the question from the [Wiki40b](https://research.google/pubs/pub49029/) dataset, a pre-processed fixed snapshot of Wikipedia. """ st.sidebar.markdown(description, unsafe_allow_html=True) snake_case = [ """Answer the question""", """View the retrieved document only""", """View the most similar ELI5 question and answer""", """Show me everything, please!""", ] snake_case = st.sidebar.checkbox("""Demo options""") if demo_options: snake_case = st.sidebar.selectbox( """""", action_list, index=3, ) snake_case = action_list.index(action_st) snake_case = st.sidebar.selectbox( """""", ["""Show full text of passages""", """Show passage section titles"""], index=0, ) snake_case = show_type == """Show full text of passages""" else: snake_case = 3 snake_case = True snake_case = st.sidebar.checkbox("""Retrieval options""") if retrieval_options: snake_case = """ ### Information retriever options The **sparse** retriever uses ElasticSearch, while the **dense** retriever uses max-inner-product search between a question and passage embedding trained using the [ELI5](https://arxiv.org/abs/1907.09190) questions-answer pairs. The answer is then generated by sequence to sequence model which takes the question and retrieved document as input. """ st.sidebar.markdown(retriever_info) snake_case = st.sidebar.selectbox("""Which Wikipedia format should the model use?""", ["""wiki40b""", """none"""]) snake_case = st.sidebar.selectbox("""Which Wikipedia indexer should the model use?""", ["""dense""", """sparse""", """mixed"""]) else: snake_case = """wiki40b""" snake_case = """dense""" snake_case = """beam""" snake_case = 2 snake_case = 64 snake_case = 256 snake_case = None snake_case = None snake_case = st.sidebar.checkbox("""Generation options""") if generate_options: snake_case = """ ### Answer generation options The sequence-to-sequence model was initialized with [BART](https://huggingface.co/facebook/bart-large) weights and fine-tuned on the ELI5 QA pairs and retrieved documents. You can use the model for greedy decoding with **beam** search, or **sample** from the decoder's output probabilities. """ st.sidebar.markdown(generate_info) snake_case = st.sidebar.selectbox("""Would you like to use beam search or sample an answer?""", ["""beam""", """sampled"""]) snake_case = st.sidebar.slider( """Minimum generation length""", min_value=8, max_value=256, value=64, step=8, format=None, key=None ) snake_case = st.sidebar.slider( """Maximum generation length""", min_value=64, max_value=512, value=256, step=16, format=None, key=None ) if sampled == "beam": snake_case = st.sidebar.slider("""Beam size""", min_value=1, max_value=8, value=2, step=None, format=None, key=None) else: snake_case = st.sidebar.slider( """Nucleus sampling p""", min_value=0.1, max_value=1.0, value=0.95, step=0.01, format=None, key=None ) snake_case = st.sidebar.slider( """Temperature""", min_value=0.1, max_value=1.0, value=0.7, step=0.01, format=None, key=None ) snake_case = None # start main text snake_case = [ """<MY QUESTION>""", """How do people make chocolate?""", """Why do we get a fever when we are sick?""", """How can different animals perceive different colors?""", """What is natural language processing?""", """What's the best way to treat a sunburn?""", """What exactly are vitamins ?""", """How does nuclear energy provide electricity?""", """What's the difference between viruses and bacteria?""", """Why are flutes classified as woodwinds when most of them are made out of metal ?""", """Why do people like drinking coffee even though it tastes so bad?""", """What happens when wine ages? How does it make the wine taste better?""", """If an animal is an herbivore, where does it get the protein that it needs to survive if it only eats grass?""", """How can we set a date to the beginning or end of an artistic period? Doesn't the change happen gradually?""", """How does New Zealand have so many large bird predators?""", ] snake_case = st.selectbox( """What would you like to ask? ---- select <MY QUESTION> to enter a new query""", questions_list, index=1, ) if question_s == "<MY QUESTION>": snake_case = st.text_input("""Enter your question here:""", """""") else: snake_case = question_s if st.button("""Show me!"""): if action in [0, 1, 3]: if index_type == "mixed": snake_case , snake_case = make_support(question, source=wiki_source, method="""dense""", n_results=10) snake_case , snake_case = make_support(question, source=wiki_source, method="""sparse""", n_results=10) snake_case = [] for res_d, res_s in zip(support_list_dense, support_list_sparse): if tuple(res_d) not in support_list: support_list += [tuple(res_d)] if tuple(res_s) not in support_list: support_list += [tuple(res_s)] snake_case = support_list[:10] snake_case = """<P> """ + """ <P> """.join([res[-1] for res in support_list]) else: snake_case , snake_case = make_support(question, source=wiki_source, method=index_type, n_results=10) if action in [0, 3]: snake_case , snake_case = answer_question( question_doc, sas_model, sas_tokenizer, min_len=min_len, max_len=int(max_len), sampling=(sampled == """sampled"""), n_beams=n_beams, top_p=top_p, temp=temp, ) st.markdown("""### The model generated answer is:""") st.write(answer) if action in [0, 1, 3] and wiki_source != "none": st.markdown("""--- \n ### The model is drawing information from the following Wikipedia passages:""") for i, res in enumerate(support_list): snake_case = """https://en.wikipedia.org/wiki/{}""".format(res[0].replace(""" """, """_""")) snake_case = res[1].strip() if sec_titles == "": snake_case = """[{}]({})""".format(res[0], wiki_url) else: snake_case = sec_titles.split(""" & """) snake_case = """ & """.join( ["""[{}]({}#{})""".format(sec.strip(), wiki_url, sec.strip().replace(""" """, """_""")) for sec in sec_list] ) st.markdown( """{0:02d} - **Article**: {1:<18} <br> _Section_: {2}""".format(i + 1, res[0], sections), unsafe_allow_html=True, ) if show_passages: st.write( """> <span style=\"font-family:arial; font-size:10pt;\">""" + res[-1] + """</span>""", unsafe_allow_html=True ) if action in [2, 3]: snake_case = find_nearest_training(question) snake_case = nn_train_list[0] st.markdown( """--- \n ### The most similar question in the ELI5 training set was: \n\n {}""".format(train_exple["""title"""]) ) snake_case = [ """{}. {}""".format(i + 1, """ \n""".join([line.strip() for line in ans.split("""\n""") if line.strip() != """"""])) for i, (ans, sc) in enumerate(zip(train_exple["""answers"""]["""text"""], train_exple["""answers"""]["""score"""])) if i == 0 or sc > 2 ] st.markdown("""##### Its answers were: \n\n {}""".format("""\n""".join(answers_st))) snake_case = """ --- **Disclaimer** *The intent of this app is to provide some (hopefully entertaining) insights into the behavior of a current LFQA system. Evaluating biases of such a model and ensuring factual generations are still very much open research problems. Therefore, until some significant progress is achieved, we caution against using the generated answers for practical purposes.* """ st.sidebar.markdown(disclaimer, unsafe_allow_html=True)
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : List[str] , UpperCAmelCase_ : int = 6 ): SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None self.create_linked_list(UpperCAmelCase_ ) def _A ( self : List[Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[int] = Node() SCREAMING_SNAKE_CASE : str = current_node SCREAMING_SNAKE_CASE : Optional[int] = current_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node for _ in range(1 , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Tuple = Node() SCREAMING_SNAKE_CASE : Dict = current_node SCREAMING_SNAKE_CASE : Optional[Any] = previous_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node SCREAMING_SNAKE_CASE : Union[str, Any] = self.front SCREAMING_SNAKE_CASE : List[str] = previous_node def _A ( self : Union[str, Any] ): return ( self.front == self.rear and self.front is not None and self.front.data is None ) def _A ( self : Optional[int] ): self.check_can_perform_operation() return self.front.data if self.front else None def _A ( self : Optional[int] , UpperCAmelCase_ : Any ): if self.rear is None: return self.check_is_full() if not self.is_empty(): SCREAMING_SNAKE_CASE : List[str] = self.rear.next if self.rear: SCREAMING_SNAKE_CASE : Dict = data def _A ( self : List[str] ): self.check_can_perform_operation() if self.rear is None or self.front is None: return None if self.front == self.rear: SCREAMING_SNAKE_CASE : List[str] = self.front.data SCREAMING_SNAKE_CASE : Optional[int] = None return data SCREAMING_SNAKE_CASE : List[str] = self.front SCREAMING_SNAKE_CASE : List[str] = old_front.next SCREAMING_SNAKE_CASE : Optional[int] = old_front.data SCREAMING_SNAKE_CASE : List[str] = None return data def _A ( self : Any ): if self.is_empty(): raise Exception("Empty Queue" ) def _A ( self : Optional[Any] ): if self.rear and self.rear.next == self.front: raise Exception("Full Queue" ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any | None = None SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None if __name__ == "__main__": import doctest doctest.testmod()
62
1
import importlib.util import json import os import warnings from dataclasses import dataclass, field import torch from ..training_args import TrainingArguments from ..utils import cached_property, is_sagemaker_dp_enabled, logging snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = os.getenv("SM_HP_MP_PARAMETERS" , "{}" ) try: # Parse it and check the field "partitions" is included, it is required for model parallel. SCREAMING_SNAKE_CASE : List[str] = json.loads(lowercase ) if "partitions" not in smp_options: return False except json.JSONDecodeError: return False # Get the sagemaker specific framework parameters from mpi_options variable. SCREAMING_SNAKE_CASE : Optional[Any] = os.getenv("SM_FRAMEWORK_PARAMS" , "{}" ) try: # Parse it and check the field "sagemaker_distributed_dataparallel_enabled". SCREAMING_SNAKE_CASE : Tuple = json.loads(lowercase ) if not mpi_options.get("sagemaker_mpi_enabled" , lowercase ): return False except json.JSONDecodeError: return False # Lastly, check if the `smdistributed` module is present. return importlib.util.find_spec("smdistributed" ) is not None if is_sagemaker_model_parallel_available(): import smdistributed.modelparallel.torch as smp smp.init() @dataclass class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : str = field( default='''''' , metadata={'''help''': '''Used by the SageMaker launcher to send mp-specific args. Ignored in SageMakerTrainer'''} , ) def _A ( self : Dict ): super().__post_init__() warnings.warn( "`SageMakerTrainingArguments` is deprecated and will be removed in v5 of Transformers. You can use " "`TrainingArguments` instead." , UpperCAmelCase_ , ) @cached_property def _A ( self : List[str] ): logger.info("PyTorch: setting up devices" ) if torch.distributed.is_available() and torch.distributed.is_initialized() and self.local_rank == -1: logger.warning( "torch.distributed process group is initialized, but local_rank == -1. " "In order to use Torch DDP, launch your script with `python -m torch.distributed.launch" ) if self.no_cuda: SCREAMING_SNAKE_CASE : List[str] = torch.device("cpu" ) SCREAMING_SNAKE_CASE : List[Any] = 0 elif is_sagemaker_model_parallel_available(): SCREAMING_SNAKE_CASE : str = smp.local_rank() SCREAMING_SNAKE_CASE : List[str] = torch.device("cuda" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = 1 elif is_sagemaker_dp_enabled(): import smdistributed.dataparallel.torch.torch_smddp # noqa: F401 torch.distributed.init_process_group(backend="smddp" , timeout=self.ddp_timeout_delta ) SCREAMING_SNAKE_CASE : Union[str, Any] = int(os.getenv("SMDATAPARALLEL_LOCAL_RANK" ) ) SCREAMING_SNAKE_CASE : Tuple = torch.device("cuda" , self.local_rank ) SCREAMING_SNAKE_CASE : Optional[int] = 1 elif self.local_rank == -1: # if n_gpu is > 1 we'll use nn.DataParallel. # If you only want to use a specific subset of GPUs use `CUDA_VISIBLE_DEVICES=0` # Explicitly set CUDA to the first (index 0) CUDA device, otherwise `set_device` will # trigger an error that a device index is missing. Index 0 takes into account the # GPUs available in the environment, so `CUDA_VISIBLE_DEVICES=1,2` with `cuda:0` # will use the first GPU in that env, i.e. GPU#1 SCREAMING_SNAKE_CASE : Dict = torch.device("cuda:0" if torch.cuda.is_available() else "cpu" ) # Sometimes the line in the postinit has not been run before we end up here, so just checking we're not at # the default value. SCREAMING_SNAKE_CASE : Dict = torch.cuda.device_count() else: # Here, we'll use torch.distributed. # Initializes the distributed backend which will take care of synchronizing nodes/GPUs if not torch.distributed.is_initialized(): torch.distributed.init_process_group(backend="nccl" , timeout=self.ddp_timeout_delta ) SCREAMING_SNAKE_CASE : Tuple = torch.device("cuda" , self.local_rank ) SCREAMING_SNAKE_CASE : List[str] = 1 if device.type == "cuda": torch.cuda.set_device(UpperCAmelCase_ ) return device @property def _A ( self : List[Any] ): if is_sagemaker_model_parallel_available(): return smp.dp_size() return super().world_size @property def _A ( self : Union[str, Any] ): return not is_sagemaker_model_parallel_available() @property def _A ( self : Tuple ): return False
62
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return 1 if input_a == input_a else 0 def lowerCamelCase__ ( ): """simple docstring""" assert xnor_gate(0 , 0 ) == 1 assert xnor_gate(0 , 1 ) == 0 assert xnor_gate(1 , 0 ) == 0 assert xnor_gate(1 , 1 ) == 1 if __name__ == "__main__": print(xnor_gate(0, 0)) print(xnor_gate(0, 1)) print(xnor_gate(1, 0)) print(xnor_gate(1, 1))
62
1
from typing import TYPE_CHECKING from ...utils import ( OptionalDependencyNotAvailable, _LazyModule, is_torch_available, ) snake_case = { """configuration_mega""": ["""MEGA_PRETRAINED_CONFIG_ARCHIVE_MAP""", """MegaConfig""", """MegaOnnxConfig"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """MEGA_PRETRAINED_MODEL_ARCHIVE_LIST""", """MegaForCausalLM""", """MegaForMaskedLM""", """MegaForMultipleChoice""", """MegaForQuestionAnswering""", """MegaForSequenceClassification""", """MegaForTokenClassification""", """MegaModel""", """MegaPreTrainedModel""", ] if TYPE_CHECKING: from .configuration_mega import MEGA_PRETRAINED_CONFIG_ARCHIVE_MAP, MegaConfig, MegaOnnxConfig try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_mega import ( MEGA_PRETRAINED_MODEL_ARCHIVE_LIST, MegaForCausalLM, MegaForMaskedLM, MegaForMultipleChoice, MegaForQuestionAnswering, MegaForSequenceClassification, MegaForTokenClassification, MegaModel, MegaPreTrainedModel, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
import math import flax.linen as nn import jax.numpy as jnp def lowerCamelCase__ ( lowercase , lowercase , lowercase = 1 , lowercase = 1 , lowercase = 1.0E4 , lowercase = False , lowercase = 1.0 , ): """simple docstring""" assert timesteps.ndim == 1, "Timesteps should be a 1d-array" assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even''' SCREAMING_SNAKE_CASE : Union[str, Any] = float(embedding_dim // 2 ) SCREAMING_SNAKE_CASE : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift) SCREAMING_SNAKE_CASE : Optional[Any] = min_timescale * jnp.exp(jnp.arange(lowercase , dtype=jnp.floataa ) * -log_timescale_increment ) SCREAMING_SNAKE_CASE : Optional[int] = jnp.expand_dims(lowercase , 1 ) * jnp.expand_dims(lowercase , 0 ) # scale embeddings SCREAMING_SNAKE_CASE : Optional[int] = scale * emb if flip_sin_to_cos: SCREAMING_SNAKE_CASE : List[Any] = jnp.concatenate([jnp.cos(lowercase ), jnp.sin(lowercase )] , axis=1 ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = jnp.concatenate([jnp.sin(lowercase ), jnp.cos(lowercase )] , axis=1 ) SCREAMING_SNAKE_CASE : Tuple = jnp.reshape(lowercase , [jnp.shape(lowercase )[0], embedding_dim] ) return signal class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : jnp.dtype = jnp.floataa @nn.compact def __call__( self : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_1" )(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.silu(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_2" )(UpperCAmelCase_ ) return temb class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : bool = False UpperCamelCase_ : float = 1 @nn.compact def __call__( self : Optional[int] , UpperCAmelCase_ : int ): return get_sinusoidal_embeddings( UpperCAmelCase_ , embedding_dim=self.dim , flip_sin_to_cos=self.flip_sin_to_cos , freq_shift=self.freq_shift )
62
1
from random import randint from tempfile import TemporaryFile import numpy as np def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : str = 0 if start < end: SCREAMING_SNAKE_CASE : int = randint(lowercase , lowercase ) SCREAMING_SNAKE_CASE : int = a[end] SCREAMING_SNAKE_CASE : str = a[pivot] SCREAMING_SNAKE_CASE : Dict = temp SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : str = _in_place_partition(lowercase , lowercase , lowercase ) count += _in_place_quick_sort(lowercase , lowercase , p - 1 ) count += _in_place_quick_sort(lowercase , p + 1 , lowercase ) return count def lowerCamelCase__ ( lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = 0 SCREAMING_SNAKE_CASE : Any = randint(lowercase , lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = a[end] SCREAMING_SNAKE_CASE : Any = a[pivot] SCREAMING_SNAKE_CASE : List[str] = temp SCREAMING_SNAKE_CASE : List[Any] = start - 1 for index in range(lowercase , lowercase ): count += 1 if a[index] < a[end]: # check if current val is less than pivot value SCREAMING_SNAKE_CASE : Optional[Any] = new_pivot_index + 1 SCREAMING_SNAKE_CASE : Dict = a[new_pivot_index] SCREAMING_SNAKE_CASE : List[Any] = a[index] SCREAMING_SNAKE_CASE : Optional[int] = temp SCREAMING_SNAKE_CASE : Any = a[new_pivot_index + 1] SCREAMING_SNAKE_CASE : Optional[Any] = a[end] SCREAMING_SNAKE_CASE : Tuple = temp return new_pivot_index + 1, count snake_case = TemporaryFile() snake_case = 100 # 1000 elements are to be sorted snake_case , snake_case = 0, 1 # mean and standard deviation snake_case = np.random.normal(mu, sigma, p) np.save(outfile, X) print("""The array is""") print(X) outfile.seek(0) # using the same array snake_case = np.load(outfile) snake_case = len(M) - 1 snake_case = _in_place_quick_sort(M, 0, r) print( """No of Comparisons for 100 elements selected from a standard normal distribution""" """is :""" ) print(z)
62
# Copyright 2023 The HuggingFace Inc. team. All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. import numpy as np import torch from ..models.clipseg import CLIPSegForImageSegmentation from ..utils import is_vision_available, requires_backends from .base import PipelineTool if is_vision_available(): from PIL import Image class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = ( '''This is a tool that creates a segmentation mask of an image according to a label. It cannot create an image.''' '''It takes two arguments named `image` which should be the original image, and `label` which should be a text ''' '''describing the elements what should be identified in the segmentation mask. The tool returns the mask.''' ) UpperCamelCase_ : Union[str, Any] = '''CIDAS/clipseg-rd64-refined''' UpperCamelCase_ : Any = '''image_segmenter''' UpperCamelCase_ : int = CLIPSegForImageSegmentation UpperCamelCase_ : Optional[Any] = ['''image''', '''text'''] UpperCamelCase_ : int = ['''image'''] def __init__( self : str , *UpperCAmelCase_ : Union[str, Any] , **UpperCAmelCase_ : Union[str, Any] ): requires_backends(self , ["vision"] ) super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : "Image" , UpperCAmelCase_ : str ): return self.pre_processor(text=[label] , images=[image] , padding=UpperCAmelCase_ , return_tensors="pt" ) def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): with torch.no_grad(): SCREAMING_SNAKE_CASE : str = self.model(**UpperCAmelCase_ ).logits return logits def _A ( self : Union[str, Any] , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = outputs.cpu().detach().numpy() SCREAMING_SNAKE_CASE : Optional[int] = 0 SCREAMING_SNAKE_CASE : str = 1 return Image.fromarray((array * 255).astype(np.uinta ) )
62
1
import warnings from .generation import TFGenerationMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' warnings.warn( '''Importing `TFGenerationMixin` from `src/transformers/generation_tf_utils.py` is deprecated and will ''' '''be removed in Transformers v5. Import as `from transformers import TFGenerationMixin` instead.''' , lowerCAmelCase , )
62
import json import os import unittest from transformers.models.blenderbot_small.tokenization_blenderbot_small import ( VOCAB_FILES_NAMES, BlenderbotSmallTokenizer, ) from ...test_tokenization_common import TokenizerTesterMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = BlenderbotSmallTokenizer UpperCamelCase_ : int = False def _A ( self : Union[str, Any] ): super().setUp() SCREAMING_SNAKE_CASE : List[Any] = ["__start__", "adapt", "act", "ap@@", "te", "__end__", "__unk__"] SCREAMING_SNAKE_CASE : Optional[Any] = dict(zip(UpperCAmelCase_ , range(len(UpperCAmelCase_ ) ) ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = ["#version: 0.2", "a p", "t e</w>", "ap t</w>", "a d", "ad apt</w>", "a c", "ac t</w>", ""] SCREAMING_SNAKE_CASE : int = {"unk_token": "__unk__", "bos_token": "__start__", "eos_token": "__end__"} SCREAMING_SNAKE_CASE : Union[str, Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["vocab_file"] ) SCREAMING_SNAKE_CASE : str = 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(UpperCAmelCase_ ) + "\n" ) with open(self.merges_file , "w" , encoding="utf-8" ) as fp: fp.write("\n".join(UpperCAmelCase_ ) ) def _A ( self : List[Any] , **UpperCAmelCase_ : str ): kwargs.update(self.special_tokens_map ) return BlenderbotSmallTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] , UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : int = "adapt act apte" return input_text, output_text def _A ( self : str ): SCREAMING_SNAKE_CASE : int = BlenderbotSmallTokenizer(self.vocab_file , self.merges_file , **self.special_tokens_map ) SCREAMING_SNAKE_CASE : Tuple = "adapt act apte" SCREAMING_SNAKE_CASE : List[str] = ["adapt", "act", "ap@@", "te"] SCREAMING_SNAKE_CASE : Optional[Any] = tokenizer.tokenize(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = [tokenizer.bos_token] + tokens + [tokenizer.eos_token] SCREAMING_SNAKE_CASE : Tuple = [0, 1, 2, 3, 4, 5] self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , UpperCAmelCase_ ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Union[str, Any] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) assert tok("sam" ).input_ids == [1384] SCREAMING_SNAKE_CASE : str = "I am a small frog." SCREAMING_SNAKE_CASE : List[Any] = tok([src_text] , padding=UpperCAmelCase_ , truncation=UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : int = tok.batch_decode(UpperCAmelCase_ , skip_special_tokens=UpperCAmelCase_ , clean_up_tokenization_spaces=UpperCAmelCase_ )[0] assert src_text != decoded # I wish it did! assert decoded == "i am a small frog ." def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : List[str] = BlenderbotSmallTokenizer.from_pretrained("facebook/blenderbot-90M" ) SCREAMING_SNAKE_CASE : Tuple = "I am a small frog ." SCREAMING_SNAKE_CASE : Optional[int] = "." SCREAMING_SNAKE_CASE : Dict = tok(UpperCAmelCase_ )["input_ids"] SCREAMING_SNAKE_CASE : Optional[Any] = tok(UpperCAmelCase_ )["input_ids"] assert encoded[-1] == encoded_dot[0]
62
1
from ...processing_utils import ProcessorMixin class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : str = '''WhisperFeatureExtractor''' UpperCamelCase_ : str = '''WhisperTokenizer''' def __init__( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[str] ): super().__init__(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = self.feature_extractor SCREAMING_SNAKE_CASE : int = False def _A ( self : Union[str, Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : Optional[int]=True ): return self.tokenizer.get_decoder_prompt_ids(task=UpperCAmelCase_ , language=UpperCAmelCase_ , no_timestamps=UpperCAmelCase_ ) def __call__( self : str , *UpperCAmelCase_ : Any , **UpperCAmelCase_ : Any ): # For backward compatibility if self._in_target_context_manager: return self.current_processor(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = kwargs.pop("audio" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = kwargs.pop("sampling_rate" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = kwargs.pop("text" , UpperCAmelCase_ ) if len(UpperCAmelCase_ ) > 0: SCREAMING_SNAKE_CASE : str = args[0] SCREAMING_SNAKE_CASE : Union[str, Any] = 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: SCREAMING_SNAKE_CASE : List[Any] = self.feature_extractor(UpperCAmelCase_ , *UpperCAmelCase_ , sampling_rate=UpperCAmelCase_ , **UpperCAmelCase_ ) if text is not None: SCREAMING_SNAKE_CASE : Union[str, Any] = self.tokenizer(UpperCAmelCase_ , **UpperCAmelCase_ ) if text is None: return inputs elif audio is None: return encodings else: SCREAMING_SNAKE_CASE : int = encodings["input_ids"] return inputs def _A ( self : Dict , *UpperCAmelCase_ : Union[str, Any] , **UpperCAmelCase_ : str ): return self.tokenizer.batch_decode(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : List[str] , *UpperCAmelCase_ : Optional[Any] , **UpperCAmelCase_ : Dict ): return self.tokenizer.decode(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : str="np" ): return self.tokenizer.get_prompt_ids(UpperCAmelCase_ , return_tensors=UpperCAmelCase_ )
62
import argparse from collections import OrderedDict from pathlib import Path import torch from transformers import ( VisualBertConfig, VisualBertForMultipleChoice, VisualBertForPreTraining, VisualBertForQuestionAnswering, VisualBertForVisualReasoning, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) snake_case = [ ("""bert.bert""", """visual_bert"""), ("""bert.cls""", """cls"""), ("""bert.classifier""", """cls"""), ("""token_type_embeddings_visual""", """visual_token_type_embeddings"""), ("""position_embeddings_visual""", """visual_position_embeddings"""), ("""projection""", """visual_projection"""), ] snake_case = [ """nlvr2_coco_pre_trained.th""", """nlvr2_fine_tuned.th""", """nlvr2_pre_trained.th""", """vcr_coco_pre_train.th""", """vcr_fine_tune.th""", """vcr_pre_train.th""", """vqa_coco_pre_trained.th""", """vqa_fine_tuned.th""", """vqa_pre_trained.th""", ] def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = torch.load(lowercase , map_location="cpu" ) return sd def lowerCamelCase__ ( lowercase , lowercase , lowercase=rename_keys_prefix ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = OrderedDict() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.arange(config.max_position_embeddings ).expand((1, -1) ) # detector_d = OrderedDict() for key in d: if "detector" in key: # detector_d[key.replace('detector.','')] = d[key] continue SCREAMING_SNAKE_CASE : Union[str, Any] = key for name_pair in rename_keys_prefix: SCREAMING_SNAKE_CASE : Any = new_key.replace(name_pair[0] , name_pair[1] ) SCREAMING_SNAKE_CASE : Dict = d[key] if key == "bert.cls.predictions.decoder.weight": # Old bert code didn't have `decoder.bias`, but was added separately SCREAMING_SNAKE_CASE : Union[str, Any] = new_d["cls.predictions.bias"] return new_d @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( checkpoint_path.split("/" )[-1] in ACCEPTABLE_CHECKPOINTS ), F'''The checkpoint provided must be in {ACCEPTABLE_CHECKPOINTS}.''' # Get Config if "pre" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = "pretraining" if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = {"visual_embedding_dim": 512} elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[Any] = {"visual_embedding_dim": 2048} elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Any = {"visual_embedding_dim": 2048} elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 1024} else: raise NotImplementedError(F'''No implementation found for `{checkpoint_path}`.''' ) else: if "vcr" in checkpoint_path: SCREAMING_SNAKE_CASE : str = {"visual_embedding_dim": 512} SCREAMING_SNAKE_CASE : Tuple = "multichoice" elif "vqa_advanced" in checkpoint_path: SCREAMING_SNAKE_CASE : List[str] = {"visual_embedding_dim": 2048} SCREAMING_SNAKE_CASE : str = "vqa_advanced" elif "vqa" in checkpoint_path: SCREAMING_SNAKE_CASE : Tuple = {"visual_embedding_dim": 2048, "num_labels": 3129} SCREAMING_SNAKE_CASE : Optional[Any] = "vqa" elif "nlvr" in checkpoint_path: SCREAMING_SNAKE_CASE : int = { "visual_embedding_dim": 1024, "num_labels": 2, } SCREAMING_SNAKE_CASE : Tuple = "nlvr" SCREAMING_SNAKE_CASE : List[Any] = VisualBertConfig(**lowercase ) # Load State Dict SCREAMING_SNAKE_CASE : List[str] = load_state_dict(lowercase ) SCREAMING_SNAKE_CASE : Any = get_new_dict(lowercase , lowercase ) if model_type == "pretraining": SCREAMING_SNAKE_CASE : int = VisualBertForPreTraining(lowercase ) elif model_type == "vqa": SCREAMING_SNAKE_CASE : Tuple = VisualBertForQuestionAnswering(lowercase ) elif model_type == "nlvr": SCREAMING_SNAKE_CASE : List[Any] = VisualBertForVisualReasoning(lowercase ) elif model_type == "multichoice": SCREAMING_SNAKE_CASE : Optional[Any] = VisualBertForMultipleChoice(lowercase ) model.load_state_dict(lowercase ) # Save Checkpoints Path(lowercase ).mkdir(exist_ok=lowercase ) model.save_pretrained(lowercase ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument("""orig_checkpoint_path""", type=str, help="""A path to .th on local filesystem.""") parser.add_argument("""pytorch_dump_folder_path""", type=str, help="""Path to the output PyTorch model.""") snake_case = parser.parse_args() convert_visual_bert_checkpoint(args.orig_checkpoint_path, args.pytorch_dump_folder_path)
62
1
from math import sqrt def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and ( number >= 0 ), "'number' must been an int and positive" SCREAMING_SNAKE_CASE : Any = True # 0 and 1 are none primes. if number <= 1: SCREAMING_SNAKE_CASE : Optional[Any] = False for divisor in range(2 , int(round(sqrt(lowercase ) ) ) + 1 ): # if 'number' divisible by 'divisor' then sets 'status' # of false and break up the loop. if number % divisor == 0: SCREAMING_SNAKE_CASE : Union[str, Any] = False break # precondition assert isinstance(lowercase , lowercase ), "'status' must been from type bool" return status def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and (n > 2), "'N' must been an int and > 2" # beginList: contains all natural numbers from 2 up to N SCREAMING_SNAKE_CASE : Union[str, Any] = list(range(2 , n + 1 ) ) SCREAMING_SNAKE_CASE : List[str] = [] # this list will be returns. # actual sieve of erathostenes for i in range(len(lowercase ) ): for j in range(i + 1 , len(lowercase ) ): if (begin_list[i] != 0) and (begin_list[j] % begin_list[i] == 0): SCREAMING_SNAKE_CASE : Dict = 0 # filters actual prime numbers. SCREAMING_SNAKE_CASE : Optional[int] = [x for x in begin_list if x != 0] # precondition assert isinstance(lowercase , lowercase ), "'ans' must been from type list" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and (n > 2), "'N' must been an int and > 2" SCREAMING_SNAKE_CASE : int = [] # iterates over all numbers between 2 up to N+1 # if a number is prime then appends to list 'ans' for number in range(2 , n + 1 ): if is_prime(lowercase ): ans.append(lowercase ) # precondition assert isinstance(lowercase , lowercase ), "'ans' must been from type list" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and number >= 0, "'number' must been an int and >= 0" SCREAMING_SNAKE_CASE : List[Any] = [] # this list will be returns of the function. # potential prime number factors. SCREAMING_SNAKE_CASE : Dict = 2 SCREAMING_SNAKE_CASE : int = number if number == 0 or number == 1: ans.append(lowercase ) # if 'number' not prime then builds the prime factorization of 'number' elif not is_prime(lowercase ): while quotient != 1: if is_prime(lowercase ) and (quotient % factor == 0): ans.append(lowercase ) quotient /= factor else: factor += 1 else: ans.append(lowercase ) # precondition assert isinstance(lowercase , lowercase ), "'ans' must been from type list" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and ( number >= 0 ), "'number' bust been an int and >= 0" SCREAMING_SNAKE_CASE : List[str] = 0 # prime factorization of 'number' SCREAMING_SNAKE_CASE : Any = prime_factorization(lowercase ) SCREAMING_SNAKE_CASE : List[str] = max(lowercase ) # precondition assert isinstance(lowercase , lowercase ), "'ans' must been from type int" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and ( number >= 0 ), "'number' bust been an int and >= 0" SCREAMING_SNAKE_CASE : List[str] = 0 # prime factorization of 'number' SCREAMING_SNAKE_CASE : Optional[int] = prime_factorization(lowercase ) SCREAMING_SNAKE_CASE : List[Any] = min(lowercase ) # precondition assert isinstance(lowercase , lowercase ), "'ans' must been from type int" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ), "'number' must been an int" assert isinstance(number % 2 == 0 , lowercase ), "compare bust been from type bool" return number % 2 == 0 def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ), "'number' must been an int" assert isinstance(number % 2 != 0 , lowercase ), "compare bust been from type bool" return number % 2 != 0 def lowerCamelCase__ ( lowercase ): """simple docstring""" assert ( isinstance(lowercase , lowercase ) and (number > 2) and is_even(lowercase ) ), "'number' must been an int, even and > 2" SCREAMING_SNAKE_CASE : List[str] = [] # this list will returned # creates a list of prime numbers between 2 up to 'number' SCREAMING_SNAKE_CASE : Tuple = get_prime_numbers(lowercase ) SCREAMING_SNAKE_CASE : Tuple = len(lowercase ) # run variable for while-loops. SCREAMING_SNAKE_CASE : List[str] = 0 SCREAMING_SNAKE_CASE : Dict = None # exit variable. for break up the loops SCREAMING_SNAKE_CASE : Any = True while i < len_pn and loop: SCREAMING_SNAKE_CASE : Any = i + 1 while j < len_pn and loop: if prime_numbers[i] + prime_numbers[j] == number: SCREAMING_SNAKE_CASE : str = False ans.append(prime_numbers[i] ) ans.append(prime_numbers[j] ) j += 1 i += 1 # precondition assert ( isinstance(lowercase , lowercase ) and (len(lowercase ) == 2) and (ans[0] + ans[1] == number) and is_prime(ans[0] ) and is_prime(ans[1] ) ), "'ans' must contains two primes. And sum of elements must been eq 'number'" return ans def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( isinstance(lowercase , lowercase ) and isinstance(lowercase , lowercase ) and (numbera >= 0) and (numbera >= 0) ), "'number1' and 'number2' must been positive integer." SCREAMING_SNAKE_CASE : List[str] = 0 while numbera != 0: SCREAMING_SNAKE_CASE : List[str] = numbera % numbera SCREAMING_SNAKE_CASE : Dict = numbera SCREAMING_SNAKE_CASE : Optional[Any] = rest # precondition assert isinstance(lowercase , lowercase ) and ( numbera >= 0 ), "'number' must been from type int and positive" return numbera def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( isinstance(lowercase , lowercase ) and isinstance(lowercase , lowercase ) and (numbera >= 1) and (numbera >= 1) ), "'number1' and 'number2' must been positive integer." SCREAMING_SNAKE_CASE : Union[str, Any] = 1 # actual answer that will be return. # for kgV (x,1) if numbera > 1 and numbera > 1: # builds the prime factorization of 'number1' and 'number2' SCREAMING_SNAKE_CASE : Optional[int] = prime_factorization(lowercase ) SCREAMING_SNAKE_CASE : Union[str, Any] = prime_factorization(lowercase ) elif numbera == 1 or numbera == 1: SCREAMING_SNAKE_CASE : Union[str, Any] = [] SCREAMING_SNAKE_CASE : Tuple = [] SCREAMING_SNAKE_CASE : Optional[int] = max(lowercase , lowercase ) SCREAMING_SNAKE_CASE : List[str] = 0 SCREAMING_SNAKE_CASE : List[Any] = 0 SCREAMING_SNAKE_CASE : Any = [] # captured numbers int both 'primeFac1' and 'primeFac2' # iterates through primeFac1 for n in prime_fac_a: if n not in done: if n in prime_fac_a: SCREAMING_SNAKE_CASE : Optional[int] = prime_fac_a.count(lowercase ) SCREAMING_SNAKE_CASE : Tuple = prime_fac_a.count(lowercase ) for _ in range(max(lowercase , lowercase ) ): ans *= n else: SCREAMING_SNAKE_CASE : List[Any] = prime_fac_a.count(lowercase ) for _ in range(lowercase ): ans *= n done.append(lowercase ) # iterates through primeFac2 for n in prime_fac_a: if n not in done: SCREAMING_SNAKE_CASE : int = prime_fac_a.count(lowercase ) for _ in range(lowercase ): ans *= n done.append(lowercase ) # precondition assert isinstance(lowercase , lowercase ) and ( ans >= 0 ), "'ans' must been from type int and positive" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and (n >= 0), "'number' must been a positive int" SCREAMING_SNAKE_CASE : List[str] = 0 SCREAMING_SNAKE_CASE : Optional[int] = 2 # this variable holds the answer while index < n: index += 1 ans += 1 # counts to the next number # if ans not prime then # runs to the next prime number. while not is_prime(lowercase ): ans += 1 # precondition assert isinstance(lowercase , lowercase ) and is_prime( lowercase ), "'ans' must been a prime number and from type int" return ans def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( is_prime(lowercase ) and is_prime(lowercase ) and (p_number_a < p_number_a) ), "The arguments must been prime numbers and 'pNumber1' < 'pNumber2'" SCREAMING_SNAKE_CASE : str = p_number_a + 1 # jump to the next number SCREAMING_SNAKE_CASE : Optional[Any] = [] # this list will be returns. # if number is not prime then # fetch the next prime number. while not is_prime(lowercase ): number += 1 while number < p_number_a: ans.append(lowercase ) number += 1 # fetch the next prime number. while not is_prime(lowercase ): number += 1 # precondition assert ( isinstance(lowercase , lowercase ) and ans[0] != p_number_a and ans[len(lowercase ) - 1] != p_number_a ), "'ans' must been a list without the arguments" # 'ans' contains not 'pNumber1' and 'pNumber2' ! return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and (n >= 1), "'n' must been int and >= 1" SCREAMING_SNAKE_CASE : Any = [] # will be returned. for divisor in range(1 , n + 1 ): if n % divisor == 0: ans.append(lowercase ) # precondition assert ans[0] == 1 and ans[len(lowercase ) - 1] == n, "Error in function getDivisiors(...)" return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and ( number > 1 ), "'number' must been an int and >= 1" SCREAMING_SNAKE_CASE : Optional[int] = get_divisors(lowercase ) # precondition assert ( isinstance(lowercase , lowercase ) and (divisors[0] == 1) and (divisors[len(lowercase ) - 1] == number) ), "Error in help-function getDivisiors(...)" # summed all divisors up to 'number' (exclusive), hence [:-1] return sum(divisors[:-1] ) == number def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert ( isinstance(lowercase , lowercase ) and isinstance(lowercase , lowercase ) and (denominator != 0) ), "The arguments must been from type int and 'denominator' != 0" # build the greatest common divisor of numerator and denominator. SCREAMING_SNAKE_CASE : List[str] = gcd(abs(lowercase ) , abs(lowercase ) ) # precondition assert ( isinstance(lowercase , lowercase ) and (numerator % gcd_of_fraction == 0) and (denominator % gcd_of_fraction == 0) ), "Error in function gcd(...,...)" return (numerator // gcd_of_fraction, denominator // gcd_of_fraction) def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and (n >= 0), "'n' must been a int and >= 0" SCREAMING_SNAKE_CASE : Union[str, Any] = 1 # this will be return. for factor in range(1 , n + 1 ): ans *= factor return ans def lowerCamelCase__ ( lowercase ): """simple docstring""" assert isinstance(lowercase , lowercase ) and (n >= 0), "'n' must been an int and >= 0" SCREAMING_SNAKE_CASE : Tuple = 0 SCREAMING_SNAKE_CASE : str = 1 SCREAMING_SNAKE_CASE : int = 1 # this will be return for _ in range(n - 1 ): SCREAMING_SNAKE_CASE : str = ans ans += fiba SCREAMING_SNAKE_CASE : Tuple = tmp return ans
62
from typing import TYPE_CHECKING from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available snake_case = { """configuration_jukebox""": [ """JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP""", """JukeboxConfig""", """JukeboxPriorConfig""", """JukeboxVQVAEConfig""", ], """tokenization_jukebox""": ["""JukeboxTokenizer"""], } try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: snake_case = [ """JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST""", """JukeboxModel""", """JukeboxPreTrainedModel""", """JukeboxVQVAE""", """JukeboxPrior""", ] if TYPE_CHECKING: from .configuration_jukebox import ( JUKEBOX_PRETRAINED_CONFIG_ARCHIVE_MAP, JukeboxConfig, JukeboxPriorConfig, JukeboxVQVAEConfig, ) from .tokenization_jukebox import JukeboxTokenizer try: if not is_torch_available(): raise OptionalDependencyNotAvailable() except OptionalDependencyNotAvailable: pass else: from .modeling_jukebox import ( JUKEBOX_PRETRAINED_MODEL_ARCHIVE_LIST, JukeboxModel, JukeboxPreTrainedModel, JukeboxPrior, JukeboxVQVAE, ) else: import sys snake_case = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
62
1
from typing import List, Optional, Union import torch from ...models import UNetaDConditionModel, VQModel from ...pipelines import DiffusionPipeline from ...pipelines.pipeline_utils import ImagePipelineOutput from ...schedulers import DDPMScheduler from ...utils import ( is_accelerate_available, is_accelerate_version, logging, randn_tensor, replace_example_docstring, ) snake_case = logging.get_logger(__name__) # pylint: disable=invalid-name snake_case = """ Examples: ```py >>> from diffusers import KandinskyV22Pipeline, KandinskyV22PriorPipeline >>> import torch >>> pipe_prior = KandinskyV22PriorPipeline.from_pretrained(\"kandinsky-community/kandinsky-2-2-prior\") >>> pipe_prior.to(\"cuda\") >>> prompt = \"red cat, 4k photo\" >>> out = pipe_prior(prompt) >>> image_emb = out.image_embeds >>> zero_image_emb = out.negative_image_embeds >>> pipe = KandinskyV22Pipeline.from_pretrained(\"kandinsky-community/kandinsky-2-2-decoder\") >>> pipe.to(\"cuda\") >>> image = pipe( ... image_embeds=image_emb, ... negative_image_embeds=zero_image_emb, ... height=768, ... width=768, ... num_inference_steps=50, ... ).images >>> image[0].save(\"cat.png\") ``` """ def lowerCamelCase__ ( lowercase , lowercase , lowercase=8 ): """simple docstring""" SCREAMING_SNAKE_CASE : int = height // scale_factor**2 if height % scale_factor**2 != 0: new_height += 1 SCREAMING_SNAKE_CASE : List[Any] = width // scale_factor**2 if width % scale_factor**2 != 0: new_width += 1 return new_height * scale_factor, new_width * scale_factor class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : Union[str, Any] , UpperCAmelCase_ : UNetaDConditionModel , UpperCAmelCase_ : DDPMScheduler , UpperCAmelCase_ : VQModel , ): super().__init__() self.register_modules( unet=UpperCAmelCase_ , scheduler=UpperCAmelCase_ , movq=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Union[str, Any] = 2 ** (len(self.movq.config.block_out_channels ) - 1) def _A ( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Union[str, Any] ): if latents is None: SCREAMING_SNAKE_CASE : List[str] = randn_tensor(UpperCAmelCase_ , generator=UpperCAmelCase_ , device=UpperCAmelCase_ , dtype=UpperCAmelCase_ ) else: if latents.shape != shape: raise ValueError(f'''Unexpected latents shape, got {latents.shape}, expected {shape}''' ) SCREAMING_SNAKE_CASE : Tuple = latents.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = latents * scheduler.init_noise_sigma return latents def _A ( self : List[str] , UpperCAmelCase_ : Optional[int]=0 ): if is_accelerate_available(): from accelerate import cpu_offload else: raise ImportError("Please install accelerate via `pip install accelerate`" ) SCREAMING_SNAKE_CASE : Tuple = torch.device(f'''cuda:{gpu_id}''' ) SCREAMING_SNAKE_CASE : int = [ self.unet, self.movq, ] for cpu_offloaded_model in models: if cpu_offloaded_model is not None: cpu_offload(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : str=0 ): if is_accelerate_available() and is_accelerate_version(">=" , "0.17.0.dev0" ): from accelerate import cpu_offload_with_hook else: raise ImportError("`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher." ) SCREAMING_SNAKE_CASE : Tuple = torch.device(f'''cuda:{gpu_id}''' ) if self.device.type != "cpu": self.to("cpu" , silence_dtype_warnings=UpperCAmelCase_ ) torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist) SCREAMING_SNAKE_CASE : Optional[Any] = None for cpu_offloaded_model in [self.unet, self.movq]: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = cpu_offload_with_hook(UpperCAmelCase_ , UpperCAmelCase_ , prev_module_hook=UpperCAmelCase_ ) # We'll offload the last model manually. SCREAMING_SNAKE_CASE : List[str] = hook @property # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._execution_device def _A ( self : List[Any] ): if not hasattr(self.unet , "_hf_hook" ): return self.device for module in self.unet.modules(): if ( hasattr(UpperCAmelCase_ , "_hf_hook" ) and hasattr(module._hf_hook , "execution_device" ) and module._hf_hook.execution_device is not None ): return torch.device(module._hf_hook.execution_device ) return self.device @torch.no_grad() @replace_example_docstring(UpperCAmelCase_ ) def __call__( self : int , UpperCAmelCase_ : Union[torch.FloatTensor, List[torch.FloatTensor]] , UpperCAmelCase_ : Union[torch.FloatTensor, List[torch.FloatTensor]] , UpperCAmelCase_ : int = 512 , UpperCAmelCase_ : int = 512 , UpperCAmelCase_ : int = 100 , UpperCAmelCase_ : float = 4.0 , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : Optional[Union[torch.Generator, List[torch.Generator]]] = None , UpperCAmelCase_ : Optional[torch.FloatTensor] = None , UpperCAmelCase_ : Optional[str] = "pil" , UpperCAmelCase_ : bool = True , ): SCREAMING_SNAKE_CASE : str = self._execution_device SCREAMING_SNAKE_CASE : str = guidance_scale > 1.0 if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Dict = torch.cat(UpperCAmelCase_ , dim=0 ) SCREAMING_SNAKE_CASE : List[str] = image_embeds.shape[0] * num_images_per_prompt if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = torch.cat(UpperCAmelCase_ , dim=0 ) if do_classifier_free_guidance: SCREAMING_SNAKE_CASE : Dict = image_embeds.repeat_interleave(UpperCAmelCase_ , dim=0 ) SCREAMING_SNAKE_CASE : Any = negative_image_embeds.repeat_interleave(UpperCAmelCase_ , dim=0 ) SCREAMING_SNAKE_CASE : List[Any] = torch.cat([negative_image_embeds, image_embeds] , dim=0 ).to(dtype=self.unet.dtype , device=UpperCAmelCase_ ) self.scheduler.set_timesteps(UpperCAmelCase_ , device=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.scheduler.timesteps SCREAMING_SNAKE_CASE : Tuple = self.unet.config.in_channels SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = downscale_height_and_width(UpperCAmelCase_ , UpperCAmelCase_ , self.movq_scale_factor ) # create initial latent SCREAMING_SNAKE_CASE : Tuple = self.prepare_latents( (batch_size, num_channels_latents, height, width) , image_embeds.dtype , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , self.scheduler , ) for i, t in enumerate(self.progress_bar(UpperCAmelCase_ ) ): # expand the latents if we are doing classifier free guidance SCREAMING_SNAKE_CASE : List[str] = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents SCREAMING_SNAKE_CASE : List[str] = {"image_embeds": image_embeds} SCREAMING_SNAKE_CASE : Union[str, Any] = self.unet( sample=UpperCAmelCase_ , timestep=UpperCAmelCase_ , encoder_hidden_states=UpperCAmelCase_ , added_cond_kwargs=UpperCAmelCase_ , return_dict=UpperCAmelCase_ , )[0] if do_classifier_free_guidance: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = noise_pred.split(latents.shape[1] , dim=1 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = noise_pred.chunk(2 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = variance_pred.chunk(2 ) SCREAMING_SNAKE_CASE : int = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) SCREAMING_SNAKE_CASE : Union[str, Any] = torch.cat([noise_pred, variance_pred_text] , dim=1 ) if not ( hasattr(self.scheduler.config , "variance_type" ) and self.scheduler.config.variance_type in ["learned", "learned_range"] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = noise_pred.split(latents.shape[1] , dim=1 ) # compute the previous noisy sample x_t -> x_t-1 SCREAMING_SNAKE_CASE : List[Any] = self.scheduler.step( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , generator=UpperCAmelCase_ , )[0] # post-processing SCREAMING_SNAKE_CASE : List[str] = self.movq.decode(UpperCAmelCase_ , force_not_quantize=UpperCAmelCase_ )["sample"] if output_type not in ["pt", "np", "pil"]: raise ValueError(f'''Only the output types `pt`, `pil` and `np` are supported not output_type={output_type}''' ) if output_type in ["np", "pil"]: SCREAMING_SNAKE_CASE : Dict = image * 0.5 + 0.5 SCREAMING_SNAKE_CASE : Dict = image.clamp(0 , 1 ) SCREAMING_SNAKE_CASE : str = image.cpu().permute(0 , 2 , 3 , 1 ).float().numpy() if output_type == "pil": SCREAMING_SNAKE_CASE : List[Any] = self.numpy_to_pil(UpperCAmelCase_ ) if not return_dict: return (image,) return ImagePipelineOutput(images=UpperCAmelCase_ )
62
from typing import Dict, List from nltk.translate import gleu_score import datasets from datasets import MetricInfo snake_case = """\ @misc{wu2016googles, title={Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation}, author={Yonghui Wu and Mike Schuster and Zhifeng Chen and Quoc V. Le and Mohammad Norouzi and Wolfgang Macherey and Maxim Krikun and Yuan Cao and Qin Gao and Klaus Macherey and Jeff Klingner and Apurva Shah and Melvin Johnson and Xiaobing Liu and Łukasz Kaiser and Stephan Gouws and Yoshikiyo Kato and Taku Kudo and Hideto Kazawa and Keith Stevens and George Kurian and Nishant Patil and Wei Wang and Cliff Young and Jason Smith and Jason Riesa and Alex Rudnick and Oriol Vinyals and Greg Corrado and Macduff Hughes and Jeffrey Dean}, year={2016}, eprint={1609.08144}, archivePrefix={arXiv}, primaryClass={cs.CL} } """ snake_case = """\ The BLEU score has some undesirable properties when used for single sentences, as it was designed to be a corpus measure. We therefore use a slightly different score for our RL experiments which we call the 'GLEU score'. For the GLEU score, we record all sub-sequences of 1, 2, 3 or 4 tokens in output and target sequence (n-grams). We then compute a recall, which is the ratio of the number of matching n-grams to the number of total n-grams in the target (ground truth) sequence, and a precision, which is the ratio of the number of matching n-grams to the number of total n-grams in the generated output sequence. Then GLEU score is simply the minimum of recall and precision. This GLEU score's range is always between 0 (no matches) and 1 (all match) and it is symmetrical when switching output and target. According to our experiments, GLEU score correlates quite well with the BLEU metric on a corpus level but does not have its drawbacks for our per sentence reward objective. """ snake_case = """\ Computes corpus-level Google BLEU (GLEU) score of translated segments against one or more references. Instead of averaging the sentence level GLEU scores (i.e. macro-average precision), Wu et al. (2016) sum up the matching tokens and the max of hypothesis and reference tokens for each sentence, then compute using the aggregate values. Args: predictions (list of str): list of translations to score. Each translation should be tokenized into a list of tokens. references (list of list of str): list of lists of references for each translation. Each reference should be tokenized into a list of tokens. min_len (int): The minimum order of n-gram this function should extract. Defaults to 1. max_len (int): The maximum order of n-gram this function should extract. Defaults to 4. Returns: 'google_bleu': google_bleu score Examples: Example 1: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.44 Example 2: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references) >>> print(round(results[\"google_bleu\"], 2)) 0.61 Example 3: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses, references=list_of_references, min_len=2) >>> print(round(results[\"google_bleu\"], 2)) 0.53 Example 4: >>> hyp1 = ['It', 'is', 'a', 'guide', 'to', 'action', 'which', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'always', ... 'disobeys', 'the', 'commands', 'of', 'the', 'cat'] >>> ref1a = ['It', 'is', 'the', 'guiding', 'principle', 'which', ... 'guarantees', 'the', 'rubber', 'duck', 'forces', 'never', ... 'being', 'under', 'the', 'command', 'of', 'the', 'cat'] >>> ref1b = ['It', 'is', 'a', 'guide', 'to', 'action', 'that', ... 'ensures', 'that', 'the', 'rubber', 'duck', 'will', 'never', ... 'heed', 'the', 'cat', 'commands'] >>> ref1c = ['It', 'is', 'the', 'practical', 'guide', 'for', 'the', ... 'rubber', 'duck', 'army', 'never', 'to', 'heed', 'the', 'directions', ... 'of', 'the', 'cat'] >>> hyp2 = ['he', 'read', 'the', 'book', 'because', 'he', 'was', ... 'interested', 'in', 'world', 'history'] >>> ref2a = ['he', 'was', 'interested', 'in', 'world', 'history', ... 'because', 'he', 'read', 'the', 'book'] >>> list_of_references = [[ref1a, ref1b, ref1c], [ref2a]] >>> hypotheses = [hyp1, hyp2] >>> google_bleu = datasets.load_metric(\"google_bleu\") >>> results = google_bleu.compute(predictions=hypotheses,references=list_of_references, min_len=2, max_len=6) >>> print(round(results[\"google_bleu\"], 2)) 0.4 """ @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : Optional[Any] ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { "predictions": datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ), "references": datasets.Sequence( datasets.Sequence(datasets.Value("string" , id="token" ) , id="sequence" ) , id="references" ), } ) , ) def _A ( self : List[Any] , UpperCAmelCase_ : List[List[List[str]]] , UpperCAmelCase_ : List[List[str]] , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : int = 4 , ): return { "google_bleu": gleu_score.corpus_gleu( list_of_references=UpperCAmelCase_ , hypotheses=UpperCAmelCase_ , min_len=UpperCAmelCase_ , max_len=UpperCAmelCase_ ) }
62
1
import os import tempfile from functools import partial from unittest import TestCase from unittest.mock import patch import datasets import datasets.config from .utils import require_beam class SCREAMING_SNAKE_CASE ( datasets.BeamBasedBuilder ): '''simple docstring''' def _A ( self : Any ): return datasets.DatasetInfo( features=datasets.Features({"content": datasets.Value("string" )} ) , supervised_keys=UpperCAmelCase_ , ) def _A ( self : Any , UpperCAmelCase_ : Any , UpperCAmelCase_ : str ): return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={"examples": get_test_dummy_examples()} )] def _A ( self : List[Any] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : List[Any] ): import apache_beam as beam return pipeline | "Load Examples" >> beam.Create(UpperCAmelCase_ ) class SCREAMING_SNAKE_CASE ( datasets.BeamBasedBuilder ): '''simple docstring''' def _A ( self : str ): return datasets.DatasetInfo( features=datasets.Features({"a": datasets.Sequence({"b": datasets.Value("string" )} )} ) , supervised_keys=UpperCAmelCase_ , ) def _A ( self : str , UpperCAmelCase_ : Any , UpperCAmelCase_ : Optional[Any] ): return [ datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={"examples": get_test_nested_examples()} ) ] def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : Tuple ): import apache_beam as beam return pipeline | "Load Examples" >> beam.Create(UpperCAmelCase_ ) def lowerCamelCase__ ( ): """simple docstring""" return [(i, {"content": content}) for i, content in enumerate(["foo", "bar", "foobar"] )] def lowerCamelCase__ ( ): """simple docstring""" return [(i, {"a": {"b": [content]}}) for i, content in enumerate(["foo", "bar", "foobar"] )] class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @require_beam def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = len(get_test_dummy_examples() ) with tempfile.TemporaryDirectory() as tmp_cache_dir: SCREAMING_SNAKE_CASE : Tuple = DummyBeamDataset(cache_dir=UpperCAmelCase_ , beam_runner="DirectRunner" ) builder.download_and_prepare() self.assertTrue( os.path.exists( os.path.join(UpperCAmelCase_ , builder.name , "default" , "0.0.0" , f'''{builder.name}-train.arrow''' ) ) ) self.assertDictEqual(builder.info.features , datasets.Features({"content": datasets.Value("string" )} ) ) SCREAMING_SNAKE_CASE : int = builder.as_dataset() self.assertEqual(dset["train"].num_rows , UpperCAmelCase_ ) self.assertEqual(dset["train"].info.splits["train"].num_examples , UpperCAmelCase_ ) self.assertDictEqual(dset["train"][0] , get_test_dummy_examples()[0][1] ) self.assertDictEqual( dset["train"][expected_num_examples - 1] , get_test_dummy_examples()[expected_num_examples - 1][1] ) self.assertTrue( os.path.exists(os.path.join(UpperCAmelCase_ , builder.name , "default" , "0.0.0" , "dataset_info.json" ) ) ) del dset @require_beam def _A ( self : List[Any] ): import apache_beam as beam SCREAMING_SNAKE_CASE : List[Any] = beam.io.parquetio.WriteToParquet SCREAMING_SNAKE_CASE : Any = len(get_test_dummy_examples() ) with tempfile.TemporaryDirectory() as tmp_cache_dir: SCREAMING_SNAKE_CASE : Any = DummyBeamDataset(cache_dir=UpperCAmelCase_ , beam_runner="DirectRunner" ) with patch("apache_beam.io.parquetio.WriteToParquet" ) as write_parquet_mock: SCREAMING_SNAKE_CASE : Tuple = partial(UpperCAmelCase_ , num_shards=2 ) builder.download_and_prepare() self.assertTrue( os.path.exists( os.path.join( UpperCAmelCase_ , builder.name , "default" , "0.0.0" , f'''{builder.name}-train-00000-of-00002.arrow''' ) ) ) self.assertTrue( os.path.exists( os.path.join( UpperCAmelCase_ , builder.name , "default" , "0.0.0" , f'''{builder.name}-train-00000-of-00002.arrow''' ) ) ) self.assertDictEqual(builder.info.features , datasets.Features({"content": datasets.Value("string" )} ) ) SCREAMING_SNAKE_CASE : Optional[Any] = builder.as_dataset() self.assertEqual(dset["train"].num_rows , UpperCAmelCase_ ) self.assertEqual(dset["train"].info.splits["train"].num_examples , UpperCAmelCase_ ) # Order is not preserved when sharding, so we just check that all the elements are there self.assertListEqual(sorted(dset["train"]["content"] ) , sorted(["foo", "bar", "foobar"] ) ) self.assertTrue( os.path.exists(os.path.join(UpperCAmelCase_ , builder.name , "default" , "0.0.0" , "dataset_info.json" ) ) ) del dset @require_beam def _A ( self : List[str] ): with tempfile.TemporaryDirectory() as tmp_cache_dir: SCREAMING_SNAKE_CASE : Optional[Any] = DummyBeamDataset(cache_dir=UpperCAmelCase_ ) self.assertRaises(datasets.builder.MissingBeamOptions , builder.download_and_prepare ) @require_beam def _A ( self : Dict ): SCREAMING_SNAKE_CASE : int = len(get_test_nested_examples() ) with tempfile.TemporaryDirectory() as tmp_cache_dir: SCREAMING_SNAKE_CASE : Any = NestedBeamDataset(cache_dir=UpperCAmelCase_ , beam_runner="DirectRunner" ) builder.download_and_prepare() self.assertTrue( os.path.exists( os.path.join(UpperCAmelCase_ , builder.name , "default" , "0.0.0" , f'''{builder.name}-train.arrow''' ) ) ) self.assertDictEqual( builder.info.features , datasets.Features({"a": datasets.Sequence({"b": datasets.Value("string" )} )} ) ) SCREAMING_SNAKE_CASE : Optional[Any] = builder.as_dataset() self.assertEqual(dset["train"].num_rows , UpperCAmelCase_ ) self.assertEqual(dset["train"].info.splits["train"].num_examples , UpperCAmelCase_ ) self.assertDictEqual(dset["train"][0] , get_test_nested_examples()[0][1] ) self.assertDictEqual( dset["train"][expected_num_examples - 1] , get_test_nested_examples()[expected_num_examples - 1][1] ) self.assertTrue( os.path.exists(os.path.join(UpperCAmelCase_ , builder.name , "default" , "0.0.0" , "dataset_info.json" ) ) ) del dset
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : int , UpperCAmelCase_ : float = 0 ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = row, column SCREAMING_SNAKE_CASE : Optional[Any] = [[default_value for c in range(UpperCAmelCase_ )] for r in range(UpperCAmelCase_ )] def __str__( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = f'''Matrix consist of {self.row} rows and {self.column} columns\n''' # Make string identifier SCREAMING_SNAKE_CASE : Dict = 0 for row_vector in self.array: for obj in row_vector: SCREAMING_SNAKE_CASE : Optional[Any] = max(UpperCAmelCase_ , len(str(UpperCAmelCase_ ) ) ) SCREAMING_SNAKE_CASE : Optional[int] = f'''%{max_element_length}s''' # Make string and return def single_line(UpperCAmelCase_ : list[float] ) -> str: nonlocal string_format_identifier SCREAMING_SNAKE_CASE : Optional[int] = "[" line += ", ".join(string_format_identifier % (obj,) for obj in row_vector ) line += "]" return line s += "\n".join(single_line(UpperCAmelCase_ ) for row_vector in self.array ) return s def __repr__( self : Dict ): return str(self ) def _A ( self : Optional[int] , UpperCAmelCase_ : tuple[int, int] ): if not (isinstance(UpperCAmelCase_ , (list, tuple) ) and len(UpperCAmelCase_ ) == 2): return False elif not (0 <= loc[0] < self.row and 0 <= loc[1] < self.column): return False else: return True def __getitem__( self : Any , UpperCAmelCase_ : tuple[int, int] ): assert self.validate_indicies(UpperCAmelCase_ ) return self.array[loc[0]][loc[1]] def __setitem__( self : int , UpperCAmelCase_ : tuple[int, int] , UpperCAmelCase_ : float ): assert self.validate_indicies(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = value def __add__( self : Optional[int] , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == another.row and self.column == another.column # Add SCREAMING_SNAKE_CASE : str = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : Any = self[r, c] + another[r, c] return result def __neg__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = -self[r, c] return result def __sub__( self : Optional[Any] , UpperCAmelCase_ : Matrix ): return self + (-another) def __mul__( self : Dict , UpperCAmelCase_ : int | float | Matrix ): if isinstance(UpperCAmelCase_ , (int, float) ): # Scalar multiplication SCREAMING_SNAKE_CASE : Any = Matrix(self.row , self.column ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : str = self[r, c] * another return result elif isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): # Matrix multiplication assert self.column == another.row SCREAMING_SNAKE_CASE : Any = Matrix(self.row , another.column ) for r in range(self.row ): for c in range(another.column ): for i in range(self.column ): result[r, c] += self[r, i] * another[i, c] return result else: SCREAMING_SNAKE_CASE : List[str] = f'''Unsupported type given for another ({type(UpperCAmelCase_ )})''' raise TypeError(UpperCAmelCase_ ) def _A ( self : int ): SCREAMING_SNAKE_CASE : List[str] = Matrix(self.column , self.row ) for r in range(self.row ): for c in range(self.column ): SCREAMING_SNAKE_CASE : List[str] = self[r, c] return result def _A ( self : Union[str, Any] , UpperCAmelCase_ : Matrix , UpperCAmelCase_ : Matrix ): assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) and isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) assert self.row == self.column == u.row == v.row # u, v should be column vector assert u.column == v.column == 1 # u, v should be column vector # Calculate SCREAMING_SNAKE_CASE : Tuple = v.transpose() SCREAMING_SNAKE_CASE : int = (v_t * self * u)[0, 0] + 1 if numerator_factor == 0: return None # It's not invertable return self - ((self * u) * (v_t * self) * (1.0 / numerator_factor)) # Testing if __name__ == "__main__": def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 3 , 0 ) for i in range(3 ): SCREAMING_SNAKE_CASE : str = 1 print(F'''a^(-1) is {ainv}''' ) # u, v SCREAMING_SNAKE_CASE : Optional[int] = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = 1, 2, -3 SCREAMING_SNAKE_CASE : Tuple = Matrix(3 , 1 , 0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = 4, -2, 5 print(F'''u is {u}''' ) print(F'''v is {v}''' ) print(F'''uv^T is {u * v.transpose()}''' ) # Sherman Morrison print(F'''(a + uv^T)^(-1) is {ainv.sherman_morrison(lowercase , lowercase )}''' ) def lowerCamelCase__ ( ): """simple docstring""" import doctest doctest.testmod() testa()
62
1
import inspect from typing import List, Optional, Tuple, Union import torch from ...models import UNetaDModel, VQModel from ...schedulers import DDIMScheduler from ...utils import randn_tensor from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , UpperCAmelCase_ : VQModel , UpperCAmelCase_ : UNetaDModel , UpperCAmelCase_ : DDIMScheduler ): super().__init__() self.register_modules(vqvae=UpperCAmelCase_ , unet=UpperCAmelCase_ , scheduler=UpperCAmelCase_ ) @torch.no_grad() def __call__( self : Optional[Any] , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : Optional[Union[torch.Generator, List[torch.Generator]]] = None , UpperCAmelCase_ : float = 0.0 , UpperCAmelCase_ : int = 50 , UpperCAmelCase_ : Optional[str] = "pil" , UpperCAmelCase_ : bool = True , **UpperCAmelCase_ : Tuple , ): SCREAMING_SNAKE_CASE : str = randn_tensor( (batch_size, self.unet.config.in_channels, self.unet.config.sample_size, self.unet.config.sample_size) , generator=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = latents.to(self.device ) # scale the initial noise by the standard deviation required by the scheduler SCREAMING_SNAKE_CASE : Tuple = latents * self.scheduler.init_noise_sigma self.scheduler.set_timesteps(UpperCAmelCase_ ) # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature SCREAMING_SNAKE_CASE : Dict = "eta" in set(inspect.signature(self.scheduler.step ).parameters.keys() ) SCREAMING_SNAKE_CASE : List[Any] = {} if accepts_eta: SCREAMING_SNAKE_CASE : Optional[int] = eta for t in self.progress_bar(self.scheduler.timesteps ): SCREAMING_SNAKE_CASE : Optional[Any] = self.scheduler.scale_model_input(UpperCAmelCase_ , UpperCAmelCase_ ) # predict the noise residual SCREAMING_SNAKE_CASE : str = self.unet(UpperCAmelCase_ , UpperCAmelCase_ ).sample # compute the previous noisy sample x_t -> x_t-1 SCREAMING_SNAKE_CASE : int = self.scheduler.step(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ ).prev_sample # decode the image latents with the VAE SCREAMING_SNAKE_CASE : int = self.vqvae.decode(UpperCAmelCase_ ).sample SCREAMING_SNAKE_CASE : int = (image / 2 + 0.5).clamp(0 , 1 ) SCREAMING_SNAKE_CASE : Union[str, Any] = image.cpu().permute(0 , 2 , 3 , 1 ).numpy() if output_type == "pil": SCREAMING_SNAKE_CASE : Optional[int] = self.numpy_to_pil(UpperCAmelCase_ ) if not return_dict: return (image,) return ImagePipelineOutput(images=UpperCAmelCase_ )
62
import json from typing import List, Optional, Tuple from tokenizers import normalizers from tokenizers.pre_tokenizers import BertPreTokenizer, PreTokenizer from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import logging from .tokenization_roformer import RoFormerTokenizer from .tokenization_utils import JiebaPreTokenizer snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """vocab.txt""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """junnyu/roformer_chinese_small""": """https://huggingface.co/junnyu/roformer_chinese_small/resolve/main/vocab.txt""", """junnyu/roformer_chinese_base""": """https://huggingface.co/junnyu/roformer_chinese_base/resolve/main/vocab.txt""", """junnyu/roformer_chinese_char_small""": ( """https://huggingface.co/junnyu/roformer_chinese_char_small/resolve/main/vocab.txt""" ), """junnyu/roformer_chinese_char_base""": ( """https://huggingface.co/junnyu/roformer_chinese_char_base/resolve/main/vocab.txt""" ), """junnyu/roformer_small_discriminator""": ( """https://huggingface.co/junnyu/roformer_small_discriminator/resolve/main/vocab.txt""" ), """junnyu/roformer_small_generator""": ( """https://huggingface.co/junnyu/roformer_small_generator/resolve/main/vocab.txt""" ), } } snake_case = { """junnyu/roformer_chinese_small""": 1_536, """junnyu/roformer_chinese_base""": 1_536, """junnyu/roformer_chinese_char_small""": 512, """junnyu/roformer_chinese_char_base""": 512, """junnyu/roformer_small_discriminator""": 128, """junnyu/roformer_small_generator""": 128, } snake_case = { """junnyu/roformer_chinese_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_base""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_small""": {"""do_lower_case""": True}, """junnyu/roformer_chinese_char_base""": {"""do_lower_case""": True}, """junnyu/roformer_small_discriminator""": {"""do_lower_case""": True}, """junnyu/roformer_small_generator""": {"""do_lower_case""": True}, } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = VOCAB_FILES_NAMES UpperCamelCase_ : int = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Tuple = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : List[Any] = PRETRAINED_INIT_CONFIGURATION UpperCamelCase_ : Any = RoFormerTokenizer def __init__( self : Tuple , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : List[Any]=None , UpperCAmelCase_ : Optional[Any]=True , UpperCAmelCase_ : List[str]="[UNK]" , UpperCAmelCase_ : Any="[SEP]" , UpperCAmelCase_ : Any="[PAD]" , UpperCAmelCase_ : List[str]="[CLS]" , UpperCAmelCase_ : str="[MASK]" , UpperCAmelCase_ : Any=True , UpperCAmelCase_ : Optional[Any]=None , **UpperCAmelCase_ : List[str] , ): super().__init__( UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , do_lower_case=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , tokenize_chinese_chars=UpperCAmelCase_ , strip_accents=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : int = json.loads(self.backend_tokenizer.normalizer.__getstate__() ) if ( pre_tok_state.get("lowercase" , UpperCAmelCase_ ) != do_lower_case or pre_tok_state.get("strip_accents" , UpperCAmelCase_ ) != strip_accents ): SCREAMING_SNAKE_CASE : Optional[Any] = getattr(UpperCAmelCase_ , pre_tok_state.pop("type" ) ) SCREAMING_SNAKE_CASE : Any = do_lower_case SCREAMING_SNAKE_CASE : List[str] = strip_accents SCREAMING_SNAKE_CASE : Tuple = pre_tok_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = do_lower_case def __getstate__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = self.__dict__.copy() SCREAMING_SNAKE_CASE : Optional[Any] = BertPreTokenizer() return state def __setstate__( self : Tuple , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Dict = d SCREAMING_SNAKE_CASE : Dict = self.__dict__["_tokenizer"].get_vocab() SCREAMING_SNAKE_CASE : Any = PreTokenizer.custom(JiebaPreTokenizer(UpperCAmelCase_ ) ) def _A ( self : str , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Union[str, Any]=None ): SCREAMING_SNAKE_CASE : List[Any] = [self.cls_token_id] + token_ids_a + [self.sep_token_id] if token_ids_a: output += token_ids_a + [self.sep_token_id] return output def _A ( self : Tuple , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : List[Any] = [self.sep_token_id] SCREAMING_SNAKE_CASE : Optional[int] = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep ) * [0] + len(token_ids_a + sep ) * [1] def _A ( self : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): SCREAMING_SNAKE_CASE : Optional[int] = self._tokenizer.model.save(UpperCAmelCase_ , name=UpperCAmelCase_ ) return tuple(UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Any=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Any=False , **UpperCAmelCase_ : str , ): SCREAMING_SNAKE_CASE : Union[str, Any] = BertPreTokenizer() return super().save_pretrained(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , **UpperCAmelCase_ )
62
1
import functools import operator from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """microsoft/unispeech-large-1500h-cv""": ( """https://huggingface.co/microsoft/unispeech-large-1500h-cv/resolve/main/config.json""" ), # See all UniSpeech models at https://huggingface.co/models?filter=unispeech } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Union[str, Any] = '''unispeech''' def __init__( self : List[str] , UpperCAmelCase_ : str=32 , UpperCAmelCase_ : Optional[Any]=768 , UpperCAmelCase_ : Tuple=12 , UpperCAmelCase_ : Dict=12 , UpperCAmelCase_ : Any=3072 , UpperCAmelCase_ : List[Any]="gelu" , UpperCAmelCase_ : Optional[Any]=0.1 , UpperCAmelCase_ : Dict=0.1 , UpperCAmelCase_ : Optional[Any]=0.1 , UpperCAmelCase_ : List[str]=0.0 , UpperCAmelCase_ : Optional[int]=0.0 , UpperCAmelCase_ : List[Any]=0.1 , UpperCAmelCase_ : List[str]=0.1 , UpperCAmelCase_ : Union[str, Any]=0.02 , UpperCAmelCase_ : Optional[int]=1E-5 , UpperCAmelCase_ : List[Any]="group" , UpperCAmelCase_ : Optional[Any]="gelu" , UpperCAmelCase_ : List[str]=(512, 512, 512, 512, 512, 512, 512) , UpperCAmelCase_ : Tuple=(5, 2, 2, 2, 2, 2, 2) , UpperCAmelCase_ : Dict=(10, 3, 3, 3, 3, 2, 2) , UpperCAmelCase_ : Any=False , UpperCAmelCase_ : List[str]=128 , UpperCAmelCase_ : Any=16 , UpperCAmelCase_ : str=False , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : int=0.05 , UpperCAmelCase_ : List[str]=10 , UpperCAmelCase_ : List[str]=2 , UpperCAmelCase_ : Union[str, Any]=0.0 , UpperCAmelCase_ : int=10 , UpperCAmelCase_ : Optional[Any]=0 , UpperCAmelCase_ : int=320 , UpperCAmelCase_ : Optional[Any]=2 , UpperCAmelCase_ : Optional[Any]=0.1 , UpperCAmelCase_ : Any=100 , UpperCAmelCase_ : int=256 , UpperCAmelCase_ : Any=256 , UpperCAmelCase_ : str=0.1 , UpperCAmelCase_ : Dict="mean" , UpperCAmelCase_ : int=False , UpperCAmelCase_ : List[str]=False , UpperCAmelCase_ : List[str]=256 , UpperCAmelCase_ : str=80 , UpperCAmelCase_ : List[str]=0 , UpperCAmelCase_ : Tuple=1 , UpperCAmelCase_ : Optional[int]=2 , UpperCAmelCase_ : Dict=0.5 , **UpperCAmelCase_ : Tuple , ): super().__init__(**UpperCAmelCase_ , pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = hidden_size SCREAMING_SNAKE_CASE : Optional[int] = feat_extract_norm SCREAMING_SNAKE_CASE : Optional[int] = feat_extract_activation SCREAMING_SNAKE_CASE : List[Any] = list(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = list(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = list(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = conv_bias SCREAMING_SNAKE_CASE : Dict = num_conv_pos_embeddings SCREAMING_SNAKE_CASE : Optional[int] = num_conv_pos_embedding_groups SCREAMING_SNAKE_CASE : List[Any] = len(self.conv_dim ) SCREAMING_SNAKE_CASE : Dict = num_hidden_layers SCREAMING_SNAKE_CASE : Optional[int] = intermediate_size SCREAMING_SNAKE_CASE : List[str] = hidden_act SCREAMING_SNAKE_CASE : Optional[Any] = num_attention_heads SCREAMING_SNAKE_CASE : List[Any] = hidden_dropout SCREAMING_SNAKE_CASE : Dict = attention_dropout SCREAMING_SNAKE_CASE : Tuple = activation_dropout SCREAMING_SNAKE_CASE : Dict = feat_proj_dropout SCREAMING_SNAKE_CASE : str = final_dropout SCREAMING_SNAKE_CASE : Dict = layerdrop SCREAMING_SNAKE_CASE : Tuple = layer_norm_eps SCREAMING_SNAKE_CASE : int = initializer_range SCREAMING_SNAKE_CASE : Any = num_ctc_classes SCREAMING_SNAKE_CASE : Dict = vocab_size SCREAMING_SNAKE_CASE : Tuple = do_stable_layer_norm SCREAMING_SNAKE_CASE : Union[str, Any] = use_weighted_layer_sum SCREAMING_SNAKE_CASE : str = classifier_proj_size if ( (len(self.conv_stride ) != self.num_feat_extract_layers) or (len(self.conv_kernel ) != self.num_feat_extract_layers) or (len(self.conv_dim ) != self.num_feat_extract_layers) ): raise ValueError( "Configuration for convolutional layers is incorrect. It is required that `len(config.conv_dim)` ==" " `len(config.conv_stride)` == `len(config.conv_kernel)`, but is `len(config.conv_dim) =" f''' {len(self.conv_dim )}`, `len(config.conv_stride) = {len(self.conv_stride )}`,''' f''' `len(config.conv_kernel) = {len(self.conv_kernel )}`.''' ) # fine-tuning config parameters for SpecAugment: https://arxiv.org/abs/1904.08779 SCREAMING_SNAKE_CASE : int = apply_spec_augment SCREAMING_SNAKE_CASE : Tuple = mask_time_prob SCREAMING_SNAKE_CASE : Any = mask_time_length SCREAMING_SNAKE_CASE : int = mask_time_min_masks SCREAMING_SNAKE_CASE : int = mask_feature_prob SCREAMING_SNAKE_CASE : List[Any] = mask_feature_length SCREAMING_SNAKE_CASE : Dict = mask_feature_min_masks # parameters for pretraining with codevector quantized representations SCREAMING_SNAKE_CASE : Any = num_codevectors_per_group SCREAMING_SNAKE_CASE : Union[str, Any] = num_codevector_groups SCREAMING_SNAKE_CASE : Optional[int] = contrastive_logits_temperature SCREAMING_SNAKE_CASE : int = feat_quantizer_dropout SCREAMING_SNAKE_CASE : List[Any] = num_negatives SCREAMING_SNAKE_CASE : str = codevector_dim SCREAMING_SNAKE_CASE : Tuple = proj_codevector_dim SCREAMING_SNAKE_CASE : Tuple = diversity_loss_weight # ctc loss SCREAMING_SNAKE_CASE : List[Any] = ctc_loss_reduction SCREAMING_SNAKE_CASE : List[Any] = ctc_zero_infinity # pretraining loss SCREAMING_SNAKE_CASE : str = replace_prob @property def _A ( self : int ): return functools.reduce(operator.mul , self.conv_stride , 1 )
62
def lowerCamelCase__ ( lowercase ): """simple docstring""" if not isinstance(lowercase , lowercase ): raise TypeError("only integers accepted as input" ) else: SCREAMING_SNAKE_CASE : Optional[int] = str(abs(lowercase ) ) SCREAMING_SNAKE_CASE : str = [list(lowercase ) for char in range(len(lowercase ) )] for index in range(len(lowercase ) ): num_transpositions[index].pop(lowercase ) return max( int("".join(list(lowercase ) ) ) for transposition in num_transpositions ) if __name__ == "__main__": __import__("""doctest""").testmod()
62
1
import json from typing import Dict, List, Optional, Tuple, Union from tokenizers import pre_tokenizers, processors from ...tokenization_utils_base import AddedToken, BatchEncoding, EncodedInput from ...tokenization_utils_fast import PreTrainedTokenizerFast from ...utils import PaddingStrategy, logging from .tokenization_led import LEDTokenizer snake_case = logging.get_logger(__name__) snake_case = {"""vocab_file""": """vocab.json""", """merges_file""": """merges.txt""", """tokenizer_file""": """tokenizer.json"""} snake_case = { """vocab_file""": { """allenai/led-base-16384""": """https://huggingface.co/allenai/led-base-16384/resolve/main/vocab.json""", }, """merges_file""": { """allenai/led-base-16384""": """https://huggingface.co/allenai/led-base-16384/resolve/main/merges.txt""", }, """tokenizer_file""": { """allenai/led-base-16384""": """https://huggingface.co/allenai/led-base-16384/resolve/main/tokenizer.json""", }, } snake_case = { """allenai/led-base-16384""": 16_384, } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[int] = VOCAB_FILES_NAMES UpperCamelCase_ : Optional[Any] = PRETRAINED_VOCAB_FILES_MAP UpperCamelCase_ : Union[str, Any] = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES UpperCamelCase_ : Optional[Any] = LEDTokenizer UpperCamelCase_ : Union[str, Any] = ['''input_ids''', '''attention_mask'''] def __init__( self : List[str] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : Dict=None , UpperCAmelCase_ : Optional[int]=None , UpperCAmelCase_ : Dict="replace" , UpperCAmelCase_ : Any="<s>" , UpperCAmelCase_ : Optional[int]="</s>" , UpperCAmelCase_ : Optional[int]="</s>" , UpperCAmelCase_ : Union[str, Any]="<s>" , UpperCAmelCase_ : Tuple="<unk>" , UpperCAmelCase_ : Tuple="<pad>" , UpperCAmelCase_ : Optional[int]="<mask>" , UpperCAmelCase_ : Any=False , UpperCAmelCase_ : Dict=True , **UpperCAmelCase_ : Union[str, Any] , ): super().__init__( UpperCAmelCase_ , UpperCAmelCase_ , tokenizer_file=UpperCAmelCase_ , errors=UpperCAmelCase_ , bos_token=UpperCAmelCase_ , eos_token=UpperCAmelCase_ , sep_token=UpperCAmelCase_ , cls_token=UpperCAmelCase_ , unk_token=UpperCAmelCase_ , pad_token=UpperCAmelCase_ , mask_token=UpperCAmelCase_ , add_prefix_space=UpperCAmelCase_ , trim_offsets=UpperCAmelCase_ , **UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Any = json.loads(self.backend_tokenizer.pre_tokenizer.__getstate__() ) if pre_tok_state.get("add_prefix_space" , UpperCAmelCase_ ) != add_prefix_space: SCREAMING_SNAKE_CASE : Union[str, Any] = getattr(UpperCAmelCase_ , pre_tok_state.pop("type" ) ) SCREAMING_SNAKE_CASE : Union[str, Any] = add_prefix_space SCREAMING_SNAKE_CASE : Dict = pre_tok_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = add_prefix_space # the pre_tokenizer is already updated in the GPT2TokenizerFast `__init__` SCREAMING_SNAKE_CASE : int = "post_processor" SCREAMING_SNAKE_CASE : List[str] = getattr(self.backend_tokenizer , UpperCAmelCase_ , UpperCAmelCase_ ) if tokenizer_component_instance: SCREAMING_SNAKE_CASE : Tuple = json.loads(tokenizer_component_instance.__getstate__() ) # The lists 'sep' and 'cls' must be cased in tuples for the object `post_processor_class` if "sep" in state: SCREAMING_SNAKE_CASE : Optional[Any] = tuple(state["sep"] ) if "cls" in state: SCREAMING_SNAKE_CASE : str = tuple(state["cls"] ) SCREAMING_SNAKE_CASE : List[str] = False if state.get("add_prefix_space" , UpperCAmelCase_ ) != add_prefix_space: SCREAMING_SNAKE_CASE : Union[str, Any] = add_prefix_space SCREAMING_SNAKE_CASE : Any = True if state.get("trim_offsets" , UpperCAmelCase_ ) != trim_offsets: SCREAMING_SNAKE_CASE : Any = trim_offsets SCREAMING_SNAKE_CASE : Optional[int] = True if changes_to_apply: SCREAMING_SNAKE_CASE : List[str] = getattr(UpperCAmelCase_ , state.pop("type" ) ) SCREAMING_SNAKE_CASE : Any = component_class(**UpperCAmelCase_ ) setattr(self.backend_tokenizer , UpperCAmelCase_ , UpperCAmelCase_ ) @property # Copied from transformers.models.bart.tokenization_bart_fast.BartTokenizerFast.mask_token with BART->LED def _A ( self : Tuple ): if self._mask_token is None: if self.verbose: logger.error("Using mask_token, but it is not set yet." ) return None return str(self._mask_token ) @mask_token.setter def _A ( self : int , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : Optional[int] = AddedToken(UpperCAmelCase_ , lstrip=UpperCAmelCase_ , rstrip=UpperCAmelCase_ ) if isinstance(UpperCAmelCase_ , UpperCAmelCase_ ) else value SCREAMING_SNAKE_CASE : Tuple = value def _A ( self : Optional[Any] , *UpperCAmelCase_ : int , **UpperCAmelCase_ : List[Any] ): SCREAMING_SNAKE_CASE : Union[str, Any] = kwargs.get("is_split_into_words" , UpperCAmelCase_ ) if is_split_into_words and not self.add_prefix_space: raise ValueError( f'''You need to instantiate {self.__class__.__name__} with add_prefix_space=True ''' "to use it with pretokenized inputs." ) return super()._batch_encode_plus(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , *UpperCAmelCase_ : Any , **UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : Tuple = kwargs.get("is_split_into_words" , UpperCAmelCase_ ) if is_split_into_words and not self.add_prefix_space: raise ValueError( f'''You need to instantiate {self.__class__.__name__} with add_prefix_space=True ''' "to use it with pretokenized inputs." ) return super()._encode_plus(*UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : str , UpperCAmelCase_ : Optional[str] = None ): SCREAMING_SNAKE_CASE : str = self._tokenizer.model.save(UpperCAmelCase_ , name=UpperCAmelCase_ ) return tuple(UpperCAmelCase_ ) def _A ( self : Dict , UpperCAmelCase_ : int , UpperCAmelCase_ : Union[str, Any]=None ): SCREAMING_SNAKE_CASE : Union[str, Any] = [self.bos_token_id] + token_ids_a + [self.eos_token_id] if token_ids_a is None: return output return output + [self.eos_token_id] + token_ids_a + [self.eos_token_id] def _A ( self : Optional[int] , UpperCAmelCase_ : List[int] , UpperCAmelCase_ : Optional[List[int]] = None ): SCREAMING_SNAKE_CASE : Dict = [self.sep_token_id] SCREAMING_SNAKE_CASE : Dict = [self.cls_token_id] if token_ids_a is None: return len(cls + token_ids_a + sep ) * [0] return len(cls + token_ids_a + sep + sep + token_ids_a + sep ) * [0] def _A ( self : str , UpperCAmelCase_ : Union[Dict[str, EncodedInput], BatchEncoding] , UpperCAmelCase_ : Optional[int] = None , UpperCAmelCase_ : PaddingStrategy = PaddingStrategy.DO_NOT_PAD , UpperCAmelCase_ : Optional[int] = None , UpperCAmelCase_ : Optional[bool] = None , ): SCREAMING_SNAKE_CASE : List[Any] = super()._pad( encoded_inputs=UpperCAmelCase_ , max_length=UpperCAmelCase_ , padding_strategy=UpperCAmelCase_ , pad_to_multiple_of=UpperCAmelCase_ , return_attention_mask=UpperCAmelCase_ , ) # Load from model defaults if return_attention_mask is None: SCREAMING_SNAKE_CASE : int = "attention_mask" in self.model_input_names if return_attention_mask and "global_attention_mask" in encoded_inputs: SCREAMING_SNAKE_CASE : int = encoded_inputs[self.model_input_names[0]] # `global_attention_mask` need to have the same length as other (sequential) inputs. SCREAMING_SNAKE_CASE : Optional[int] = len(encoded_inputs["global_attention_mask"] ) != len(UpperCAmelCase_ ) if needs_to_be_padded: SCREAMING_SNAKE_CASE : Optional[Any] = len(UpperCAmelCase_ ) - len(encoded_inputs["global_attention_mask"] ) if self.padding_side == "right": # Use `-1` since `0` in `global_attention_mask` means `local attention` instead of `not to attend` SCREAMING_SNAKE_CASE : Any = ( encoded_inputs["global_attention_mask"] + [-1] * difference ) elif self.padding_side == "left": SCREAMING_SNAKE_CASE : str = [-1] * difference + encoded_inputs[ "global_attention_mask" ] else: raise ValueError("Invalid padding strategy:" + str(self.padding_side ) ) return encoded_inputs
62
# This model implementation is heavily inspired by https://github.com/haofanwang/ControlNet-for-Diffusers/ import gc import random import tempfile import unittest import numpy as np import torch from PIL import Image from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, ControlNetModel, DDIMScheduler, StableDiffusionControlNetImgaImgPipeline, UNetaDConditionModel, ) from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_controlnet import MultiControlNetModel from diffusers.utils import floats_tensor, load_image, load_numpy, randn_tensor, slow, torch_device from diffusers.utils.import_utils import is_xformers_available from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..pipeline_params import ( IMAGE_TO_IMAGE_IMAGE_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_PARAMS, ) from ..test_pipelines_common import ( PipelineKarrasSchedulerTesterMixin, PipelineLatentTesterMixin, PipelineTesterMixin, ) enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : List[Any] = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[Any] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Tuple = IMAGE_TO_IMAGE_IMAGE_PARAMS.union({'''control_image'''} ) UpperCamelCase_ : Any = IMAGE_TO_IMAGE_IMAGE_PARAMS def _A ( self : List[str] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : str = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : str , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Any = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Optional[Any] = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = 2 SCREAMING_SNAKE_CASE : Union[str, Any] = randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ) SCREAMING_SNAKE_CASE : Tuple = floats_tensor(control_image.shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : str = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : List[str] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : int ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : Union[str, Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = StableDiffusionControlNetImgaImgPipeline UpperCamelCase_ : str = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {'''height''', '''width'''} UpperCamelCase_ : Optional[int] = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS UpperCamelCase_ : Dict = frozenset([] ) # TO_DO: add image_params once refactored VaeImageProcessor.preprocess def _A ( self : Optional[Any] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = UNetaDConditionModel( block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , up_block_types=("CrossAttnUpBlock2D", "UpBlock2D") , cross_attention_dim=32 , ) torch.manual_seed(0 ) def init_weights(UpperCAmelCase_ : List[Any] ): if isinstance(UpperCAmelCase_ , torch.nn.Convad ): torch.nn.init.normal(m.weight ) m.bias.data.fill_(1.0 ) SCREAMING_SNAKE_CASE : List[str] = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = ControlNetModel( block_out_channels=(32, 64) , layers_per_block=2 , in_channels=4 , down_block_types=("DownBlock2D", "CrossAttnDownBlock2D") , cross_attention_dim=32 , conditioning_embedding_out_channels=(16, 32) , ) controlneta.controlnet_down_blocks.apply(UpperCAmelCase_ ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Any = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Dict = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , ) SCREAMING_SNAKE_CASE : Any = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : Tuple = MultiControlNetModel([controlneta, controlneta] ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "controlnet": controlnet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, "safety_checker": None, "feature_extractor": None, } return components def _A ( self : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Any=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : Dict = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = 2 SCREAMING_SNAKE_CASE : Tuple = [ randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), randn_tensor( (1, 3, 32 * controlnet_embedder_scale_factor, 32 * controlnet_embedder_scale_factor) , generator=UpperCAmelCase_ , device=torch.device(UpperCAmelCase_ ) , ), ] SCREAMING_SNAKE_CASE : Optional[int] = floats_tensor(control_image[0].shape , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = image.cpu().permute(0 , 2 , 3 , 1 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = Image.fromarray(np.uinta(UpperCAmelCase_ ) ).convert("RGB" ).resize((64, 64) ) SCREAMING_SNAKE_CASE : Optional[Any] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "numpy", "image": image, "control_image": control_image, } return inputs def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : Any = self.get_dummy_components() SCREAMING_SNAKE_CASE : str = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = 10.0 SCREAMING_SNAKE_CASE : Any = 4 SCREAMING_SNAKE_CASE : Optional[int] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : Optional[int] = pipe(**UpperCAmelCase_ )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = steps SCREAMING_SNAKE_CASE : Any = scale SCREAMING_SNAKE_CASE : List[str] = pipe(**UpperCAmelCase_ , control_guidance_start=0.1 , control_guidance_end=0.2 )[0] SCREAMING_SNAKE_CASE : Union[str, Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = steps SCREAMING_SNAKE_CASE : int = scale SCREAMING_SNAKE_CASE : List[Any] = pipe(**UpperCAmelCase_ , control_guidance_start=[0.1, 0.3] , control_guidance_end=[0.2, 0.7] )[0] SCREAMING_SNAKE_CASE : Optional[Any] = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = steps SCREAMING_SNAKE_CASE : Dict = scale SCREAMING_SNAKE_CASE : Dict = pipe(**UpperCAmelCase_ , control_guidance_start=0.4 , control_guidance_end=[0.5, 0.8] )[0] # make sure that all outputs are different assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 assert np.sum(np.abs(output_a - output_a ) ) > 1E-3 def _A ( self : Union[str, Any] ): return self._test_attention_slicing_forward_pass(expected_max_diff=2E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : str ): self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=2E-3 ) def _A ( self : List[Any] ): self._test_inference_batch_single_identical(expected_max_diff=2E-3 ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = self.get_dummy_components() SCREAMING_SNAKE_CASE : Dict = self.pipeline_class(**UpperCAmelCase_ ) pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) with tempfile.TemporaryDirectory() as tmpdir: try: # save_pretrained is not implemented for Multi-ControlNet pipe.save_pretrained(UpperCAmelCase_ ) except NotImplementedError: pass @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Optional[Any] ): super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : str = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny" ) SCREAMING_SNAKE_CASE : Union[str, Any] = StableDiffusionControlNetImgaImgPipeline.from_pretrained( "runwayml/stable-diffusion-v1-5" , safety_checker=UpperCAmelCase_ , controlnet=UpperCAmelCase_ ) pipe.enable_model_cpu_offload() pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.Generator(device="cpu" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : str = "evil space-punk bird" SCREAMING_SNAKE_CASE : Optional[Any] = load_image( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/bird_canny.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : Optional[int] = load_image( "https://huggingface.co/lllyasviel/sd-controlnet-canny/resolve/main/images/bird.png" ).resize((512, 512) ) SCREAMING_SNAKE_CASE : str = pipe( UpperCAmelCase_ , UpperCAmelCase_ , control_image=UpperCAmelCase_ , generator=UpperCAmelCase_ , output_type="np" , num_inference_steps=50 , strength=0.6 , ) SCREAMING_SNAKE_CASE : int = output.images[0] assert image.shape == (512, 512, 3) SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/img2img.npy" ) assert np.abs(expected_image - image ).max() < 9E-2
62
1
import sys from typing import Tuple import numpy as np import torch from PIL import Image from torch import nn from transformers.image_utils import PILImageResampling from utils import img_tensorize class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[Any]=sys.maxsize ): SCREAMING_SNAKE_CASE : str = "bilinear" SCREAMING_SNAKE_CASE : str = max_size SCREAMING_SNAKE_CASE : int = short_edge_length def __call__( self : Dict , UpperCAmelCase_ : Optional[int] ): SCREAMING_SNAKE_CASE : Tuple = [] for img in imgs: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = img.shape[:2] # later: provide list and randomly choose index for resize SCREAMING_SNAKE_CASE : Optional[int] = np.random.randint(self.short_edge_length[0] , self.short_edge_length[1] + 1 ) if size == 0: return img SCREAMING_SNAKE_CASE : Any = size * 1.0 / min(UpperCAmelCase_ , UpperCAmelCase_ ) if h < w: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = size, scale * w else: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = scale * h, size if max(UpperCAmelCase_ , UpperCAmelCase_ ) > self.max_size: SCREAMING_SNAKE_CASE : Union[str, Any] = self.max_size * 1.0 / max(UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = newh * scale SCREAMING_SNAKE_CASE : List[str] = neww * scale SCREAMING_SNAKE_CASE : List[str] = int(neww + 0.5 ) SCREAMING_SNAKE_CASE : str = int(newh + 0.5 ) if img.dtype == np.uinta: SCREAMING_SNAKE_CASE : Optional[Any] = Image.fromarray(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = pil_image.resize((neww, newh) , PILImageResampling.BILINEAR ) SCREAMING_SNAKE_CASE : Union[str, Any] = np.asarray(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : int = img.permute(2 , 0 , 1 ).unsqueeze(0 ) # 3, 0, 1) # hw(c) -> nchw SCREAMING_SNAKE_CASE : Optional[int] = nn.functional.interpolate( UpperCAmelCase_ , (newh, neww) , mode=self.interp_method , align_corners=UpperCAmelCase_ ).squeeze(0 ) img_augs.append(UpperCAmelCase_ ) return img_augs class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Optional[int] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = ResizeShortestEdge([cfg.INPUT.MIN_SIZE_TEST, cfg.INPUT.MIN_SIZE_TEST] , cfg.INPUT.MAX_SIZE_TEST ) SCREAMING_SNAKE_CASE : Optional[Any] = cfg.INPUT.FORMAT SCREAMING_SNAKE_CASE : List[str] = cfg.SIZE_DIVISIBILITY SCREAMING_SNAKE_CASE : List[str] = cfg.PAD_VALUE SCREAMING_SNAKE_CASE : List[str] = cfg.INPUT.MAX_SIZE_TEST SCREAMING_SNAKE_CASE : int = cfg.MODEL.DEVICE SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(cfg.MODEL.PIXEL_STD ).to(self.device ).view(len(cfg.MODEL.PIXEL_STD ) , 1 , 1 ) SCREAMING_SNAKE_CASE : Tuple = torch.tensor(cfg.MODEL.PIXEL_MEAN ).to(self.device ).view(len(cfg.MODEL.PIXEL_STD ) , 1 , 1 ) SCREAMING_SNAKE_CASE : List[Any] = lambda UpperCAmelCase_ : (x - self.pixel_mean) / self.pixel_std def _A ( self : List[str] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : Tuple = tuple(max(UpperCAmelCase_ ) for s in zip(*[img.shape for img in images] ) ) SCREAMING_SNAKE_CASE : Dict = [im.shape[-2:] for im in images] SCREAMING_SNAKE_CASE : str = [ nn.functional.pad( UpperCAmelCase_ , [0, max_size[-1] - size[1], 0, max_size[-2] - size[0]] , value=self.pad_value , ) for size, im in zip(UpperCAmelCase_ , UpperCAmelCase_ ) ] return torch.stack(UpperCAmelCase_ ), torch.tensor(UpperCAmelCase_ ) def __call__( self : Any , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Union[str, Any]=False ): with torch.no_grad(): if not isinstance(UpperCAmelCase_ , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Optional[int] = [images] if single_image: assert len(UpperCAmelCase_ ) == 1 for i in range(len(UpperCAmelCase_ ) ): if isinstance(images[i] , torch.Tensor ): images.insert(UpperCAmelCase_ , images.pop(UpperCAmelCase_ ).to(self.device ).float() ) elif not isinstance(images[i] , torch.Tensor ): images.insert( UpperCAmelCase_ , torch.as_tensor(img_tensorize(images.pop(UpperCAmelCase_ ) , input_format=self.input_format ) ) .to(self.device ) .float() , ) # resize smallest edge SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([im.shape[:2] for im in images] ) SCREAMING_SNAKE_CASE : Any = self.aug(UpperCAmelCase_ ) # transpose images and convert to torch tensors # images = [torch.as_tensor(i.astype("float32")).permute(2, 0, 1).to(self.device) for i in images] # now normalize before pad to avoid useless arithmetic SCREAMING_SNAKE_CASE : Union[str, Any] = [self.normalizer(UpperCAmelCase_ ) for x in images] # now pad them to do the following operations SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.pad(UpperCAmelCase_ ) # Normalize if self.size_divisibility > 0: raise NotImplementedError() # pad SCREAMING_SNAKE_CASE : Union[str, Any] = torch.true_divide(UpperCAmelCase_ , UpperCAmelCase_ ) if single_image: return images[0], sizes[0], scales_yx[0] else: return images, sizes, scales_yx def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" boxes[:, 0::2] *= scale_yx[:, 1] boxes[:, 1::2] *= scale_yx[:, 0] return boxes def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" assert torch.isfinite(lowercase ).all(), "Box tensor contains infinite or NaN!" SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = box_size tensor[:, 0].clamp_(min=0 , max=lowercase ) tensor[:, 1].clamp_(min=0 , max=lowercase ) tensor[:, 2].clamp_(min=0 , max=lowercase ) tensor[:, 3].clamp_(min=0 , max=lowercase )
62
import argparse import json from pathlib import Path import requests import torch from huggingface_hub import hf_hub_download from PIL import Image from transformers import ( MobileViTConfig, MobileViTForImageClassification, MobileViTForSemanticSegmentation, MobileViTImageProcessor, ) from transformers.utils import logging logging.set_verbosity_info() snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : int = MobileViTConfig() # size of the architecture if "mobilevit_s" in mobilevit_name: SCREAMING_SNAKE_CASE : List[Any] = [144, 192, 240] SCREAMING_SNAKE_CASE : Tuple = [16, 32, 64, 96, 128, 160, 640] elif "mobilevit_xs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [96, 120, 144] SCREAMING_SNAKE_CASE : Dict = [16, 32, 48, 64, 80, 96, 384] elif "mobilevit_xxs" in mobilevit_name: SCREAMING_SNAKE_CASE : List[str] = [64, 80, 96] SCREAMING_SNAKE_CASE : List[str] = [16, 16, 24, 48, 64, 80, 320] SCREAMING_SNAKE_CASE : int = 0.05 SCREAMING_SNAKE_CASE : int = 2.0 if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : str = 512 SCREAMING_SNAKE_CASE : List[str] = 16 SCREAMING_SNAKE_CASE : Union[str, Any] = 21 SCREAMING_SNAKE_CASE : Dict = "pascal-voc-id2label.json" else: SCREAMING_SNAKE_CASE : Optional[Any] = 1000 SCREAMING_SNAKE_CASE : Optional[Any] = "imagenet-1k-id2label.json" SCREAMING_SNAKE_CASE : Any = "huggingface/label-files" SCREAMING_SNAKE_CASE : Tuple = json.load(open(hf_hub_download(lowercase , lowercase , repo_type="dataset" ) , "r" ) ) SCREAMING_SNAKE_CASE : List[str] = {int(lowercase ): v for k, v in idalabel.items()} SCREAMING_SNAKE_CASE : Optional[Any] = idalabel SCREAMING_SNAKE_CASE : str = {v: k for k, v in idalabel.items()} return config def lowerCamelCase__ ( lowercase , lowercase=False ): """simple docstring""" for i in range(1 , 6 ): if F'''layer_{i}.''' in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(F'''layer_{i}.''' , F'''encoder.layer.{i - 1}.''' ) if "conv_1." in name: SCREAMING_SNAKE_CASE : Dict = name.replace("conv_1." , "conv_stem." ) if ".block." in name: SCREAMING_SNAKE_CASE : List[str] = name.replace(".block." , "." ) if "exp_1x1" in name: SCREAMING_SNAKE_CASE : str = name.replace("exp_1x1" , "expand_1x1" ) if "red_1x1" in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("red_1x1" , "reduce_1x1" ) if ".local_rep.conv_3x3." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".local_rep.conv_3x3." , ".conv_kxk." ) if ".local_rep.conv_1x1." in name: SCREAMING_SNAKE_CASE : int = name.replace(".local_rep.conv_1x1." , ".conv_1x1." ) if ".norm." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".norm." , ".normalization." ) if ".conv." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(".conv." , ".convolution." ) if ".conv_proj." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".conv_proj." , ".conv_projection." ) for i in range(0 , 2 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.layer.{j}.''' ) for i in range(2 , 6 ): for j in range(0 , 4 ): if F'''.{i}.{j}.''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.{i}.{j}.''' , F'''.{i}.''' ) if "expand_1x1" in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace("expand_1x1" , "downsampling_layer.expand_1x1" ) if "conv_3x3" in name: SCREAMING_SNAKE_CASE : str = name.replace("conv_3x3" , "downsampling_layer.conv_3x3" ) if "reduce_1x1" in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("reduce_1x1" , "downsampling_layer.reduce_1x1" ) for i in range(2 , 5 ): if F'''.global_rep.{i}.weight''' in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace(F'''.global_rep.{i}.weight''' , ".layernorm.weight" ) if F'''.global_rep.{i}.bias''' in name: SCREAMING_SNAKE_CASE : str = name.replace(F'''.global_rep.{i}.bias''' , ".layernorm.bias" ) if ".global_rep." in name: SCREAMING_SNAKE_CASE : Dict = name.replace(".global_rep." , ".transformer." ) if ".pre_norm_mha.0." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_mha.0." , ".layernorm_before." ) if ".pre_norm_mha.1.out_proj." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_mha.1.out_proj." , ".attention.output.dense." ) if ".pre_norm_ffn.0." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace(".pre_norm_ffn.0." , ".layernorm_after." ) if ".pre_norm_ffn.1." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".pre_norm_ffn.1." , ".intermediate.dense." ) if ".pre_norm_ffn.4." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".pre_norm_ffn.4." , ".output.dense." ) if ".transformer." in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".transformer." , ".transformer.layer." ) if ".aspp_layer." in name: SCREAMING_SNAKE_CASE : int = name.replace(".aspp_layer." , "." ) if ".aspp_pool." in name: SCREAMING_SNAKE_CASE : Tuple = name.replace(".aspp_pool." , "." ) if "seg_head." in name: SCREAMING_SNAKE_CASE : Optional[int] = name.replace("seg_head." , "segmentation_head." ) if "segmentation_head.classifier.classifier." in name: SCREAMING_SNAKE_CASE : Optional[Any] = name.replace("segmentation_head.classifier.classifier." , "segmentation_head.classifier." ) if "classifier.fc." in name: SCREAMING_SNAKE_CASE : List[Any] = name.replace("classifier.fc." , "classifier." ) elif (not base_model) and ("segmentation_head." not in name): SCREAMING_SNAKE_CASE : List[Any] = "mobilevit." + name return name def lowerCamelCase__ ( lowercase , lowercase , lowercase=False ): """simple docstring""" if base_model: SCREAMING_SNAKE_CASE : Optional[int] = "" else: SCREAMING_SNAKE_CASE : Any = "mobilevit." for key in orig_state_dict.copy().keys(): SCREAMING_SNAKE_CASE : Union[str, Any] = orig_state_dict.pop(lowercase ) if key[:8] == "encoder.": SCREAMING_SNAKE_CASE : int = key[8:] if "qkv" in key: SCREAMING_SNAKE_CASE : Optional[int] = key.split("." ) SCREAMING_SNAKE_CASE : Any = int(key_split[0][6:] ) - 1 SCREAMING_SNAKE_CASE : List[Any] = int(key_split[3] ) SCREAMING_SNAKE_CASE : List[Any] = model.get_submodule(F'''{model_prefix}encoder.layer.{layer_num}''' ) SCREAMING_SNAKE_CASE : int = layer.transformer.layer[transformer_num].attention.attention.all_head_size SCREAMING_SNAKE_CASE : Union[str, Any] = ( F'''{model_prefix}encoder.layer.{layer_num}.transformer.layer.{transformer_num}.attention.attention.''' ) if "weight" in key: SCREAMING_SNAKE_CASE : Optional[int] = val[:dim, :] SCREAMING_SNAKE_CASE : Tuple = val[dim : dim * 2, :] SCREAMING_SNAKE_CASE : Dict = val[-dim:, :] else: SCREAMING_SNAKE_CASE : str = val[:dim] SCREAMING_SNAKE_CASE : Union[str, Any] = val[dim : dim * 2] SCREAMING_SNAKE_CASE : Union[str, Any] = val[-dim:] else: SCREAMING_SNAKE_CASE : List[Any] = val return orig_state_dict def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = "http://images.cocodataset.org/val2017/000000039769.jpg" SCREAMING_SNAKE_CASE : Union[str, Any] = Image.open(requests.get(lowercase , stream=lowercase ).raw ) return im @torch.no_grad() def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase=False ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = get_mobilevit_config(lowercase ) # load original state_dict SCREAMING_SNAKE_CASE : Union[str, Any] = torch.load(lowercase , map_location="cpu" ) # load 🤗 model if mobilevit_name.startswith("deeplabv3_" ): SCREAMING_SNAKE_CASE : List[str] = MobileViTForSemanticSegmentation(lowercase ).eval() else: SCREAMING_SNAKE_CASE : str = MobileViTForImageClassification(lowercase ).eval() SCREAMING_SNAKE_CASE : Any = convert_state_dict(lowercase , lowercase ) model.load_state_dict(lowercase ) # Check outputs on an image, prepared by MobileViTImageProcessor SCREAMING_SNAKE_CASE : List[Any] = MobileViTImageProcessor(crop_size=config.image_size , size=config.image_size + 32 ) SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=prepare_img() , return_tensors="pt" ) SCREAMING_SNAKE_CASE : List[Any] = model(**lowercase ) SCREAMING_SNAKE_CASE : str = outputs.logits if mobilevit_name.startswith("deeplabv3_" ): assert logits.shape == (1, 21, 32, 32) if mobilevit_name == "deeplabv3_mobilevit_s": SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor( [ [[6.2065, 6.1292, 6.2070], [6.1079, 6.1254, 6.1747], [6.0042, 6.1071, 6.1034]], [[-6.9253, -6.8653, -7.0398], [-7.3218, -7.3983, -7.3670], [-7.1961, -7.2482, -7.1569]], [[-4.4723, -4.4348, -4.3769], [-5.3629, -5.4632, -5.4598], [-5.1587, -5.3402, -5.5059]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor( [ [[5.4449, 5.5733, 5.6314], [5.1815, 5.3930, 5.5963], [5.1656, 5.4333, 5.4853]], [[-9.4423, -9.7766, -9.6714], [-9.1581, -9.5720, -9.5519], [-9.1006, -9.6458, -9.5703]], [[-7.7721, -7.3716, -7.1583], [-8.4599, -8.0624, -7.7944], [-8.4172, -7.8366, -7.5025]], ] ) elif mobilevit_name == "deeplabv3_mobilevit_xxs": SCREAMING_SNAKE_CASE : List[Any] = torch.tensor( [ [[6.9811, 6.9743, 7.3123], [7.1777, 7.1931, 7.3938], [7.5633, 7.8050, 7.8901]], [[-10.5536, -10.2332, -10.2924], [-10.2336, -9.8624, -9.5964], [-10.8840, -10.8158, -10.6659]], [[-3.4938, -3.0631, -2.8620], [-3.4205, -2.8135, -2.6875], [-3.4179, -2.7945, -2.8750]], ] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3, :3, :3] , lowercase , atol=1E-4 ) else: assert logits.shape == (1, 1000) if mobilevit_name == "mobilevit_s": SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor([-0.9866, 0.2392, -1.1241] ) elif mobilevit_name == "mobilevit_xs": SCREAMING_SNAKE_CASE : Dict = torch.tensor([-2.4761, -0.9399, -1.9587] ) elif mobilevit_name == "mobilevit_xxs": SCREAMING_SNAKE_CASE : Tuple = torch.tensor([-1.9364, -1.2327, -0.4653] ) else: raise ValueError(F'''Unknown mobilevit_name: {mobilevit_name}''' ) assert torch.allclose(logits[0, :3] , lowercase , atol=1E-4 ) Path(lowercase ).mkdir(exist_ok=lowercase ) print(F'''Saving model {mobilevit_name} to {pytorch_dump_folder_path}''' ) model.save_pretrained(lowercase ) print(F'''Saving image processor to {pytorch_dump_folder_path}''' ) image_processor.save_pretrained(lowercase ) if push_to_hub: SCREAMING_SNAKE_CASE : List[str] = { "mobilevit_s": "mobilevit-small", "mobilevit_xs": "mobilevit-x-small", "mobilevit_xxs": "mobilevit-xx-small", "deeplabv3_mobilevit_s": "deeplabv3-mobilevit-small", "deeplabv3_mobilevit_xs": "deeplabv3-mobilevit-x-small", "deeplabv3_mobilevit_xxs": "deeplabv3-mobilevit-xx-small", } print("Pushing to the hub..." ) SCREAMING_SNAKE_CASE : int = model_mapping[mobilevit_name] image_processor.push_to_hub(lowercase , organization="apple" ) model.push_to_hub(lowercase , organization="apple" ) if __name__ == "__main__": snake_case = argparse.ArgumentParser() # Required parameters parser.add_argument( """--mobilevit_name""", default="""mobilevit_s""", type=str, help=( """Name of the MobileViT model you'd like to convert. Should be one of 'mobilevit_s', 'mobilevit_xs',""" """ 'mobilevit_xxs', 'deeplabv3_mobilevit_s', 'deeplabv3_mobilevit_xs', 'deeplabv3_mobilevit_xxs'.""" ), ) parser.add_argument( """--checkpoint_path""", required=True, type=str, help="""Path to the original state dict (.pt file).""" ) parser.add_argument( """--pytorch_dump_folder_path""", required=True, type=str, help="""Path to the output PyTorch model directory.""" ) parser.add_argument( """--push_to_hub""", action="""store_true""", help="""Whether or not to push the converted model to the 🤗 hub.""" ) snake_case = parser.parse_args() convert_movilevit_checkpoint( args.mobilevit_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub )
62
1
import ast import os import re import shutil import tempfile import unittest from unittest import mock import torch from accelerate.test_utils.examples import compare_against_test from accelerate.test_utils.testing import TempDirTestCase, require_trackers, run_command, slow from accelerate.utils import write_basic_config # DataLoaders built from `test_samples/MRPC` for quick testing # Should mock `{script_name}.get_dataloaders` via: # @mock.patch("{script_name}.get_dataloaders", mocked_dataloaders) snake_case = [ """cross_validation.py""", """gradient_accumulation.py""", """local_sgd.py""", """multi_process_metrics.py""", """memory.py""", """automatic_gradient_accumulation.py""", """fsdp_with_peak_mem_tracking.py""", """deepspeed_with_config_support.py""", """megatron_lm_gpt_pretraining.py""", ] class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Dict , UpperCAmelCase_ : str , UpperCAmelCase_ : bool , UpperCAmelCase_ : str = None , UpperCAmelCase_ : list = None ): SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : Any = os.path.abspath(os.path.join("examples" , "by_feature" ) ) SCREAMING_SNAKE_CASE : Optional[Any] = os.path.abspath("examples" ) for item in os.listdir(UpperCAmelCase_ ): if item not in EXCLUDE_EXAMPLES: SCREAMING_SNAKE_CASE : str = os.path.join(UpperCAmelCase_ , UpperCAmelCase_ ) if os.path.isfile(UpperCAmelCase_ ) and ".py" in item_path: with self.subTest( tested_script=UpperCAmelCase_ , feature_script=UpperCAmelCase_ , tested_section="main()" if parser_only else "training_function()" , ): SCREAMING_SNAKE_CASE : Dict = compare_against_test( os.path.join(UpperCAmelCase_ , UpperCAmelCase_ ) , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = "\n".join(UpperCAmelCase_ ) if special_strings is not None: for string in special_strings: SCREAMING_SNAKE_CASE : Dict = diff.replace(UpperCAmelCase_ , "" ) self.assertEqual(UpperCAmelCase_ , "" ) def _A ( self : List[Any] ): self.one_complete_example("complete_nlp_example.py" , UpperCAmelCase_ ) self.one_complete_example("complete_nlp_example.py" , UpperCAmelCase_ ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Any = os.path.abspath(os.path.join("examples" , "cv_example.py" ) ) SCREAMING_SNAKE_CASE : List[Any] = [ " " * 16 + "{\n\n", " " * 20 + "\"accuracy\": eval_metric[\"accuracy\"],\n\n", " " * 20 + "\"f1\": eval_metric[\"f1\"],\n\n", " " * 20 + "\"train_loss\": total_loss.item() / len(train_dataloader),\n\n", " " * 20 + "\"epoch\": epoch,\n\n", " " * 16 + "},\n\n", " " * 16 + "step=epoch,\n", " " * 12, " " * 8 + "for step, batch in enumerate(active_dataloader):\n", ] self.one_complete_example("complete_cv_example.py" , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) self.one_complete_example("complete_cv_example.py" , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) @mock.patch.dict(os.environ , {'''TESTING_MOCKED_DATALOADERS''': '''1'''} ) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Optional[int] = False @classmethod def _A ( cls : Union[str, Any] ): super().setUpClass() SCREAMING_SNAKE_CASE : Tuple = tempfile.mkdtemp() SCREAMING_SNAKE_CASE : List[Any] = os.path.join(cls._tmpdir , "default_config.yml" ) write_basic_config(save_location=cls.configPath ) SCREAMING_SNAKE_CASE : Optional[Any] = ["accelerate", "launch", "--config_file", cls.configPath] @classmethod def _A ( cls : Union[str, Any] ): super().tearDownClass() shutil.rmtree(cls._tmpdir ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = f''' examples/by_feature/checkpointing.py --checkpointing_steps epoch --output_dir {self.tmpdir} '''.split() run_command(self._launch_args + testargs ) self.assertTrue(os.path.exists(os.path.join(self.tmpdir , "epoch_0" ) ) ) def _A ( self : str ): SCREAMING_SNAKE_CASE : List[Any] = f''' examples/by_feature/checkpointing.py --checkpointing_steps 1 --output_dir {self.tmpdir} '''.split() SCREAMING_SNAKE_CASE : str = run_command(self._launch_args + testargs ) self.assertTrue(os.path.exists(os.path.join(self.tmpdir , "step_2" ) ) ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : str = f''' examples/by_feature/checkpointing.py --resume_from_checkpoint {os.path.join(self.tmpdir , 'epoch_0' )} '''.split() SCREAMING_SNAKE_CASE : Union[str, Any] = run_command(self._launch_args + testargs , return_stdout=UpperCAmelCase_ ) self.assertNotIn("epoch 0:" , UpperCAmelCase_ ) self.assertIn("epoch 1:" , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : str = f''' examples/by_feature/checkpointing.py --resume_from_checkpoint {os.path.join(self.tmpdir , 'step_2' )} '''.split() SCREAMING_SNAKE_CASE : Optional[int] = run_command(self._launch_args + testargs , return_stdout=UpperCAmelCase_ ) if torch.cuda.is_available(): SCREAMING_SNAKE_CASE : str = torch.cuda.device_count() else: SCREAMING_SNAKE_CASE : Dict = 1 if num_processes > 1: self.assertNotIn("epoch 0:" , UpperCAmelCase_ ) self.assertIn("epoch 1:" , UpperCAmelCase_ ) else: self.assertIn("epoch 0:" , UpperCAmelCase_ ) self.assertIn("epoch 1:" , UpperCAmelCase_ ) @slow def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "\n examples/by_feature/cross_validation.py\n --num_folds 2\n ".split() with mock.patch.dict(os.environ , {"TESTING_MOCKED_DATALOADERS": "0"} ): SCREAMING_SNAKE_CASE : int = run_command(self._launch_args + testargs , return_stdout=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = re.findall("({.+})" , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = [r for r in results if "accuracy" in r][-1] SCREAMING_SNAKE_CASE : List[Any] = ast.literal_eval(UpperCAmelCase_ ) self.assertGreaterEqual(results["accuracy"] , 0.75 ) def _A ( self : Dict ): SCREAMING_SNAKE_CASE : Optional[Any] = ["examples/by_feature/multi_process_metrics.py"] run_command(self._launch_args + testargs ) @require_trackers @mock.patch.dict(os.environ , {"WANDB_MODE": "offline"} ) def _A ( self : Union[str, Any] ): with tempfile.TemporaryDirectory() as tmpdir: SCREAMING_SNAKE_CASE : Any = f''' examples/by_feature/tracking.py --with_tracking --project_dir {tmpdir} '''.split() run_command(self._launch_args + testargs ) self.assertTrue(os.path.exists(os.path.join(UpperCAmelCase_ , "tracking" ) ) ) def _A ( self : str ): SCREAMING_SNAKE_CASE : List[str] = ["examples/by_feature/gradient_accumulation.py"] run_command(self._launch_args + testargs ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : str = ["examples/by_feature/local_sgd.py"] run_command(self._launch_args + testargs )
62
from collections import Counter import numpy as np from sklearn import datasets from sklearn.model_selection import train_test_split snake_case = datasets.load_iris() snake_case = np.array(data["""data"""]) snake_case = np.array(data["""target"""]) snake_case = data["""target_names"""] snake_case , snake_case , snake_case , snake_case = train_test_split(X, y) def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return np.linalg.norm(np.array(lowercase ) - np.array(lowercase ) ) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase=5 ): """simple docstring""" SCREAMING_SNAKE_CASE : List[str] = zip(lowercase , lowercase ) # List of distances of all points from the point to be classified SCREAMING_SNAKE_CASE : Optional[int] = [] for data_point in data: SCREAMING_SNAKE_CASE : Tuple = euclidean_distance(data_point[0] , lowercase ) distances.append((distance, data_point[1]) ) # Choosing 'k' points with the least distances. SCREAMING_SNAKE_CASE : List[Any] = [i[1] for i in sorted(lowercase )[:k]] # Most commonly occurring class among them # is the class into which the point is classified SCREAMING_SNAKE_CASE : List[Any] = Counter(lowercase ).most_common(1 )[0][0] return classes[result] if __name__ == "__main__": print(classifier(X_train, y_train, classes, [4.4, 3.1, 1.3, 1.4]))
62
1
import operator as op def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Any = [] SCREAMING_SNAKE_CASE : Any = lambda lowercase , lowercase : int(x / y ) # noqa: E731 integer division operation SCREAMING_SNAKE_CASE : str = { "^": op.pow, "*": op.mul, "/": div, "+": op.add, "-": op.sub, } # operators & their respective operation # print table header print("Symbol".center(8 ) , "Action".center(12 ) , "Stack" , sep=" | " ) print("-" * (30 + len(lowercase )) ) for x in post_fix: if x.isdigit(): # if x in digit stack.append(lowercase ) # append x to stack # output in tabular format print(x.rjust(8 ) , ("push(" + x + ")").ljust(12 ) , ",".join(lowercase ) , sep=" | " ) else: SCREAMING_SNAKE_CASE : Any = stack.pop() # pop stack # output in tabular format print("".rjust(8 ) , ("pop(" + b + ")").ljust(12 ) , ",".join(lowercase ) , sep=" | " ) SCREAMING_SNAKE_CASE : int = stack.pop() # pop stack # output in tabular format print("".rjust(8 ) , ("pop(" + a + ")").ljust(12 ) , ",".join(lowercase ) , sep=" | " ) stack.append( str(opr[x](int(lowercase ) , int(lowercase ) ) ) ) # evaluate the 2 values popped from stack & push result to stack # output in tabular format print( x.rjust(8 ) , ("push(" + a + x + b + ")").ljust(12 ) , ",".join(lowercase ) , sep=" | " , ) return int(stack[0] ) if __name__ == "__main__": snake_case = input("""\n\nEnter a Postfix Equation (space separated) = """).split(""" """) print("""\n\tResult = """, solve(Postfix))
62
from typing import Callable, Optional, Union from ...configuration_utils import PretrainedConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """microsoft/xprophetnet-large-wiki100-cased""": ( """https://huggingface.co/microsoft/xprophetnet-large-wiki100-cased/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Tuple = '''xlm-prophetnet''' UpperCamelCase_ : Tuple = ['''past_key_values'''] UpperCamelCase_ : int = { '''num_attention_heads''': '''num_encoder_attention_heads''', } def __init__( self : Dict , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[Union[str, Callable]] = "gelu" , UpperCAmelCase_ : Optional[int] = 3_0522 , UpperCAmelCase_ : Optional[int] = 1024 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[int] = 4096 , UpperCAmelCase_ : Optional[int] = 12 , UpperCAmelCase_ : Optional[int] = 16 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[float] = 0.1 , UpperCAmelCase_ : Optional[int] = 512 , UpperCAmelCase_ : Optional[float] = 0.02 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 2 , UpperCAmelCase_ : Optional[int] = 32 , UpperCAmelCase_ : Optional[int] = 128 , UpperCAmelCase_ : Optional[bool] = False , UpperCAmelCase_ : Optional[float] = 0.0 , UpperCAmelCase_ : Optional[bool] = True , UpperCAmelCase_ : Optional[int] = 0 , UpperCAmelCase_ : Optional[int] = 1 , UpperCAmelCase_ : Optional[int] = 2 , **UpperCAmelCase_ : Dict , ): SCREAMING_SNAKE_CASE : Any = vocab_size SCREAMING_SNAKE_CASE : str = hidden_size SCREAMING_SNAKE_CASE : Tuple = encoder_ffn_dim SCREAMING_SNAKE_CASE : List[Any] = num_encoder_layers SCREAMING_SNAKE_CASE : Any = num_encoder_attention_heads SCREAMING_SNAKE_CASE : Dict = decoder_ffn_dim SCREAMING_SNAKE_CASE : List[str] = num_decoder_layers SCREAMING_SNAKE_CASE : List[Any] = num_decoder_attention_heads SCREAMING_SNAKE_CASE : List[Any] = max_position_embeddings SCREAMING_SNAKE_CASE : Dict = init_std # Normal(0, this parameter) SCREAMING_SNAKE_CASE : Union[str, Any] = activation_function # parameters for xlmprophetnet SCREAMING_SNAKE_CASE : Dict = ngram SCREAMING_SNAKE_CASE : Any = num_buckets SCREAMING_SNAKE_CASE : str = relative_max_distance SCREAMING_SNAKE_CASE : str = disable_ngram_loss SCREAMING_SNAKE_CASE : Dict = eps # 3 Types of Dropout SCREAMING_SNAKE_CASE : Union[str, Any] = attention_dropout SCREAMING_SNAKE_CASE : Optional[int] = activation_dropout SCREAMING_SNAKE_CASE : List[Any] = dropout SCREAMING_SNAKE_CASE : int = use_cache super().__init__( pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , is_encoder_decoder=UpperCAmelCase_ , add_cross_attention=UpperCAmelCase_ , decoder_start_token_id=UpperCAmelCase_ , **UpperCAmelCase_ , ) @property def _A ( self : int ): return self.num_encoder_layers + self.num_decoder_layers @num_hidden_layers.setter def _A ( self : str , UpperCAmelCase_ : Optional[Any] ): raise NotImplementedError( "This model does not support the setting of `num_hidden_layers`. Please set `num_encoder_layers` and" " `num_decoder_layers`." )
62
1
import numpy as np from nltk.translate import meteor_score import datasets from datasets.config import importlib_metadata, version snake_case = version.parse(importlib_metadata.version("""nltk""")) if NLTK_VERSION >= version.Version("""3.6.4"""): from nltk import word_tokenize snake_case = """\ @inproceedings{banarjee2005, title = {{METEOR}: An Automatic Metric for {MT} Evaluation with Improved Correlation with Human Judgments}, author = {Banerjee, Satanjeev and Lavie, Alon}, booktitle = {Proceedings of the {ACL} Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization}, month = jun, year = {2005}, address = {Ann Arbor, Michigan}, publisher = {Association for Computational Linguistics}, url = {https://www.aclweb.org/anthology/W05-0909}, pages = {65--72}, } """ snake_case = """\ METEOR, an automatic metric for machine translation evaluation that is based on a generalized concept of unigram matching between the machine-produced translation and human-produced reference translations. Unigrams can be matched based on their surface forms, stemmed forms, and meanings; furthermore, METEOR can be easily extended to include more advanced matching strategies. Once all generalized unigram matches between the two strings have been found, METEOR computes a score for this matching using a combination of unigram-precision, unigram-recall, and a measure of fragmentation that is designed to directly capture how well-ordered the matched words in the machine translation are in relation to the reference. METEOR gets an R correlation value of 0.347 with human evaluation on the Arabic data and 0.331 on the Chinese data. This is shown to be an improvement on using simply unigram-precision, unigram-recall and their harmonic F1 combination. """ snake_case = """ Computes METEOR score of translated segments against one or more references. Args: predictions: list of predictions to score. Each prediction should be a string with tokens separated by spaces. references: list of reference for each prediction. Each reference should be a string with tokens separated by spaces. alpha: Parameter for controlling relative weights of precision and recall. default: 0.9 beta: Parameter for controlling shape of penalty as a function of fragmentation. default: 3 gamma: Relative weight assigned to fragmentation penalty. default: 0.5 Returns: 'meteor': meteor score. Examples: >>> meteor = datasets.load_metric('meteor') >>> predictions = [\"It is a guide to action which ensures that the military always obeys the commands of the party\"] >>> references = [\"It is a guide to action that ensures that the military will forever heed Party commands\"] >>> results = meteor.compute(predictions=predictions, references=references) >>> print(round(results[\"meteor\"], 4)) 0.6944 """ @datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION ) class SCREAMING_SNAKE_CASE ( datasets.Metric ): '''simple docstring''' def _A ( self : str ): return datasets.MetricInfo( description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features( { "predictions": datasets.Value("string" , id="sequence" ), "references": datasets.Value("string" , id="sequence" ), } ) , codebase_urls=["https://github.com/nltk/nltk/blob/develop/nltk/translate/meteor_score.py"] , reference_urls=[ "https://www.nltk.org/api/nltk.translate.html#module-nltk.translate.meteor_score", "https://en.wikipedia.org/wiki/METEOR", ] , ) def _A ( self : Any , UpperCAmelCase_ : Dict ): import nltk nltk.download("wordnet" ) if NLTK_VERSION >= version.Version("3.6.5" ): nltk.download("punkt" ) if NLTK_VERSION >= version.Version("3.6.6" ): nltk.download("omw-1.4" ) def _A ( self : List[str] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int , UpperCAmelCase_ : Union[str, Any]=0.9 , UpperCAmelCase_ : Optional[int]=3 , UpperCAmelCase_ : str=0.5 ): if NLTK_VERSION >= version.Version("3.6.5" ): SCREAMING_SNAKE_CASE : Dict = [ meteor_score.single_meteor_score( word_tokenize(UpperCAmelCase_ ) , word_tokenize(UpperCAmelCase_ ) , alpha=UpperCAmelCase_ , beta=UpperCAmelCase_ , gamma=UpperCAmelCase_ ) for ref, pred in zip(UpperCAmelCase_ , UpperCAmelCase_ ) ] else: SCREAMING_SNAKE_CASE : List[Any] = [ meteor_score.single_meteor_score(UpperCAmelCase_ , UpperCAmelCase_ , alpha=UpperCAmelCase_ , beta=UpperCAmelCase_ , gamma=UpperCAmelCase_ ) for ref, pred in zip(UpperCAmelCase_ , UpperCAmelCase_ ) ] return {"meteor": np.mean(UpperCAmelCase_ )}
62
import gc import random import unittest import numpy as np import torch from diffusers import DDIMScheduler, KandinskyVaaPipeline, KandinskyVaaPriorPipeline, UNetaDConditionModel, VQModel from diffusers.utils import floats_tensor, load_numpy, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference enable_full_determinism() class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : List[str] = KandinskyVaaPipeline UpperCamelCase_ : List[Any] = [ '''image_embeds''', '''negative_image_embeds''', ] UpperCamelCase_ : Tuple = ['''image_embeds''', '''negative_image_embeds'''] UpperCamelCase_ : Any = [ '''generator''', '''height''', '''width''', '''latents''', '''guidance_scale''', '''num_inference_steps''', '''return_dict''', '''guidance_scale''', '''num_images_per_prompt''', '''output_type''', '''return_dict''', ] UpperCamelCase_ : List[str] = False @property def _A ( self : List[Any] ): return 32 @property def _A ( self : List[Any] ): return 32 @property def _A ( self : Any ): return self.time_input_dim @property def _A ( self : Union[str, Any] ): return self.time_input_dim * 4 @property def _A ( self : Tuple ): return 100 @property def _A ( self : Optional[int] ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : int = { "in_channels": 4, # Out channels is double in channels because predicts mean and variance "out_channels": 8, "addition_embed_type": "image", "down_block_types": ("ResnetDownsampleBlock2D", "SimpleCrossAttnDownBlock2D"), "up_block_types": ("SimpleCrossAttnUpBlock2D", "ResnetUpsampleBlock2D"), "mid_block_type": "UNetMidBlock2DSimpleCrossAttn", "block_out_channels": (self.block_out_channels_a, self.block_out_channels_a * 2), "layers_per_block": 1, "encoder_hid_dim": self.text_embedder_hidden_size, "encoder_hid_dim_type": "image_proj", "cross_attention_dim": self.cross_attention_dim, "attention_head_dim": 4, "resnet_time_scale_shift": "scale_shift", "class_embed_type": None, } SCREAMING_SNAKE_CASE : str = UNetaDConditionModel(**UpperCAmelCase_ ) return model @property def _A ( self : int ): return { "block_out_channels": [32, 64], "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"], "in_channels": 3, "latent_channels": 4, "layers_per_block": 1, "norm_num_groups": 8, "norm_type": "spatial", "num_vq_embeddings": 12, "out_channels": 3, "up_block_types": [ "AttnUpDecoderBlock2D", "UpDecoderBlock2D", ], "vq_embed_dim": 4, } @property def _A ( self : Any ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : List[str] = VQModel(**self.dummy_movq_kwargs ) return model def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = self.dummy_unet SCREAMING_SNAKE_CASE : str = self.dummy_movq SCREAMING_SNAKE_CASE : Optional[int] = DDIMScheduler( num_train_timesteps=1000 , beta_schedule="linear" , beta_start=0.00_085 , beta_end=0.012 , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , steps_offset=1 , prediction_type="epsilon" , thresholding=UpperCAmelCase_ , ) SCREAMING_SNAKE_CASE : Optional[int] = { "unet": unet, "scheduler": scheduler, "movq": movq, } return components def _A ( self : List[str] , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Optional[Any]=0 ): SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(UpperCAmelCase_ ) ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = floats_tensor((1, self.text_embedder_hidden_size) , rng=random.Random(seed + 1 ) ).to( UpperCAmelCase_ ) if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : List[Any] = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = { "image_embeds": image_embeds, "negative_image_embeds": negative_image_embeds, "generator": generator, "height": 64, "width": 64, "guidance_scale": 4.0, "num_inference_steps": 2, "output_type": "np", } return inputs def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : Any = "cpu" SCREAMING_SNAKE_CASE : Tuple = self.get_dummy_components() SCREAMING_SNAKE_CASE : List[str] = self.pipeline_class(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = pipe.to(UpperCAmelCase_ ) pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(**self.get_dummy_inputs(UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : str = output.images SCREAMING_SNAKE_CASE : Tuple = pipe( **self.get_dummy_inputs(UpperCAmelCase_ ) , return_dict=UpperCAmelCase_ , )[0] SCREAMING_SNAKE_CASE : str = image[0, -3:, -3:, -1] SCREAMING_SNAKE_CASE : Any = image_from_tuple[0, -3:, -3:, -1] assert image.shape == (1, 64, 64, 3) SCREAMING_SNAKE_CASE : List[Any] = np.array( [0.6_237_976, 1.0, 0.36_441_332, 1.0, 0.70_639_634, 0.29_877_186, 0.85_652_125, 0.5_216_843, 0.54_454_046] ) assert ( np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_slice.flatten()}''' assert ( np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2 ), f''' expected_slice {expected_slice}, but got {image_from_tuple_slice.flatten()}''' @slow @require_torch_gpu class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): # clean up the VRAM after each test super().tearDown() gc.collect() torch.cuda.empty_cache() def _A ( self : str ): SCREAMING_SNAKE_CASE : Dict = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main" "/kandinskyv22/kandinskyv22_text2img_cat_fp16.npy" ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPriorPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-prior" , torch_dtype=torch.floataa ) pipe_prior.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = KandinskyVaaPipeline.from_pretrained( "kandinsky-community/kandinsky-2-2-decoder" , torch_dtype=torch.floataa ) SCREAMING_SNAKE_CASE : Tuple = pipeline.to(UpperCAmelCase_ ) pipeline.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = "red cat, 4k photo" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Tuple = pipe_prior( UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=5 , negative_prompt="" , ).to_tuple() SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Generator(device="cuda" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Tuple = pipeline( image_embeds=UpperCAmelCase_ , negative_image_embeds=UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=100 , output_type="np" , ) SCREAMING_SNAKE_CASE : List[Any] = output.images[0] assert image.shape == (512, 512, 3) assert_mean_pixel_difference(UpperCAmelCase_ , UpperCAmelCase_ )
62
1
import unittest import numpy as np import torch from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer from diffusers import ( AutoencoderKL, DDIMScheduler, DPMSolverMultistepScheduler, TextToVideoSDPipeline, UNetaDConditionModel, ) from diffusers.utils import is_xformers_available, load_numpy, skip_mps, slow, torch_device from diffusers.utils.testing_utils import enable_full_determinism from ..pipeline_params import TEXT_TO_IMAGE_BATCH_PARAMS, TEXT_TO_IMAGE_PARAMS from ..test_pipelines_common import PipelineTesterMixin enable_full_determinism() @skip_mps class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Optional[Any] = TextToVideoSDPipeline UpperCamelCase_ : Optional[Any] = TEXT_TO_IMAGE_PARAMS UpperCamelCase_ : Union[str, Any] = TEXT_TO_IMAGE_BATCH_PARAMS # No `output_type`. UpperCamelCase_ : Tuple = frozenset( [ '''num_inference_steps''', '''generator''', '''latents''', '''return_dict''', '''callback''', '''callback_steps''', ] ) def _A ( self : str ): torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = UNetaDConditionModel( block_out_channels=(32, 64, 64, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=("CrossAttnDownBlock3D", "CrossAttnDownBlock3D", "CrossAttnDownBlock3D", "DownBlock3D") , up_block_types=("UpBlock3D", "CrossAttnUpBlock3D", "CrossAttnUpBlock3D", "CrossAttnUpBlock3D") , cross_attention_dim=32 , attention_head_dim=4 , ) SCREAMING_SNAKE_CASE : Optional[Any] = DDIMScheduler( beta_start=0.00_085 , beta_end=0.012 , beta_schedule="scaled_linear" , clip_sample=UpperCAmelCase_ , set_alpha_to_one=UpperCAmelCase_ , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : str = AutoencoderKL( block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"] , up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"] , latent_channels=4 , sample_size=128 , ) torch.manual_seed(0 ) SCREAMING_SNAKE_CASE : str = CLIPTextConfig( bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , hidden_act="gelu" , projection_dim=512 , ) SCREAMING_SNAKE_CASE : Any = CLIPTextModel(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" ) SCREAMING_SNAKE_CASE : str = { "unet": unet, "scheduler": scheduler, "vae": vae, "text_encoder": text_encoder, "tokenizer": tokenizer, } return components def _A ( self : Tuple , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Optional[int]=0 ): if str(UpperCAmelCase_ ).startswith("mps" ): SCREAMING_SNAKE_CASE : int = torch.manual_seed(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Optional[Any] = torch.Generator(device=UpperCAmelCase_ ).manual_seed(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = { "prompt": "A painting of a squirrel eating a burger", "generator": generator, "num_inference_steps": 2, "guidance_scale": 6.0, "output_type": "pt", } return inputs def _A ( self : int ): SCREAMING_SNAKE_CASE : List[str] = "cpu" # ensure determinism for the device-dependent torch.Generator SCREAMING_SNAKE_CASE : List[str] = self.get_dummy_components() SCREAMING_SNAKE_CASE : Union[str, Any] = TextToVideoSDPipeline(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = sd_pipe.to(UpperCAmelCase_ ) sd_pipe.set_progress_bar_config(disable=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = self.get_dummy_inputs(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = "np" SCREAMING_SNAKE_CASE : Any = sd_pipe(**UpperCAmelCase_ ).frames SCREAMING_SNAKE_CASE : Optional[Any] = frames[0][-3:, -3:, -1] assert frames[0].shape == (64, 64, 3) SCREAMING_SNAKE_CASE : Optional[Any] = np.array([158.0, 160.0, 153.0, 125.0, 100.0, 121.0, 111.0, 93.0, 113.0] ) assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2 def _A ( self : List[str] ): self._test_attention_slicing_forward_pass(test_mean_pixel_difference=UpperCAmelCase_ , expected_max_diff=3E-3 ) @unittest.skipIf( torch_device != "cuda" or not is_xformers_available() , reason="XFormers attention is only available with CUDA and `xformers` installed" , ) def _A ( self : List[Any] ): self._test_xformers_attention_forwardGenerator_pass(test_mean_pixel_difference=UpperCAmelCase_ , expected_max_diff=1E-2 ) @unittest.skip(reason="Batching needs to be properly figured out first for this pipeline." ) def _A ( self : Dict ): pass @unittest.skip(reason="Batching needs to be properly figured out first for this pipeline." ) def _A ( self : Union[str, Any] ): pass @unittest.skip(reason="`num_images_per_prompt` argument is not supported for this pipeline." ) def _A ( self : Dict ): pass def _A ( self : Any ): return super().test_progress_bar() @slow @skip_mps class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : int ): SCREAMING_SNAKE_CASE : Any = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/text_to_video/video.npy" ) SCREAMING_SNAKE_CASE : Dict = TextToVideoSDPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b" ) SCREAMING_SNAKE_CASE : str = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config ) SCREAMING_SNAKE_CASE : Any = pipe.to("cuda" ) SCREAMING_SNAKE_CASE : List[Any] = "Spiderman is surfing" SCREAMING_SNAKE_CASE : Dict = torch.Generator(device="cpu" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[Any] = pipe(UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=25 , output_type="pt" ).frames SCREAMING_SNAKE_CASE : Any = video_frames.cpu().numpy() assert np.abs(expected_video - video ).mean() < 5E-2 def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : str = load_numpy( "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/text_to_video/video_2step.npy" ) SCREAMING_SNAKE_CASE : int = TextToVideoSDPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b" ) SCREAMING_SNAKE_CASE : Optional[Any] = pipe.to("cuda" ) SCREAMING_SNAKE_CASE : Tuple = "Spiderman is surfing" SCREAMING_SNAKE_CASE : Tuple = torch.Generator(device="cpu" ).manual_seed(0 ) SCREAMING_SNAKE_CASE : Optional[int] = pipe(UpperCAmelCase_ , generator=UpperCAmelCase_ , num_inference_steps=2 , output_type="pt" ).frames SCREAMING_SNAKE_CASE : str = video_frames.cpu().numpy() assert np.abs(expected_video - video ).mean() < 5E-2
62
import logging import os import quant_trainer import torch from torch.utils.data import DataLoader from transformers import Trainer, is_torch_tpu_available from transformers.trainer_utils import PredictionOutput snake_case = logging.getLogger(__name__) if is_torch_tpu_available(check_device=False): import torch_xla.core.xla_model as xm import torch_xla.debug.metrics as met class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def __init__( self : int , *UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : str=None , UpperCAmelCase_ : str=None , UpperCAmelCase_ : Tuple=None , **UpperCAmelCase_ : Optional[Any] ): super().__init__(*UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = eval_examples SCREAMING_SNAKE_CASE : List[Any] = post_process_function SCREAMING_SNAKE_CASE : Any = quant_trainer_args SCREAMING_SNAKE_CASE : Optional[Any] = 128 # default number of calibration samples def _A ( self : Optional[Any] , UpperCAmelCase_ : Tuple=None ): if calib_dataset is None and self.calib_dataset is None: raise ValueError("Trainer: calibration requires an calib_dataset." ) SCREAMING_SNAKE_CASE : str = calib_dataset if calib_dataset is not None else self.calib_dataset SCREAMING_SNAKE_CASE : str = self._remove_unused_columns(UpperCAmelCase_ , description="Calibration" ) return DataLoader( UpperCAmelCase_ , batch_size=self.args.eval_batch_size , collate_fn=self.data_collator , drop_last=self.args.dataloader_drop_last , num_workers=self.args.dataloader_num_workers , pin_memory=self.args.dataloader_pin_memory , shuffle=UpperCAmelCase_ , ) def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[int]=None ): SCREAMING_SNAKE_CASE : Any = self.train_dataset if calib_dataset is None else calib_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_calib_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args , calib=UpperCAmelCase_ ) model.eval() quant_trainer.enable_calibration(UpperCAmelCase_ ) logger.info("***** Running calibration *****" ) logger.info(f''' Num examples = {self.calib_num}''' ) logger.info(f''' Batch size = {calib_dataloader.batch_size}''' ) for step, inputs in enumerate(UpperCAmelCase_ ): # Prediction step SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[Any] = self.prediction_step(UpperCAmelCase_ , UpperCAmelCase_ , prediction_loss_only=UpperCAmelCase_ ) if (step + 1) * calib_dataloader.batch_size >= self.calib_num: break quant_trainer.finish_calibration(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : Optional[int] = model def _A ( self : List[Any] , UpperCAmelCase_ : int=None , UpperCAmelCase_ : Optional[Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : str = "eval" ): SCREAMING_SNAKE_CASE : List[str] = self.eval_dataset if eval_dataset is None else eval_dataset SCREAMING_SNAKE_CASE : Tuple = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = self.eval_examples if eval_examples is None else eval_examples # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Dict = self.compute_metrics SCREAMING_SNAKE_CASE : Tuple = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : int = eval_loop( UpperCAmelCase_ , description="Evaluation" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : int = compute_metrics if self.post_process_function is not None and self.compute_metrics is not None: SCREAMING_SNAKE_CASE : List[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions ) SCREAMING_SNAKE_CASE : Any = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : Union[str, Any] = metrics.pop(UpperCAmelCase_ ) self.log(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : List[Any] = {} if self.args.tpu_metrics_debug or self.args.debug: # tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.) xm.master_print(met.metrics_report() ) SCREAMING_SNAKE_CASE : List[str] = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCAmelCase_ ) return metrics def _A ( self : Optional[Any] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple=None , UpperCAmelCase_ : str = "test" ): SCREAMING_SNAKE_CASE : str = self.get_test_dataloader(UpperCAmelCase_ ) # Temporarily disable metric computation, we will do it in the loop here. SCREAMING_SNAKE_CASE : Tuple = self.compute_metrics SCREAMING_SNAKE_CASE : Union[str, Any] = None SCREAMING_SNAKE_CASE : Optional[int] = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop try: SCREAMING_SNAKE_CASE : str = eval_loop( UpperCAmelCase_ , description="Prediction" , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCAmelCase_ , ) finally: SCREAMING_SNAKE_CASE : Union[str, Any] = compute_metrics if self.post_process_function is None or self.compute_metrics is None: return output SCREAMING_SNAKE_CASE : Optional[Any] = self.post_process_function(UpperCAmelCase_ , UpperCAmelCase_ , output.predictions , "predict" ) SCREAMING_SNAKE_CASE : str = self.compute_metrics(UpperCAmelCase_ ) # Prefix all keys with metric_key_prefix + '_' for key in list(metrics.keys() ): if not key.startswith(f'''{metric_key_prefix}_''' ): SCREAMING_SNAKE_CASE : str = metrics.pop(UpperCAmelCase_ ) return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCAmelCase_ ) def _A ( self : Any , UpperCAmelCase_ : int="./" ): SCREAMING_SNAKE_CASE : List[Any] = self.eval_dataset SCREAMING_SNAKE_CASE : List[Any] = self.get_eval_dataloader(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = next(iter(UpperCAmelCase_ ) ) # saving device - to make it consistent SCREAMING_SNAKE_CASE : int = torch.device("cuda" if torch.cuda.is_available() else "cpu" ) # convert to tuple SCREAMING_SNAKE_CASE : Tuple = tuple(v.to(UpperCAmelCase_ ) for k, v in batch.items() ) logger.info("Converting model to be onnx compatible" ) from pytorch_quantization.nn import TensorQuantizer SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Any = self.model.to(UpperCAmelCase_ ) model.eval() model.float() SCREAMING_SNAKE_CASE : str = model.module if hasattr(UpperCAmelCase_ , "module" ) else model quant_trainer.configure_model(UpperCAmelCase_ , self.quant_trainer_args ) SCREAMING_SNAKE_CASE : List[Any] = os.path.join(UpperCAmelCase_ , "model.onnx" ) logger.info(f'''exporting model to {output_model_file}''' ) SCREAMING_SNAKE_CASE : int = {0: "batch_size", 1: "seq_len"} torch.onnx.export( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , export_params=UpperCAmelCase_ , opset_version=13 , do_constant_folding=UpperCAmelCase_ , input_names=["input_ids", "attention_mask", "token_type_ids"] , output_names=["output_start_logits", "output_end_logits"] , dynamic_axes={ "input_ids": axes, "attention_mask": axes, "token_type_ids": axes, "output_start_logits": axes, "output_end_logits": axes, } , verbose=UpperCAmelCase_ , ) logger.info("onnx export finished" )
62
1
class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : int , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Any = val SCREAMING_SNAKE_CASE : int = None SCREAMING_SNAKE_CASE : Optional[int] = None def _A ( self : str , UpperCAmelCase_ : Optional[int] ): if self.val: if val < self.val: if self.left is None: SCREAMING_SNAKE_CASE : Any = Node(UpperCAmelCase_ ) else: self.left.insert(UpperCAmelCase_ ) elif val > self.val: if self.right is None: SCREAMING_SNAKE_CASE : List[Any] = Node(UpperCAmelCase_ ) else: self.right.insert(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : Dict = val def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" if root: inorder(root.left , lowercase ) res.append(root.val ) inorder(root.right , lowercase ) def lowerCamelCase__ ( lowercase ): """simple docstring""" if len(lowercase ) == 0: return arr SCREAMING_SNAKE_CASE : Tuple = Node(arr[0] ) for i in range(1 , len(lowercase ) ): root.insert(arr[i] ) # Traverse BST in order. SCREAMING_SNAKE_CASE : Tuple = [] inorder(lowercase , lowercase ) return res if __name__ == "__main__": print(tree_sort([10, 1, 3, 2, 9, 14, 13]))
62
import os import unittest from transformers import LayoutLMTokenizer, LayoutLMTokenizerFast from transformers.models.layoutlm.tokenization_layoutlm import VOCAB_FILES_NAMES from transformers.testing_utils import require_tokenizers from ...test_tokenization_common import TokenizerTesterMixin @require_tokenizers class SCREAMING_SNAKE_CASE ( lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = LayoutLMTokenizer UpperCamelCase_ : str = LayoutLMTokenizerFast UpperCamelCase_ : Any = True UpperCamelCase_ : Optional[Any] = True def _A ( self : Any ): super().setUp() SCREAMING_SNAKE_CASE : Optional[Any] = [ "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest", ] SCREAMING_SNAKE_CASE : Optional[Any] = 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 _A ( self : str , **UpperCAmelCase_ : Optional[int] ): return LayoutLMTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Tuple , UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Any = "UNwant\u00E9d,running" SCREAMING_SNAKE_CASE : Union[str, Any] = "unwanted, running" return input_text, output_text def _A ( self : int ): SCREAMING_SNAKE_CASE : Optional[int] = self.tokenizer_class(self.vocab_file ) SCREAMING_SNAKE_CASE : Dict = tokenizer.tokenize("UNwant\u00E9d,running" ) self.assertListEqual(UpperCAmelCase_ , ["un", "##want", "##ed", ",", "runn", "##ing"] ) self.assertListEqual(tokenizer.convert_tokens_to_ids(UpperCAmelCase_ ) , [7, 4, 5, 10, 8, 9] ) def _A ( self : List[str] ): pass
62
1
import os from typing import Dict, List, Union import tensorflow as tf from keras_nlp.tokenizers import BytePairTokenizer from tensorflow_text import pad_model_inputs from .tokenization_gpta import GPTaTokenizer class SCREAMING_SNAKE_CASE ( tf.keras.layers.Layer ): '''simple docstring''' def __init__( self : List[Any] , UpperCAmelCase_ : Dict[str, int] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : int = None , UpperCAmelCase_ : int = None ): super().__init__() SCREAMING_SNAKE_CASE : Union[str, Any] = pad_token_id SCREAMING_SNAKE_CASE : Optional[Any] = max_length SCREAMING_SNAKE_CASE : Optional[int] = vocab SCREAMING_SNAKE_CASE : Optional[Any] = merges SCREAMING_SNAKE_CASE : int = BytePairTokenizer(UpperCAmelCase_ , UpperCAmelCase_ , sequence_length=UpperCAmelCase_ ) @classmethod def _A ( cls : Tuple , UpperCAmelCase_ : GPTaTokenizer , *UpperCAmelCase_ : Tuple , **UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : Optional[Any] = [" ".join(UpperCAmelCase_ ) for m in tokenizer.bpe_ranks.keys()] SCREAMING_SNAKE_CASE : str = tokenizer.get_vocab() return cls(UpperCAmelCase_ , UpperCAmelCase_ , *UpperCAmelCase_ , **UpperCAmelCase_ ) @classmethod def _A ( cls : str , UpperCAmelCase_ : Union[str, os.PathLike] , *UpperCAmelCase_ : Dict , **UpperCAmelCase_ : List[str] ): SCREAMING_SNAKE_CASE : Tuple = GPTaTokenizer.from_pretrained(UpperCAmelCase_ , *UpperCAmelCase_ , **UpperCAmelCase_ ) return cls.from_tokenizer(UpperCAmelCase_ , *UpperCAmelCase_ , **UpperCAmelCase_ ) @classmethod def _A ( cls : Union[str, Any] , UpperCAmelCase_ : str ): return cls(**UpperCAmelCase_ ) def _A ( self : List[Any] ): return { "vocab": self.vocab, "merges": self.merges, "max_length": self.max_length, "pad_token_id": self.pad_token_id, } def _A ( self : Tuple , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : int = None ): SCREAMING_SNAKE_CASE : Any = self.tf_tokenizer(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = tf.ones_like(UpperCAmelCase_ ) if self.pad_token_id is not None: # pad the tokens up to max length SCREAMING_SNAKE_CASE : Tuple = max_length if max_length is not None else self.max_length if max_length is not None: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Any = pad_model_inputs( UpperCAmelCase_ , max_seq_length=UpperCAmelCase_ , pad_value=self.pad_token_id ) return {"attention_mask": attention_mask, "input_ids": input_ids}
62
import argparse import json import os from collections import OrderedDict import numpy as np import tensorflow as tf import torch def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Dict = os.path.join(args.tf_model_dir , "parameters.json" ) SCREAMING_SNAKE_CASE : Union[str, Any] = json.loads(open(lowercase ).read() ) if not params: raise ValueError( F'''It seems that the json file at {parameter_file} is empty. Make sure you have a correct json file.''' ) if not args.output.endswith(".pt" ): SCREAMING_SNAKE_CASE : Optional[int] = args.output + ".pt" SCREAMING_SNAKE_CASE : Any = OrderedDict() with tf.device("/CPU:0" ): SCREAMING_SNAKE_CASE : Union[str, Any] = tf.train.load_checkpoint(args.tf_model_dir ) SCREAMING_SNAKE_CASE : Union[str, Any] = reader.get_variable_to_shape_map() for key_name in shapes.keys(): SCREAMING_SNAKE_CASE : Any = reader.get_tensor(lowercase ).astype(np.floataa ) if key_name.endswith("/adam_m" ) or key_name.endswith("/adam_v" ): continue if key_name.startswith("pasts/" ): if key_name.startswith("pasts/mlp" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9] ) elif key_name.startswith("pasts/out" ): SCREAMING_SNAKE_CASE : Optional[int] = 8 SCREAMING_SNAKE_CASE : List[Any] = "model.sqout.%d.weight" % (player * 2) # enter to nn.Sequencial with Tanh, so 2 at a time SCREAMING_SNAKE_CASE : Optional[int] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.startswith("model/moe" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/switch_gating/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.router.classifier.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/softmlp/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.soft_bypass_mlp.weight" % player SCREAMING_SNAKE_CASE : Any = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/wo/kernel" ) or key_name.endswith("/wi/kernel" ): SCREAMING_SNAKE_CASE : Optional[int] = key_name[-9:-7] for i in range(16 ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.feed_forward.mlp.experts.expert_%d.%s.weight" % (player, i, nlayer) SCREAMING_SNAKE_CASE : List[str] = ( vnp[i].transpose([1, 0] ).copy() ) # In Mesh-Tensorflow, it is one array, so it is divided SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name.startswith("model/mlp" ): SCREAMING_SNAKE_CASE : str = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/p1/kernel" ): SCREAMING_SNAKE_CASE : Dict = "model.blocks.%d.feed_forward.mlp.wi.weight" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Any = torch.tensor(lowercase ) elif key_name.endswith("/p1/bias" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.mlp.wi.bias" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/kernel" ): SCREAMING_SNAKE_CASE : str = "model.blocks.%d.feed_forward.mlp.wo.weight" % player SCREAMING_SNAKE_CASE : List[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) elif key_name.endswith("/p2/bias" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.feed_forward.mlp.wo.bias" % player SCREAMING_SNAKE_CASE : str = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) elif key_name.startswith("model/ln" ): SCREAMING_SNAKE_CASE : Union[str, Any] = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.feed_forward.norm.bias" % player SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Union[str, Any] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : List[str] = "model.blocks.%d.feed_forward.norm.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/att" ): SCREAMING_SNAKE_CASE : Optional[int] = int(key_name[9:].split("/" )[0] ) if key_name.endswith("/qkv/kernel" ): SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # Compute same dimension as Mesh-tensorflow using einsum SCREAMING_SNAKE_CASE : List[str] = state[:, 0, :, :] SCREAMING_SNAKE_CASE : Tuple = state[:, 1, :, :] SCREAMING_SNAKE_CASE : List[Any] = state[:, 2, :, :] SCREAMING_SNAKE_CASE : Tuple = ( state_q.reshape([state_q.shape[0], state_q.shape[1] * state_q.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : List[Any] = ( state_k.reshape([state_k.shape[0], state_k.shape[1] * state_k.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Union[str, Any] = ( state_v.reshape([state_v.shape[0], state_v.shape[1] * state_v.shape[2]] ) .transpose([1, 0] ) .copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : int = "model.blocks.%d.self_attn.self_attn.q_proj.weight" % player SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = "model.blocks.%d.self_attn.self_attn.k_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[Any] = torch.tensor(lowercase ) SCREAMING_SNAKE_CASE : Optional[Any] = "model.blocks.%d.self_attn.self_attn.v_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = torch.tensor(lowercase ) elif key_name.endswith("/o/kernel" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "model.blocks.%d.self_attn.self_attn.out_proj.weight" % player SCREAMING_SNAKE_CASE : Optional[int] = ( vnp.reshape([vnp.shape[0] * vnp.shape[1], vnp.shape[2]] ).transpose([1, 0] ).copy() ) # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif key_name.startswith("model/an" ): SCREAMING_SNAKE_CASE : int = int(key_name[8:].split("/" )[0] ) if key_name.endswith("/b" ): SCREAMING_SNAKE_CASE : List[Any] = "model.blocks.%d.self_attn.norm.bias" % player SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : List[str] = torch.tensor(lowercase ) elif key_name.endswith("/g" ): SCREAMING_SNAKE_CASE : Tuple = "model.blocks.%d.self_attn.norm.weight" % player SCREAMING_SNAKE_CASE : List[str] = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) elif ( key_name.startswith("model/wte" ) or key_name.startswith("model/wpe" ) or key_name.startswith("model/ete" ) ): SCREAMING_SNAKE_CASE : str = {"wte": "embed_tokens", "wpe": "position_embeddings", "ete": "extra_position_embeddings"}[ key_name[-3:] ] SCREAMING_SNAKE_CASE : List[str] = "model.%s.weight" % nlayer SCREAMING_SNAKE_CASE : Union[str, Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) if key_name.startswith("model/wte" ): SCREAMING_SNAKE_CASE : Union[str, Any] = "lm_head.weight" SCREAMING_SNAKE_CASE : List[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : Dict = torch.tensor(lowercase ) elif key_name.startswith("model/wob" ): SCREAMING_SNAKE_CASE : List[Any] = "final_logits_bias" SCREAMING_SNAKE_CASE : Optional[Any] = vnp.copy() # same in embedded SCREAMING_SNAKE_CASE : List[str] = state.reshape((1, -1) ) SCREAMING_SNAKE_CASE : int = torch.tensor(lowercase ) elif key_name == "model/dense/kernel": SCREAMING_SNAKE_CASE : Optional[int] = "model.last_project.weight" SCREAMING_SNAKE_CASE : Dict = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix SCREAMING_SNAKE_CASE : Tuple = torch.tensor(lowercase ) elif key_name == "model/dense_1/bias": SCREAMING_SNAKE_CASE : str = "model.last_project.bias" SCREAMING_SNAKE_CASE : int = vnp.copy() # same because it is one dimensional SCREAMING_SNAKE_CASE : str = torch.tensor(lowercase ) torch.save(lowercase , args.output ) if __name__ == "__main__": snake_case = argparse.ArgumentParser( description="""model converter.""", formatter_class=argparse.ArgumentDefaultsHelpFormatter ) parser.add_argument("""--tf_model_dir""", metavar="""PATH""", type=str, required=True, help="""import model""") parser.add_argument("""--output""", metavar="""PATH""", type=str, required=True, help="""output model""") snake_case = parser.parse_args() convert_tf_gptsan_to_pt(args)
62
1
import logging import os from typing import Dict, List, Optional, Union import torch import torch.nn as nn from accelerate.utils.imports import ( is_abit_bnb_available, is_abit_bnb_available, is_bnb_available, ) from ..big_modeling import dispatch_model, init_empty_weights from .dataclasses import BnbQuantizationConfig from .modeling import ( find_tied_parameters, get_balanced_memory, infer_auto_device_map, load_checkpoint_in_model, offload_weight, set_module_tensor_to_device, ) if is_bnb_available(): import bitsandbytes as bnb from copy import deepcopy snake_case = logging.getLogger(__name__) def lowerCamelCase__ ( lowercase , lowercase , lowercase = None , lowercase = None , lowercase = None , lowercase = None , lowercase = None , lowercase = False , ): """simple docstring""" SCREAMING_SNAKE_CASE : int = bnb_quantization_config.load_in_abit SCREAMING_SNAKE_CASE : List[Any] = bnb_quantization_config.load_in_abit if load_in_abit and not is_abit_bnb_available(): raise ImportError( "You have a version of `bitsandbytes` that is not compatible with 8bit quantization," " make sure you have the latest version of `bitsandbytes` installed." ) if load_in_abit and not is_abit_bnb_available(): raise ValueError( "You have a version of `bitsandbytes` that is not compatible with 4bit quantization," "make sure you have the latest version of `bitsandbytes` installed." ) SCREAMING_SNAKE_CASE : Tuple = [] # custom device map if isinstance(lowercase , lowercase ) and len(device_map.keys() ) > 1: SCREAMING_SNAKE_CASE : int = [key for key, value in device_map.items() if value in ["disk", "cpu"]] # We keep some modules such as the lm_head in their original dtype for numerical stability reasons if bnb_quantization_config.skip_modules is None: SCREAMING_SNAKE_CASE : Optional[Any] = get_keys_to_not_convert(lowercase ) # add cpu modules to skip modules only for 4-bit modules if load_in_abit: bnb_quantization_config.skip_modules.extend(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = bnb_quantization_config.skip_modules # We add the modules we want to keep in full precision if bnb_quantization_config.keep_in_fpaa_modules is None: SCREAMING_SNAKE_CASE : int = [] SCREAMING_SNAKE_CASE : int = bnb_quantization_config.keep_in_fpaa_modules modules_to_not_convert.extend(lowercase ) # compatibility with peft SCREAMING_SNAKE_CASE : int = load_in_abit SCREAMING_SNAKE_CASE : str = load_in_abit SCREAMING_SNAKE_CASE : List[str] = get_parameter_device(lowercase ) if model_device.type != "meta": # quantization of an already loaded model logger.warning( "It is not recommended to quantize a loaded model. " "The model should be instantiated under the `init_empty_weights` context manager." ) SCREAMING_SNAKE_CASE : Any = replace_with_bnb_layers(lowercase , lowercase , modules_to_not_convert=lowercase ) # convert param to the right dtype SCREAMING_SNAKE_CASE : Optional[int] = bnb_quantization_config.torch_dtype for name, param in model.state_dict().items(): if any(module_to_keep_in_fpaa in name for module_to_keep_in_fpaa in keep_in_fpaa_modules ): param.to(torch.floataa ) if param.dtype != torch.floataa: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(".weight" , "" ).replace(".bias" , "" ) SCREAMING_SNAKE_CASE : Optional[int] = getattr(lowercase , lowercase , lowercase ) if param is not None: param.to(torch.floataa ) elif torch.is_floating_point(lowercase ): param.to(lowercase ) if model_device.type == "cuda": # move everything to cpu in the first place because we can't do quantization if the weights are already on cuda model.cuda(torch.cuda.current_device() ) torch.cuda.empty_cache() elif torch.cuda.is_available(): model.to(torch.cuda.current_device() ) else: raise RuntimeError("No GPU found. A GPU is needed for quantization." ) logger.info( F'''The model device type is {model_device.type}. However, cuda is needed for quantization.''' "We move the model to cuda." ) return model elif weights_location is None: raise RuntimeError( F'''`weights_location` needs to be the folder path containing the weights of the model, but we found {weights_location} ''' ) else: with init_empty_weights(): SCREAMING_SNAKE_CASE : str = replace_with_bnb_layers( lowercase , lowercase , modules_to_not_convert=lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = get_quantized_model_device_map( lowercase , lowercase , lowercase , max_memory=lowercase , no_split_module_classes=lowercase , ) if offload_state_dict is None and device_map is not None and "disk" in device_map.values(): SCREAMING_SNAKE_CASE : List[str] = True SCREAMING_SNAKE_CASE : Optional[Any] = any(x in list(device_map.values() ) for x in ["cpu", "disk"] ) load_checkpoint_in_model( lowercase , lowercase , lowercase , dtype=bnb_quantization_config.torch_dtype , offload_folder=lowercase , offload_state_dict=lowercase , keep_in_fpaa_modules=bnb_quantization_config.keep_in_fpaa_modules , offload_abit_bnb=load_in_abit and offload , ) return dispatch_model(lowercase , device_map=lowercase , offload_dir=lowercase ) def lowerCamelCase__ ( lowercase , lowercase , lowercase=None , lowercase=None , lowercase=None ): """simple docstring""" if device_map is None: if torch.cuda.is_available(): SCREAMING_SNAKE_CASE : Optional[Any] = {"": torch.cuda.current_device()} else: raise RuntimeError("No GPU found. A GPU is needed for quantization." ) logger.info("The device_map was not initialized." "Setting device_map to `{'':torch.cuda.current_device()}`." ) if isinstance(lowercase , lowercase ): if device_map not in ["auto", "balanced", "balanced_low_0", "sequential"]: raise ValueError( "If passing a string for `device_map`, please choose 'auto', 'balanced', 'balanced_low_0' or " "'sequential'." ) SCREAMING_SNAKE_CASE : Tuple = {} special_dtypes.update( { name: bnb_quantization_config.torch_dtype for name, _ in model.named_parameters() if any(m in name for m in bnb_quantization_config.skip_modules ) } ) special_dtypes.update( { name: torch.floataa for name, _ in model.named_parameters() if any(m in name for m in bnb_quantization_config.keep_in_fpaa_modules ) } ) SCREAMING_SNAKE_CASE : Optional[Any] = {} SCREAMING_SNAKE_CASE : Optional[Any] = special_dtypes SCREAMING_SNAKE_CASE : List[str] = no_split_module_classes SCREAMING_SNAKE_CASE : str = bnb_quantization_config.target_dtype # get max_memory for each device. if device_map != "sequential": SCREAMING_SNAKE_CASE : List[Any] = get_balanced_memory( lowercase , low_zero=(device_map == "balanced_low_0") , max_memory=lowercase , **lowercase , ) SCREAMING_SNAKE_CASE : int = max_memory SCREAMING_SNAKE_CASE : Tuple = infer_auto_device_map(lowercase , **lowercase ) if isinstance(lowercase , lowercase ): # check if don't have any quantized module on the cpu SCREAMING_SNAKE_CASE : List[str] = bnb_quantization_config.skip_modules + bnb_quantization_config.keep_in_fpaa_modules SCREAMING_SNAKE_CASE : List[Any] = { key: device_map[key] for key in device_map.keys() if key not in modules_not_to_convert } for device in ["cpu", "disk"]: if device in device_map_without_some_modules.values(): if bnb_quantization_config.load_in_abit: raise ValueError( "\n Some modules are dispatched on the CPU or the disk. Make sure you have enough GPU RAM to fit\n the quantized model. If you want to dispatch the model on the CPU or the disk while keeping\n these modules in `torch_dtype`, you need to pass a custom `device_map` to\n `load_and_quantize_model`. Check\n https://huggingface.co/docs/accelerate/main/en/usage_guides/quantization#offload-modules-to-cpu-and-disk\n for more details.\n " ) else: logger.info( "Some modules are are offloaded to the CPU or the disk. Note that these modules will be converted to 8-bit" ) del device_map_without_some_modules return device_map def lowerCamelCase__ ( lowercase , lowercase , lowercase=None , lowercase=None ): """simple docstring""" if modules_to_not_convert is None: SCREAMING_SNAKE_CASE : int = [] SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Dict = _replace_with_bnb_layers( lowercase , lowercase , lowercase , lowercase ) if not has_been_replaced: logger.warning( "You are loading your model in 8bit or 4bit but no linear modules were found in your model." " this can happen for some architectures such as gpt2 that uses Conv1D instead of Linear layers." " Please double check your model architecture, or submit an issue on github if you think this is" " a bug." ) return model def lowerCamelCase__ ( lowercase , lowercase , lowercase=None , lowercase=None , ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[Any] = False for name, module in model.named_children(): if current_key_name is None: SCREAMING_SNAKE_CASE : Any = [] current_key_name.append(lowercase ) if isinstance(lowercase , nn.Linear ) and name not in modules_to_not_convert: # Check if the current key is not in the `modules_to_not_convert` SCREAMING_SNAKE_CASE : Dict = ".".join(lowercase ) SCREAMING_SNAKE_CASE : Optional[int] = True for key in modules_to_not_convert: if ( (key in current_key_name_str) and (key + "." in current_key_name_str) ) or key == current_key_name_str: SCREAMING_SNAKE_CASE : Union[str, Any] = False break if proceed: # Load bnb module with empty weight and replace ``nn.Linear` module if bnb_quantization_config.load_in_abit: SCREAMING_SNAKE_CASE : Optional[int] = bnb.nn.LinearabitLt( module.in_features , module.out_features , module.bias is not None , has_fpaa_weights=lowercase , threshold=bnb_quantization_config.llm_inta_threshold , ) elif bnb_quantization_config.load_in_abit: SCREAMING_SNAKE_CASE : int = bnb.nn.Linearabit( module.in_features , module.out_features , module.bias is not None , bnb_quantization_config.bnb_abit_compute_dtype , compress_statistics=bnb_quantization_config.bnb_abit_use_double_quant , quant_type=bnb_quantization_config.bnb_abit_quant_type , ) else: raise ValueError("load_in_8bit and load_in_4bit can't be both False" ) SCREAMING_SNAKE_CASE : int = module.weight.data if module.bias is not None: SCREAMING_SNAKE_CASE : List[str] = module.bias.data bnb_module.requires_grad_(lowercase ) setattr(lowercase , lowercase , lowercase ) SCREAMING_SNAKE_CASE : Any = True if len(list(module.children() ) ) > 0: SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = _replace_with_bnb_layers( lowercase , lowercase , lowercase , lowercase ) SCREAMING_SNAKE_CASE : Tuple = has_been_replaced | _has_been_replaced # Remove the last key for recursion current_key_name.pop(-1 ) return model, has_been_replaced def lowerCamelCase__ ( lowercase ): """simple docstring""" with init_empty_weights(): SCREAMING_SNAKE_CASE : List[str] = deepcopy(lowercase ) # this has 0 cost since it is done inside `init_empty_weights` context manager` SCREAMING_SNAKE_CASE : Dict = find_tied_parameters(lowercase ) # For compatibility with Accelerate < 0.18 if isinstance(lowercase , lowercase ): SCREAMING_SNAKE_CASE : List[Any] = sum(list(tied_params.values() ) , [] ) + list(tied_params.keys() ) else: SCREAMING_SNAKE_CASE : Optional[Any] = sum(lowercase , [] ) SCREAMING_SNAKE_CASE : int = len(lowercase ) > 0 # Check if it is a base model SCREAMING_SNAKE_CASE : Any = False if hasattr(lowercase , "base_model_prefix" ): SCREAMING_SNAKE_CASE : str = not hasattr(lowercase , model.base_model_prefix ) # Ignore this for base models (BertModel, GPT2Model, etc.) if (not has_tied_params) and is_base_model: return [] # otherwise they have an attached head SCREAMING_SNAKE_CASE : Optional[int] = list(model.named_children() ) SCREAMING_SNAKE_CASE : Optional[int] = [list_modules[-1][0]] # add last module together with tied weights SCREAMING_SNAKE_CASE : str = set(lowercase ) - set(lowercase ) SCREAMING_SNAKE_CASE : List[str] = list(set(lowercase ) ) + list(lowercase ) # remove ".weight" from the keys SCREAMING_SNAKE_CASE : str = [".weight", ".bias"] SCREAMING_SNAKE_CASE : List[Any] = [] for name in list_untouched: for name_to_remove in names_to_remove: if name_to_remove in name: SCREAMING_SNAKE_CASE : Union[str, Any] = name.replace(lowercase , "" ) filtered_module_names.append(lowercase ) return filtered_module_names def lowerCamelCase__ ( lowercase ): """simple docstring""" for m in model.modules(): if isinstance(lowercase , bnb.nn.Linearabit ): return True return False def lowerCamelCase__ ( lowercase ): """simple docstring""" return next(parameter.parameters() ).device def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase , lowercase , lowercase , lowercase ): """simple docstring""" if fpaa_statistics is None: set_module_tensor_to_device(lowercase , lowercase , 0 , dtype=lowercase , value=lowercase ) SCREAMING_SNAKE_CASE : str = param_name SCREAMING_SNAKE_CASE : List[str] = model if "." in tensor_name: SCREAMING_SNAKE_CASE : Dict = tensor_name.split("." ) for split in splits[:-1]: SCREAMING_SNAKE_CASE : Tuple = getattr(lowercase , lowercase ) if new_module is None: raise ValueError(F'''{module} has no attribute {split}.''' ) SCREAMING_SNAKE_CASE : Dict = new_module SCREAMING_SNAKE_CASE : List[str] = splits[-1] # offload weights SCREAMING_SNAKE_CASE : Tuple = False offload_weight(module._parameters[tensor_name] , lowercase , lowercase , index=lowercase ) if hasattr(module._parameters[tensor_name] , "SCB" ): offload_weight( module._parameters[tensor_name].SCB , param_name.replace("weight" , "SCB" ) , lowercase , index=lowercase , ) else: offload_weight(lowercase , lowercase , lowercase , index=lowercase ) offload_weight(lowercase , param_name.replace("weight" , "SCB" ) , lowercase , index=lowercase ) set_module_tensor_to_device(lowercase , lowercase , "meta" , dtype=lowercase , value=torch.empty(*param.size() ) )
62
import json import os import shutil import tempfile import unittest import numpy as np from transformers import BertTokenizerFast from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer from transformers.testing_utils import require_tokenizers, require_vision from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available if is_vision_available(): from PIL import Image from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor @require_tokenizers @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' def _A ( self : Any ): SCREAMING_SNAKE_CASE : Dict = tempfile.mkdtemp() # fmt: off SCREAMING_SNAKE_CASE : Any = ["[UNK]", "[CLS]", "[SEP]", "[PAD]", "[MASK]", "want", "##want", "##ed", "wa", "un", "runn", "##ing", ",", "low", "lowest"] # fmt: on SCREAMING_SNAKE_CASE : Dict = 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] ) ) SCREAMING_SNAKE_CASE : Optional[int] = { "do_resize": True, "size": {"height": 18, "width": 18}, "do_normalize": True, "image_mean": [0.5, 0.5, 0.5], "image_std": [0.5, 0.5, 0.5], } SCREAMING_SNAKE_CASE : Optional[Any] = os.path.join(self.tmpdirname , UpperCAmelCase_ ) with open(self.image_processor_file , "w" , encoding="utf-8" ) as fp: json.dump(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] , **UpperCAmelCase_ : List[str] ): return BertTokenizer.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Union[str, Any] , **UpperCAmelCase_ : Any ): return ViTImageProcessor.from_pretrained(self.tmpdirname , **UpperCAmelCase_ ) def _A ( self : Optional[int] ): shutil.rmtree(self.tmpdirname ) def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[Any] = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )] SCREAMING_SNAKE_CASE : Union[str, Any] = [Image.fromarray(np.moveaxis(UpperCAmelCase_ , 0 , -1 ) ) for x in image_inputs] return image_inputs def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : int = VisionTextDualEncoderProcessor( tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() ) processor.save_pretrained(self.tmpdirname ) SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer(bos_token="(BOS)" , eos_token="(EOS)" ) SCREAMING_SNAKE_CASE : Any = self.get_image_processor(do_normalize=UpperCAmelCase_ , padding_value=1.0 ) SCREAMING_SNAKE_CASE : str = VisionTextDualEncoderProcessor.from_pretrained( self.tmpdirname , bos_token="(BOS)" , eos_token="(EOS)" , do_normalize=UpperCAmelCase_ , padding_value=1.0 ) self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() ) self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) ) self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() ) self.assertIsInstance(processor.image_processor , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : int = image_processor(UpperCAmelCase_ , return_tensors="np" ) SCREAMING_SNAKE_CASE : Optional[int] = processor(images=UpperCAmelCase_ , 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 _A ( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Optional[Any] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = "lower newer" SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = tokenizer(UpperCAmelCase_ ) for key in encoded_tok.keys(): self.assertListEqual(encoded_tok[key] , encoded_processor[key] ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : int = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[Any] = self.get_tokenizer() SCREAMING_SNAKE_CASE : List[str] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = "lower newer" SCREAMING_SNAKE_CASE : Union[str, Any] = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Optional[Any] = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , ["input_ids", "token_type_ids", "attention_mask", "pixel_values"] ) # test if it raises when no input is passed with self.assertRaises(UpperCAmelCase_ ): processor() def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.get_image_processor() SCREAMING_SNAKE_CASE : List[str] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Tuple = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]] SCREAMING_SNAKE_CASE : Optional[Any] = processor.batch_decode(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = tokenizer.batch_decode(UpperCAmelCase_ ) self.assertListEqual(UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE : Optional[int] = self.get_image_processor() SCREAMING_SNAKE_CASE : Optional[int] = self.get_tokenizer() SCREAMING_SNAKE_CASE : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = "lower newer" SCREAMING_SNAKE_CASE : Any = self.prepare_image_inputs() SCREAMING_SNAKE_CASE : Dict = processor(text=UpperCAmelCase_ , images=UpperCAmelCase_ ) self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
62
1
import hashlib import unittest from transformers import MODEL_FOR_DEPTH_ESTIMATION_MAPPING, is_torch_available, is_vision_available from transformers.pipelines import DepthEstimationPipeline, pipeline from transformers.testing_utils import ( is_pipeline_test, nested_simplify, require_tf, require_timm, require_torch, require_vision, slow, ) from .test_pipelines_common import ANY if is_torch_available(): import torch if is_vision_available(): from PIL import Image else: class SCREAMING_SNAKE_CASE : '''simple docstring''' @staticmethod def _A ( *UpperCAmelCase_ : Optional[Any] , **UpperCAmelCase_ : Union[str, Any] ): pass def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Optional[int] = hashlib.mda(image.tobytes() ) return m.hexdigest() @is_pipeline_test @require_vision @require_timm @require_torch class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : int = MODEL_FOR_DEPTH_ESTIMATION_MAPPING def _A ( self : Optional[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Dict ): SCREAMING_SNAKE_CASE : Tuple = DepthEstimationPipeline(model=UpperCAmelCase_ , image_processor=UpperCAmelCase_ ) return depth_estimator, [ "./tests/fixtures/tests_samples/COCO/000000039769.png", "./tests/fixtures/tests_samples/COCO/000000039769.png", ] def _A ( self : Optional[int] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : List[str] = depth_estimator("./tests/fixtures/tests_samples/COCO/000000039769.png" ) self.assertEqual({"predicted_depth": ANY(torch.Tensor ), "depth": ANY(Image.Image )} , UpperCAmelCase_ ) import datasets SCREAMING_SNAKE_CASE : int = datasets.load_dataset("hf-internal-testing/fixtures_image_utils" , "image" , split="test" ) SCREAMING_SNAKE_CASE : List[Any] = depth_estimator( [ Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" ), "http://images.cocodataset.org/val2017/000000039769.jpg", # RGBA dataset[0]["file"], # LA dataset[1]["file"], # L dataset[2]["file"], ] ) self.assertEqual( [ {"predicted_depth": ANY(torch.Tensor ), "depth": ANY(Image.Image )}, {"predicted_depth": ANY(torch.Tensor ), "depth": ANY(Image.Image )}, {"predicted_depth": ANY(torch.Tensor ), "depth": ANY(Image.Image )}, {"predicted_depth": ANY(torch.Tensor ), "depth": ANY(Image.Image )}, {"predicted_depth": ANY(torch.Tensor ), "depth": ANY(Image.Image )}, ] , UpperCAmelCase_ , ) @require_tf @unittest.skip("Depth estimation is not implemented in TF" ) def _A ( self : Tuple ): pass @slow @require_torch def _A ( self : int ): SCREAMING_SNAKE_CASE : List[Any] = "Intel/dpt-large" SCREAMING_SNAKE_CASE : Optional[Any] = pipeline("depth-estimation" , model=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = depth_estimator("http://images.cocodataset.org/val2017/000000039769.jpg" ) SCREAMING_SNAKE_CASE : Optional[int] = hashimage(outputs["depth"] ) # This seems flaky. # self.assertEqual(outputs["depth"], "1a39394e282e9f3b0741a90b9f108977") self.assertEqual(nested_simplify(outputs["predicted_depth"].max().item() ) , 29.304 ) self.assertEqual(nested_simplify(outputs["predicted_depth"].min().item() ) , 2.662 ) @require_torch def _A ( self : List[Any] ): # This is highly irregular to have no small tests. self.skipTest("There is not hf-internal-testing tiny model for either GLPN nor DPT" )
62
import inspect import unittest from transformers import MobileNetVaConfig from transformers.testing_utils import require_torch, require_vision, slow, torch_device from transformers.utils import cached_property, is_torch_available, is_vision_available from ...test_configuration_common import ConfigTester from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor from ...test_pipeline_mixin import PipelineTesterMixin if is_torch_available(): import torch from transformers import MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation, MobileNetVaModel from transformers.models.mobilenet_va.modeling_mobilenet_va import MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST if is_vision_available(): from PIL import Image from transformers import MobileNetVaImageProcessor class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' def _A ( self : List[str] ): SCREAMING_SNAKE_CASE : List[str] = self.config_class(**self.inputs_dict ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "tf_padding" ) ) self.parent.assertTrue(hasattr(UpperCAmelCase_ , "depth_multiplier" ) ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : str , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : List[str]=13 , UpperCAmelCase_ : Any=3 , UpperCAmelCase_ : Union[str, Any]=32 , UpperCAmelCase_ : int=0.25 , UpperCAmelCase_ : Union[str, Any]=8 , UpperCAmelCase_ : Dict=8 , UpperCAmelCase_ : Optional[int]=6 , UpperCAmelCase_ : Optional[Any]=32 , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : str=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : str="relu6" , UpperCAmelCase_ : List[str]=1280 , UpperCAmelCase_ : Optional[int]=0.1 , UpperCAmelCase_ : Any=0.02 , UpperCAmelCase_ : Union[str, Any]=True , UpperCAmelCase_ : Tuple=True , UpperCAmelCase_ : Any=10 , UpperCAmelCase_ : Optional[Any]=None , ): SCREAMING_SNAKE_CASE : Union[str, Any] = parent SCREAMING_SNAKE_CASE : int = batch_size SCREAMING_SNAKE_CASE : Dict = num_channels SCREAMING_SNAKE_CASE : Dict = image_size SCREAMING_SNAKE_CASE : int = depth_multiplier SCREAMING_SNAKE_CASE : str = depth_divisible_by SCREAMING_SNAKE_CASE : Union[str, Any] = min_depth SCREAMING_SNAKE_CASE : int = expand_ratio SCREAMING_SNAKE_CASE : Tuple = tf_padding SCREAMING_SNAKE_CASE : List[str] = output_stride SCREAMING_SNAKE_CASE : Optional[int] = first_layer_is_expansion SCREAMING_SNAKE_CASE : Any = finegrained_output SCREAMING_SNAKE_CASE : Optional[Any] = hidden_act SCREAMING_SNAKE_CASE : Optional[int] = last_hidden_size if finegrained_output else int(last_hidden_size * depth_multiplier ) SCREAMING_SNAKE_CASE : Any = classifier_dropout_prob SCREAMING_SNAKE_CASE : Dict = use_labels SCREAMING_SNAKE_CASE : int = is_training SCREAMING_SNAKE_CASE : Dict = num_labels SCREAMING_SNAKE_CASE : Optional[Any] = initializer_range SCREAMING_SNAKE_CASE : Union[str, Any] = scope def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] ) SCREAMING_SNAKE_CASE : Optional[int] = None SCREAMING_SNAKE_CASE : List[Any] = None if self.use_labels: SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size] , self.num_labels ) SCREAMING_SNAKE_CASE : List[Any] = ids_tensor([self.batch_size, self.image_size, self.image_size] , self.num_labels ) SCREAMING_SNAKE_CASE : Tuple = self.get_config() return config, pixel_values, labels, pixel_labels def _A ( self : Optional[int] ): return MobileNetVaConfig( num_channels=self.num_channels , image_size=self.image_size , depth_multiplier=self.depth_multiplier , depth_divisible_by=self.depth_divisible_by , min_depth=self.min_depth , expand_ratio=self.expand_ratio , output_stride=self.output_stride , first_layer_is_expansion=self.first_layer_is_expansion , finegrained_output=self.finegrained_output , hidden_act=self.hidden_act , tf_padding=self.tf_padding , classifier_dropout_prob=self.classifier_dropout_prob , initializer_range=self.initializer_range , ) def _A ( self : int , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaModel(config=UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ ) self.parent.assertEqual( result.last_hidden_state.shape , ( self.batch_size, self.last_hidden_size, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) self.parent.assertEqual( result.pooler_output.shape , (self.batch_size, self.last_hidden_size) , ) def _A ( self : List[str] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : Any ): SCREAMING_SNAKE_CASE : str = self.num_labels SCREAMING_SNAKE_CASE : List[Any] = MobileNetVaForImageClassification(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) ) def _A ( self : Union[str, Any] , UpperCAmelCase_ : Union[str, Any] , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : Optional[Any] , UpperCAmelCase_ : Optional[Any] ): SCREAMING_SNAKE_CASE : int = self.num_labels SCREAMING_SNAKE_CASE : Dict = MobileNetVaForSemanticSegmentation(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() SCREAMING_SNAKE_CASE : Optional[int] = model(UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) SCREAMING_SNAKE_CASE : Any = model(UpperCAmelCase_ , labels=UpperCAmelCase_ ) self.parent.assertEqual( result.logits.shape , ( self.batch_size, self.num_labels, self.image_size // self.output_stride, self.image_size // self.output_stride, ) , ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : List[Any] = self.prepare_config_and_inputs() SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : List[str] = config_and_inputs SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": pixel_values} return config, inputs_dict @require_torch class SCREAMING_SNAKE_CASE ( lowerCAmelCase , lowerCAmelCase , unittest.TestCase ): '''simple docstring''' UpperCamelCase_ : Any = ( (MobileNetVaModel, MobileNetVaForImageClassification, MobileNetVaForSemanticSegmentation) if is_torch_available() else () ) UpperCamelCase_ : List[Any] = ( { '''feature-extraction''': MobileNetVaModel, '''image-classification''': MobileNetVaForImageClassification, '''image-segmentation''': MobileNetVaForSemanticSegmentation, } if is_torch_available() else {} ) UpperCamelCase_ : Any = False UpperCamelCase_ : List[str] = False UpperCamelCase_ : int = False UpperCamelCase_ : str = False def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[str] = MobileNetVaModelTester(self ) SCREAMING_SNAKE_CASE : Optional[Any] = MobileNetVaConfigTester(self , config_class=UpperCAmelCase_ , has_text_modality=UpperCAmelCase_ ) def _A ( self : Optional[Any] ): self.config_tester.run_common_tests() @unittest.skip(reason="MobileNetV2 does not use inputs_embeds" ) def _A ( self : List[Any] ): pass @unittest.skip(reason="MobileNetV2 does not support input and output embeddings" ) def _A ( self : Dict ): pass @unittest.skip(reason="MobileNetV2 does not output attentions" ) def _A ( self : Union[str, Any] ): pass def _A ( self : List[Any] ): SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : Tuple = model_class(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = inspect.signature(model.forward ) # signature.parameters is an OrderedDict => so arg_names order is deterministic SCREAMING_SNAKE_CASE : Dict = [*signature.parameters.keys()] SCREAMING_SNAKE_CASE : Union[str, Any] = ["pixel_values"] self.assertListEqual(arg_names[:1] , UpperCAmelCase_ ) def _A ( self : Optional[int] ): SCREAMING_SNAKE_CASE : Dict = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_model(*UpperCAmelCase_ ) def _A ( self : List[Any] ): def check_hidden_states_output(UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any = model_class(UpperCAmelCase_ ) model.to(UpperCAmelCase_ ) model.eval() with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**self._prepare_for_class(UpperCAmelCase_ , UpperCAmelCase_ ) ) SCREAMING_SNAKE_CASE : List[Any] = outputs.hidden_states SCREAMING_SNAKE_CASE : Any = 16 self.assertEqual(len(UpperCAmelCase_ ) , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common() for model_class in self.all_model_classes: SCREAMING_SNAKE_CASE : str = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) # check that output_hidden_states also work using config del inputs_dict["output_hidden_states"] SCREAMING_SNAKE_CASE : List[Any] = True check_hidden_states_output(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) def _A ( self : Any ): SCREAMING_SNAKE_CASE : Tuple = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_image_classification(*UpperCAmelCase_ ) def _A ( self : Optional[Any] ): SCREAMING_SNAKE_CASE : List[Any] = self.model_tester.prepare_config_and_inputs() self.model_tester.create_and_check_for_semantic_segmentation(*UpperCAmelCase_ ) @slow def _A ( self : Optional[Any] ): for model_name in MOBILENET_V2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: SCREAMING_SNAKE_CASE : int = MobileNetVaModel.from_pretrained(UpperCAmelCase_ ) self.assertIsNotNone(UpperCAmelCase_ ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : int = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" ) return image @require_torch @require_vision class SCREAMING_SNAKE_CASE ( unittest.TestCase ): '''simple docstring''' @cached_property def _A ( self : Optional[int] ): return ( MobileNetVaImageProcessor.from_pretrained("google/mobilenet_v2_1.0_224" ) if is_vision_available() else None ) @slow def _A ( self : Tuple ): SCREAMING_SNAKE_CASE : int = MobileNetVaForImageClassification.from_pretrained("google/mobilenet_v2_1.0_224" ).to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Tuple = self.default_image_processor SCREAMING_SNAKE_CASE : Optional[int] = prepare_img() SCREAMING_SNAKE_CASE : Optional[int] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : str = model(**UpperCAmelCase_ ) # verify the logits SCREAMING_SNAKE_CASE : Union[str, Any] = torch.Size((1, 1001) ) self.assertEqual(outputs.logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = torch.tensor([0.2_445, -1.1_993, 0.1_905] ).to(UpperCAmelCase_ ) self.assertTrue(torch.allclose(outputs.logits[0, :3] , UpperCAmelCase_ , atol=1E-4 ) ) @slow def _A ( self : str ): SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaForSemanticSegmentation.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : int = model.to(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Optional[int] = MobileNetVaImageProcessor.from_pretrained("google/deeplabv3_mobilenet_v2_1.0_513" ) SCREAMING_SNAKE_CASE : Optional[Any] = prepare_img() SCREAMING_SNAKE_CASE : Union[str, Any] = image_processor(images=UpperCAmelCase_ , return_tensors="pt" ).to(UpperCAmelCase_ ) # forward pass with torch.no_grad(): SCREAMING_SNAKE_CASE : Optional[Any] = model(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = outputs.logits # verify the logits SCREAMING_SNAKE_CASE : Dict = torch.Size((1, 21, 65, 65) ) self.assertEqual(logits.shape , UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = torch.tensor( [ [[17.5_790, 17.7_581, 18.3_355], [18.3_257, 18.4_230, 18.8_973], [18.6_169, 18.8_650, 19.2_187]], [[-2.1_595, -2.0_977, -2.3_741], [-2.4_226, -2.3_028, -2.6_835], [-2.7_819, -2.5_991, -2.7_706]], [[4.2_058, 4.8_317, 4.7_638], [4.4_136, 5.0_361, 4.9_383], [4.5_028, 4.9_644, 4.8_734]], ] , device=UpperCAmelCase_ , ) self.assertTrue(torch.allclose(logits[0, :3, :3, :3] , UpperCAmelCase_ , atol=1E-4 ) )
62
1
import PIL.Image import PIL.ImageOps from packaging import version from PIL import Image if version.parse(version.parse(PIL.__version__).base_version) >= version.parse("""9.1.0"""): snake_case = { """linear""": PIL.Image.Resampling.BILINEAR, """bilinear""": PIL.Image.Resampling.BILINEAR, """bicubic""": PIL.Image.Resampling.BICUBIC, """lanczos""": PIL.Image.Resampling.LANCZOS, """nearest""": PIL.Image.Resampling.NEAREST, } else: snake_case = { """linear""": PIL.Image.LINEAR, """bilinear""": PIL.Image.BILINEAR, """bicubic""": PIL.Image.BICUBIC, """lanczos""": PIL.Image.LANCZOS, """nearest""": PIL.Image.NEAREST, } def lowerCamelCase__ ( lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE : Union[str, Any] = (images / 2 + 0.5).clamp(0 , 1 ) SCREAMING_SNAKE_CASE : List[str] = images.cpu().permute(0 , 2 , 3 , 1 ).float().numpy() SCREAMING_SNAKE_CASE : Tuple = numpy_to_pil(lowercase ) return images def lowerCamelCase__ ( lowercase ): """simple docstring""" if images.ndim == 3: SCREAMING_SNAKE_CASE : Tuple = images[None, ...] SCREAMING_SNAKE_CASE : Tuple = (images * 255).round().astype("uint8" ) if images.shape[-1] == 1: # special case for grayscale (single channel) images SCREAMING_SNAKE_CASE : Tuple = [Image.fromarray(image.squeeze() , mode="L" ) for image in images] else: SCREAMING_SNAKE_CASE : Any = [Image.fromarray(lowercase ) for image in images] return pil_images
62
import math from typing import Dict, Iterable, List, Optional, Tuple, Union import numpy as np from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format from ...image_utils import ( IMAGENET_STANDARD_MEAN, IMAGENET_STANDARD_STD, ChannelDimension, ImageInput, PILImageResampling, get_image_size, is_torch_available, is_torch_tensor, make_list_of_images, to_numpy_array, valid_images, ) from ...utils import TensorType, is_vision_available, logging if is_torch_available(): import torch if is_vision_available(): import PIL snake_case = logging.get_logger(__name__) def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" def constraint_to_multiple_of(lowercase , lowercase , lowercase=0 , lowercase=None ): SCREAMING_SNAKE_CASE : Any = round(val / multiple ) * multiple if max_val is not None and x > max_val: SCREAMING_SNAKE_CASE : Optional[int] = math.floor(val / multiple ) * multiple if x < min_val: SCREAMING_SNAKE_CASE : int = math.ceil(val / multiple ) * multiple return x SCREAMING_SNAKE_CASE : Tuple = (output_size, output_size) if isinstance(lowercase , lowercase ) else output_size SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : int = get_image_size(lowercase ) SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = output_size # determine new height and width SCREAMING_SNAKE_CASE : Tuple = output_height / input_height SCREAMING_SNAKE_CASE : List[Any] = output_width / input_width if keep_aspect_ratio: # scale as little as possible if abs(1 - scale_width ) < abs(1 - scale_height ): # fit width SCREAMING_SNAKE_CASE : List[str] = scale_width else: # fit height SCREAMING_SNAKE_CASE : Optional[Any] = scale_height SCREAMING_SNAKE_CASE : int = constraint_to_multiple_of(scale_height * input_height , multiple=lowercase ) SCREAMING_SNAKE_CASE : Dict = constraint_to_multiple_of(scale_width * input_width , multiple=lowercase ) return (new_height, new_width) class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : Any = ['''pixel_values'''] def __init__( self : Any , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Dict[str, int] = None , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BILINEAR , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Union[int, float] = 1 / 255 , UpperCAmelCase_ : bool = True , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , **UpperCAmelCase_ : str , ): super().__init__(**UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Dict = size if size is not None else {"height": 384, "width": 384} SCREAMING_SNAKE_CASE : List[Any] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[str] = do_resize SCREAMING_SNAKE_CASE : Optional[Any] = size SCREAMING_SNAKE_CASE : str = keep_aspect_ratio SCREAMING_SNAKE_CASE : int = ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample SCREAMING_SNAKE_CASE : List[str] = do_rescale SCREAMING_SNAKE_CASE : Tuple = rescale_factor SCREAMING_SNAKE_CASE : Union[str, Any] = do_normalize SCREAMING_SNAKE_CASE : Union[str, Any] = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN SCREAMING_SNAKE_CASE : int = image_std if image_std is not None else IMAGENET_STANDARD_STD def _A ( self : Tuple , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Dict[str, int] , UpperCAmelCase_ : bool = False , UpperCAmelCase_ : int = 1 , UpperCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : int , ): SCREAMING_SNAKE_CASE : List[str] = get_size_dict(UpperCAmelCase_ ) if "height" not in size or "width" not in size: raise ValueError(f'''The size dictionary must contain the keys \'height\' and \'width\'. Got {size.keys()}''' ) SCREAMING_SNAKE_CASE : Optional[Any] = get_resize_output_image_size( UpperCAmelCase_ , output_size=(size["height"], size["width"]) , keep_aspect_ratio=UpperCAmelCase_ , multiple=UpperCAmelCase_ , ) return resize(UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[int, float] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : Dict , ): return rescale(UpperCAmelCase_ , scale=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Dict , UpperCAmelCase_ : np.ndarray , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Union[float, List[float]] , UpperCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **UpperCAmelCase_ : List[str] , ): return normalize(UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ , data_format=UpperCAmelCase_ , **UpperCAmelCase_ ) def _A ( self : Optional[Any] , UpperCAmelCase_ : ImageInput , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : int = None , UpperCAmelCase_ : PILImageResampling = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : float = None , UpperCAmelCase_ : bool = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[float, List[float]]] = None , UpperCAmelCase_ : Optional[Union[str, TensorType]] = None , UpperCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **UpperCAmelCase_ : Optional[Any] , ): SCREAMING_SNAKE_CASE : Optional[int] = do_resize if do_resize is not None else self.do_resize SCREAMING_SNAKE_CASE : Dict = size if size is not None else self.size SCREAMING_SNAKE_CASE : Optional[int] = get_size_dict(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : List[Any] = keep_aspect_ratio if keep_aspect_ratio is not None else self.keep_aspect_ratio SCREAMING_SNAKE_CASE : Optional[Any] = ensure_multiple_of if ensure_multiple_of is not None else self.ensure_multiple_of SCREAMING_SNAKE_CASE : Any = resample if resample is not None else self.resample SCREAMING_SNAKE_CASE : Dict = do_rescale if do_rescale is not None else self.do_rescale SCREAMING_SNAKE_CASE : Optional[Any] = rescale_factor if rescale_factor is not None else self.rescale_factor SCREAMING_SNAKE_CASE : Optional[int] = do_normalize if do_normalize is not None else self.do_normalize SCREAMING_SNAKE_CASE : int = image_mean if image_mean is not None else self.image_mean SCREAMING_SNAKE_CASE : Tuple = image_std if image_std is not None else self.image_std SCREAMING_SNAKE_CASE : List[str] = make_list_of_images(UpperCAmelCase_ ) if not valid_images(UpperCAmelCase_ ): raise ValueError( "Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, " "torch.Tensor, tf.Tensor or jax.ndarray." ) if do_resize and size is None or resample is None: raise ValueError("Size and resample must be specified if do_resize is True." ) if do_rescale and rescale_factor is None: raise ValueError("Rescale factor must be specified if do_rescale is True." ) if do_normalize and (image_mean is None or image_std is None): raise ValueError("Image mean and std must be specified if do_normalize is True." ) # All transformations expect numpy arrays. SCREAMING_SNAKE_CASE : Any = [to_numpy_array(UpperCAmelCase_ ) for image in images] if do_resize: SCREAMING_SNAKE_CASE : Tuple = [self.resize(image=UpperCAmelCase_ , size=UpperCAmelCase_ , resample=UpperCAmelCase_ ) for image in images] if do_rescale: SCREAMING_SNAKE_CASE : Tuple = [self.rescale(image=UpperCAmelCase_ , scale=UpperCAmelCase_ ) for image in images] if do_normalize: SCREAMING_SNAKE_CASE : Optional[Any] = [self.normalize(image=UpperCAmelCase_ , mean=UpperCAmelCase_ , std=UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = [to_channel_dimension_format(UpperCAmelCase_ , UpperCAmelCase_ ) for image in images] SCREAMING_SNAKE_CASE : Optional[Any] = {"pixel_values": images} return BatchFeature(data=UpperCAmelCase_ , tensor_type=UpperCAmelCase_ ) def _A ( self : int , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : List[Tuple] = None ): SCREAMING_SNAKE_CASE : int = outputs.logits # Resize logits and compute semantic segmentation maps if target_sizes is not None: if len(UpperCAmelCase_ ) != len(UpperCAmelCase_ ): raise ValueError( "Make sure that you pass in as many target sizes as the batch dimension of the logits" ) if is_torch_tensor(UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : List[Any] = target_sizes.numpy() SCREAMING_SNAKE_CASE : str = [] for idx in range(len(UpperCAmelCase_ ) ): SCREAMING_SNAKE_CASE : Dict = torch.nn.functional.interpolate( logits[idx].unsqueeze(dim=0 ) , size=target_sizes[idx] , mode="bilinear" , align_corners=UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : str = resized_logits[0].argmax(dim=0 ) semantic_segmentation.append(UpperCAmelCase_ ) else: SCREAMING_SNAKE_CASE : str = logits.argmax(dim=1 ) SCREAMING_SNAKE_CASE : str = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )] return semantic_segmentation
62
1
def lowerCamelCase__ ( lowercase , lowercase , lowercase , lowercase ): """simple docstring""" SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE : Optional[Any] = len(lowercase ), len(grid[0] ) if ( min(lowercase , lowercase ) < 0 or row == row_length or col == col_length or (row, col) in visit or grid[row][col] == 1 ): return 0 if row == row_length - 1 and col == col_length - 1: return 1 visit.add((row, col) ) SCREAMING_SNAKE_CASE : Union[str, Any] = 0 count += depth_first_search(lowercase , row + 1 , lowercase , lowercase ) count += depth_first_search(lowercase , row - 1 , lowercase , lowercase ) count += depth_first_search(lowercase , lowercase , col + 1 , lowercase ) count += depth_first_search(lowercase , lowercase , col - 1 , lowercase ) visit.remove((row, col) ) return count if __name__ == "__main__": import doctest doctest.testmod()
62
from __future__ import annotations from typing import Any class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : List[str] , UpperCAmelCase_ : int = 6 ): SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None self.create_linked_list(UpperCAmelCase_ ) def _A ( self : List[Any] , UpperCAmelCase_ : int ): SCREAMING_SNAKE_CASE : Optional[int] = Node() SCREAMING_SNAKE_CASE : str = current_node SCREAMING_SNAKE_CASE : Optional[int] = current_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node for _ in range(1 , UpperCAmelCase_ ): SCREAMING_SNAKE_CASE : Tuple = Node() SCREAMING_SNAKE_CASE : Dict = current_node SCREAMING_SNAKE_CASE : Optional[Any] = previous_node SCREAMING_SNAKE_CASE : Optional[Any] = current_node SCREAMING_SNAKE_CASE : Union[str, Any] = self.front SCREAMING_SNAKE_CASE : List[str] = previous_node def _A ( self : Union[str, Any] ): return ( self.front == self.rear and self.front is not None and self.front.data is None ) def _A ( self : Optional[int] ): self.check_can_perform_operation() return self.front.data if self.front else None def _A ( self : Optional[int] , UpperCAmelCase_ : Any ): if self.rear is None: return self.check_is_full() if not self.is_empty(): SCREAMING_SNAKE_CASE : List[str] = self.rear.next if self.rear: SCREAMING_SNAKE_CASE : Dict = data def _A ( self : List[str] ): self.check_can_perform_operation() if self.rear is None or self.front is None: return None if self.front == self.rear: SCREAMING_SNAKE_CASE : List[str] = self.front.data SCREAMING_SNAKE_CASE : Optional[int] = None return data SCREAMING_SNAKE_CASE : List[str] = self.front SCREAMING_SNAKE_CASE : List[str] = old_front.next SCREAMING_SNAKE_CASE : Optional[int] = old_front.data SCREAMING_SNAKE_CASE : List[str] = None return data def _A ( self : Any ): if self.is_empty(): raise Exception("Empty Queue" ) def _A ( self : Optional[Any] ): if self.rear and self.rear.next == self.front: raise Exception("Full Queue" ) class SCREAMING_SNAKE_CASE : '''simple docstring''' def __init__( self : Union[str, Any] ): SCREAMING_SNAKE_CASE : Any | None = None SCREAMING_SNAKE_CASE : Node | None = None SCREAMING_SNAKE_CASE : Node | None = None if __name__ == "__main__": import doctest doctest.testmod()
62
1
import json import os import torch from diffusers import UNetaDModel os.makedirs("""hub/hopper-medium-v2/unet/hor32""", exist_ok=True) os.makedirs("""hub/hopper-medium-v2/unet/hor128""", exist_ok=True) os.makedirs("""hub/hopper-medium-v2/value_function""", exist_ok=True) def lowerCamelCase__ ( lowercase ): """simple docstring""" if hor == 128: SCREAMING_SNAKE_CASE : List[str] = ("DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D") SCREAMING_SNAKE_CASE : List[Any] = (32, 128, 256) SCREAMING_SNAKE_CASE : str = ("UpResnetBlock1D", "UpResnetBlock1D") elif hor == 32: SCREAMING_SNAKE_CASE : Tuple = ("DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D") SCREAMING_SNAKE_CASE : List[Any] = (32, 64, 128, 256) SCREAMING_SNAKE_CASE : Union[str, Any] = ("UpResnetBlock1D", "UpResnetBlock1D", "UpResnetBlock1D") SCREAMING_SNAKE_CASE : Optional[Any] = torch.load(F'''/Users/bglickenhaus/Documents/diffuser/temporal_unet-hopper-mediumv2-hor{hor}.torch''' ) SCREAMING_SNAKE_CASE : Optional[Any] = model.state_dict() SCREAMING_SNAKE_CASE : List[str] = { "down_block_types": down_block_types, "block_out_channels": block_out_channels, "up_block_types": up_block_types, "layers_per_block": 1, "use_timestep_embedding": True, "out_block_type": "OutConv1DBlock", "norm_num_groups": 8, "downsample_each_block": False, "in_channels": 14, "out_channels": 14, "extra_in_channels": 0, "time_embedding_type": "positional", "flip_sin_to_cos": False, "freq_shift": 1, "sample_size": 65536, "mid_block_type": "MidResTemporalBlock1D", "act_fn": "mish", } SCREAMING_SNAKE_CASE : Dict = UNetaDModel(**lowercase ) print(F'''length of state dict: {len(state_dict.keys() )}''' ) print(F'''length of value function dict: {len(hf_value_function.state_dict().keys() )}''' ) SCREAMING_SNAKE_CASE : List[Any] = dict(zip(model.state_dict().keys() , hf_value_function.state_dict().keys() ) ) for k, v in mapping.items(): SCREAMING_SNAKE_CASE : List[Any] = state_dict.pop(lowercase ) hf_value_function.load_state_dict(lowercase ) torch.save(hf_value_function.state_dict() , F'''hub/hopper-medium-v2/unet/hor{hor}/diffusion_pytorch_model.bin''' ) with open(F'''hub/hopper-medium-v2/unet/hor{hor}/config.json''' , "w" ) as f: json.dump(lowercase , lowercase ) def lowerCamelCase__ ( ): """simple docstring""" SCREAMING_SNAKE_CASE : str = { "in_channels": 14, "down_block_types": ("DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D"), "up_block_types": (), "out_block_type": "ValueFunction", "mid_block_type": "ValueFunctionMidBlock1D", "block_out_channels": (32, 64, 128, 256), "layers_per_block": 1, "downsample_each_block": True, "sample_size": 65536, "out_channels": 14, "extra_in_channels": 0, "time_embedding_type": "positional", "use_timestep_embedding": True, "flip_sin_to_cos": False, "freq_shift": 1, "norm_num_groups": 8, "act_fn": "mish", } SCREAMING_SNAKE_CASE : Dict = torch.load("/Users/bglickenhaus/Documents/diffuser/value_function-hopper-mediumv2-hor32.torch" ) SCREAMING_SNAKE_CASE : Union[str, Any] = model SCREAMING_SNAKE_CASE : Optional[Any] = UNetaDModel(**lowercase ) print(F'''length of state dict: {len(state_dict.keys() )}''' ) print(F'''length of value function dict: {len(hf_value_function.state_dict().keys() )}''' ) SCREAMING_SNAKE_CASE : Tuple = dict(zip(state_dict.keys() , hf_value_function.state_dict().keys() ) ) for k, v in mapping.items(): SCREAMING_SNAKE_CASE : List[str] = state_dict.pop(lowercase ) hf_value_function.load_state_dict(lowercase ) torch.save(hf_value_function.state_dict() , "hub/hopper-medium-v2/value_function/diffusion_pytorch_model.bin" ) with open("hub/hopper-medium-v2/value_function/config.json" , "w" ) as f: json.dump(lowercase , lowercase ) if __name__ == "__main__": unet(32) # unet(128) value_function()
62
def lowerCamelCase__ ( lowercase , lowercase ): """simple docstring""" return 1 if input_a == input_a else 0 def lowerCamelCase__ ( ): """simple docstring""" assert xnor_gate(0 , 0 ) == 1 assert xnor_gate(0 , 1 ) == 0 assert xnor_gate(1 , 0 ) == 0 assert xnor_gate(1 , 1 ) == 1 if __name__ == "__main__": print(xnor_gate(0, 0)) print(xnor_gate(0, 1)) print(xnor_gate(1, 0)) print(xnor_gate(1, 1))
62
1
from collections import OrderedDict from typing import Mapping from ...configuration_utils import PretrainedConfig from ...onnx import OnnxConfig from ...utils import logging snake_case = logging.get_logger(__name__) snake_case = { """xlm-roberta-base""": """https://huggingface.co/xlm-roberta-base/resolve/main/config.json""", """xlm-roberta-large""": """https://huggingface.co/xlm-roberta-large/resolve/main/config.json""", """xlm-roberta-large-finetuned-conll02-dutch""": ( """https://huggingface.co/xlm-roberta-large-finetuned-conll02-dutch/resolve/main/config.json""" ), """xlm-roberta-large-finetuned-conll02-spanish""": ( """https://huggingface.co/xlm-roberta-large-finetuned-conll02-spanish/resolve/main/config.json""" ), """xlm-roberta-large-finetuned-conll03-english""": ( """https://huggingface.co/xlm-roberta-large-finetuned-conll03-english/resolve/main/config.json""" ), """xlm-roberta-large-finetuned-conll03-german""": ( """https://huggingface.co/xlm-roberta-large-finetuned-conll03-german/resolve/main/config.json""" ), } class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' UpperCamelCase_ : str = '''xlm-roberta''' def __init__( self : List[Any] , UpperCAmelCase_ : List[str]=3_0522 , UpperCAmelCase_ : str=768 , UpperCAmelCase_ : str=12 , UpperCAmelCase_ : List[Any]=12 , UpperCAmelCase_ : int=3072 , UpperCAmelCase_ : Optional[int]="gelu" , UpperCAmelCase_ : List[str]=0.1 , UpperCAmelCase_ : int=0.1 , UpperCAmelCase_ : Tuple=512 , UpperCAmelCase_ : Tuple=2 , UpperCAmelCase_ : Optional[Any]=0.02 , UpperCAmelCase_ : List[str]=1E-12 , UpperCAmelCase_ : List[Any]=1 , UpperCAmelCase_ : List[str]=0 , UpperCAmelCase_ : Optional[int]=2 , UpperCAmelCase_ : List[str]="absolute" , UpperCAmelCase_ : Optional[int]=True , UpperCAmelCase_ : int=None , **UpperCAmelCase_ : List[Any] , ): super().__init__(pad_token_id=UpperCAmelCase_ , bos_token_id=UpperCAmelCase_ , eos_token_id=UpperCAmelCase_ , **UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : int = vocab_size SCREAMING_SNAKE_CASE : Tuple = hidden_size SCREAMING_SNAKE_CASE : str = num_hidden_layers SCREAMING_SNAKE_CASE : Any = num_attention_heads SCREAMING_SNAKE_CASE : Optional[int] = hidden_act SCREAMING_SNAKE_CASE : List[str] = intermediate_size SCREAMING_SNAKE_CASE : List[Any] = hidden_dropout_prob SCREAMING_SNAKE_CASE : Optional[int] = attention_probs_dropout_prob SCREAMING_SNAKE_CASE : Any = max_position_embeddings SCREAMING_SNAKE_CASE : int = type_vocab_size SCREAMING_SNAKE_CASE : Tuple = initializer_range SCREAMING_SNAKE_CASE : str = layer_norm_eps SCREAMING_SNAKE_CASE : Optional[Any] = position_embedding_type SCREAMING_SNAKE_CASE : Union[str, Any] = use_cache SCREAMING_SNAKE_CASE : Optional[int] = classifier_dropout class SCREAMING_SNAKE_CASE ( lowerCAmelCase ): '''simple docstring''' @property def _A ( self : Tuple ): if self.task == "multiple-choice": SCREAMING_SNAKE_CASE : str = {0: "batch", 1: "choice", 2: "sequence"} else: SCREAMING_SNAKE_CASE : Union[str, Any] = {0: "batch", 1: "sequence"} return OrderedDict( [ ("input_ids", dynamic_axis), ("attention_mask", dynamic_axis), ] )
62
import math import flax.linen as nn import jax.numpy as jnp def lowerCamelCase__ ( lowercase , lowercase , lowercase = 1 , lowercase = 1 , lowercase = 1.0E4 , lowercase = False , lowercase = 1.0 , ): """simple docstring""" assert timesteps.ndim == 1, "Timesteps should be a 1d-array" assert embedding_dim % 2 == 0, F'''Embedding dimension {embedding_dim} should be even''' SCREAMING_SNAKE_CASE : Union[str, Any] = float(embedding_dim // 2 ) SCREAMING_SNAKE_CASE : Dict = math.log(max_timescale / min_timescale ) / (num_timescales - freq_shift) SCREAMING_SNAKE_CASE : Optional[Any] = min_timescale * jnp.exp(jnp.arange(lowercase , dtype=jnp.floataa ) * -log_timescale_increment ) SCREAMING_SNAKE_CASE : Optional[int] = jnp.expand_dims(lowercase , 1 ) * jnp.expand_dims(lowercase , 0 ) # scale embeddings SCREAMING_SNAKE_CASE : Optional[int] = scale * emb if flip_sin_to_cos: SCREAMING_SNAKE_CASE : List[Any] = jnp.concatenate([jnp.cos(lowercase ), jnp.sin(lowercase )] , axis=1 ) else: SCREAMING_SNAKE_CASE : Union[str, Any] = jnp.concatenate([jnp.sin(lowercase ), jnp.cos(lowercase )] , axis=1 ) SCREAMING_SNAKE_CASE : Tuple = jnp.reshape(lowercase , [jnp.shape(lowercase )[0], embedding_dim] ) return signal class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : jnp.dtype = jnp.floataa @nn.compact def __call__( self : Tuple , UpperCAmelCase_ : str ): SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_1" )(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Union[str, Any] = nn.silu(UpperCAmelCase_ ) SCREAMING_SNAKE_CASE : Any = nn.Dense(self.time_embed_dim , dtype=self.dtype , name="linear_2" )(UpperCAmelCase_ ) return temb class SCREAMING_SNAKE_CASE ( nn.Module ): '''simple docstring''' UpperCamelCase_ : int = 3_2 UpperCamelCase_ : bool = False UpperCamelCase_ : float = 1 @nn.compact def __call__( self : Optional[int] , UpperCAmelCase_ : int ): return get_sinusoidal_embeddings( UpperCAmelCase_ , embedding_dim=self.dim , flip_sin_to_cos=self.flip_sin_to_cos , freq_shift=self.freq_shift )
62
1